Consumer-oriented interventions for evidence-based prescribing and medicines use: an overview of systematic reviews

  • Review
  • Overview

Authors


Abstract

Background

Numerous systematic reviews exist on interventions to improve consumers’ medicines use, but this research is distributed across diseases, populations and settings. The scope and focus of reviews on consumers’ medicines use also varies widely. Such differences create challenges for decision makers seeking review-level evidence to inform decisions about medicines use.

Objectives

To synthesise the evidence from systematic reviews on the effects of interventions which target healthcare consumers to promote evidence-based prescribing for, and medicines use, by consumers. We sought evidence on the effects on health and other outcomes for healthcare consumers, professionals and services.

Methods

We included systematic reviews published on the Cochrane Database of Systematic Reviews and the Database of Abstracts of Reviews of Effects. We identified relevant reviews by handsearching both databases from start date to Issue 3 2008. We screened and ranked reviews based on relevance to consumers’ medicines use, using criteria developed for this overview. Standardised forms were used to extract data, and reviews were assessed for methodological quality using the AMSTAR instrument. We used standardised language to summarise results within and across reviews; and a further synthesis step was used to give bottom-line statements about intervention effectiveness. Two review authors selected reviews, extracted and analysed data. We used a taxonomy of interventions to categorise reviews.

Main results

We included 37 reviews (18 Cochrane, 19 non-Cochrane), of varied methodological quality.

Reviews assessed interventions with diverse aims including support for behaviour change, risk minimisation, skills acquisition and information provision. No reviews aimed to promote systems-level consumer participation in medicines-related activities. Medicines adherence was the most commonly reported outcome, but others such as clinical (health and wellbeing), service use and knowledge outcomes were also reported. Reviews rarely reported adverse events or harms, and the evidence was sparse for several populations, including children and young people, carers, and people with multimorbidity.

Promising interventions to improve adherence and other key medicines use outcomes (eg adverse events, knowledge) included self-monitoring and self-management, simplified dosing and interventions directly involving pharmacists. Other strategies showed promise in relation to adherence but their effects were less consistent. These included reminders; education combined with self-management skills training, counselling or support; financial incentives; and lay health worker interventions.

No interventions were effective to improve all medicines use outcomes across all diseases, populations or settings. For some interventions, such as information or education provided alone, the evidence suggests ineffectiveness; for many others there is insufficient evidence to determine effects on medicines use outcomes.

Authors' conclusions

Systematically assembling the evidence across reviews allows identification of effective or promising interventions to improve consumers’ medicines use, as well as those for which the evidence indicates ineffectiveness or uncertainty.

Decision makers faced with implementing interventions to improve consumers’ medicines use can use this overview to inform these decisions and also to consider the range of interventions available; while researchers and funders can use this overview to determine where research is needed. However, the limitations of the literature relating to the lack of evidence for important outcomes and specific populations, such as people with multimorbidity, should also be considered.

Resumen

Antecedentes

Intervenciones orientadas a los consumidores para el uso de medicamentos y prescripciones basado en la evidencia: un resumen de revisiones sistemáticas

Existen numerosas revisiones sistemáticas sobre intervenciones para mejorar el uso de medicamentos por parte de los consumidores, pero esta investigación está distribuida entre enfermedades, poblaciones y contextos. El alcance y el propósito de las revisiones sobre el uso de medicamentos por parte de los consumidores también varían ampliamente. Dichas diferencias crean retos para los responsables de tomar decisiones que buscan pruebas a nivel de revisiones para informar las decisiones acerca del uso de medicamentos.

Objetivos

Resumir las pruebas de las revisiones sistemáticas sobre los efectos de las intervenciones que apuntan a los consumidores de asistencia sanitaria para promover el uso de medicamentos y prescripciones basado en la evidencia por parte de los consumidores. Se buscaron pruebas sobre los efectos en la salud y otros resultados para los consumidores, los profesionales y los servicios de asistencia sanitaria.

Estrategia de búsqueda

 

Criterios de selección

 

Obtención y análisis de los datos

 

Resultados principales

Se incluyeron 37 revisiones (18 Cochrane, 19 no Cochrane), de calidad metodológica variada.

Las revisiones evaluaron intervenciones con objetivos diversos incluido el apoyo al cambio en el comportamiento, la minimización del riesgo, la adquisición de habilidades y la provisión de información. Ninguna revisión procuró promover la participación de los consumidores a nivel de los sistemas en las actividades relacionadas con los medicamentos. El cumplimiento con los medicamentos fue el resultado informado más comúnmente, aunque también se informaron otros resultados como los clínicos (salud y bienestar), del uso del servicio y del conocimiento. Las revisiones rara vez informaron los eventos adversos o efectos perjudiciales, y las pruebas fueron escasas para varias poblaciones, incluidos los niños y los jóvenes, los cuidadores y las personas con multimorbilidad.

Las intervenciones prometedoras en cuanto a la mejoría del cumplimiento y otros resultados importantes del uso de medicamentos (p.ej. eventos adversos, conocimiento) incluyeron automonitorización y autocuidado, dosificación simplificada e intervenciones que incluían directamente a los farmacéuticos. Otras estrategias parecieron prometedoras con relación al cumplimiento, aunque sus efectos fueron menos consistentes. Las mismas incluyeron recordatorios; educación combinada con adiestramiento de las habilidades de autocuidado, asesoramiento o apoyo; incentivos económicos; e intervenciones con trabajadores sanitarios no profesionales.

Ninguna intervención fue efectiva para mejorar todos los resultados del uso de medicamentos en todas las enfermedades, las poblaciones o los contextos. Para algunas intervenciones, como la información o la educación proporcionadas solas, las pruebas indican ineficacia; para muchas otras no hay pruebas suficientes para determinar los efectos sobre los resultados del uso de medicamentos.

Conclusiones de los autores

Al reunir sistemáticamente las pruebas de las revisiones es posible identificar las intervenciones efectivas o prometedoras para mejorar el uso de medicamentos por parte de los consumidores, así como aquellas cuyas pruebas indican ineficacia o incertidumbre.

Los responsables de tomar decisiones, enfrentados con la ejecución de intervenciones para mejorar el uso de medicamentos por parte de los consumidores, pueden usar este resumen para informar estas decisiones y también para considerar el rango de intervenciones disponibles; mientras que los investigadores y los patrocinadores pueden usar este resumen para determinar dónde se necesita investigación. Sin embargo, también deben considerarse las limitaciones de la bibliografía en relación con la ausencia de pruebas sobre los resultados importantes y las poblaciones específicas, como las personas con multimorbilidad.

Traducción

Traducción realizada por el Centro Cochrane Iberoamericano

Plain language summary

Strategies to improve safe and effective medicines use by consumers: an overview of systematic reviews

Medicines are a cornerstone of treatment for many health problems. Many strategies exist to help people to use medicines safely and effectively, but research in the area is not well organised across diseases, populations and settings. This can make it difficult for policy makers, health professionals and others to find and use the evidence about what works and what does not.

This overview summarised the evidence contained in 37 systematic reviews on consumers' medicine use. A wide range of strategies to improve medicines use, including information provision, support for behaviour change, risk minimisation and skills acquisition, was included. No one type of strategy improved medicines use outcomes across all diseases, populations or settings, or for all outcomes.

Strategies that appear promising to improve medicines use included medicines self-monitoring and self-management, simplified dosing and direct involvement of pharmacists in medicines management. Other strategies such as reminders; education combined with self-management skills training, counselling or support; financial incentives; and strategies involving lay health workers may also show promise, although effects were less consistent. Some strategies, such as providing information or education as single interventions, may be ineffective; while for many strategies there is not enough evidence to decide how effective or ineffective they are.

Reviews included in this overview often had methodological limitations, meaning results should be interpreted with caution. Despite the large number of included reviews there are many gaps in the assembled evidence on medicines use strategies, such as those focussing on children, young people or carers, or those for people with more than one coexisting health problem.

Background

Numerous systematic reviews have examined interventions to improve medicines use. Some reviews include various interventions related to medicines for a specific disease, such as diabetes or schizophrenia (Vermeire 2005; Zygmunt 2002), while others focus on one type of intervention (eg written information, directly observed therapy, reminders) across different diseases (Heneghan 2006a; Nicolson 2009). Still others focus on one primary goal, for example, improving medicines adherence (Haynes 2008) or immunisation rates (Jacobson 2005; Stone 2002). These differences in the foci of systematic reviews can make it difficult for decision makers to access the review-level evidence, and for researchers to know where gaps in the evidence exist.

This overview synthesises evidence from systematic reviews of studies of 'consumer-oriented medicines interventions', also described as 'consumers' use of medicines'. These comprise interventions targeting consumers to promote evidence-based prescribing for, and medicines use, by them. It considers interventions applied across different diseases, settings and populations.

A consumer perspective on evidence-based prescribing and medicines use

In this overview, we define consumers to include patients, their family members or carers. We define consumer-oriented interventions as those principally directed to consumers, in recognition of their central role in decision making and management of medicines, alone or in partnership with healthcare professionals. Since it is ultimately the consumer who decides whether and how to take medicines, the purpose of such interventions might include promoting consumers' knowledge and ability to make informed decisions about medicines, and providing them with sufficient skills and support to take medicines safely and effectively. This overview adopts this inclusive perspective on consumers’ medicines use. Practically, this means considering a wide range of specific interventions targeting consumers, such as purposeful communication, information provision, education, skills training, strategies promoting participation, and support for medicines use.

Organisations working worldwide to regulate and optimise medicines use and availability for individuals and populations have defined evidence-based prescribing and medicines use in different ways, but definitions rest broadly on the principles of rational use of medicines and evidence-based health care. These principles specify that medicines are considered as only one among many options for treatment; that the medicine chosen is the safest and most effective option available; and that the medicine is the most appropriate option based on the individual's need. Such principles aim to enable healthcare consumers, professionals and systems to make the best possible use of available medicines and minimise harms. Internationally, many policies and strategies based on these broad principles have been used to inform, educate, support and communicate with consumers to help them understand and use their medicines in ways that are consistent with healthcare evidence(Chetley 2007; NICE 2009).

Quality and safety in the use of medicines: issues of adherence

Internationally, the pursuit of safe, high-quality health care is a goal, yet major problems in achieving this have been documented across countries (Coulter 2006; Schoen 2005). Amongst other areas of concern, attention has focussed on medicines use, and particularly on high rates of errors by prescribers and patients, on preventable adverse effects (Coulter 2006; Feldstein 2006; Schoen 2005), on inconsistent medicines review, and on difficulties in communication and transitional care (Coleman 2006). Even when medicines are used appropriately, adverse events may occur. The chance of medicines problems occurring is increased by errors such as administration of the wrong drug or dose, overlooked allergies or interactions, inadequate monitoring, and insufficient communication of key information to consumers.

Medicines management is only one aspect of managing health, yet it is an important area for decision making by consumers. The developing area of patient-centred care has promoted greater awareness of the role of consumers as self-managers and as shared managers of health and illness with healthcare providers, and of the principles and practices of shared and informed decision making (Coulter 2006; Dickinson 2003; Little 2001). Despite these conceptual shifts, much research on consumers' medicines use has focussed primarily or exclusively on adherence, and so involved consumers in largely passive roles.

Medicines adherence has major implications for the effective treatment of many diseases (Haynes 2008; Munro 2007; Van Dulmen 2007). If medicines are required, and are selected from available treatment options, they must be taken appropriately to be effective and safe. This might involve ensuring the correct medicine is used, avoiding interactions and identifying contraindications, and taking the medicine according to the appropriate dose, schedule and duration. These activities form a complex set of processes which can be disrupted at any point and so contribute to poor adherence (Coulter 2006).

Poor medicines adherence is of major concern with good cause: studies consistently show that up to half of patients do not take their medicines as prescribed (Haynes 2008; Heneghan 2006a), and more than 85% of patients are occasionally non-adherent (O'Connor 2006). Research suggests that taking less of a medicine than prescribed, rather than more, is most common, although both occur (Britten 2004), and many health problems stem from failing to take medicines properly. Taking too little can dilute any possible therapeutic benefit, but taking medicines in the wrong dose or frequency can also cause problems if, for example, a person tries to compensate for a missed dose by taking more of the medicine when they remember, or takes doses too close together. Poor or inconsistent adherence can therefore cause a range of problems, including increased adverse events, overdose, unnecessary hospitalisations and prescriptions, antibiotic resistance, rising costs, progression of disease and treatment failure or death (Haynes 2008; Tarn 2006a).

Adherence, however, is complex. A recent systematic review concluded that high levels of adherence to medicines (variously defined but including continued medicine use, or use above a threshold such as 80% of pills taken) was associated in many cases with positive health outcomes, such as decreased mortality (Simpson 2006). However, the review also suggested that good adherence to potentially harmful treatments can lead to adverse outcomes. Other authors have also stressed that increased adherence may have various effects, some of them harmful. These include harmful effects of the medicine itself, as well as harms associated with a loss of patient autonomy and choice (Haynes 2008; Pound 2005).

Many factors affect adherence to medicines. Previous research has concentrated largely on factors affecting consumers' behaviour. These include consumers' ability to remember to take medicines appropriately, the quality of instructions about the medicine, and the demands of complex treatment regimens (Dickinson 2003; Haynes 2008; Heneghan 2006a; Van Dulmen 2007). Such factors largely reflect unintentional non-adherence, but there is now growing recognition that intentional non-adherence may also play a role. Factors affecting intentional non-adherence are complex, and include those associated with cost, adverse effects, patient preferences, disagreement with the need for treatment, or communication breakdown between patient and provider (Britten 2004; Coleman 2006; Munro 2007; Pound 2005; Soumerai 2006; Tarn 2006a).

Recent qualitative research gives further insight into the many complex factors that interact to affect how and why people take medicines. For example, a so-called 'aversion' to medicines use has been documented (Britten 2004; Pound 2005; Townsend 2003). Consumers may use medicines only when symptoms demand it, or in ways that least disrupt their daily routines, rather than as prescribed. Sometimes people adjust or halt the regimen to minimise adverse effects or financial costs, or simply because they do not like taking medicines regularly or continuously (Pound 2005). These choices reflect the realities of daily medicines use, and the influence of perceptions of health and illness, such as self-identity and the stigma of having an illness dependent on medicines (Britten 2004; Pound 2005; Townsend 2003). Consumers' concerns about the medicines themselves, including adverse effects, tolerance and dependence, can also affect adherence (Pound 2005).

There is also growing awareness that factors beyond consumers' control can affect adherence (Coulter 2006; Munro 2007;O'Connor 2006; Soumerai 2006). Research suggests, for example, that healthcare providers' behaviour can affect patients' medicines use, with documented examples of communication breakdown including failure to:

  • adequately explain how to take a medicine or provide information about new prescription medicines (Tarn 2006a; Tarn 2006b);

  • review medicines, even where needs are complex (Schoen 2005);

  • raise and discuss with consumers any reluctance to take medicines (Britten 2004; Pound 2005); and

  • discuss with consumers their knowledge and beliefs about health and treatment (Munro 2007).

Other factors that can affect medicine use but which may be largely beyond consumers' control include:

  • financial costs or burden (Munro 2007; Pound 2005; Tarn 2006b);

  • the co-existence of problems (co-morbidity or multimorbidity) (Soumerai 2006; Tarn 2006a);

  • features of health service organisation, such as access to and availability of services, and requirements of the treatment itself; and

  • the social and cultural context in which treatment occurs, including the influence of community, family members and peers (Garner 2007; Munro 2007).

Given these compounding factors, communication in its entirety is critical, and there is increasingly a view that interventions to improve adherence should focus not just on consumers but on the wider patient context and healthcare system. An emerging theme among recent research is an emphasis on adopting patient-centred care and shared decision-making principles in order to achieve better adherence, together with attention to barriers that may be targets for interventions (Garner 2007; Munro 2007).

Why it is important to do this overview

A recent overview of systematic reviews examined the evidence on interventions to improve adherence to medicines (Van Dulmen 2007). This work identified many simple and complex interventions aiming to improve adherence, typically with mixed effects. Measuring adherence and seeking to understand it as a key aspect of medicines use is important, but taken in isolation fails to consider the wider medicines management, communication and decision-making roles for consumers.

In this overview we seek to extend previous research in the area beyond adherence, and to systematically identify and organise this literature. There is a need to deliberately consider interventions on consumers' medicines use which have purposes other than, or in addition to, adherence. This includes interventions with broader aims in relation to medicines, such as informed decision making or information to improve medicines awareness, communication about medicines, support for medicines use and minimisation of adverse events. It also includes collecting information on a comprehensive range of outcomes in addition to adherence, such as consumers' knowledge, skills, capacity and their ability to minimise harms, as well as outcomes for healthcare professionals and systems which are fundamental to understanding and supporting consumers' medicines use.

Taking this wider view of consumers’ medicines use is essential if we are to better understand why - or why not - interventions aiming to improve adherence and medicines use are effective. Partnerships in which consumers are involved actively as managers of medicines and decision makers with health professionals are important. Involving consumers in their choices about medicines could also be central to the sustainability of evidence-informed treatments. However, since interventions span diseases, populations and treatment settings, it may be difficult for decision makers, healthcare professionals and researchers to find and use the evidence. We therefore undertook this overview to systematically gather, evaluate and organise the review-level evidence on consumer-oriented medicines interventions, to improve access to the evidence in order to inform decision making.

Objectives

To synthesise the evidence from systematic reviews examining the effects of interventions which target healthcare consumers to promote evidence-based prescribing for, and medicines use, by consumers. We sought evidence on the effects on health and other outcomes for consumers, healthcare professionals and health services. Interventions can target consumers alone, or may enable them to work more effectively with their healthcare providers to improve safe and effective medicines use. This overview also aimed to provide an overall structure and synthesis of the evidence on the range of interventions with which it is possible to target consumers' medicines use.

Methods

Criteria for considering reviews for inclusion

Types of reviews

We included all reviews published in English that met our selection criteria and which were published in the:

  • Cochrane Database of Systematic Reviews (CDSR); and

  • Database of Abstracts of Reviews of Effects (DARE).

Reviews eligible for inclusion were those of randomised controlled trials (RCTs), quasi-randomised controlled trials (CCTs), controlled before-and-after studies (CBAs), interrupted time series (ITS) or before-and-after (BA) studies. Reviews of other study designs or of qualitative studies were excluded, although issues raised by reviews of qualitative studies were considered in the Background.

Types of participants

We included consumers, defined as any person using medicine(s), either a patient, carer or both, and targeted as individuals or as a group. We also included healthcare professionals who prescribed or monitored medicines. To be included, interventions must have explicitly targeted consumers as primary recipients. We excluded reviews which focussed solely or primarily on interventions for healthcare professionals, services or systems, as these are the focus of work undertaken by the Cochrane Effective Practice and Organisation of Care (EPOC) Review Group (Weir 2009).

There were no restrictions based on the medicines being used or prescribed (type of medicine, indication, number of concurrent medicines), the number or type of health problems, or other participant features.

'Medicines' were defined as any prescribed or over the counter medicine, taken acutely, chronically or intermittently. We also considered vaccines as medicines.

Types of interventions

There are many interventions to influence the use of medicines by consumers. To help to organise and provide a framework for selecting and evaluating interventions we developed a taxonomy based on the purpose of interventions (for details of the intervention taxonomy see Additional Table 1; Lowe 2010). We included interventions which fell into one of the eight categories below:

Table 1. Taxonomy of interventions and reviews mapped to intervention categories
Providing information or education (20 reviews)

Definition: Strategies to enable consumers to know about their treatment and their health.

Interventions include those to educate, provide information or to promote health or treatment. Interventions can be provided to individuals or groups, in print or verbally, or face to face or remotely. Interventions may be simple, such as those seeking solely to educate or provide information; or complex, such as those to promote or manage health or treatment as part of a multifaceted strategy.

Examples of interventions:

Reviews mapped to this category:

Amico 2006; Beney 2000; Halpern 2006; Haynes 2008; Koshman 2008; Lewin 2005; Maglione 2002; Morrison 2001; Nicolson 2009; Olthoff 2005; Rueda 2006; Russell 2006; Schedlbauer 2004; Schroeder 2004; Stevenson 2004; Stone 2002; Van Wijk 2005; Vergouwen 2003; Vermeire 2005; Zygmunt 2002

Bottom-line statements of effectiveness:

Overall, there is insufficient evidence to support the use of interventions that provide information or education as a single component to improve adherence or clinical outcomes - it is generally ineffective. However, there is some evidence that patient education as a single component or as part of a complex intervention may be effective to improve immunisation rates. There is insufficient evidence to determine whether this intervention alone improves other outcomes such as knowledge or adverse effects. When used in combination with other interventions, such as self-management skills training or counselling, there is some evidence that it may improve adherence and other outcomes such as clinical outcomes, but results are mixed.

Supporting behaviour change (26 reviews)

Definition: Strategies focussing on the adoption or promotion of health behaviours and treatment behaviours, such as adherence to medicines.

Included are interventions at an individual level that address behaviour change for the under-use, overuse or misuse of medicines, and may include practical strategies to assist consumers in taking their medicines correctly such as reminder devices, pre-packaging of multiple medicines, or different or simplified medicines formulations.

Examples of interventions:

Reviews mapped to this category:

Amico 2006; Beney 2000; Bhogal 2006; Halpern 2006; Haynes 2008; Heneghan 2006a; Heneghan 2006b; Jacobson 2005; Koshman 2008; Lewin 2005; Maglione 2002; McIntosh 2006; Morrison 2001; Nicolson 2009; Olthoff 2005; Orton 2005; Rueda 2006; Russell 2006; Schedlbauer 2004; Schroeder 2004; Stone 2002; van Eijken 2003; Vergouwen 2003; Vermeire 2005; Volmink 2007; Zygmunt 2002

Bottom-line statements of effectiveness:

Overall, there were mixed effects in general of interventions to support behaviour change in relation to medicines use. In general, there is some evidence of the effectiveness of simple interventions for short-term treatments, and complex interventions for long-term treatments to improve adherence and clinical outcomes.

More specifically, there is sufficient evidence that self-monitoring or self-management improve medicines adherence and some evidence that reminders and lay health worker interventions improve immunisation rates. There is some evidence that simplified dosing regimens are generally effective to improve medicines adherence. There is also some evidence that reminders, support and education or support and motivation interventions, and those involving pharmacists directly, are effective to improve adherence, although results are mixed. 

Acquiring skills and competencies (9 reviews)

Definition: Strategies focussing on the acquisition of skills relevant to medicines use.

These interventions aim to assist consumers to develop a broad set of competencies around medicines use and health, such as medicines management or monitoring; or training consumers in the correct use of devices to deliver treatment, or the correct use of treatments.

Examples of interventions:

Reviews mapped to this category:

Amico 2006; Bhogal 2006; Haynes 2008; Heneghan 2006b; Morrison 2001; Rueda 2006; Roughead 2005; Russell 2006; Vermeire 2005

Bottom-line statements of effectiveness:

There is some evidence that strategies which focus on the acquisition of skills and competencies may improve adherence, clinical outcomes and other outcomes, but results are mixed. Regarding specific types of interventions, there is sufficient evidence that self-monitoring decreases adverse events, it is generally effective. There is insufficient evidence to support the provision of training by pharmacists to improve adherence, but some evidence that it improves knowledge and medicines use. There is some evidence to support the provision of counselling of patients and/or physicians by pharmacists to improve adherence, but insufficient evidence to support more intensive patient care by pharmacists.

Support (12 reviews)

Definition: Strategies to provide assistance and encouragement to help consumers to cope with and manage their health and related medicines use.

Interventions can target patients or carers, as individuals or in groups, and may be delivered face to face or remotely.

Examples of interventions:

  • Counselling (group or individual, structured) and support (Amico 2006; Halpern 2006)

  • Therapy (family intervention, psychological therapy, cognitive behavioural therapy, motivational interviewing (Haynes 2008; Rueda 2006)

  • Group programmes (peer support and shared identification) (Zygmunt 2002)

Reviews mapped to this category:

Amico 2006; Halpern 2006; Haynes 2008; Lummis 2006; McIntosh 2006; Rueda 2006; Russell 2006; Schroeder 2004; Stevenson 2004; Van Wijk 2005; Vergouwen 2003; Zygmunt 2002

Bottom-line statements of effectiveness:

Due to the mixed results from studies found in most reviews, there is some evidence that interventions that provide support alone or in combination with other strategies may be effective to improve adherence and other outcomes . There is insufficient evidence to determine for which conditions support may be effective, or who should provide the support for greatest effect.

Facilitating communication and/or decision making (8 reviews)

Definition: Strategies to involve consumers in decision making about medicines.

Interventions include those that aim to help consumers make decisions about medicines use; to encourage consumers to express their beliefs, values and preferences about treatments and care; and/or to optimise communication with consumers about medicines use and related issues.

Examples of interventions:

  • Written action plans (Bhogal 2006)

  • Pharmaceutical care services including on-to-one consultation to manage medicines-related problems and develop a care plan (Roughead 2005

  • Delayed antibiotic prescriptions (Spurling 2007)

  • Written question lists for pharmacists; doctor and patient communication skills training (Stevenson 2004)

Reviews mapped to this category:

Bhogal 2006; Halpern 2006; Haynes 2008; McIntosh 2006; Roughead 2005; Spurling 2007; Stevenson 2004; Zygmunt 2002

Bottom-line statements of effectiveness:

There is insufficient evidence from one key review to determine whether interventions specifically focussed on promoting communication between patients and professionals is effective. 

Overall, there is some evidence to support the use of interventions which do not have a specific focus on facilitating decision making and/or communication, but effects are mixed. Delayed prescribing is effective to decrease antibiotic use, but also decreases satisfaction and has mixed effects on clinical outcomes and adverse events. In general, there is some evidence of effect for adherence, knowledge or other outcomes with education and enhanced follow-up, and pharmaceutical care services, but insufficient evidence to support the use of psychosocial interventions, which are generally ineffective. There is insufficient evidence to determine the effectiveness of structured counselling or compliance therapy.

Minimising risks or harms (15 reviews)

Definition: Strategies with a specific focus on preventing or managing adverse events of treatment and complications of disease.

Interventions can be for ongoing treatment or related to emergency or crisis events. Strategies can be to minimise risks or harms at an individual level or a population level (eg reducing antibiotic use; or augmenting immunisation uptake).

Examples of interventions:

  • Harm reduction training (Amico 2006)

  • Mass mailings (such as personalised letters, postcards, brochures) for immunisation uptake (Maglione 2002)

  • Self-monitoring, with or without self-adjustment of medicines (Heneghan 2006b)

  • Directly observed therapy (Amico 2006; Volmink 2007)

  • Complex interventions incorporating medicines review to reduce adverse events and/or falls (Royal 2006)

Reviews mapped to this category:

Amico 2006; Bhogal 2006; Haynes 2008; Heneghan 2006b; Jacobson 2005; Koshman 2008; Lewin 2005; Lummis 2006; Maglione 2002; Roughead 2005; Royal 2006; Spurling 2007; Stevenson 2004; Stone 2002; Volmink 2007

Bottom-line statements of effectiveness:

There is sufficient evidence that self-monitoring, with or without self-adjustment, is effective at decreasing adverse events of treatment. There is also some evidence that strategies to improve interactions between healthcare professionals and patients may decrease adverse events and improve other outcomes, but results are mixed. In particular, there is insufficient evidence to determine whether the use of patients’ own medicines in hospital is effective. There is also some evidence that educational strategies to minimise risks and harms may be effective, and telling patients about adverse effects of medicines does not negatively influence adherence.

For immunisation uptake, there is sufficient-to-some evidence that organisational change, reminders and recall, financial incentives, education and lay health worker interventions are each generally effective, while effects of mass mailings are mixed, and reminders with outreach are ineffective. There is some evidence that directly observed therapy for tuberculosis is generally ineffective to improve cure or treatment completion. There is sufficient evidence that delayed antibiotic prescription decreases antibiotic use without increases in complications, but it may increase supplementary medicines use and results are mixed for clinical outcomes and adverse effects.

Improving quality (17 reviews)

Definition: Strategies to improve the total package, coordination or integration of care delivered.

Interventions can involve substitution or expansion of one type of care, such as interventions that aim to overcome system barriers to medicines use, including access and financial barriers.

Examples of interventions:

  • Financial interventions (reference pricing, index pricing) (Aaserud 2006), caps and co-payments (Austvoll-Dahlgren 2008; Maio 2005), financial incentives (Giuffrida 1997)

  • Collaborative care interventions (Bower 2006)

  • Lay health mentoring, comprehensive pharmaceutical care services (Haynes 2008)

  • Pharmaceutical care services including one-to-one consultation to develop care plan and provide follow-up (Roughead 2005)

  • Pharmacist-directed care including medicines assessment and review, self-monitoring education and GP liaison (Koshman 2008)

Reviews mapped to this category:

Aaserud 2006; Austvoll-Dahlgren 2008; Beney 2000; Bower 2006; Giuffrida 1997; Haynes 2008; Koshman 2008; Lewin 2005; Lummis 2006; Maio 2005; Roughead 2005; Royal 2006; Schroeder 2004; Stevenson 2004; Stone 2002; van Eijken 2003; Vergouwen 2003

Bottom-line statements of effectiveness:

Because this overview did not specifically include reviews which targeted organisational or structural interventions to change consumer medicines use, provisional conclusions about the effectiveness of those interventions are provided here. There is some evidence that changing the coordination of care (eg changing roles of healthcare professionals to interact with patients or to provide additional services to patients) may improve adherence and other outcomes related to medicines use, however the results from most reviews are mixed. In depression, there appears to be some evidence that these interventions are generally effective. There is also some evidence that financial interventions are effective. In addition, the review of using a patient's own medicines (POMs) included less rigorous study designs which provides insufficient evidence to determine whether supporting POMs in hospital is effective. We did however search broadly for reviews in immunisation uptake and found that there is sufficient evidence that organisational interventions are effective at improving uptake. There is also some evidence that pharmaceutical pricing policies to indirectly influence consumers’ use of medicines are effective to improve medicines use and costs, but results are mixed for effects on health status and health service use.

Consumer system participation (0 reviews)

Definition: Strategies to involve consumers in decision making processes on medicines prescribing and use at a system level, such as in research planning, formulary and policy decisions.

Interventions can involve consumers in different roles, such as planning, research, audit and review and governance.

Examples of interventions:

  • Consumer involvement in developing patient medicines information

  • Medicines policy or guideline committee involvement

Reviews mapped to this category:

None.

Bottom-line statements of effectiveness:

There is insufficient evidence to determine the effects of consumer system participation because no reviews were identified.

  • Providing information or education

  • Facilitating communication and/or decision making

  • Acquiring skills and competencies

  • Supporting behaviour change

  • Support

  • Minimising risks or harms

  • Improving quality

  • Consumer system participation.

Many systematic reviews may be relevant to understanding the effects of interventions relevant to the use of medicines by consumers. We developed selection criteria to help us to identify the most highly-relevant reviews. These selection criteria were used to rank reviews as high, moderate or low/ very low relevance. In this overview, we included only those reviews rated as high relevance to consumers' medicines use, based on meeting the following criteria:

  • The main objective of review focussed exclusively on evidence-based medicines use by and prescribing for consumers; and

  • The interventions in the review were directed to consumers and exclusively focussed on evidence-based medicines use by and prescribing for consumers; and

  • The outcomes of studies in the review were related to medicines use by and prescribing for consumers (searched for and/or found and/or reported).

We included reviews evaluating both 'direct to consumer' interventions and ‘indirect to consumer’ interventions, or a combination. Direct to consumer interventions were defined as those with a direct interface or line of communication with consumers, for example, through education or counselling. In comparison, indirect to consumer interventions were not immediate to the consumer, but still aimed to influence their medicines use, for example, through structural, organisational, financing or system of care delivery strategies.

There were no restrictions according to the: medical condition(s); type(s) of medicine prescribed, taken or targeted; intervention setting; or duration of treatment.

Reviews were also unrestricted based on comparisons examined, therefore all of the following were eligible for inclusion, whether interventions were simple or complex:

  • Intervention versus control (any - no intervention, usual care, placebo or other control)

  • One intervention versus another.

Types of outcome measures

We sought data for outcomes in the following categories:

  • Consumer-oriented outcomes, such as knowledge and understanding, skills acquisition, and health status and wellbeing;

  • Provider-oriented outcomes, including knowledge and understanding and evaluation of care; and

  • Health service-oriented outcomes, including service use outcomes and costs.

For a full list of outcomes sought, see Appendix 1.

Search methods for identification of reviews

Handsearching for reviews

Reviews of consumer-oriented medicines interventions cannot be reliably identified by key word or subject heading searching as they typically encompass a diverse range of interventions. Additionally, relevant interventions may be disease-specific but also have applicability across diseases, and so are not reliably captured using systematic database searches.

Relevant reviews were therefore identified in two steps:

Step 1: Identification of all reviews on consumer-oriented interventions.

One investigator handsearched CDSR and DARE up to Issue 3 2008 (inclusive) of The Cochrane Library, screening by review title and abstract to identify all reviews published in English and relevant to communicating with consumers and improving their participation in health care (irrespective of relevance to medicines). This involved identifying all reviews of interventions to communicate with, inform or educate, support or seek the participation of consumers.

Step 2: Selection of reviews relevant to medicines use.

Two investigators independently screened the set of reviews identified in step 1, by title and abstract to identify all reviews of any relevance (high, moderate, low/very low) to consumers' medicines use. All reviews identified as relevant in this step were obtained in full text for further assessment.

Data collection and analysis

Selection of reviews identified by handsearching

We categorised reviews of any relevance to consumers' medicines use by assessing the full-text review. Two investigators working independently applied the criteria for ranking reviews (outlined in Criteria for considering reviews for inclusion - Types of interventions) to identify those of high relevance. Differences were resolved by discussion or by consultation with a third party to reach consensus.We excluded from this overview those reviews ranked as moderate relevance or lower (see Additional Table 2).

Table 2. Excluded reviews and reasons for exclusion
Author DateReason for exclusion
Ara 2004Moderate or lower relevance
Armour 2005Moderate or lower relevance  
Arnold 2005Moderate or lower relevance  (indirect)
Arroll 2003Overlap
Bailey 2009Moderate or lower relevance
Blackstock 2006Moderate or lower relevance  
Bordley 2000Too indirect to consumer
Bosch-Capblanch 2007Moderate or lower relevance
Bradley 2008Moderate or lower relevance
Bravata 2007Moderate or lower relevance
Bridle 2005Moderate or lower relevance
Briss 2000Overlap
Brown 2004Moderate or lower relevance
Cabana 2004Moderate or lower relevance
Chang 2007Moderate or lower relevance
Christensen 2007Overlap
Clar 2007Moderate or lower relevance
Clark 2007Moderate or lower relevance
Connor 2004Overlap
Cooper 2006Moderate or lower relevance
Costello 2004Overlap
Cote 2005Overlap
Craven 2006Moderate or lower relevance
Currell 2000Moderate or lower relevance
Davey 2005Moderate or lower relevance
Deakin 2005Moderate or lower relevance
Desplenter 2006Overlap
Doggett 2005Moderate or lower relevance
Dolder 2003Overlap
Donald 2005Moderate or lower relevance
Ebrahim 2006Moderate or lower relevance
Effing 2007Moderate or lower relevance
Ersser 2007Moderate or lower relevance
Fahey 2005Moderate or lower relevance
Fahey 2006Moderate or lower relevance
Finley 2003Low quality (AMSTAR < 4)
Forster 2008Moderate or lower relevance
Foster 2007Moderate or lower relevance
Gagnon 2007Moderate or lower relevance
Gaston 2005Moderate or lower relevance
Gensichen 2006Moderate or lower relevance 
Gibson 2002Moderate or lower relevance
Gibson 2003Moderate or lower relevance
Gilbody 2003Moderate or lower relevance
Gill 1999Moderate or lower relevance  (indirect)
Gillespie 2001Moderate or lower relevance
Glazier 2006Moderate or lower relevance
Grilli 2002Moderate or lower relevance  (indirect)
Gunn 2006Moderate or lower relevance  
Haby 2001Moderate or lower relevance
Handford 2006Moderate or lower relevance
Harding 2005Moderate or lower relevance
Harris 2005Moderate or lower relevance
Harris 2006Moderate or lower relevance
Haynes 1996Overlap
Heideman 2005Moderate or lower relevance
Henderson 1999Moderate or lower relevance
Hersh 2006Moderate or lower relevance
Hey 2005Moderate or lower relevance
Higgins 2004Overlap
Hiller 2002Moderate or lower relevance
Hodnett 2000Moderate or lower relevance
Holland 2005Moderate or lower relevance
Hulscher 1999Moderate or lower relevance
Hwang 2005Moderate or lower relevance
Ilott 2005Overlap
Ioannidis 2001Moderate or lower relevance (indirect)
Jaber 2006Moderate or lower relevance
Jamtvedt 2006Moderate or lower relevance
Johnson 2003Moderate or lower relevance
Jones 2002Moderate or lower relevance
Jovicic 2006Moderate or lower relevance
Kaboli 2006Low quality (AMSTAR < 4)
Kaper 2005Moderate or lower relevance
Karjalainen 1999Moderate or lower relevance
Karjalainen 2003aModerate or lower relevance
Karjalainen 2003bModerate or lower relevance
Kendrick 2000Overlap, outdated               
Kripalani 2007Overlap
Krueger 2003Overlap, low quality (AMSTAR < 4)
Kwan 2004Moderate or lower relevance
Lagarde 2007Moderate or lower relevance
Lancaster 2002Moderate or lower relevance
Lancaster 2004Moderate or lower relevance
Leach 2006Moderate or lower relevance
Lewin 2001Moderate or lower relevance
Liss 2002Moderate or lower relevance
Loveman 2003Moderate or lower relevance
Lussier 2006Moderate or lower relevance
Malone 2007Moderate or lower relevance
Marshall 1998Moderate or lower relevance
Martire 2005Moderate or lower relevance
McClure 2005Moderate or lower relevance
McDonald 2002Overlap
McGraw 2004Overlap
McKinstry 2006Moderate or lower relevance
Michie 2003Moderate or lower relevance
Mistiaen 2006Moderate or lower relevance
Murray 2005Moderate or lower relevance
Ndiaye 2005Overlap
Newell 1999Overlap, outdated
Niesink 2007Moderate or lower relevance
Nilsen 2006Moderate or lower relevance
Norris 2006Moderate or lower relevance
Nose 2003Overlap
O'Connor 2009Moderate or lower relevance
Odegaard 2004Overlap
Page 2005Moderate or lower relevance
Pampallona 2002Overlap
Pegurri 2005Overlap
Pekkala 2002Moderate or lower relevance
Pelletier 2001Moderate or lower relevance
Petersen 2006Moderate or lower relevance
Pharoah 2006Moderate or lower relevance
Pignone 2005Moderate or lower relevance
Ploeg 2005Moderate or lower relevance
Ponniah 2007Overlap
Powell 2003Moderate or lower relevance
Raymond 2007Moderate or lower relevance
Raynor 2007Overlap
Reda 2001Moderate or lower relevance
Renders 2000Moderate or lower relevance 
Rice 2004Moderate or lower relevance
Riemsma 2003Moderate or lower relevance
Roccaforte 2005Moderate or lower relevance
Rollason 2003Low quality (AMSTAR < 4)
Sajatovic 2004Overlap
Sanders 2006Moderate or lower relevance
Sangani 2001Moderate or lower relevance
Schroeder 2004aOverlap
Scott 2003Moderate or lower relevance
Scott 2008Moderate or lower relevance
Shaw 2007Moderate or lower relevance
Shepperd 2004Moderate or lower relevance
Siebenhofer 2004 Overlap
Simpson 2005Too indirect to consumer
Sinclair 2004Moderate or lower relevance
Smetana 2007Moderate or lower relevance
Smith 2001Moderate or lower relevance
Smith 2005Moderate or lower relevance
Smith 2007aModerate or lower relevance
Smith 2007bModerate or lower relevance
Stokes 2007Moderate or lower relevance
Szilagyi 2000Overlap
Tapp 2007Moderate or lower relevance
Taylor 2005Moderate or lower relevance
Toelle 2004Moderate or lower relevance
Tsai 2005Moderate or lower relevance
Tully 2000Low quality (AMSTAR < 4)
Turnock 2005Moderate or lower relevance
Urquhart 2005Qualitative review
Valk 2005Moderate or lower relevance
Van der Wal 2005Moderate or lower relevance
Volmink 1997Overlap
Walburn 2001Qualitative review
Walton 2008Too indirect to consumer
Wendt 1998Overlap, outdated
Werawatganon 2005Moderate or lower relevance
Whittaker 2002Overlap
Winkley 2006Moderate or lower relevance
Winterbottom 2008Moderate or lower relevance
Wise 2007Moderate or lower relevance
Wofford 2005Moderate or lower relevance
Wolf 2003Moderate or lower relevance
Woolacott 2006Moderate or lower relevance
Yu 2006Moderate or lower relevance
Zhang 2007Moderate or lower relevance

To minimise duplication amongst reviews, we developed a second set of selection criteria which were applied in two steps (see below) to high relevance non-Cochrane systematic reviews identified in DARE. This was in order to remove reviews that were considered to duplicate Cochrane reviews, while retaining reviews whose scope was not covered by Cochrane reviews.The rationale for this decision was twofold: first, Cochrane reviews are, in the main, of higher quality than systematic reviews from other sources (Moher 2007); and second, Cochrane reviews are regularly updated to reflect the state of the evidence, whereas reviews from other sources typically are not.

High relevance non-Cochrane reviews were therefore screened in two further selection steps and were excluded from this overview if:

Step 1: The review was of low quality. The non-Cochrane review was excluded if it was rated as low quality or had serious methodological flaws according to the Centre for Reviews and Dissemination assessment of the review published as part of the DARE abstract; and as assessed by the reviewers (NS, DL, RR) using the AMSTAR assessment tool (Shea 2007; rating of less than 4 of a possible 11 points; see Data collection and analysis - Quality assessment of included reviews, for details of the AMSTAR tool and assessment).

Step 2: The review had substantial overlap with Cochrane reviews. For each non-Cochrane review, we identified all Cochrane reviews with a similar scope and the degree of overlap with these Cochrane review(s) assessed to determine how many unique studies would be contributed by inclusion of the non-Cochrane review. Non-Cochrane reviews with approximately 50% or more of their studies already captured by Cochrane reviews were generally excluded. Where we identified two non-Cochrane reviews with similar scope (duplicative reviews) the higher-quality review was included in this overview.  Two investigators working independently assessed these reviews, with differences resolved by discussion or by consultation with a third party to reach consensus. We provide details of high relevance, non-Cochrane reviews excluded at the full-text stage based on these two further screening steps in Additional Table 2, with reasons for exclusion.

Data extraction

We developed and piloted a data extraction form to summarise the key characteristics of reviews, including information about the objectives, participants, intervention features, outcomes assessed and comparisons performed; as well as the quality of included studies, quality of the review and the review's results. One investigator extracted data and a second investigator verified the extracted data. Differences were resolved by discussion to reach consensus.

Quality assessment of included reviews

We assessed the quality of included systematic reviews using the AMSTAR instrument (Shea 2007). AMSTAR assesses the degree to which review methods avoided bias by evaluating the methods against 11 distinct criteria, including:

  • use of an 'a priori’ design;

  • duplicate study selection and data extraction;

  • comprehensive searching of the literature;

  • use of publication status as an exclusion criterion;

  • provision of (included and excluded) studies;

  • provision of characteristics of included studies;

  • assessment of methodological quality of included studies;

  • appropriate use of quality of included studies in formulating conclusions;

  • appropriate methods for combining results of studies;

  • assessment of publication bias; and

  • conflict of interest (both review and included studies) stated.

Each AMSTAR item was rated as yes (clearly done), no (clearly not done), can't answer, or not applicable, based on the published review report. A review that adequately met all of the 11 criteria was considered to be a review of the highest quality. Quality rating was as follows:

AMSTAR score (out of 11 criteria)Rating
8 to 11high quality
4 to 7medium quality
3 or lowerlow quality

One investigator assessed review quality and a second investigator verified this assessment. Differences were resolved by discussion to reach consensus.

Quality assessment of included studies within reviews

We did not reassess the quality of included studies within reviews but instead reported study quality according to the review authors' assessment. We collected this information during the data extraction process. We used ratings of study quality in the synthesis and interpretation of results; for example, to downplay the certainty of conclusions and ratings of effectiveness where studies were all of poor methodological quality or had serious methodological shortcomings that may have predisposed the review's results to bias. For example, finding intervention 'x' effective but with serious methodological limitations, conclusions of 'sufficient evidence' would be downgraded to 'some evidence' to reflect a lower degree of confidence in the findings from the review overall.

Extracting data and identifying relevant outcomes

We identified outcomes for data extraction by screening against the medicines outcome taxonomy developed for this overview (Appendix 1), agreed upon by two investigators reaching consensus.

We developed the medicines outcome taxonomy by assessing the range and types of interventions on consumers' medicines use (and used to develop the intervention taxonomy; see Lowe 2010; Ryan 2010); and by those outcomes relevant to these medicines interventions. The identified medicines-related outcomes were then mapped back onto Cochrane Consumers and Communication Review Group (CC&CRG) outcome taxonomy categories, which comprehensively articulates and organises outcomes on communication with and participation by consumers. It maps outcomes at different levels within the health system: consumers, providers and systems (available at: www.latrobe.edu.au/chcp/assets/downloads/Outcomes.pdf).

These steps were undertaken by two investigators working together (NS, RR), in consultation with a third investigator with expertise in research on consumer communication and participation (SH). These steps were used to identify iteratively broad and specific medicines outcomes and to organise them into a meaningful taxonomy.

Statistical presentation of results from reviews

For each included review, we extracted all results for medicines-related outcomes.

Within individual reviews, we extracted and reported, where available, pooled effect sizes for outcomes meta-analysed in reviews; or a range of effect sizes from their included studies. We preferred absolute rather than relative effect sizes, and calculated these wherever possible (Akl 2011). In all cases, one investigator extracted results and performed conversions to absolute effect sizes, and this was verified by a second investigator, with disagreements resolved by discussion to reach consensus.

If the above information was not available, we used vote counting by direction of effect or by statistical significance, in order to allow us to report results consistently across included reviews. Vote counting sums and compares the numbers of studies reporting particular outcomes, for example: the numbers of studies reporting positive results compared with the number reporting negative results for a particular outcome; or the number of studies reporting statistically significant results compared with the number reporting no statistically significant results for a certain outcome. Where none of these forms of reporting were possible, for example, where outcomes were reported descriptively by single studies, we reported these results using standardised language indicating direction of effect and statistical significance. It should also be noted that vote counting as a synthesis method is by no means universally accepted, and often researchers are advised to adopt other ways of summarising results (Higgins 2009). We acknowledge that vote counting has limitations, but we adopted it for this overview as there were few other robust alternative methods by which to summarise such diverse results across reviews.

Summaries of main results

As well as numerical data, we extracted and descriptively summarised each study's results using standardised language, in order to allow consistent reporting of results across reviews. Two investigators analysed and summarised results of the included reviews, and reported them narratively to enable identification of broad conclusions within and across reviews.

Synthesis of results and rating the evidence of effectiveness

We formulated standardised ‘effectiveness statements’ to rate the evidence arising from reviews, using a further synthesis step that went beyond a simple summary of the main results of each review. These statements were based on the rating scheme (see Appendix 2) developed by the CC&CRG to help synthesise and rate the evidence across systematic reviews where interventions are complex and diverse (Ryan 2005; Ryan 2009a).

The effectiveness statements give bottom-line statements about the main effects of interventions assessed within each intervention category, using standardised language and based on a set of decision rules that take into account the results, statistical significance and the quality and number of studies on which the result is based. For example, the result from a review reported as 'education significantly improved adherence in 1 study' would be given as the bottom-line statement 'there is insufficient evidence to determine the effects of education on adherence'.

One investigator systematically rated the review's results and a second investigator verified the rating, with disagreements resolved by discussion to reach consensus.

Mapping of reviews to the intervention taxonomy and summarising results across reviews

First, we organised the evidence using the taxonomy of interventions developed in parallel with this overview (Lowe 2010); see Criteria for considering reviews for inclusion - Types of interventions and Additional Table 1.

As part of the data extraction process one investigator assigned reviews to one or more categories of the taxonomy, based on the review's aims. A second investigator verified this mapping, with differences resolved by discussion to reach consensus. As a result of differences in the scope (range) of included interventions within reviews, intervention categories were not treated as being mutually exclusive: some reviews were mapped to multiple categories, while others appear in only one, but this was performed in a deliberate attempt to disaggregate the evidence contributing towards different medicines use objectives. This step was designed to deal with differences across reviews in the way that interventions were split or lumped together. Reviews in which interventions were lumped (for instance, all interventions to promote adherence in Haynes 2008) needed to be unpacked into constituent interventions, based on their purposes, to allow interpretation of the extracted data and results.

One investigator systematically synthesised each review's extracted data, mapped to an intervention category, to produce an overall summary of the evidence for that intervention category. We developed overall summaries of the standardised statements of effectiveness for each intervention category by systematically summarising the assembled statements for all reviews mapped to that category. Summaries were written by one investigator and checked by a second investigator, with differences resolved by discussion to reach consensus.

Consumer participation

A consumer representative reviewed the protocol and a consumer peer-reviewed this overview to ensure that consumers' views are adequately and accurately represented.

Results

Description of included reviews

We handsearched over 3000 Cochrane systematic reviews of interventions for health care and 8000 reviews in the DARE database, to identify reviews relevant to prescribing for, and medicines use by, consumers.

After screening titles and/or abstracts, we retrieved 204 reviews in full text for further assessment.

After further selection, quality assessment, and categorisation as 'high' or 'other' relevance to consumers' medicines use, we excluded 167 reviews for the following reasons:

  • Moderate or lower relevance to consumers' medicines use (127/167, 76%).

  • Significant degree of overlap with other reviews (31/167, 19%).

  • Low quality review (total AMSTAR score < 4) (4/167, 2%).

  • Too indirect to consumer (interventions and/or outcomes) (3/167, 2%)

  • Review of qualitative studies (2/167, 1%).

See Additional Table 2 for full list of reviews and reasons for exclusion. Figure 1 gives a flow diagram outlining the selection process and review numbers at each stage.

Figure 1.

Review selection process and numbers

After all selection and categorisation steps, we identified 37 reviews for inclusion. Almost half (18/37, 49%) were Cochrane systematic reviews of high relevance to consumers' use of medicines.

Objectives and scope of the reviews

The objectives and scope of included reviews varied, although just over half (20/37, 54%) primarily aimed to improve adherence to medicines or immunisations (Amico 2006; Halpern 2006; Haynes 2008; Heneghan 2006a; Jacobson 2005; Lewin 2005, McIntosh 2006; Olthoff 2005; Orton 2005; Rueda 2006; Russell 2006; Schedlbauer 2004; Schroeder 2004; Stone 2002, van Eijken 2003; Van Wijk 2005; Vergouwen 2003; Vermeire 2005; Volmink 2007; Zygmunt 2002). These reviews included a wide range of specific strategies, and although reviews most consistently reported on the interventions' effects on adherence, additional effects (outcomes) were also reported, such as clinical outcomes, adverse events, satisfaction, attitudes to medicines, quality of life, and costs.

A further 12 reviews took a slightly wider focus, for example considering medicines use within a clinical or self-management context, while still targeting consumers directly (Beney 2000; Bhogal 2006; Bower 2006; Heneghan 2006b; Koshman 2008; Lummis 2006; Morrison 2001; Nicolson 2009; Roughead 2005; Royal 2006; Spurling 2007; Stevenson 2004). These reviews also reported on the effects on adherence, but more consistently reported a range of additional outcomes, such as knowledge, understanding and recall, adverse events, medicines errors, health service use and professionals' workload, dropouts and withdrawals, costs, clinical outcomes, and quality of life.

Some reviews evaluated interventions which targeted consumers both directly and indirectly. For example, Vergouwen 2003 reviewed interventions to improve adherence to antidepressants in which education (direct to consumers) and collaborative care (indirect to consumers through changes to the organisation of care) were evaluated. Similarly, Stone 2002 reviewed interventions which targeted different levels of the health system to improve adult immunisation, including direct-to-consumer interventions such as patient reminders or financial incentives, as well as indirect to consumer interventions such as organisational change, provider financial incentives and provider education. In such reviews, the combined data, plus any separate data from the direct and indirect interventions, were collected and reported wherever possible.

Five reviews evaluated interventions which were aimed at consumers indirectly (Aaserud 2006; Austvoll-Dahlgren 2008; Giuffrida 1997; Maio 2005; Maglione 2002). All assessed the effects of financial interventions to indirectly influence consumers' use of medicines, except Maglione 2002 which assessed the effects of mass mailing strategies on immunisation uptake. These indirect-to-consumer reviews reported a range of consumer outcomes relevant to prescribing and medicines use, and this distinguishes the reviews included in this overview from those failing to report consumer outcomes, and which were therefore excluded on the basis of being rated as 'too indirect' to consumer.

Study characteristics and populations

Almost half of the included reviews (18/37 reviews, 49%) included only randomised controlled trials (RCTs) (Bower 2006; Giuffrida 1997; Halpern 2006; Haynes 2008; Heneghan 2006a; Heneghan 2006b; Koshman 2008; Lewin 2005; McIntosh 2006; Nicolson 2009; Roughead 2005; Rueda 2006; Russell 2006; Schedlbauer 2004; Schroeder 2004; Spurling 2007; van Eijken 2003; Vergouwen 2003). This reflects in part the wide recognition that RCTs represent the 'gold standard' study design for evaluating intervention effects. Study designs other than RCTs may be more prone to bias. Nonetheless, selected studies other than RCTs may be appropriate for assessing the effects of complex interventions (EPOC 2007; Higgins 2009; Ryan 2009b).

We explicitly attempted to avoid duplication by excluding non-Cochrane reviews that had substantial overlap with Cochrane reviews, seeing little advantage in including multiple reviews assessing the same set of trials split and organised in different ways. On the other hand, we also wished to maximise the spread of the interventions under evaluation, and so our assessment of overlap between reviews was formalised and systematic. The result of this trade-off between duplication and coverage is that there is still some overlap between included reviews. To some extent, overlap is unavoidable where large, general reviews looking across diseases (eg Haynes 2008) are assembled together with more focussed reviews (eg in terms of disease, population, setting or intervention), but with different selection criteria (eg study design, methodological quality, setting, participants eligible).

We therefore report the range of studies in each review as opposed to a total number of included studies, as we cannot be certain that there are not duplicated studies across included reviews. There were between 1 and 78 intervention studies included in each review in this overview. The most recent search date in the reviews was September 2007 and the oldest April 1997, but most reviews (24/37, 65%) had conducted their searches before the end of 2004.

Most studies involved adult participants. Three reviews (Maio 2005; Russell 2006; van Eijken 2003) focussed on older adults (60 years or older), while another six reviews incorporated a wide range of ages that included older with younger adults (Aaserud 2006; Haynes 2008; Heneghan 2006a; Jacobson 2005; Koshman 2008; Nicolson 2009). Eight reviews included child participants (Bhogal 2006; Giuffrida 1997; Heneghan 2006b; Jacobson 2005; Lewin 2005; Roughead 2005; Rueda 2006; Spurling 2007).

While most studies focussed on people with a condition and/or taking medicines, as opposed to carers, 10 reviews included studies with carers (Amico 2006; Bhogal 2006; Giuffrida 1997; Haynes 2008; Heneghan 2006b; Jacobson 2005; Lewin 2005; Orton 2005; Spurling 2007; Zygmunt 2002), the majority of whom were family members. In nine reviews, healthcare professionals were included as recipients of the intervention alongside consumers (Beney 2000; Jacobson 2005; Lummis 2006; Morrison 2001; Royal 2006; Stevenson 2004; Stone 2002; Van Wijk 2005; Vergouwen 2003). In the majority of cases at least some of the professional population were pharmacists, but others including physicians, general practitioners, nurses, psychologists and psychiatrists were also included in a small number of reviews.

Seventeen reviews evaluated interventions for medicines use in relation to a particular medical condition, including:

Five reviews assessed the effects of interventions for immunisation uptake (Giuffrida 1997; Jacobson 2005; Lewin 2005; Maglione 2002; Stone 2002), although Giuffrida 1997 also included people with a range of chronic and other health problems and treatment aims. The remaining reviews evaluated interventions for medicines use more generally across populations (and diseases), for example including consumers taking medicines for any reason (Aaserud 2006; Austvoll-Dahlgren 2008; Beney 2000; Haynes 2008; Heneghan 2006a; Maio 2005; Morrison 2001; Nicolson 2009; Royal 2006; Russell 2006; Stevenson 2004; van Eijken 2003), although some were setting-specific (eg hospital inpatients (Lummis 2006); and community dwelling older adults (van Eijken 2003)).

One other review included adults and children with a range of chronic conditions but also specified that they must have been at high risk of medicines misadventure, eg those requiring polypharmacy (Roughead 2005). Most other reviews did not provide information on people requiring polypharmacy and/or with co-morbid conditions, although several reviews reported inclusion criteria and/or included studies in which people with more than one concurrent health problem or taking multiple medicines were represented (Austvoll-Dahlgren 2008; Beney 2000; Haynes 2008; Heneghan 2006a; Nicolson 2009; Royal 2006; Russell 2006; Roughead 2005; Volmink 2007).

Interventions

Strategies directly targeting consumers ranged from the simple (such as reducing the frequency of dosing, changing the medicine formulation, sending a postcard reminder) to the complex (different combinations of education, counselling and self-monitoring, or self-management programmes). To organise the evidence, we categorised included reviews according to the intervention taxonomy and mapped reviews to intervention categories based on the underlying aim(s) of the included interventions.

Reviews dealt with interventions in very different ways, reflecting underlying differences in the lumping or splitting of interventions. Some reviews were very focussed, assessing the effects of a single type of intervention with a single aim - such as reminder packaging (Heneghan 2006a) or unit-dose packaging (Orton 2005); these interventions aiming to improve medicines adherence were mapped to the category 'supporting behaviour change.'

Other reviews were similarly focussed on a single type of intervention, but the intervention itself had multiple purposes. For example, a review on delayed antibiotic prescription interventions (Spurling 2007) had dual aims, and was mapped to provide evidence for interventions to facilitate communication and decision making (ie enabling decision making about whether to take antibiotics or not) as well as to minimise the risks or harms of medicines use (ie to decrease antibiotic overuse/ emergence of antibiotic resistance in the community).

Other reviews incorporated interventions with a broader set of aims, and hence were mapped to a number of intervention categories. For example, the review by Haynes 2008 assessed all interventions to improve medicines adherence, including practical medicines management interventions such as reminders, packaging or dose simplification (mapped to the supporting behaviour change category); but also included a large range of other interventions with different aims which were mapped to different categories, such as counselling (support), self-monitoring (acquiring skills and competencies), instruction (providing information or education), and comprehensive pharmaceutical care services (improving quality).

Of note, we identified no reviews on 'consumer system participation' related to medicines, a relatively new area of consumer involvement in health care (see Additional Table 1). Similarly, relatively few reviews addressed the acquisition of skills and competencies, or aimed to facilitate communication and decision making, also relatively new areas in relation to consumers' use of medicines. In contrast, other categories were well populated with evidence from included reviews, in particular, both the 'supporting behaviour change' and 'providing information or education' intervention categories.

Outcomes

Most reviews, taken individually and collectively, reported on a relatively narrow range of outcomes, despite the fact that the range of included interventions and scope of reviews was diverse.

Additional Table 3 shows the range of outcomes reported by included reviews and how these relate to categories of the medicines outcome taxonomy. There was a large range of outcomes reported across the included reviews, taken collectively. However, many individual reviews focussed on adherence, often reporting a very small set of outcomes. Many reviews also did not report outcomes which are necessary for understanding the interventions under consideration, for example, reviews of educational interventions did not routinely report outcomes such as knowledge.

Table 3. Taxonomy of medicines outcomes and specific outcomes reported by included reviews
Outcome taxonomy categoryOutcomes reported
Health behaviour
  • Adherence or concordance (eg pill counts, prescription refill, blood levels, immunisation rate or uptake, discontinuation)

  • Appropriate use (eg initiation of medicines, use of appropriate medicine and/or dose)

Health status and wellbeing
  • Clinical and physiological outcomes of treatment with medicines (eg blood pressure, cure of tuberculosis, symptoms, quality of life, mortality, time in therapeutic range)

  • Personal cost of medicines (eg out of pocket expenses, patient medicines expenditure)

Consumer adverse events
  • Complications

  • Adverse effects of medicines, medicines-related problems

  • Withdrawals due to adverse events

  • Inability to complete monitoring/ treatment

Consumer evaluation of care
  • Patient satisfaction

  • Attitudes (towards condition, therapy, health professionals), preferences

  • Perceived barriers to use

  • Intentions to use monitoring strategy or medicines

Consumer involvement in the care process
  • Patient preferences

  • Communication or discussions with healthcare professionals

  • Medicines questions asked

  • Decision to take medicines

Support and consumer skills acquisition
  • Use and/or performance of self-management techniques or devices

  • Medicines question asking skill, number of questions asked

  • Correct application of medicines information

  • Self-efficacy

Knowledge and understanding
  • Knowledge (of medicines, health, adverse effects)

  • Recall

System benefits
  • Service use (eg hospital admission/ readmission, discharge time, emergency department visits, physician visits)

  • Costs (eg medicines pricing, cost containment)

  • Medicines errors (eg identified, administration errors)

  • Professional workload (eg pharmacist time, discharge time

Consultation and communication by provider
  • Repeated patient complaint

  • Asked patient to repeat instructions or demonstrate use

  • Addressed patient fears about new medicines

We present characteristics of included reviews in Additional Table 4; Table 5; Table 6; Table 7; Table 8.

Table 4. Characteristics of included reviews: Part A
  1. RCT: Randomised controlled trial

    CCT: Controlled clinical trial

    CBA: Controlled before-and-after study

    BA: Before-and-after study (uncontrolled)

    ITS: Interrupted time series

Aaserud 2006
Review question/objective: What are the effects of pharmaceutical pricing and purchasing policies on medicines use, healthcare utilisation, patient outcomes and costs?
StudiesSearch date up to: September 2005
 Number of studies related to medicines use:  11
 Study design:  ITS (simple and repeated measure designs; some with controls), CBA
ParticipantsPatients: elderly people aged 65 years and older; otherwise not specified. Medicines involved included nitrates, beta-blockers, ACE inhibitors, calcium channel blockers, histamine H2 receptor antagonists, proton pump inhibitors, antidiabetic agents, antibiotics, and antidepressants. Carers: not specified. Professionals: not specified.
SettingNot specified
InterventionsReference pricing; index pricing; other
 Maps to intervention taxonomy categories: Improving quality
OutcomesConsumer adverse events, system benefits, health status and wellbeing
Quality of the review (AMSTAR)10
Quality of included studiesOverall, included studies were generally well designed but some had serious limitations in design and implementation. Transferability across populations and settings may also be limited.
Amico 2006
Review question/objective: What are the effects of interventions to improve adherence to antiretroviral therapy (ART) for people living with HIV?
StudiesSearch date: from 1996 up to December 2004
 Number of studies related to medicines use:  24
 Study design:  RCT, CCT, CBA
ParticipantsPatients: people with Human Immunodeficiency Virus (HIV) and receiving antiretroviral therapy. Carers: informal caregivers. Professionals: none.
SettingCommunity, not specified
InterventionsAny intervention to improve adherence (support and referral interventions; education; feedback on viral load; reminder or calendar packaging or pill boxes; alarms; information provision, counselling and support; problem solving skills training; self-management medication training; harm reduction training; directly observed therapy; incentives; medication diaries); control
 Maps to intervention taxonomy categories: Providing information or education, Supporting behaviour change, Acquiring skills and competencies, Support, Minimising risks or harms
OutcomesHealth behaviour
Quality of the review (AMSTAR)4
Quality of included studiesIncluded study populations were generally small and may have been too small (based on power calculations) to detect effects of interventions. Half (52%) of included studies were RCTs, others included were of non-randomised or within-group design. Methodological quality was not formally assessed so the risk of bias is unknown.
Austvoll-Dahlgren 2008
Review question/objective: What are the effects of cap and co-payment (cost-sharing) policies on medicines use, healthcare utilisation, health outcomes and costs?
StudiesSearch date up to: September 2007
 Number of studies related to medicines use:  21
 Study design:  RCT, ITS (simple and repeated measure designs), CBA
ParticipantsPatients: families; employees in large companies; community mental health service users; people with schizophrenia; elderly people (high, low and mixed income groups); nursing home residents; low income populations (including those receiving social security, families with dependent children). Medicines involved included antihypertensives, anticoagulants, antithrombotics, nitrates, corticosteroids, anticonvulsants, neuroleptics, antibiotics, diabetic agents, thyroid agents, beta-blockers, antiparkinsonian drugs, antipsychotics, mood stabilizers and antidepressants. Carers: not specified. Professionals: not specified.
SettingPrimary care, hospital, long term care, community, home, private organisation, not specified
InterventionsCap (limits on: number of prescriptions reimbursed, number of repeat prescriptions, or number of days before prescriptions can be re-supplied); fixed co-payments (fixed co-payment per branded or generic medicine, income based partial co-payments up to limit, co-payments in different schedules, phased co-payment increases); ceiling (based on proportion of income), including fixed co-payments with ceiling; co-insurance with ceiling (where co-payment was based on income, or ceiling was income based); fixed co-payments and co-insurance with ceiling; tier co-payments (based on different numbers of tiers according to medicine types); no restrictions; full medicine coverage; no medicine coverage; alternate medicine cap and co-payment policies (different schedules, tiers, ceilings)
 Maps to intervention taxonomy categories: Improving quality
OutcomesHealth status and wellbeing, system benefits, health behaviour
Quality of the review (AMSTAR)10
Quality of included studiesIndividual comparisons were typically based on small numbers of studies. Only 1 included study was randomised, while the majority (2/3rds) of included studies had some methodological limitations that may introduce bias, with 3 studies having serious limitations in design and implementation.
Beney 2000
Review question/objective: Does expanding the role of outpatient pharmacists affect patient outcomes, health services use and costs?
StudiesSearch date up to: March 1999
 Number of studies related to medicines use: 16
 Study design: RCT, CCT, ITS, CBA
ParticipantsPatients: any person receiving medicines and including those with diabetes, hypertension, high cholesterol, or those at-risk for medicines-related problems. Carers: none. Professionals: physicians, not specified.
SettingOutpatient, pharmacy, academic institution, primary care
InterventionsPharmacist services targeted at patients; services delivered by other health professionals (physicians); usual care 
 Maps to intervention taxonomy categories: Providing information or education, Supporting behaviour change, Improving quality 
OutcomesKnowledge and understanding, consumer evaluation of care, health status and wellbeing, health behaviour 
Quality of the review (AMSTAR)9
Quality of included studiesConclusions are limited by the small number of included studies, most of which lacked the power to differentiate between the effects of interventions. All included studies had methodological limitations leading to some risk of bias. Of the included randomised studies: 4 adequately concealed allocation; 2 were quasi-randomised; 4 had unit-of-analysis errors. The majority of trials had good follow up, blinded outcome assessment, and reliable measurement of at least some key outcomes. For included CBA studies: 3 of 6 had control site similar to intervention site; baseline measurement and blinded outcome assessment occurred in 3 studies; and protection against contamination in was performed in 2 studies.
Bhogal 2006
Review question/objective: Do written action plans improve the management of asthma in children and adolescents?
StudiesSearch date up to: November 2004
 Number of studies related to medicines use:  4
 Study design: RCT, CCT
ParticipantsPatients: school-aged children and adolescents with mild to severe asthma. Carers: parents of children or adolescents with asthma. Professionals: none.
SettingPrimary care, secondary care, home
InterventionsSymptom-based written action plan; peak flow-based written action plan
 Maps to intervention taxonomy categories: Supporting behaviour change, Facilitating communication and/or decision making, Acquiring skills and competencies, Minimising risks or harms
OutcomesSystem benefits, health status and wellbeing, support and consumer skills acquisition, health behaviour
Quality of the review (AMSTAR)11
Quality of included studiesOverall of included trials only 1 was of good quality, 2 were assessed as of fair quality, and 1 poor quality, and these limitations may introduce bias. Of included trials, 3 were truly randomised, with allocation concealment inadequate in 1 trial and unclear in 2 trials. All but 1 trial assessed baseline comparability and adequately followed up participants. None used intention-to-treat analysis.
Bower 2006
Review question/objective: Do collaborative care interventions improve the symptoms of depression and use of antidepressants in patients in primary care settings?
StudiesSearch date up to: November 2005
 Number of studies related to medicines use:  32
 Study design: RCT
ParticipantsPatients: adults with depressive symptoms or depression managed in primary care. Carers: none. Professionals: none. 
SettingPrimary care 
InterventionsCollaborative care; usual care 
 Maps to intervention taxonomy categories: Improving quality 
OutcomesHealth status and wellbeing, health behaviour
Quality of the review (AMSTAR)4
Quality of included studiesAllocation concealment was unclear in the majority of studies and other aspects of methodological quality were not assessed; therefore the risk of bias is unclear. All results of meta-regression analysis should be interpreted with caution as they rely on observational comparisons between groups.
Gilbody 2006
This review is a duplicate of Bower 2006.
Table 5. Characteristics of included reviews: Part B
  1. RCT: Randomised controlled trial

    CCT: Controlled clinical trial

    CBA: Controlled before-and-after study

    BA: Before-and-after study (uncontrolled)

    ITS: Interrupted time series

Giuffrida 1997
Review question/objective: Do financial incentives improve adherence to healthcare interventions or treatments?
StudiesSearch date up to: April 1997
 Number of studies related to medicines use:  4
 Study design:  RCT
ParticipantsPatients: with hypertension, tuberculosis, cocaine dependence or overweight; pregnant teenagers, or teenage mothers. Carers: parents considering dental care or immunisation for children; parents for paediatric outpatient clinic attendance. Professionals: none.
SettingCommunity, primary care, outpatient, not specified
InterventionsFinancial incentives; other interventions; usual care/ no intervention
 Maps to intervention taxonomy category: Improving quality
OutcomesHealth behaviour
Quality of the review (AMSTAR)6
Quality of included studiesMost studies in the review were small, none performed a sample size calculation to justify choice of numbers in sample, and none indicated that allocation was adequately concealed.
Halpern 2006
Review question/objective: Do enhanced counselling techniques or other client-provider interventions increase adherence to and continuation of hormonal contraceptives?
StudiesSearch date: unknown
 Number of studies related to medicines use:  6
 Study design:  RCT
ParticipantsPatients: women of reproductive age (no contraindications for hormone use); women who wanted or were willing to use hormonal contraception, who requested an abortion or had an abortion. Carers: none.  Professionals: none.
SettingPrimary care, hospital, outpatient
InterventionsGroup motivational counselling; structured counselling; multi-component intervention; peer or nurse individual counselling; reminders or appointment cards; routine counselling
 Maps to intervention taxonomy categories: Providing information or education, Facilitating communication and/or decision making, Supporting behaviour change, Support
OutcomesHealth behaviour, health status and well being
Quality of the review (AMSTAR)10
Quality of included studiesMost study populations in the review were small, none performed a sample size calculation to justify choice of sample size, and no study indicated that allocation to groups was concealed adequately.
Haynes 2008
Review question/objective: What are the effects of interventions to help patients follow prescriptions for medical problems?
StudiesSearch date up to: February 2007
 Number of studies related to medicines use:  78
 Study design:  RCT
ParticipantsPatients: all ages, acute infections and long-term conditions (including heart disease and related conditions, HIV, mental health, asthma/ chronic obstructive pulmonary disease (COPD), arthritis, epilepsy, diabetes, tuberculosis, contraception). Carers: parents, carers or legal guardians of children were included; as were carers of elderly people. Professionals: none.
SettingCommunity, outpatient, primary care, hospital, home
InterventionsInstruction; counselling; automated telephone monitoring and counselling; manual telephone follow-up; family intervention; increasing the convenience of care; simplified dosing; self-monitoring; reminders; special 'reminder' pill packaging; dose-dispensing units and medicines charts; appointment and prescription refill reminders; reinforcement/rewards; medicines formulations; crisis intervention; direct observation of treatment; lay health mentoring; comprehensive pharmaceutical care services; psychological therapy
 Maps to intervention taxonomy categories: Providing information or education, Facilitating communication and/or decision making, Acquiring skills and competencies, Supporting behaviour change, Support, Minimising risks or harms, Improving quality 
OutcomesHealth behaviour, health status and well being
Quality of the review (AMSTAR)10
Quality of included studiesMost included study populations were small and there is a high possibility that no difference in adherence was found by studies when in truth there was one. Only a minority of included studies adequately concealed allocation; however studies with high drop out (> 20%) or those with confounded comparisons were excluded by the review. Only published studies were included, this may overestimate intervention effects. Interventions for long-term treatments were complex and labour-intensive, and feasibility of implementation in ‘real world’ settings is unclear. Elements of the interventions were also not described well in many studies, and effectiveness of the individual components is also not clear.
Heneghan 2006a
Review question/objective: What are the effects of reminder packaging aids to enhance patient adherence to self-administered medicines taken for one month or more?
StudiesSearch date up to: September 2004
 Number of studies related to medicines use:  8
 Study design:  RCT
ParticipantsPatients: with hypertension, type II diabetes, chronic mental illness, elderly with variety of illnesses, healthy adults. Medicines for at least one month; at least 80% follow up; direct observation of therapy by health professional excluded. Carers: administration by carer included. Professionals: none.
SettingCommunity, academic institution, outpatient
InterventionsReminder packaging included the use of a pillbox; blister packaging; reminder pill pack; sequentially numbered 30-days supply inventory tray; medicines provided with pill organiser; and calendar blister pack; usual care
 Maps to intervention taxonomy categories: Supporting behaviour change
OutcomesHealth behaviour, health status and well being, consumer evaluation of care, system benefits
Quality of the review (AMSTAR)10
Quality of included studiesOf the 8 included randomised trials, only 1 was rated as of low risk of bias. The remaining 7 trials were rated as having a high risk of bias: 2/8 adequately randomised and concealed allocation; 2/8 adequately blinded outcome assessors; and half (4/8) conducted analysis based on intention-to-treat principles.
Heneghan 2006b
Review question/objective: Does self-monitoring and self-management of oral anticoagulation improve the quality of anticoagulation and patient outcomes?
StudiesSearch date up to: April 2005
 Number of studies related to medicines use:  14
 Study design:  RCT
ParticipantsPatients: adults or children requiring anticoagulant therapy for any indication (such as valve replacement, atrial fibrillation, venous thromboembolism). Carers: none. Professionals: none.
SettingPrimary care, home, outpatient, hospital
InterventionsSelf-monitoring in which patient self-tests and then self-adjusts treatment based on a predetermined dose schedule (self-adjusted); or the patient self-tests and then calls a clinic to receive the appropriate dose adjustment (non-adjusted by patient); standard monitoring
 Maps to intervention taxonomy categories: Acquiring skills and competencies, Supporting behaviour change, Minimising risks or harms
OutcomesHealth behaviour, health status and wellbeing, consumer adverse events
Quality of the review (AMSTAR)8
Quality of included studiesA minority (4/14) of included trials were of poor methodological quality; however removing these trials from meta-analysis did not substantially alter results (effect estimates). Some included trials did not blind outcome assessment, and intention-to-treat analysis was not used in all trials.
Jacobson 2005
Review question/objective: Do patient reminder and recall systems improve immunisation rates?
StudiesSearch date up to: May 2007
 Number of studies related to medicines use:  47
 Study design:  RCT, CBA
ParticipantsPatients: children and adolescents (birth to 18 years); adults 65 years and older or those with chronic illnesses; adults. Carers: family members. Professionals: healthcare providers/ physicians/ community residents who deliver immunisations.
SettingPrimary care, community, academic institution, private organisation
InterventionsPatient reminder and recall systems (letters, postcards, person-to-person phone calls, autodialer computer phone messages, reminders with outreach or with provider reminder, and reminders in combination); usual care
 Maps to intervention taxonomy categories: Supporting behaviour change, Minimising risks or harms
OutcomesHealth behaviour, system benefits
Quality of the review (AMSTAR)10
Quality of included studiesSeveral included studies had methodological limitations that may introduce bias. Allocation concealment was unclear in over half of included trials; follow-up was unclear in almost half of studies (21/47); blinding of outcome assessment was done in half of studies; while protection against contamination was implemented in only a minority (6/47) of included trials. Only papers published in English were included, but publication bias was assessed, and did not appear likely.
Koshman 2008
Review question/objective: Does pharmacist care improve outcomes for people with heart failure?
StudiesSearch date up to: August 2007
 Number of studies related to medicines use: 12
 Study design: RCT
ParticipantsPatients: adults (majority over 65) with heart failure. Carers: none. Professionals: general practitioners, community pharmacists.
SettingOutpatient, community, home, hospital, pharmacy
InterventionsPharmacist directed care (including medicines assessment and recommendations, self-monitoring education, General Practitioner (GP) liaison, written information, adherence assessment, medicines review and organizers, adherence aids); pharmacist collaborative care (including medicines assessment, education and recommendations, self-monitoring education, referrals to community pharmacist, telephone follow-up, GP liaison, written and audio information); usual care; no education; no intervention; general information
 Maps to intervention taxonomy categories: Providing information or education, Supporting behaviour change, Minimising risks or harms, Improving quality
OutcomesHealth behaviour, health status and wellbeing, consumer adverse events, system benefits
Quality of the review (AMSTAR)6
Quality of included studiesIncluded studies were of variable quality, and the majority of studies did not adequately conceal allocation or blind different aspects of the study, which may introduce bias. Authors note that analysis based on study quality showed that lower quality studies were more likely to overestimate interventions' effects.
Table 6. Characteristics of included reviews: Part C
  1. RCT: Randomised controlled trial

    CCT: Controlled clinical trial

    CBA: Controlled before-and-after study

    BA: Before-and-after study (uncontrolled)

    ITS: Interrupted time series

Lewin 2005
Review question/objective: Are lay health workers effective in improving the delivery of health care and health care outcomes?
StudiesSearch date up to: August 2001
 Number of studies related to medicines use:  3
 Study design:  RCT
ParticipantsPatients: adults and children. Carers: families and mothers of children. Professionals: none.
SettingHome, primary care, community
InterventionsLay health workers (LHWs): people not formally certified as healthcare professionals; no intervention/ usual care
 Maps to intervention taxonomy categories: Providing information or education, Supporting behaviour change, Minimising risks or harms, Improving quality
OutcomesHealth behaviour, health status and wellbeing, system benefits
Quality of the review (AMSTAR)9
Quality of included studiesOverall, 15 included studies were assessed as high quality (low risk of bias); however, the methodological quality of the remainder (29/43) was low, which may introduce bias. Allocation concealment was rated as adequate in 32 studies; loss to follow up was acceptable in 21 studies, unclear in 12 and not done in 10; and analysis was performed according to intention-to-treat principles in 27 studies, was unclear in 12 and not done in 4 studies.
Lummis 2006
Review question/objective: Are there benefits, risks and other impacts when patients' own medicines (POMs) are used in hospital?
StudiesSearch date: From 1984 up to 2004
 Number of studies related to medicines use:  5
 Study design:  CCT, BA
ParticipantsPatients: patients on hospital wards (acute medical, general medical and surgical, endocrine and diabetes medicine, vascular surgery and renal medicine). Carers: none. Professionals: ward pharmacists, discharge pharmacists, dispensary staff, nurses.
SettingHospital
InterventionsUsing patients' own medicines (POM) that have been prescribed and dispensed in the community and brought to hospital; Pharmacists assessing POMs use; POM use; control
 Maps to intervention taxonomy categories: Support, Minimising risks or harms, Improving quality
OutcomesHealth status and wellbeing, consumer adverse events, system benefits
Quality of the review (AMSTAR)5
Quality of included studiesResults should be interpreted with caution due to small numbers of studies assessing relevant outcomes and comparisons. There were also serious limitations of study design that introduce the risk of bias: none were RCTs; only 1 included study was quasi-randomised and the remainder were observational studies which are prone to bias.
Maglione 2002
Review question/objective: Do mass mailings increase the uptake of influenza immunisation among people receiving Medicare?
StudiesSearch date up to: Early 1999
 Number of studies related to medicines use:  5
 Study design:  RCT, CCT
ParticipantsPatients: adult Medicare beneficiaries eligible for influenza vaccination. Carers: unclear. Professionals: none.
SettingNot specified
InterventionsMass mailings (personalised or form letters, postcards and/or brochures); control
 Maps to intervention taxonomy categories: Providing information or education, Supporting behaviour change, Minimising risks or harms
OutcomesHealth behaviour
Quality of the review (AMSTAR)5
Quality of included studiesThe quality and number of studies in the review were limited. No further details were provided and so risk of bias is unclear.
Maio 2005
Review question/objective: What is the impact of pharmacy utilisation management measures (PUM) on the care of seniors?
StudiesSearch date up to: May 2003
 Number of studies related to medicines use: 18
 Study design:  RCT, other
ParticipantsPatients: > 60 years (or mean > 60). Carers: none. Professionals: none. 
SettingCommunity, pharmacy, outpatient
InterventionsDrug benefit cap; copayment, coinsurance, deductibles; prior authorisation; closed formulary; therapeutic substitution; generic substitution; incented formulary
 Maps to intervention taxonomy categories: Improving quality 
OutcomesHealth behaviour, health status and wellbeing, consumer adverse events, system benefits
Quality of the review (AMSTAR)6
Quality of included studiesOverall, the number of included studies was small. Trial methodological quality was generally inadequately reported, and where reported trials lacked rigorous study design. It is therefore difficult to assess the impacts of interventions conclusively or to draw valid conclusions.
McIntosh 2006
Review question/objective: Does compliance therapy improve adherence to antipsychotic medication, symptoms or quality of life for people with schizophrenia?
StudiesSearch date up to: June 2005
 Number of studies related to medicines use:  1
 Study design:  RCT
ParticipantsPatients: English speaking adults with a diagnosis of schizophrenia. Carers: none. Professionals: none.
SettingPrimary care, hospital
InterventionsCompliance therapy using aspects of motivational interviewing, cognitive therapy, cognitive behavioural techniques and psychoeducation to explore with the patient their medical history and the benefits and limitations of antipsychotic treatment; non-specific counselling
 Maps to intervention taxonomy categories: Facilitating communication and/or decision making, Supporting behaviour change, Support
OutcomesHealth behaviour, health status and well being, consumer adverse events, system benefits
Quality of the review (AMSTAR)10
Quality of included studiesResults are based on a single small study. This study was at moderate risk of bias: it was rated as poor on randomisation and allocation concealment; blinding of outcome assessment was unclear; reasons for dropouts were not given, although all participants were accounted for; and it was unclear whether analysis was based on intention-to-treat principles for all reported outcomes.
Morrison 2001
Review question/objective: Do services provided by pharmacists improve patient outcomes in ambulatory care settings?
StudiesSearch date up to: May 1999
 Number of studies related to medicines use:  32
 Study design:  RCT, CCT
ParticipantsPatients: patients requiring pharmacist services. Carers: none. Professionals: physicians of patients requiring pharmacist services. 
SettingOutpatient, primary care, hospital, home, pharmacy, community 
InterventionsPharmacist counselling of patients; pharmacist counselling of physicians; pharmacist counselling of patients and physicians; pharmacist provided patient care; usual care
 Maps to intervention taxonomy categories: Providing information or education, Acquiring skills and competencies, Supporting behaviour change
OutcomesHealth behaviour, knowledge and understanding, health status and wellbeing, consumer adverse events
Quality of the review (AMSTAR)4
Quality of included studiesConclusions are limited by the small number of studies reporting several outcomes. Methodological quality of included studies overall was fair, however many had methodological limitations that may introduce bias: the majority (26/32 trials) were randomised; but observers were blinded in the minority of trials (8/32) and subjects were blinded in only 2/32 trials.
Nicolson 2009
Review question/objective: Does providing written information about individual prescription or over-the-counter medicines improve patient outcomes?
StudiesSearch date up to: June 2007
 Number of studies related to medicines use:  25
 Study design:  RCT
ParticipantsPatients: individuals of any age currently taking medicines (prescribed or over the counter medicines). Carers: none. Providers: none. 
SettingHospital, outpatient, community, long term care, primary care 
InterventionsWritten medicines information; written medicines information in different formats; no written medicines information
 Maps to intervention taxonomy categories: Providing information or education, Supporting behaviour change 
OutcomesKnowledge and understanding, consumer evaluation of care, health behaviour, consumer involvement in care process
Quality of the review (AMSTAR)9
Quality of included studiesFor many comparisons, there were only single small studies contributing to results. Included trials were of generally poor quality which may introduce bias: 10 trials reported adequate randomisation, but 15 trials failed to report this or rated it as unclear; 8 trials reported allocation concealment but this was rated as adequate in only 5 and unclear in the remaining trials; 10 trials adequately blinded outcome assessors, and in 2 this was inadequate. Loss to follow up was variable, ranging from 0 to 68% (mean loss to follow-up in the 22 trials reporting it was 16%). Withdrawals in the 11 trials reporting it was also variable, ranging from 0 to 37% (mean withdrawal was 12%). 
Olthoff 2005
Review question/objective: What are the effects of interventions to help patients adhere to medicines for glaucoma?
StudiesSearch date up to: February 2004
 Number of studies related to medicines use:  4
 Study design:  RCT, ITS, CBA
ParticipantsPatients: people with raised intraocular pressure or glaucoma. Carers: none. Professionals: none.
SettingNot specified
InterventionsCompliance aid (medicines alarm or memory aid); counselling and memory aid; education and tailoring of medicines routine; counselling only; no intervention
 Maps to intervention taxonomy categories: Providing information or education, Supporting behaviour change
OutcomesHealth behaviour
Quality of the review (AMSTAR)7
Quality of included studiesOf the 4 included intervention studies, only 1 study was rated as good quality (with 2 rated as moderate and 1 poor), and this may introduce bias.
Table 7. Characteristics of included reviews: Part D
  1. RCT: Randomised controlled trial

    CCT: Controlled clinical trial

    CBA: Controlled before-and-after study

    BA: Before-and-after study (uncontrolled)

    ITS: Interrupted time series

Orton 2005
Review question/objective: What are the effects of unit-dose packaged treatment on cure and treatment adherence for people with uncomplicated malaria?
StudiesSearch date up to: November 2004
 Number of studies related to medicines use:  4
 Study design:  RCT, CCT
ParticipantsPatients: people with uncomplicated malaria. Carers: parents. Professionals: none.
SettingPrimary care, community, home
InterventionsUnit-dose packaged medicines: labelled and boxed blister packs or labelled and sectioned polythene bags; usual care
 Maps to intervention taxonomy categories: Supporting behaviour change
OutcomesHealth behaviour, health status and well being, consumer adverse events
Quality of the review (AMSTAR)10
Quality of included studiesAll of the included studies were relatively small and had serious methodological limitations that might introduce bias. Only 1 cluster RCT adequately generated the randomisation sequence; while adequacy of allocation concealment was unclear in all included studies. Similarly, blinding of outcome assessment was not done all trials; completeness of outcome data was assessed in only 2 trials (1 assessed as adequate, 1 inadequate); and there were unit of analysis issues in cluster RCTs.
Roughead 2005
Review question/objective: Do pharmaceutical care service interventions improve patient outcomes?
StudiesSearch date up to: December 2003
 Number of studies related to medicines use:  22
 Study design:  RCT
ParticipantsPatients: adults and children with chronic conditions or at high risk of medicines misadventure (eg polypharmacy). Carers: none. Professionals: none.
SettingOutpatient, primary care, pharmacy, community
InterventionsPharmaceutical care services involving one-to-one consultation between patient and pharmacist, to manage health or resolve medicines-related problems, to develop a care plan and provide follow-up; usual care
 Maps to intervention taxonomy categories: Facilitating communication and/or decision making, Acquiring skills and competencies, Minimising risks or harms, Improving quality
OutcomesHealth behaviour, knowledge and understanding, health status and wellbeing, consumer adverse events, system benefits
Quality of the review (AMSTAR)7
Quality of included studiesThis review included only published, English-language randomised trials, and almost half (10/22) were rated as having a high risk of bias. Methodological limitations included inadequate randomisation in some included trials, allocation concealment adequacy was often unclear, as were blinding of outcome assessors and contamination between study sites. Additionally, some included studies had sample sizes that were too small to detect effects of interventions.
Royal 2006
Review question/objective: Do interventions aiming to reduce preventable medicines-related adverse events decrease morbidity, hospital admission and mortality?
StudiesSearch date up to: February 2005
 Number of studies related to medicines use: 38
 Study design: RCT, CCT, CBA, ITS
ParticipantsPatients: People taking medicines. Carers: none. Professionals: healthcare professionals and pharmacists providing care in community-based family medical services. 
SettingPrimary care, community, long term care, pharmacy
InterventionsPharmacist-led medicines review; primary healthcare professional-led interventions (nurse protocols or primary care physician education); complex interventions including medicines review to reduce falls; control 
 Maps to intervention taxonomy categories: Minimising risks or harms, Improving quality 
OutcomesConsumer adverse events, health status and wellbeing, system benefits
Quality of the review (AMSTAR)8
Quality of included studiesNone of the included studies were designed to explicitly assess patient outcomes that could be linked causally to medicines adverse events, and these studies set in primary care may not be applicable to other healthcare settings. All of the included studies had methodological limitations that are likely to introduce bias: many are subject to attrition bias, allocation concealment and blinding of assessors was unclear or not done in the majority of studies and analysis did not adjust for clusters of sites.
Rueda 2006
Review question/objective: What are the effects of interventions to support and educate people living with HIV/AIDS on adherence to highly active antiretroviral therapy (HAART)?
StudiesSearch date up to: May 2005
 Number of studies related to medicines use:  19
 Study design:  RCT
ParticipantsPatients: adults and children with HIV and receiving HAART. Carers: none. Professionals: none.
SettingOutpatient, hospital, community
InterventionsSupport and education interventions; individual or group interventions; medical management strategies; cognitive behavioural therapy; motivational interviewing; usual care
 Maps to intervention taxonomy categories: Providing information or education, Acquiring skills and competencies, Supporting behaviour change, Support
OutcomesHealth behaviour, health status and well being
Quality of the review (AMSTAR)9
Quality of included studiesOverall, the quality of studies was low, with potential for bias. Randomisation was described and adequate in only 5 trials, with allocation adequately concealed in 3. Intention-to-treat analysis was conducted in 3 included trials, while follow-up post intervention and up to 6 months was variable (3 studies up to 6 months). Only 6 studies used an objective measure of adherence.
Russell 2006
Review question/objective: Do interventions directed at older adults improve medicines adherence?
StudiesSearch date up to: 2004
 Number of studies related to medicines use:  57
 Study design:  RCT
ParticipantsPatients: older adults (mean age over 60 years) with hypertension or other cardiac, diabetes mellitus, osteoarthritis, cancer, glaucoma, receiving blood thinners, or with multiple (> 2) or other diagnoses. Carers: none. Professionals: none.
SettingHome, community, pharmacy, hospital, primary care
InterventionsCounselling and education (brief (1 to 3 days), extensive (> 3 days), or unknown duration); cues, organisers or both; simplification of dose frequency; self-medication management programs; control
 Maps to intervention taxonomy categories: Providing information or education, Supporting behaviour change, Acquiring skills and competencies, Support
OutcomesHealth behaviour
Quality of the review (AMSTAR)4
Quality of included studiesMany studies were small, with insufficient power to detected an effect of interventions in approximately 1/3rd of studies. Study quality was not formally assessed, and risk of bias is therefore unknown.
Schedlbauer 2004
Review question/objective: What is the effect of adherence-enhancing interventions to help people take prescribed self-administered lipid lowering medicines?
StudiesSearch date up to: February 2003
 Number of studies related to medicines use:  8
 Study design:  RCT
ParticipantsPatients: pre-existing cardiovascular pathology or increased cardiovascular risk; healthy with high cholesterol; taking medicines for primary and secondary prevention. Carers: none. Professionals: none.
SettingPrimary care, pharmacy, outpatient
InterventionsSimplification of medicines regime (decreasing intake from four times daily to twice daily or powder form to bar form); patient information and education (pharmacist-mediated counselling and information, handing out videotapes, booklets and newsletters, followed by educational newsletters sent via post or sending out informational/educational videotapes); intensified patient care (reminders via mail and telephone); complex behavioural approach - group sessions (small group training with information packages sent by post); decision support systems (none found); administrative improvements (none found); usual care or other intervention
 Maps to intervention taxonomy categories: Providing information or education, Supporting behaviour change
OutcomesHealth behaviour, consumer evaluation of care, health status and well being, consumer adverse events
Quality of the review (AMSTAR)10
Quality of included studiesOverall, risk of bias in included studies was rated as high. None of the included trials met all methodological quality criteria; only 2 trials were rated as having a moderate risk of bias, with the remaining trials rated as moderate to high risk of bias. Blinding of recipients was not possible due to the nature of the intervention; and while blinding of provider was theoretically possible, this was only attempted in 2 included trials.
Schroeder 2004
Review question/objective: What is the effect of adherence-enhancing interventions to help people take prescribed antihypertensive medicines?
StudiesSearch date up to: April 2002
 Number of studies related to medicines use:  38
 Study design:  RCT
ParticipantsPatients: community dwelling adults with primary hypertension, newly diagnosed or established; excluded: secondary hypertension; hospitalised (non-ambulatory) patients. Carers: none. Professionals: none.
SettingPrimary care, community, outpatient
InterventionsSimplification of medicines regimens (once daily versus twice daily; tablet to transdermal delivery; 2 tablets versus 1 tablet); patient education (programmes with slides, audiotapes, booklets, group education, written materials, visual aids, lecture, discussion and knowledge tests); complex health and organisational interventions including interventions in combination and structured hypertension management; patient motivation, support and reminders (dispensers, medicines reminder charts with pharmacist supervision, self-recording of blood pressure, home visits, nurse and psychologist teaching self-determination, counselling, nurse phone calls, social support, group training, postal reminders, reminder packaging, telephone-linked computer counselling); usual care or no treatment
 Maps to intervention taxonomy categories: Providing information or education, Supporting behaviour change, Support; Improving quality
OutcomesHealth behaviour, health status and well being
Quality of the review (AMSTAR)10
Quality of included studiesResults may be limited as study quality was generally low. No included study met all methodological quality criteria. Randomisation method and adequate allocation concealment occurred in only 10/38 studies; outcome assessors were blinded in 12/38 studies; losses to follow-up were accounted for in 33/38 studies. Only a minority (8/38 studies) reported a power calculation and the majority of the remaining trials appear too small to detect clinically important differences between groups.
Spurling 2007
Review question/objective: What are the effects of delaying antibiotic prescriptions for at least 48 hours after respiratory infection symptoms begin on antibiotic use, clinical outcomes and patient satisfaction?
StudiesSearch date up to: January 2007
 Number of studies related to medicines use:  9
 Study design:  RCT
ParticipantsPatients: adults or children with respiratory infections. Carers: parents. Professionals: none.
SettingPrimary care, outpatient, home
InterventionsDelayed antibiotics; immediate antibiotics; no antibiotics
 Maps to intervention taxonomy categories: Facilitating communication and/or decision making, Minimising risks or harms
OutcomesHealth behaviour, consumer evaluation of care, health status and wellbeing, consumer adverse events
Quality of the review (AMSTAR)9
Quality of included studiesThere were methodological limitations with some included studies that may introduce bias. Overall, 8 studies were rated as high quality. All 9 included trials were properly randomised, with 5 adequately concealing allocation. Six trials had attempted blinding some aspect of the study; and analysis was on an intention-to-treat basis in 5 trials.
Table 8. Characteristics of included reviews: Part E
  1. RCT: Randomised controlled trial

    CCT: Controlled clinical trial

    CBA: Controlled before-and-after study

    BA: Before-and-after study (uncontrolled)

    ITS: Interrupted time series

Stevenson 2004
Review question/objective: What are the effects of interventions to improve two-way communication between patients and healthcare professionals about medicines?
StudiesSearch date: From 1991 up to July 2001
 Number of studies related to medicines use: 16
 Study design:  RCT, CBA, BA
ParticipantsPatients: any patient requiring medicines. Carers: none. Professionals: pharmacists and pharmacy staff, GPs, nurses, outpatient clinic doctors and staff, staff at psychiatric inpatient units.
SettingPrimary care, outpatient, hospital, pharmacy, community, home
InterventionsTraining seminars for doctors; patient communication skills training; medicine fact sheet plus counselling; modified pharmacy services and medicines review; advertising campaign to promote communication with pharmacists; written questions for pharmacist plus counselling; nurse/ assistant telephone follow-up; nurse/ assistant face-to-face consultation; usual care; medicines education; medicines fact sheet; no control
 Maps to intervention taxonomy categories: Providing information or education, Facilitating communication and/or decision making, Improving quality, Support, Minimising risks or harms
OutcomesHealth behaviour, knowledge and understanding, consumer evaluation of care, health status and wellbeing, consumer adverse events, consumer involvement in care process, communication and consultation by provider, system benefits
Quality of the review (AMSTAR)5
Quality of included studiesMost included studies were only of moderate methodological quality that may predispose results to bias. Of the included intervention studies, 10 were RCTs, however, many included studies had methodological limitations (such as lack of randomisation, lack of numbers recruited, pre- and post-intervention data not given; attrition from study), and these may introduce bias.
Stone 2002
Review question/objective: Which interventions improve adherence to preventive cancer screening and adult immunisation guidelines?
StudiesSearch date up to: February 1999
 Number of studies related to medicines use: 29
 Study design:  RCT, CCT
ParticipantsPatients: adults eligible for immunisation or cancer screening. Carers: none. Professionals: any involved in the delivery of preventive care services.
SettingNot specified
InterventionsOrganisational change; provider reminder; patient financial incentives; provider education; patient reminder; patient education; provider financial incentive; feedback; usual care/control
 Maps to intervention taxonomy categories: Providing information or education, Supporting behaviour change, Improving quality, Minimising risks or harms
OutcomesHealth behaviour
Quality of the review (AMSTAR)9
Quality of included studiesMost included studies were high quality (although not described in any detail). The majority of included studies were RCTs, but no further details were given about assessment of risk of bias. Authors note that several cluster randomised trials suffered from unit of analysis issues which may distort the results.
van Eijken 2003
Review question/objective: What is the effectiveness of interventions, both multifaceted and tailored, that aim to improve medicines adherence in older people living in the community?
StudiesSearch date up to: June 2001
 Number of studies related to medicines use:  14
 Study design:  RCT
ParticipantsPatients: 60 years (median > 70); community-dwelling. Carers: none. Professionals: none.
SettingCommunity, pharmacy, home, primary Care
InterventionsSingle generalised intervention; multifaceted generalised intervention; multifaceted tailored intervention; control
 Maps to intervention taxonomy categories: Supporting behaviour change, Improving quality
OutcomesHealth behaviour
Quality of the review (AMSTAR)6
Quality of included studiesThe methodological quality of the studies was moderate. Although all 14 included studies were RCTs, many had methodological limitations that may introduce bias: only 3 reported power calculation to justify sample size; only 4 described randomisation explicitly; only 1 conducted intention-to-treat analysis; and proportion of patients followed up was unclear in 5 trials. 
van Wijk 2005
Review question/objective: Do interventions delivered by community pharmacists improve patient adherence to chronic medicines?
StudiesSearch date up to: November 2003
 Number of studies related to medicines use:  17
 Study design:  RCT, BA, other
ParticipantsPatients: patients prescribed medicine for a chronic disease (lasting > 3 months). Carers: none. Professionals: community pharmacists.
SettingCommunity, pharmacy
InterventionsEducation; counselling and monitoring (at prescription refill or initial fill, pharmacist incorporation of written patient questions, identification of medicines problems); chart review and identification of drug related problems; usual care
 Maps to intervention taxonomy categories: Providing information or education, Support
OutcomesHealth behaviour
Quality of the review (AMSTAR)5
Quality of included studiesStudies were generally small in size, and only a minority of studies reported conducting a power calculation and most contained methodological limitations that may introduce bias. Overall, several studies were of poor design for assessing effectiveness, and in many baseline adherence was high which may mask intervention effects. Overall quality of included studies was poor: only a minority of included studies blinded outcome assessors or had < 10% loss to follow up; randomisation was not clear in many studies; and several included studies were of non-randomised design and this may introduce bias.
Vergouwen 2003
Review question/objective: What is the effectiveness of interventions to improve adherence to antidepressant medicines in patients with unipolar depression?
StudiesSearch date up to: January 2002
 Number of studies related to medicines use:  19
 Study design:  RCT
ParticipantsPatients: with unipolar depression. Carers: none. Professionals: physicians, nurses, psychiatrists, psychologists.
SettingPrimary care, outpatient
InterventionsEducation (outpatient); education (primary care); multimodal collaborative care (primary care; including counselling, general and emotional support, psychotherapy); dosage regimen; usual care
 Maps to intervention taxonomy categories: Providing information or education, Supporting behaviour change, Support, Improving quality
OutcomesHealth behaviour, health status and wellbeing
Quality of the review (AMSTAR)5
Quality of included studiesThere were methodological limitations to included studies which may introduce bias, and several studies on patient education in particular were of poor methodological quality. Few details of quality assessment were reported, except for numbers completing, which ranged from 38% to 100% in included trials. 
Vermeire 2005
Review question/objective: What are the effects of interventions to improve adherence to treatment recommendations for people with type 2 diabetes mellitus?
StudiesSearch date up to: November 2002
 Number of studies related to medicines use:  21
 Study design:  RCT, CCT, CBA, other
ParticipantsPatients: Type 2 diabetes. Carers: none. Professionals: none.
SettingPrimary care, outpatient
InterventionsNurse led interventions; home aides; diabetes education programmes; pharmacy based interventions; dosing and frequency interventions; other: patient participation programme; oral versus injected medicines; fundus photography; patient participation consultation; counselling; usual care
 Maps to intervention taxonomy categories: Providing information or education, Acquiring skills and competencies, Supporting behaviour change
OutcomesHealth behaviour, health status and well being, knowledge and understanding
Quality of the review (AMSTAR)10
Quality of included studiesOverall, of 21 included studies, 3 were considered at low risk of bias; 13 moderate; and 5 high risk of bias. In 5 randomised trials, randomisation and allocation were both adequate; in 6 trials there was adequate randomisation but not concealment of allocation; and in 4 studies both were unclear due to lack of data. Groups were similar at baseline in 15 trials. In 3 studies blinding of patients, administrators and outcome assessors was adequate; 2 studies had adequate blinding of patients, but not of administrators and outcome assessors; in 1 study there was adequate blinding of patients, but unclear blinding of administrators or outcome assessors; in 11 studies data any blinding was unclear; and 1 study did not apply any form of blinding. In 11 studies, groups were provided with comparable care (1 study not equivalent; missing in 5 studies); analysis was on an intention-to-treat basis in 8 studies, and other losses to follow-up were adequately described in 15 (inadequately in 6 studies).
Volmink 2006
Review question/objective: Does directly observed therapy (DOT) cure or improve treatment completion in people with clinically active tuberculosis or requiring prevention of active disease?
StudiesSearch date up to: May 2007
 Number of studies related to medicines use:  11
 Study design:  RCT, CCT
ParticipantsPatients: low, middle and high-income countries; preventive therapy for tuberculosis or clinically active tuberculosis. Carers: none. Professionals: none.
SettingOutpatient, community, home, primary care
InterventionsDOT; DOT at home or at clinic; DOT by family member, community health worker, nurse, family member, lay health worker; DOT for prophylaxis with IV drug users (own location or treatment centre); self-administration
 Maps to intervention taxonomy categories: Supporting behaviour change, Minimising risks or harms
OutcomesHealth behaviour, system benefits, health status and well being
Quality of the review (AMSTAR)10
Quality of included studiesSeveral of the included studies had methodological limitations that may introduce bias. Generation of the allocation sequence was adequate in 7 trials; inadequate in 1 and unclear in the remainder. Allocation concealment was adequate in 4 trials; unclear in 3; and inadequate in those remaining. Blinding of outcome assessment occurred in only 4 trials; while completeness of follow up was adequate in all but 6 trials (2 trials with > 20% excluded from analysis; 4 trials where follow-up was rated unclear).
Zygmunt 2002
Review question/objective: Do psychosocial interventions improve adherence to antipsychotic medicines in people with schizophrenia?
StudiesSearch date up to: December 2000
 Number of studies related to medicines use:  39
 Study design:  RCT, CCT
ParticipantsPatients: people with schizophrenia requiring antipsychotic medicine. Carers: family members. Professionals: none.
SettingOutpatient, hospital, home, community
InterventionsPyschoeducation (dissemination of knowledge about disease, treatment and medicines); group programmes (peer support and shared identification); family (influence on patient illness); cognitive (attitudes and beliefs towards medicines); behavioural; and, community (support and rehabilitation); standard care; other interventions
 Maps to intervention taxonomy categories: Providing information or education, Facilitating communication and/or decision making, Supporting behaviour change, Support
OutcomesHealth behaviour, health status and wellbeing
Quality of the review (AMSTAR)4
Quality of included studiesLimited outcomes were reported. Effectiveness of components of multifaceted interventions could not be assessed. No included study rigorously assessed adherence; and methodological quality was variable, although no further details were provided so risk of bias is unknown.

Methodological quality of included reviews

Quality of included reviews

Overall, all eighteen included Cochrane reviews were rated as high quality (see Data collection and analysis). Most Cochrane reviews (13/18, 72%) scored 10 or more points out of 11 on AMSTAR, indicating that reviews were of high quality and likely to have minimal bias in their design and conduct. Common reasons for a review to lose points on AMSTAR were: not reporting conflicts of interest for both the review authors and in the review's included studies; and the review not assessing publication bias (ie the possibility that results from the review are biased because of the propensity for positive trials to be published and for negative trials not to be).

Of the nineteen included non-Cochrane reviews, few (3/19 reviews, 16%) were rated as high quality (scoring 8 or 9 on AMSTAR, Heneghan 2006b; Royal 2006; Stone 2002); the remaining reviews were rated as moderate quality (scoring between 4 and 7 points). Of the moderate quality reviews, most (10/16, 63%) received a rating of 5 or lower, suggesting that these reviews may be at risk of bias that might influence the results.

Quality of included studies

As reported by review authors, reviews included studies that ranged from high (well designed and conducted studies) to low quality (studies with serious methodological limitations). A small number of reviews were highly selective about the quality of the studies they included, for example, both Haynes 2008 and Heneghan 2006a specified that only RCTs with at least 80% follow-up and unconfounded comparisons would be eligible for inclusion. However even these measures did not ensure that included studies were of high quality: in Haynes only a minority of studies adequately concealed allocation for example; while in Heneghan all but one study were rated as being at a high risk of bias (due to inadequate allocation concealment, blinding and lack of intention-to-treat analysis).

Almost half of reviews (18/37, 49%) included only RCTs. While a few of these reviews reported that included studies were of generally high quality (eg Heneghan 2006b, Spurling 2007), more often reviews reported methodological limitations that may have introduced some degree of bias (Giuffrida 1997; Halpern 2006; Haynes 2008; Heneghan 2006a; Koshman 2008; Lewin 2005; McIntosh 2006; van Eijken 2003; Vergouwen 2003). Several other reviews of RCTs reported that included studies were of generally low quality and/or had serious methodological limitations (Nicolson 2009; Roughead 2005; Rueda 2006; Schedlbauer 2004; Schroeder 2004); while methodological quality (risk of bias) was unclear in others (Bower 2006; Russell 2006). Thus reviews including RCTs alone did not guarantee that methodological quality of included studies was high, nor that results were unaffected by bias.

A similar picture of the quality of included studies emerged in those reviews including studies other than RCTs, most often quasi-randomised controlled trials (although other study designs were included in several reviews). Included studies had methodological limitations (Jacobson 2005; Maglione 2002; Maio 2005; Morrison 2001; Royal 2006; Van Wijk 2005; Volmink 2007) or serious methodological limitations (Aaserud 2006; Austvoll-Dahlgren 2008; Beney 2000; Lummis 2006; Orton 2005) highlighted by authors. Other reviews included studies of moderate methodological quality overall (Bhogal 2006; Olthoff 2005; Stevenson 2004; Vermeire 2005), while quality was unclear or not reported in others (Amico 2006; Stone 2002; Zygmunt 2002).

Overall, included studies were of variable methodological quality and this may in some cases predispose the results to bias. Where reviews had obvious methodological shortcomings, we attempted to adjust for this by downgrading the effectiveness statements.

Effect of interventions

In the following section we give the bottom-line statements of intervention effectiveness, presented by intervention category.

For transparency we present the effects of the interventions in full in Appendix 3.

Providing information or education

Overall interventions that provide information or education as a single component may be ineffective to improve adherence or clinical outcomes. However, there is some evidence that patient education as a single component or as part of a complex intervention may be effective to improve immunisation rates. There is insufficient evidence to determine whether this intervention alone improves other outcomes such as knowledge or adverse effects. When used in combination with other interventions, such as self-management skills training or counselling, there is some evidence that it may improve adherence and other outcomes such as clinical outcomes, but results are mixed.

Facilitating communication and/or decision making

There is insufficient evidence from one key review to determine whether interventions focussed on promoting communication between patients and professionals are effective. 

Overall, there is some evidence to support the use of interventions which do not have a specific focus on facilitating decision making and/or communication, but effects are mixed. Delayed prescribing is effective to decrease antibiotic use, but also decreases satisfaction and has mixed effects on clinical outcomes and adverse events. In general, there is some evidence of effect for adherence, knowledge or other outcomes with education and enhanced follow-up, and pharmaceutical care services, but insufficient evidence to support the use of psychosocial interventions, which are generally ineffective. There is insufficient evidence to determine the effectiveness of structured counselling or compliance therapy.

Acquiring skills and competencies

There is some evidence that strategies which focus on the acquisition of skills and competencies may improve adherence, clinical outcomes and other outcomes, but results are mixed. Regarding specific types of interventions, there is sufficient evidence that self-monitoring decreases adverse events; it is generally effective. There is insufficient evidence to support the provision of training by pharmacists to improve adherence, but some evidence that it improves knowledge and medicines use. There is some evidence to support the provision of counselling of patients and/or physicians by pharmacists to improve adherence, but insufficient evidence to support more intensive patient care by pharmacists.

Supporting behaviour change

Overall, there were mixed effects in general of interventions to support behaviour change in relation to medicines use. In general, there is some evidence of the effectiveness of simple interventions for short-term treatments, and complex interventions for long-term treatments to improve adherence and clinical outcomes.

More specifically, there is sufficient evidence that self-monitoring or self-management improve medicines adherence and some evidence that reminders and lay health worker interventions improve immunisation rates. There is some evidence that simplified dosing regimens are generally effective to improve medicines adherence. There is also some evidence that reminders, support and education or support and motivation interventions, and those involving pharmacists directly, are effective to improve adherence, although results are mixed. 

Support

Due to the mixed results from studies found in most reviews, there is some evidence that interventions that provide support alone or in combination with other strategies may be effective to improve adherence and other outcomes . There is insufficient evidence to determine for which conditions support may be effective, or who should provide the support for greatest effect.

Minimising risks or harms

There is sufficient evidence that self-monitoring, with or without self-adjustment, is effective at decreasing adverse events of treatment. There is also some evidence that strategies to improve interactions between healthcare professionals and patients may decrease adverse events and improve other outcomes, but results are mixed. In particular, there is insufficient evidence to determine whether the use of patients’ own medicines (POMs) in hospital is effective. There is also some evidence that educational strategies to minimise risks and harms may be effective, and that telling patients about adverse effects of medicines does not negatively influence adherence.

For immunisation uptake, there is sufficient-to-some evidence that organisational change, reminders and recall, financial incentives, education and lay health worker interventions are each generally effective, while effects of mass mailings are mixed, and reminders with outreach are ineffective. There is some evidence that directly observed therapy for tuberculosis is generally ineffective to improve cure or treatment completion. There is sufficient evidence that delayed antibiotic prescription decreases antibiotic use without increasing complications, but it may increase supplementary medicines use and results are mixed for clinical outcomes and adverse effects.

Improving quality

Because this overview did not specifically include reviews which targeted organisational or structural interventions to change consumer medicines use, provisional conclusions about the effectiveness of those interventions are provided here. There is some evidence that changing the coordination of care (eg changing roles of healthcare professionals to interact with patients or to provide additional services to patients) may improve adherence and other outcomes related to medicines use, however the results from most reviews are mixed. In depression, there appears to be some evidence that these interventions are generally effective. There is also some evidence that financial interventions are effective. In addition, the review of POMs included less rigorous study designs which provides insufficient evidence to determine whether supporting patient’s use of their own medicines in hospital is effective. We did, however, search broadly for reviews in immunisation uptake, and found that there is sufficient evidence that organisational interventions are effective at improving uptake. There is also some evidence that pharmaceutical pricing policies to indirectly influence consumers’ use of medicines are effective to improve medicines use and costs, but results are mixed for effects on health status and health service use.

Consumer system participation

There is insufficient evidence to determine the effects of consumer system participation in medicines-related activities because no reviews were identified.

For detailed information on intervention effects refer to Appendix 3.

Results of each included review, reported individually (as quantitative results; a narrative summary of the results; and effectiveness statements) can be found in Additional Table 9; Table 10; Table 11; Table 12; Table 13.

Table 9. Results, by individual review: Part A
  1. * Numbers of individual studies contributing to results for each outcome are reported unless otherwise indicated as numbers of interventions (int) per outcome.

    RR = Relative Risk; OR = Odds Ratio; ARR = Absolute Risk Reduction; ARI = Absolute Risk Increase; MR = Mean Reduction; MI = Mean Increase; AMR = Absolute Mean Reduction; AMI = Absolute Mean Increase; 95% CI = 95% Confidence Interval

Aaserud 2006

Pharmaceutical policies: effects of reference pricing, other pricing, and purchasing policies

[Maps to: Improving quality]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results
Reference pricing policyReference medicine use (immediately following policy introduction)4Relative change = 119% (range 60% to 196%); 2 studies significant increase; 2 studies increase (significance unknown)
Reference medicine use (6 months to 1 year after policy introduction)31 study further increase relative to effects immediately after policy introduction (significance unknown); 2 studies less of an increase relative to effects immediately after policy introduction (significance unknown)
Use of cost share medicines (immediately following policy introduction)4Relative decrease = 38% (range 19% to 42%)
Total use of reference group medicines2Non-significant changes
Total use of medicines other than reference group2Non-significant changes
Patient payment share of total expenditure (immediately following policy introduction)1Increase from 0% to 16%
Medicine pricing22 studies decrease (range 11 to 26%): 1 study significant reduction in both generic and brand medicines; 1 study brand price reduction (significance unknown)
Mortality2Non-significant changes
Emergency visits and hospital admissions through emergency department10 intsRelative increase = 9% (range -41% to 49%); 1 int significant increase; 5 ints non-significant increase
Non-emergency hospital admissions10 intsRelative decrease = 12% (range -42% to 7%); 3 ints non-significant increase
Physician office visits10 intsRelative increase = 1% (range -18% to 31%); 5 ints significant increase
Index pricing policyMedicine use - brand medicines1Relative decrease = 29% immediately after policy introduction; 43% decrease at 6 months after policy introduction
Medicine use - generic medicines1Relative increase = 114% immediately after policy introduction; 55% increase at 6 months after policy introduction
Medicines pricing1Decreases immediately and long-term, with long-term decreases being larger than changes immediately post policy introduction for both brand (1.1% decrease) and generic (5.3% decrease) drugs

Summary of results:

Reference pricing increased reference medicine use (4 studies) and decreased the use of cost share medicines (4 studies) immediately following policy change, and these trends were still apparent at 6 months to 1 year, although diminished in size (3 studies). Reference pricing reduced total medicine expenditures (2 studies) but increased the patients’ share of total medicines expenditure of total (1 study). Reference pricing had no significant effects on mortality; increased emergency visits and hospital admissions through the emergency department in a minority (1 of 10 interventions) of studies; and had mixed effects on non-emergency hospital admissions and physician visits (5 of 10 comparisons significant increase). There were no significant effects of reference pricing on total reference medicines use, and use of medicines other than those in the reference group. Index pricing reduced brand medicines use and increased use of generic medicines (1 study), and decreased costs of both medicines over time, although cost reductions were larger with generic than with brand medicines over time.

Effectiveness statements:

There is some evidence that reference pricing increases use of reference medicines and decreases the use of cost share medicines and total medicines expenditure - it is generally effective. There is some evidence that reference pricing increases healthcare use - results are mixed. There is insufficient evidence to determine the effects of reference pricing on patient expenditure, or on the effects of index pricing.

Amico 2006

Efficacy of antiretroviral therapy adherence interventions: a research synthesis of trials, 1996 to 2004

[Maps to: Providing information or education, Supporting behaviour change, Acquiring skills and competencies, Support, Minimising risks or harms]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results
Any intervention to improve adherence vs controlAdherence26 intSignificant increase, standardised MI = 0.35 (95% CI 0.20 to 0.51). For people with poor adherence at baseline standardised MI = 0.62 (95% CI 0.42 to 0.82); for those with unknown adherence levels at baseline standardised MI = 0.19 (95% CI 0.10 to 0.27)

Summary of results:

Twenty four studies including 26 interventions to improve antiretroviral therapy (ART) adherence were meta-analysed and a small effect size was found. Analysis showed that the intensity of the intervention, ranging from low intensity ad hoc conversations with healthcare professionals, to moderate intensity reminders and support, to high intensity self-management training, was not related to effect size. Duration of the intervention was also not related. A larger effect was seen in those people in whom adherence problems were known or anticipated, when compared with people with unknown pre-existing adherence problems.

Effectiveness statements:

There is some evidence that interventions to improve ART adherence lead to small increases in adherence - they are generally effective.

Austvoll-Dahlgren 2008

Pharmaceutical policies: effects of cap and co-payment on rational drug use

[Maps to: Improving quality]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results
Any cap vs full drug coverageOverall prescription medicines use (general population)2 ints2 ints significant decrease: 1 int decreased by 42.7% (95%CI -50.1% to -35.4%); 1 int decreased by 17%
Overall prescription medicines use (vulnerable population)1 int1 int significant decreased by 46%
"Essential" medicines use1 intSignificant decrease by 28.0%
Discretionary medicines use2 ints2 ints significant decrease: 1 int decrease by 42.7% (95% CI -50.1% to -35.4%); 1 int decrease in "symptomatic relief drugs" by 38.0% and "limited efficacy drugs" by 58.0%
Healthcare use3 ints1 int non-significant change to hospitalisation rates (for complicated or uncomplicated peptic ulcers and non peptic ulcer conditions); 1 int significant 17% increase in psychiatric hospital admissions and significant 43.0% to 57.0% increase per month increase in number of community mental health centre visits (severe schizophrenia population); 1 int significant increase in risk of admissions to nursing homes (elderly population): RR = 1.8 (95% CI 1.2 to 2.6)
One cap (5 reimbursed scripts) vs another (6 reimbursed scripts)Overall prescription medicines use (vulnerable population)1 int1 int significant decrease by 5.9% (95% CI -9.4 to -2.4)
Out of pocket expenditure (vulnerable population)1 int1 int significant increase by 26.5% (95% CI 16.5% to 36.5%)
Fixed co-payment (US$1.50 to $3 per script filled) vs full coverageOverall medicines use general population2 ints2 ints significant decrease, range = 10.6% to 10.7% lower per person
Patient medicine expenditure2 ints2 ints significant decrease, range = 5.2% to 6.7% lower
Fixed co-payment (US$0.50 per script filled) vs full coverageOverall medicines use vulnerable populations2 ints1 int significant decrease by 12% per person; 1 int decreases (significance unclear), range = 5 to 17% lower across population subgroups
Fixed (income based) co-payment vs full coverageOverall medicines use general population1 int1 int decrease by 14.2% (significance unclear)
"Essential" medicines use1 int1 int decrease, range = 10.3% to 15.9% (significance unclear)
Discretionary medicines use1 int1 int decrease, range = 14.3% to 24.3% (significance unclear)
Fixed co-payment (US$3) plus cap vs full coverageOverall medicines use general population1 int1 int significant decrease by 12%
"Essential" medicines use1 intNon-significant change
Medicine expenditure per prescription1 int1 int significant increase by 8.5%
Patient medicine expenditure1 int1 int significant decrease by 8.8%
One fixed co-payment vs anotherOverall medicines use general population3 ints2 ints decreases, range = 21.3% to 22.5% lower per person; 1 int mixed effects
Patient medicines expenditure2 ints2 ints significant increase, range = 32.2 to 39.8% higher
Fixed co-payment with ceiling vs full coverage"Essential" medicines use general populations1 int1 int significant decrease, range = 1.3 to 3.7% lower
"Essential" medicines use vulnerable populations2 ints2 ints significant decrease, range = 2.3 to 23% lower
Discretionary medicines use general population1 int1 int significant decrease by 1.3%
Discretionary medicines use vulnerable population2 ints2 ints significant decrease, range = 1.2 to 24% lower
One fixed co-payment (income-based) with ceiling vs another“Essential” medicines use1 int1 int significant decrease by 22%
Discretionary medicines use1 int1 int significant decrease by 27%
Any co-insurance with ceiling vs full coverageOverall medicines use general population4 ints4 ints significant decreases, range 33.6% to 18.4% lower
"Essential" medicines use vulnerable populations1 int1 int significant decrease by 17.7% (95% CI -14.8 to -20.5)
Discretionary medicines use general population1 int1 int significant decrease by 19.4% (95% CI -17.4 to -21.4)
One co-insurance with ceiling vs another"Essential" medicines use general populations1 int1 int significant decrease by 6.9% (95% CI -5.5 to -8.4)
Discretionary medicines use general population1 int1 int significant decrease by 14.0% (95% CI -13.0 to -15.0)
Fixed co-payment plus coinsurance and ceiling vs fixed co-payment plus coinsuranceOverall medicines use1 int1 int mixed effects: significant decreases were seen women's use of drugs across medicines, while men's use of drugs did not show sustained significant changes
Change in tiered co-paymentOverall medicines use across tiers3 ints2 ints significant decrease, range = 5 to 24% lower; 1 int non-significant changes
Branded medicines use3 ints1 int significant decrease by 34%; 1 int significant decrease, range = 4 to 22% lower; 1 int non-significant changes
Generic medicines use1 intNon-significant decrease
Patient medicines expenditure3 ints1 int significant increase 23% above predicted levels; 1 int significant increase, range = 118% to 148%; 1 int mixed effects
Changes to healthcare use1 intNon-significant increases

Summary of results:

Any cap intervention: Compared with full coverage, overall prescription medicines use in both general (2 ints) and vulnerable populations (1 int) decreased significantly, as did discretionary medicines use (2 ints). Essential medicines use also decreased significantly (1 int), and while effects on health care were mixed there were significant increases in admissions with the majority (2 of 3) of interventions. One cap (5 reimbursed scripts, vulnerable population): Compared to another cap (6 reimbursed scripts), overall prescription medicines use (1 int) and out-of-pocket drug expenditure (1 int) significantly decreased. Fixed co-payments (US$1.50 to $3 general population;US$0.50 vulnerable population ) per script: Compared with full medicines coverage, for fixed co-payments (US$0.50 vulnerable population), overall prescription medicines use (2 ints) decreased significantly. Compared with full medicines coverage, for fixed co-payments (US$1.50 to $3 per script general population), overall prescription medicines use (2 ints) and patient medicines expenditure (2 ints) decreased significantly. Fixed (income based) co-payment interventions: Compared with full coverage, there were decreases (significance unclear) in overall prescription medicines use (1 int) and both discretionary and essential medicines use (1 int). Fixed (US$3) co-payment plus cap interventions: Compared with full coverage, overall prescription medicines use decreased significantly (1 int), as did patient medicines expenditure (1 int), however, medicines expenditure per prescription significantly increased (1 int), while essential medicines use did not change significantly. One fixed co-payment intervention: Compared with another fixed co-payment, overall prescription medicines use decreased significantly (2 of 3 ints), but patient medicines expenditure significantly increased (2 ints). Fixed co-payment with ceiling interventions: Compared with full coverage, discretionary and essential medicines use decreased significantly in both general (1 int) and vulnerable populations (2 ints). Fixed co-payment (income based) interventions: Compared with another fixed co-payment, both discretionary (1 int) and essential medicines use (1 int) significantly decreased. Any co-insurance with ceiling interventions: Compared with full coverage there were significant decreases in overall medicines use in the general population (4 ints) and discretionary medicines use (1 int), but essential medicines use in the vulnerable population (1 int) also significantly decreased. One co-insurance with ceiling intervention: Compared with another co-insurance with ceiling intervention, both discretionary and essential medicines use significantly decreased (1 int). Fixed co-payment plus co-insurance, comparing with and without ceiling had mixed effects on overall medicines use (1 int). Comparative changes in tiered co-payments significantly decreased overall medicines use (2 of 3 ints) and branded medicines use (2 of 3 ints). Generic medicines use non-significantly decreased (1 int), but patient medicines expenditure significantly increased (2 of 3 ints) and effects on health service use increased non-significantly (1 int).

Effectiveness statements:

Overall, cap and copayment policy interventions have mixed effects on medicines use and costs. There is some evidence that caps may decrease overall and discretionary medicines use but may increase healthcare use - the results are mixed. There is insufficient evidence to determine the effects of caps on essential medicines use or patient expenditure. There is some evidence that fixed co-payments, with or without a cap, decrease overall prescription medicines use, but with mixed effects on patient medicines expenditure and cost per prescription - the results are mixed; and there is insufficient evidence to determine effects on essential medicines use. There is some evidence that fixed co-insurance with ceiling interventions decrease overall medicines use in the general population; but there is insufficient evidence to determine effects on essential and discretionary medicines use in general or vulnerable populations - the results are mixed. There is some evidence that changes in tiered co-payments interventions decrease overall and branded medicines use, and increase patient medicines expenditure - the results are mixed. There is insufficient evidence to determine the effects of changes to tiered co-payments on generic medicines use or health service use. There is insufficient evidence to determine the effects of fixed (income-based) interventions or fixed co-payment plus co-insurance, with or without ceiling interventions, on overall, essential or discretionary medicines use.

Beney 2000

Expanding roles of outpatient pharmacists: effects on health services utilisation, costs, and patient outcomes

[Maps to: Providing information or education, Supporting behaviour change, Improving quality]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results
Pharmacist services targeted at patients versus services delivered by other professional (physician)

Systolic blood pressure

 

1

 

Systolic blood pressure better in control group (significance unclear)

 

Pharmacist services targeted at patients versus usual care

 

Quality of life5Non-significant changes
Knowledge32 studies significant increase; 1 study non-significant changes
Satisfaction1Non-significant changes
Attitudes towards therapy, condition or pharmacist1Significant improvement
Adherence (self report)1Non-significant changes

Adherence (pill count)

 

4

 

2 studies significant increase; 2 studies non-significant changes

Adherence (blood levels; prescription refills)

 

1

 

Significant relative increases in number of individuals with blood levels being within +/- 10% and 20% of predicted levels; significant increase in prescription refills
Medicine adverse effects1Non-significant changes

Appropriateness of medicine

 

2

 

1 study 28% relative improvement on the medication appropriateness index; 1 study significant improvement in therapy changes required and identification of wrong dose
Appropriateness of dose1Significant improvement

Cost of medicines/prescriptions

 

3

 

3 studies significant decrease: total prescription costs decreased by $142.16 (20%) in one study; annual mean drug ingredients cost per participant decreased by $0.45 in another; cost data not available in other study
Number of medicines/prescriptions

3

 

2 studies significant decrease in prescriptions; 1 study non-significant changes

Total operating cost

 

1

 

Total operating cost of intervention exceeded the savings in drug costs by $0.85 per patient, but did not take into account savings due to decreased health service use.
Total Medicaid cost1Significant decrease by $824.28

Clinical (blood pressure)

 

3

 

1 study significant improvement; 2 studies non-significant changes
Clinical (blood glucose levels)33 studies significant improvements
Clinical (symptoms)22 studies significant improvements

Summary of results:

In a single study comparing pharmacist services targeting patients with services delivered by physicians, systolic blood pressure was better controlled in the groups receiving physician services. Compared with usual care, pharmacist services targeting patients significantly improved knowledge (2 of 3 studies) and attitudes (1 study), but did not significantly change quality of life or satisfaction. Pharmacist services targeting patients significantly improved the appropriateness (2 of 2 studies) and dose (1 study) of medicines, but not adverse effects (1 study). Effects of pharmacist services to patients were mixed on adherence: adherence measured by blood levels and prescription refills was significantly improved in a single study, but only half (2 of 4 studies) showed significant increases with pill counts, and self-reported adherence showed non-significant changes in a single study. Compared with usual care, pharmacist services targeting patients significantly improved blood glucose levels (3 of 3 studies) and other symptoms (2 of 2 studies) and blood pressure in the minority (1 of 3) studies. Cost of medicines or prescriptions was significantly reduced in three studies, as was the number of medicines or prescriptions in the majority (2 of 3) of studies. The total cost to Medicaid to operate pharmacist services targeting patients was significantly reduced compared to usual care (1 study) however, in another the total operating cost of the intervention exceeded the savings in drug costs, however, this study did not take into account potential savings due to decreased health service usage.

Effectiveness statements:

There is some evidence that compared to usual care, pharmacist services targeting patients significantly improve knowledge and attitudes, medicine appropriateness and dosage, costs and number of medicines - they are generally effective. There is some evidence that compared to usual care, pharmacist services targeting patients improve adherence, clinical outcomes and total costs - the results are mixed. There is insufficient evidence that pharmacist services targeting patients to improve quality of life - they are generally ineffective. There is insufficient evidence to determine the effects of pharmacist services on adverse effects or satisfaction. There is insufficient evidence to determine the effects of pharmacist, compared with physician, service delivery.

Bhogal 2006

Written action plans for asthma in children

[Maps to: Supporting behaviour change, Facilitating communication and/or decision making, Acquiring skills and competencies, Minimising risks or harms]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results
Symptom monitoring action plans vs peak flow action plansNumber of patients with at least one acute care visit4ARR = 11 fewer patients out of 100 (95% CI 18 to 0 fewer) with symptom monitoring plans
Number of patients requiring systemic steroids (per year)3Non-significant decrease with symptom monitoring plans
 Withdrawals4Non-significant change
Change in number of symptomatic days per week2Significant decrease with peak flow written action plan MR =  0.45 (95% CI 0.04 to 0.86)
Number of symptomatic days per week2Non-significant decrease with peak flow plans
Number of parents  intending to use monitoring strategy1Non-significant decrease with symptom monitoring plans;
Number of children intending to use monitoring strategy1ARI = 14 more people out of 100 (95% CI 0 to 30 more) with symptom monitoring plans
Change in parent-reported quality of life at one year3Non-significant increase with symptom monitoring plans
Change in child-reported quality of life at one year2Non-significant increase with symptom monitoring plans

Summary of results:

There were no significant differences between symptom and peak flow monitoring written action plans for number of patients requiring systemic steroids, withdrawals, change in child or parent quality of life, or number of parents intending to use the monitoring strategy. Significantly more children intended to continue using symptom-based written action plans and had significantly lower risk of exacerbations requiring acute care than children who used peak flow-based written action plans. Children using peak flow based action plans had significantly greater change in the number of symptomatic days per week, but not overall number of symptomatic days per week than those using symptom based written action plans.

Effectiveness statements:

There is some evidence that symptom monitoring action plans reduce the number of patients with at lease one acute care visit and increase the number of children intending to use the strategy - they are generally effective. There is insufficient evidence of consistent effects of one action plan versus another on symptoms, use of systemic steroids, quality of life or withdrawals - they are generally ineffective. 

Bower 2006

Collaborative care for depression in primary care

[Maps to: Improving quality]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results

Collaborative care vs usual care

 

 

Adherence

 

28 ints

 

Significant increase, OR = 1.92 (95% CI 1.54 to 2.39)

Depressive symptoms

 

34 ints

 

Significant decrease, OR = 0.24 (95% CI 0.17 to 0.32)

Summary of results:

Collaborative care in primary care settings significantly decreased depressive symptoms and significantly increased antidepressant use, when compared with usual care.

Effectiveness statements:

There is some evidence that collaborative care interventions improve antidepressant use and depressive symptoms in adults with depression in primary care - they are generally effective.

Table 10. Results, by individual review: Part B
  1. * Numbers of individual studies contributing to results for each outcome are reported unless otherwise indicated as numbers of interventions (int) per outcome.

    RR = Relative Risk; OR = Odds Ratio; ARR = Absolute Risk Reduction; ARI = Absolute Risk Increase; MR = Mean Reduction; MI = Mean Increase; AMR = Absolute Mean Reduction; AMI = Absolute Mean Increase; 95% CI = 95% Confidence Interval

Giuffrida 1997

Should we pay the patient? Review of financial incentives to enhance patient compliance

[Maps to: Improving quality]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results
Financial incentives vs usual care/ no interventionAdherence with healthcare treatment7 int5 int non-significant increase; 2 int OR = 2.1 to 4.7
Adherence to medicines use5 int2 int non-significant increases; 3 studies OR = 3.0 to 4.7
Financial incentives vs other interventionAdherence with healthcare treatment5 intNon-significant increases;
Adherence to medicines use8 int6 int non-significant increases; 2 int (vs telephone or prompts) OR = 2.5 to 5.6

Summary of results:

A majority of financial interventions (3 of 5) found significant effects on adherence to medicines use when compared with usual care or no treatment. A minority of financial interventions (2 of 8) found significant effects when compared with other interventions.

Effectiveness statements:

There is some evidence that financial incentives improves adherence to medicines use - the results for financial interventions compared to no intervention were mixed. There is insufficient evidence to support the use of financial incentives instead of other interventions - it is generally ineffective in comparison. 

Halpern 2006

Strategies to improve adherence and acceptability of hormonal methods for contraception

[Maps to: Providing information or education, Facilitating communication and/or decision making, Supporting behaviour change, Support]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results
Group motivation vs routine counsellingDiscontinuation (6 months)1Non-significant increase

Structured counselling vs routine counselling

 

Discontinuation (6 months)1ARR = 21 fewer people out of 100 (95% CI 21 to 9 fewer)
Discontinuation (12 months)1ARR = 32 fewer people out of 100 (95% CI 36 to 24 fewer)

Multicomponent intervention vs routine counselling

 

Continuation (12 months)1Non-significant increase
Switched contraceptives (12 months)1Non-significant increase
Peer vs nurse counsellingNon-compliance (4 months)1Non-significant decrease compared to nurse counselling
Intensive reminders vs written appointment cardsDiscontinuation (12 months)1Non-significant increase compared to appointment cards
On-time injections1Non-significant decrease compared to appointment cards

Summary of results:

One out of the 6 studies showed a significant reduction in discontinuation, in other words, improved continuation of hormonal methods of contraception (oral and injection). The intervention was structured counselling, including individual counselling sessions and education (audiovisual messages). Many of the studies also measured why women discontinued the hormonal methods. One study found that women were less likely to discontinue the hormonal method due to menstrual disturbances and one study found it was due to dissatisfaction with the method.   

Effectiveness statements:

There is insufficient evidence to determine if enhanced counselling techniques or other client-provider interventions increase adherence to and continuation of hormonal contraceptives (injectable or oral).

Haynes 2008

Interventions for enhancing medication adherence

[Maps to: Providing information or education, Facilitating communication and/or decision making, Acquiring skills and competencies, Supporting behaviour change, Support, Minimising risks or harms, Improving quality] 

Summary of results:

Less that half (41 of 93) of the interventions showed significant increases in medicines adherence (5 for short-term treatments and 36 for long-term treatments). A minority of interventions (29 of 93) showed significant improvements in at least one treatment outcome (4 for short-term treatments and 25 for long-term treatments). The majority of effective interventions in short-term treatments were simple (eg counselling, written information and personal phone calls). The majority of effective interventions in long-term treatments were complex (eg combinations of more convenient care, information, counselling, reminders, self-monitoring, reinforcement, family therapy, psychological therapy, crisis intervention, manual telephone follow-up, additional supervision or attention). Of several studies examined the effects of telling patients about adverse effects of medicines, none showed significant negative effects on adherence.

Effectiveness statements:

There is some evidence that simple interventions improve adherence and treatment outcomes in short-term treatments, and complex interventions in long-term treatments - results are mixed.  

Heneghan 2006a

Reminder packaging for improving adherence to self-administered long-term medications

[Maps to: Supporting behaviour change]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results

Reminder packaging vs usual care

 

Adherence (pill count)6 intAMI = 11% more pills taken
Adherence (self report)2Non-significant reduction
Patient satisfaction21 study intervention more difficult/less convenient to use; 1 study ARI = 50 more people out of 100 (significance unknown)
Hospitalisation/readmission1No data
Costs22 studies significant increase
Clinical outcomes51 study significant improvements; 3 studies non-significant improvements; 1 study improvements (significance unknown)

Summary of results:

The majority of studies reported significantly improved adherence by pill count, but non-significant results with self report. A minority of studies (1 of 5) showed significant improvement in clinical outcomes such as blood pressure, glycated haemoglobin, vitamin C and E levels, and psychological symptoms. The study with significant changes in clinical outcomes did not measure adherence. Significant increases in costs occurred with reminder packaging (2 of 2 studies). Reminder packaging was more difficult/less convenient to use in one study but useful in another. No studies reported adverse events, with little data provided for hospitalisations.

Effectiveness statements:

There is some evidence to support the use of reminder packaging on medicines adherence - the results were mixed when measuring adherence by pill count versus self report. There is insufficient evidence to determine the effectiveness of reminder packaging to improve clinical outcomes, patient satisfaction and provide system benefits. 

Heneghan 2006b

Self-monitoring of oral anticoagulation: a systematic review and meta-analysis

[Maps to: Acquiring skills and competencies, Supporting behaviour change, Minimising risks or harms]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results

Self-monitoring vs standard monitoring

 

Mean INR (international normalised ratio) within target range116 studies significant increase; 5 studies non-significant increase
Proportion of time within therapeutic range72 studies significant increase; 2 studies non-significant increase; 3 studies non-significant reduction
Thromboembolic events13ARR = 2 fewer people out of 100 (95% CI 4 to 2 fewer)
Death10ARR = 2 fewer people out of 100 (95% CI 2 to 1 fewer)
Major haemorrhage12ARR = 1 fewer person out of 100 (95% CI 2 to 0 fewer)
Total number of tests9Increased
Dropout rates (unable to complete treatment)8Mean = 22% (range 9% to 43%)

Summary of results:

The majority of studies comparing self-monitoring with standard monitoring showed significant improvement in the frequency of mean international normalized ratio (INR) falling within the target therapeutic range and in the proportion of time the INR was within the therapeutic range. The number of tests increased in people self-monitoring. Self-monitoring (with adjustment or no adjustment) significantly decreased the risk of thromboembolic events, major haemorrhages and death from all causes, but results on the rate of minor haemorrhages were mixed. When subgroups were analysed, self-monitoring with or without self-adjustment significantly decreased thromboembolic events. Death and major haemorrhages were decreased; significantly with self-adjustment, non-significantly without self-adjustment. A significant proportion of people self-monitoring were unable to complete treatment and dropped out.

Effectiveness statements:

There is sufficient evidence that self-monitoring decreases adverse events (including major haemorrhage, death and thromboembolic events) with an increase in testing - self-monitoring is generally effective. There is some evidence that self-monitoring can improve time within the therapeutic range and average result in therapeutic range - results are mixed. There is some evidence that self-monitoring with self-adjustment may have a greater reduction in adverse events than self-monitoring without self-adjustment - results are mixed. 

Jacobson 2005

Patient reminder and recall systems to improve immunization rates

[Maps to: Supporting behaviour change, Minimising risks or harms]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results

Patient reminder and recall systems versus usual care

 

Immunisation rate35ARI = 11 more people out of 100 (95% CI 8 to 14 more)
Child influenza immunisations4ARI = 19 more people out of 100 (95% CI 6 to 29 more)
Pre-school child routine immunisations15ARI = 9 more people out of 100 (95% CI 6 to 12 more)
Adult influenza immunisations12ARI = 12 more people out of 100 (95% CI 6 to 18 more)
Adult (other vaccines)3ARI = 18 more people out of 100 (95% CI 4 to 33 more)
Adolescent immunisations1Non-significant increase
Costs16Not available

Summary of results:

Typically, immunisation rates increased within the range of 5% to 20% with patient reminders/recall systems (42 studies, range 1% to 47%), although a small number (5 studies) reported decreased immunisation rates (range 2% to 9%). Immunisation rates significantly increased for routine childhood vaccinations, influenza vaccinations for children and adults, and adult pneumococcus, tetanus and Hepatitis B vaccinations. In the single study on adolescents, there was no significant effect of a reminder intervention on immunisation rates. Person-to-person telephone calls, letters, postcards, autodialer computer reminders, postcards plus telephone calls, and patient plus provider reminders all significantly increased immunisation rates. Person-to-person calls were the most effective single intervention, but patient and provider reminders delivered together was the most effective approach overall. Patient reminders with outreach non-significantly increased immunisation rates. Cost data were mixed due to different types of reminder used (eg with telephone reminders more expensive than either letter or postcard reminders), different intensities of interventions (eg ranging from single postcard reminders to repeat reminders plus home visits), and different methods of calculating costs and resources.

Effectiveness statements:

There is some evidence that patient reminder and recall systems improve immunisation rates in adults and children - they are generally effective. There is some evidence that person-to-person telephone calls are the most effective single intervention, and that patient and provider reminders delivered together are the most effective intervention overall. There is insufficient evidence to determine the cost effectiveness of interventions; the effects of interventions in low- and middle-income countries; and the effects of reminder and recall interventions in adolescents.  

Koshman 2008

Pharmacist care of patients with heart failure

[Maps to: Providing information or education, Supporting behaviour change, Minimising risks or harms, Improving quality]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results
Pharmacist-directed care vs controlMortality7Non-significant reduction
All-cause hospitalisation 7Non-significant reduction
Hospitalisation for heart failure6Non-significant reduction
Health-related quality of life61 study significant increase; 1 study mixed effects (non-significant and significant reductions with different measures); 4 studies non-significant changes
Adherence - Medication Events Monitoring (MEM) system61 study significant increase with MEMs; 1 study significant decrease with self report; 3 studies non-significant changes (pharmacy fill records; tablet counts); 1 study mixed effects (significant increase with MEMs, non-significant changes with self report)
Pharmacist collaborative care vs controlMortality5Non-significant reduction
All-cause hospitalisation 4ARR = 12 fewer people out of 100 (95% CI 22 to 1 fewer)
Hospitalisation for heart failure5ARR = 15 fewer people out of 100 (95% CI 22 to 6 fewer)
Health-related quality of life1Mixed effects (significant increase and non-significant changes with different measures)
Adherence1Non-significant changes all medicines

Summary of results:

Pharmacist-directed care did not significantly decrease hospitalisation rates (all-case or heart failure-related) or mortality, improved health-related quality of life in only the minority (1 of 6) of studies, and had mixed effects on adherence when compared with control. Pharmacist collaborative care interventions significantly reduced hospitalisations, both for heart failure and due to any cause, when compared with control. However there were no significant changes to mortality or adherence and effects on health-related quality of life were mixed in a single study.

Effectiveness statements:

There is insufficient evidence from trials that pharmacist-directed care improves service use, clinical outcomes, quality of life or adherence in people with heart failure - it is generally ineffective. There is some evidence from trials that pharmacist collaborative care reduces hospital admissions for heart failure, and all-cause hospital admission - it is generally effective. There is insufficient evidence from trials that pharmacist collaborative care improves mortality - it is generally ineffective. There is insufficient evidence to determine the effects of pharmacist collaborative care on adherence or quality of life.

Table 11. Results, by individual review: Part C
  1. * Numbers of individual studies contributing to results for each outcome are reported unless otherwise indicated as numbers of interventions (int) per outcome.

    RR = Relative Risk; OR = Odds Ratio; ARR = Absolute Risk Reduction; ARI = Absolute Risk Increase; MR = Mean Reduction; MI = Mean Increase; AMR = Absolute Mean Reduction; AMI = Absolute Mean Increase; 95% CI = 95% Confidence Interval

Lewin 2005

Lay health workers in primary and community health care

[Maps to: Providing information or education, Supporting behaviour change, Minimising risks or harms, Improving quality]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results
Lay health worker (LHW) intervention vs usual careImmunisation uptake3Significant increase RR = 1.30 (95% CI 1.14 to 1.48)

Summary of results:

There was a significant increase in immunisation uptake in adults and in children with LHW interventions.

Effectiveness statements:

There is some evidence that LHW interventions improve immunisation uptake in adults and children - they are generally effective. 

Lummis 2006

Systematic review of the use of patients' own medications in acute care institutions

[Maps to: Support, Minimising risks or harms, Improving quality]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results

Pharmacist assessing patients' own medicines (POM) (no control)

 

Number POMs reviewed1Significant increase with intervention
Medicines errors identified1Significant increase with intervention
POM vs hospital dispensed medicinesMedicines administration errors1Non-significant change

Pharmacist assessing POM (no control)

 

Patients with medicines errors1Significant increase with intervention
Medicines errors identified using POMS1Significant increase with intervention
Workload (pharmacist time)11 study increase with intervention (significance not reported)
Workload (dispensary staff)22 studies decrease with intervention (significance not reported)
Allergies recorded11 study increase with intervention (significance not reported)

Pharmacist assessing (POM) (no control two studies)

 

Cost of medicines31 study cost saved per patient on re-use of POMS at discharge $US11 (vs control); 1 study cost saved per patient $US9 with POMs; 1 study decreased costs with intervention (significance not reported)
Discharge time31 study significant decrease with intervention (vs control); 2 studies decrease with intervention (significance not reported)

Summary of results:

Of the intervention studies included in this review only 1 of 5 was controlled and results should be interpreted with caution due to inclusion of studies of poor design for assessing intervention effectiveness. Single studies each reported that pharmacists assessing patients' own medicine (POM) use significantly increased identification of medicines errors, numbers of patients with medicines errors, and medicines errors identified amongst POMs. Allergy documentation in charts was also increased by pharmacists assessing POM use, but significance was unclear. One study assessing medicines administration errors did not find a difference between POMs use alone and hospital-dispensed medicines. One study indicated that interventions involving pharmacists assessing POMs increased workload (time requirements) for the pharmacist involved, and hospital dispensary staff workload was decreased in two studies, but significance of these results is unclear. Studies also show costs to hospitals and patients after discharge were reduced with pharmacists assessing POMs use (3 of 3 studies: significance unclear). Time taken for patient discharge was also decreased with pharmacists assessing POMs use, but was only significant in the minority (1 of 3) studies.

Effectiveness statements:

There is insufficient evidence to determine if pharmacists assessing POM use improves identification of medicines errors. There is insufficient evidence to determine if using POM alone improves medicines administration errors.

Maglione 2002

Mass mailings have little effect on utilization of influenza vaccine among Medicare beneficiaries:

[Maps to: Providing information or education, Supporting behaviour change, Minimising risks or harms]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results
Mass mailings: personalised or form letter vs controlImmunisation uptake1Absolute increase = 2 to 8 more people out of 100
Mass mailings: postcard or letter plus brochure/postcard vs controlImmunisation uptake4Absolute increase = 1 to 3 more people out of 100

Summary of results:

The majority of studies (3 of 5) examining mass mailings, compared with control, found significant increases in immunisation uptake. However, authors note that the significant results are not clinically significant.  

Effectiveness statements:

There is some evidence that mass mailing interventions increases the uptake of influenza vaccination - results of mass mailings were mixed.

Maio 2005

Pharmacy utilisation and the Medicare Modernisation Act

[Maps to: Improving quality]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results
Drug benefit capCost containment; appropriate use; system benefits52 of 2 studies reduction in medicines use; 2 of 2 studies disenrolment from healthcare plan; 1 of 1 study reduces cost; 1 of 1 study increased nursing home admission
CopaymentCost containment; appropriate use; system benefits; adverse events74 of 5 studies reduction in medicines use (1 study reduction with large copayment but not small copayment); 3 of 4 studies reduction in costs; 2 of 3 studies increased health services utilisation; 1 of 2 studies increased adverse events
Prior authorisationCost containment; system benefits1Reduced costs; system use no difference

Closed formulary

 

Appropriate use; cost containment; system benefits

 

1

 

Increased use, costs and system use

Therapeutic substitution

 

Cost containment; adverse events; health status

 

2

 

2 of 2 studies no change health status or adverse events; 1 of 1 study reduced costs

Generic substitution

 

Cost containment; health status; adverse events

 

2

 

2 of 2 studies no change health status; 1 of 1 study no change adverse events; 1 of 1 study reduced costs

Summary of results:

A majority of the studies found Pharmacy Utilisation Management (PUM) strategies decrease prescription medicines use and medicines costs. This is with the exception of the closed formulary study (very weak study design) which found increases in use and costs. Increased healthcare utilisation was found in the majority of studies, but a minority found a reduction in health status and increase in adverse events. The majority of studies for drug caps showed reduced costs, but increased system use and reduction in health status; copayment studies showed mixed results; the 1 study of prior authorisation showed reduced costs with no change in system use; and formularies showed mixed results but the studies of substitutions showed a reduction in costs without effects to health status or system use.

Effectiveness statements:

There is some evidence that PUM reduces medicines costs and improves medicines use in seniors without reducing health status it is generally effective. But there is some evidence that it increases healthcare utilisation. Specifically, there is some evidence that drug caps reduce costs and use, but increases system use and reduces health status; some evidence that copayment reduces costs and use, but increases system use and reduces health status results are mixed; insufficient evidence to determine the effect of prior authorisation; and some evidence that formularies reduce costs and use with no effect on health status and system use results are mixed. 

McIntosh 2006

Compliance therapy for schizophrenia:

[Maps to: Facilitating communication and/or decision making, Supporting behaviour change, Support]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results
Compliance therapy vs non-specific counsellingAdherence1Non-significant decrease
Attitudes to medicines1Non-significant decrease
Mental health status1Non-significant change
Quality of life1Non-significant decrease

Summary of results:

There were no significant differences in adherence to antipsychotic treatment, attitudes to medicines, quality of life or mental health status when compliance therapy and non-specific counselling were compared. There were also no significant differences for compliance therapy for clinical or service use (hospital admission) outcomes.

Effectiveness statements:

There is insufficient evidence to determine whether compliance therapy improves adherence, attitudes to antipsychotic medicines, clinical outcomes or quality of life in people with schizophrenia.

Morrison 2001

Evaluation of studies investigating the effectiveness of pharmacists' clinical services

[Maps to: Providing information or education, Supporting behaviour change, Acquiring skills and competencies]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results

Pharmacist provided patient counselling vs usual care

 

Adherence

 

6

 

4 studies significant increase; 2 increase significance unknown
Medicines errors1Significant increase

Knowledge

 

5

 

Increase favouring intervention: 2 studies non-significant; 3 significance unknown

Correct use of inhaler

 

2

 

Increase favouring intervention: 1 study non-significant, 1 study significance unknown

Clinical measure (blood sugar)

 

1

 

Decrease (significance unclear)

Pharmacist provided patient and physician counselling vs usual care

 

Adherence

 

4

 

2 studies significant increase;  2 studies non-significant increase

Clinical measures (blood cholesterol (BC), blood pressure (BP), chronic obstructive pulmonary disease symptoms)

 

4

 

2 studies significant increase (BP and BC); 2 studies increase favouring interventions significance unknown (BP and symptoms)
Adverse experiences1Significant decrease

Pharmacist provided physician counselling vs usual care

 

Clinical outcomes

 

2

 

1 study significant increase; 1 study significant decrease
Drug monitoring (time for pyrexia to abate)1Non-significant changes

Proportion of prescriptions meeting guidelines

 

1

 

Significant increase OR = 2.9 (95% CI 2.2 to 3.8)
Mean number of prescriptions22 studies non-significant changes
Cost per prescription1

Non-significant changes

 

Pharmacist provided patient care vs usual care

 

Clinical measures (symptoms, blood pressure, blood sugar)4 ints3 ints non-significant changes; 1 int significantly favours intervention 

Adherence

 

1

 

Non-significant increase

Summary of results:

Pharmacist provided patient counselling significantly increased identification of medicines errors in a single study, and significantly improved adherence in the majority of studies (4 of 6), when compared to usual care. Pharmacist provided patient counselling also improved knowledge, correct use of inhaler and blood sugar levels but significance of these results was unclear. In half of studies, counselling of both patients and physicians by pharmacists significantly improved adherence (2 of 4) and clinical outcomes (2 of 4), and significantly decreased adverse experiences in a single study, when compared to usual care. Pharmacist counselling of physicians significantly increased the proportion of prescriptions meeting guidelines (1 study) and significantly improved clinical outcomes in half (1 of 2) of studies, but had no significant effects on cost per prescription, mean number of prescriptions, or drug monitoring, compared to usual care. Pharmacist provided patient care interventions did not significantly improve adherence (1 study) and improved clinical measures significantly in only the minority (1 of 4) of studies, compared with usual care.

Effectiveness statements:

There is some evidence that pharmacist provided patient counselling improves identification of medicines errors and adherence - it is generally effective. There is insufficient evidence that pharmacist provided patient counselling improves knowledge, correct inhaler use or clinical measures - it is generally ineffective. There is some evidence that pharmacist provided patient and physician counselling improves adherence, clinical outcomes and adverse experiences - the results are mixed. There is some evidence that pharmacist provided physician counselling interventions increases the proportion of prescriptions meeting guidelines - it is generally effective. There is insufficient evidence to decide the effects of pharmacist provided physician counselling on prescription costs, mean number of prescriptions, drug monitoring or clinical outcomes. There is insufficient evidence that pharmacist provided patient care interventions improve adherence or clinical measures - it is generally ineffective.

Nicolson 2009

Written information about individual medicines for consumers

[Maps to: Providing information or education, Supporting behaviour change]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results

Written medicines information (WMI) vs none

 

Knowledge

 

12

 

6 studies significant increase; 4 studies non-significant changes; 2 studies mixed effects (increase and no changes)

Medicines recall

 

4

 

1 study significant increase; 3 studies mixed effects (increases and no changes)

Recall of side effects

 

6 ints

 

3 interventions significant increase; 1 intervention mixed effects (significance unclear); 1 intervention non-significant changes; 1 intervention no changes (significance unclear)
Satisfaction with information22 studies significant increase

Ratings of information

 

1

 

Significant increases in ratings of ease of understanding, usefulness, clarity and adequacy of information provided; significantly fewer felt information could be improved; significant decrease in worry about medicines AMR = 28 fewer people out of 100 (no CI);

Adherence - adherence to medicines instructions

 

6

 

2 studies significant increase; 3 studies non-significant changes; 1 study increase (significance unclear)
Number reporting health problems1Increase (significance unclear)
Number reporting side effects1Significant increase
Correct application of medicines information

1

 

Non-significant change
One WMI versus another: programmed instruction versus standard handoutKnowledge1Significant increase with programmed instruction

One WMI vs another: experimental leaflet versus manufacturer's leaflet

 

Knowledge

 

1

 

Increase with experimental leaflet (significance unclear)

Ratings of information

 

1

 

Significant increase with experimental leaflet in ease of understanding, completeness and containing new information; non-significant changes in ease of reading or interest of content

One WMI vs another: structured format versus easy-to-read format

 

Knowledge1Non-significant changes
Correct application of medicines information1Non-significant changes

One WMI vs another: numerical side effect risk versus descriptive side effect risk

 

Knowledge

 

1

 

Significant increase with numerical information for correct risk estimation

Decision to take medicines

 

1

 

Significant decrease with numerical information for 1 of 2 side effects (constipation); non-significant change for other side effect (pancreatitis)

Satisfaction with information

 

1

 

Significant increase with numerical information for 1 of 2 side effects (pancreatitis); non-significant change for other side effect (constipation)
One WMI vs another: evidence-based leaflet versus standard leaflet

Knowledge

 

1

 

Non-significant increase with evidence-based leaflet
One WMI vs another: risk information before benefits versus risk information after benefitsDecision to take medicines1Significantly more favourable rating of treatment with risk information presented before benefits
One WMI vs another: usual wording versus simplified wording or professional wording formatsRatings of information1Significant increase with usual wording format in length and complexity; non-significant changes in emotional response to information or evaluation of information; effects on judgement about information unclear
One WMI vs another: improved readability layout versus traditional insert

Reading of the information

 

1Non-significant changes

Summary of results:

Written Medicines Information (WMI) versus none: In half of studies, WMI significantly improved knowledge of medicines (6 of 12) and recall of side effects (3 of 6 interventions), but medicines recall significantly improved in only a minority of studies (1 of 4 studies). Two studies showed significantly improved satisfaction with WMI compared with none, and single studies each showed significant increases in numbers reporting side effects; ratings of the information clarity, adequacy and usefulness, and decreased worry about medicines with WMI. However, WMI significantly improved adherence to medicines and instructions in only a minority of studies (2 of 6), and did not improve application of medicines information in the single study reporting this outcome.

One WMI versus another: All comparisons were assessed in single studies. Numerical compared with descriptive side effect risk information significantly increased correct risk estimates, but had mixed effects on decision to take medicines and satisfaction with information. WMI with medicines risk information presented before benefits showed significantly more favourable ratings of treatment than when risk information was presented after benefits. WMI with programmed instruction significantly improved knowledge, when compared with a standard handout, whereas an evidence-based leaflet did not. A structured WMI, compared with an easy to read format, had no significant effects on knowledge or correct application of information; and usual wording versus simplified or professional wording had mixed effects on ratings of the information. An experimental leaflet compared with the manufacturer’s increased knowledge but significance was unclear, and significantly improved ratings of information on some but not all features (ease of understanding, completeness); while reading of medicines information was not significantly higher with an improved readability WMI over a traditional insert.

Effectiveness statements:

There is some evidence that using WMI, compared with none, may improve knowledge, recall of side effects and satisfaction with information - results were mixed. There is insufficient evidence to determine whether WMI, compared to none, improves outcomes related to medicines behaviours or attitudes. There is also insufficient evidence to decide whether one type of WMI is better than another with respect to medicines knowledge, attitudes or behaviours.

Olthoff 2005

Noncompliance with ocular hypotensive treatment in patients with glaucoma or ocular hypertension: an evidence-based review

[Maps to: Providing information or education, Supporting behaviour change]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results
Medicines alarm device vs no interventionAdherence (bottle weight)1Significant increase
Compliance aid (medicines alarm or memory aid) vs no interventionAdherence (self-report)2Significant increase, AMI = 13% to 26% more pills taken with intervention
Compliance aid (memory aid) vs no interventionIntraocular pressure1Non-significant change
Counselling and memory aid vs medicines counselling onlyMean number of prescription refills1Significant increase with combined intervention
Education and tailoring of medicines routine vs no interventionAdherence - proportion of time elapsed between doses > 8 hours1Significant decrease
Adherence - proportion of missed doses1Significant decrease 

Summary of results:

All studies reported significant increases in adherence to treatment with interventions (compliance devices, counselling with memory aids, or education and tailoring of medicines), whether assessed by self-report, pill counts or prescription refills. One study reported no significant effects of a memory aid intervention on intraocular pressure, despite an increase in medicines adherence.

Effectiveness statements:

There is some evidence that compliance aids (memory aids and alarms), counselling and memory aids, and education and tailoring can each improve treatment adherence in people with glaucoma - they are generally effective. There is insufficient evidence to determine whether interventions improve clinical outcomes such as intraocular pressure.

Table 12. Results, by individual review: Part D
  1. * Numbers of individual studies contributing to results for each outcome are reported unless otherwise indicated as numbers of interventions (int) per outcome.

    RR = Relative Risk; OR = Odds Ratio; ARR = Absolute Risk Reduction; ARI = Absolute Risk Increase; MR = Mean Reduction; MI = Mean Increase; AMR = Absolute Mean Reduction; AMI = Absolute Mean Increase; 95% CI = 95% Confidence Interval

Orton 2005

Unit-dose packaged drugs for treating malaria

[Maps to: Supporting behaviour change]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results

Unit-dose packaged drugs vs usual care

 

Adherence42 studies (blister packaging) ARI = 15 more people out of 100 (95% CI 10 to 21 more); 1 study (bags versus syrup) ARI = 49 more people out of 100 (95% CI 32 to 68 more); 1 study increase (significance unknown)
Adverse events2Reported vomiting, itching, dizziness, other
Cure rates after drug regimen42 studies all aparasitaemic and asymptomatic; 1 study most fully recovered; 1 study most improved

Summary of results:

All studies showed improved adherence with unit-dose packaging when combined with provider training and patient information; 3 studies were significant, 1 of unknown significance. Treatment failure was not adequately assessed in the studies; nor were adverse events systematically collected and reported.

Effectiveness statements:

There is insufficient evidence to determine if unit-dose packaging of medicines can improve adherence to medicines, treatment outcomes and adverse events for uncomplicated malaria, when supported by provider training and patient information. 

Roughead 2005

Pharmaceutical care services: A systematic review of published studies, 1990 to 2003, examining effectiveness in improving patient outcomes

[Maps to: Facilitating communication and/or decision making, Acquiring skills and competencies, Minimising risks or harms, Improving quality]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results

Pharmaceutical care vs usual care

 

Change in adherence82 studies significant improvement; 6 studies non-significant changes
Change in knowledge64 studies significant increase; 2 studies non-significant changes
Medicines use96 studies significant improvement; 3 studies non-significant changes
Medicines technique22 studies significant improvement
Pharmaceutical care issues and risk management22 studies significantly improvement
Health resource use82 studies significant reduction; 6 studies non-significant changes
Morbidity and mortality6Mixed results (increases and decreases)
Quality of life1611 studies non-significant changes
Clinical outcomes16Mixed results
Adverse events41 study significant decrease, 3 studies non-significant changes

Summary of results:

In a review of 22 studies of pharmaceutical care interventions, a minority of studies (2 of 8) showed significant improvements in adherence. However, a majority showed significant improvements in knowledge (4 of 6) and medicines use (6 of 9), including improvements (2 of 2) following education on techniques for using drugs (eg inhaler use), and improved risk management (2 of 2). There were mixed results for clinical outcomes (16 studies), and mortality and morbidity (6 studies). A minority of studies (1 of 4) showed improvement in adverse events, quality of life (5 of 16) and (2 of 8) for health resource use (hospitalisation and emergency admissions).

Effectiveness statements:

There is insufficient evidence to support the use of pharmaceutical care services to improve medicines adherence - it is generally ineffective. There is some evidence that pharmaceutical care improves knowledge and medicines use - it is generally effective. But insufficient evidence to support its use to improve health service, most morbidity outcomes and adverse events - it is generally ineffective. However, there is some evidence that it improves clinical outcomes - the results were mixed.  

Royal 2006

Interventions in primary care to reduce medication related adverse events and hospital admissions: systematic review and meta-analysis

[Maps to: Minimising risks or harms, Improving quality]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results

Pharmacist-led intervention vs control

 

Hospital admission

 

15

 

11 studies non-significant changes; 4 studies decrease (significance unclear)

Emergency department visits

 

3

 

3 studies non-significant changes

Mortality

 

4

 

2 studies significant decrease; 2 studies non-significant decrease

Adverse medicines reactions

 

3

 

1 study significant increase in resolution of adverse events; 2 studies non-significant decrease in adverse events

Primary healthcare professional-led intervention vs control

 

Hospital admission

 

7

 

2 studies non-significant decrease; 2 studies non-significant increase; 2 studies non-significant changes; 1 study significant decrease pre- to post-intervention

Emergency department visits

 

43 studies non-significant decrease; 1 study non-significant increase
Adverse drug events per patient1Non-significant increase
Nurse-led chronic disease management vs controlAdverse drug events4Non-significant increase 
Complex intervention to reduce falls vs controlHospital admission22 studies non-significant decrease
Emergency department visits1Non-significant decrease

Falls

 

11

 

10 studies non-significant decrease; 1 study significant decrease

Summary of results:

All 3 studies assessing medicines adverse events showed an improvement with pharmacist-led medicines review, compared with control, although only 1 of 3 studies was significant. A minority (4 of 15) of studies of pharmacist-led medicines review, compared with control, decreased hospital admissions, although significance was unclear. Half of studies (2 of 4) showed significantly decreased mortality with pharmacist-led interventions, with no significant changes in emergency department visits when compared with control. There were no significant changes to hospital admission, emergency department visits or adverse drug events when interventions delivered by other healthcare professionals, or complex interventions to reduce medicines-related falls, were compared with control.

Effectiveness statements:

There is some evidence that pharmacist-led interventions decrease adverse events - results are mixed. There is some evidence that pharmacist-led interventions decrease mortality - results are mixed. There is insufficient evidence that pharmacist-led interventions improve hospital admissions or emergency department visits - they are generally ineffective. There is insufficient evidence that interventions led by nurses and physicians, or complex interventions to reduce falls, improve adverse events, hospital admissions or other outcomes - they are generally ineffective.

Rueda 2006

Patient support and education for promoting adherence to highly active antiretroviral therapy for HIV/AIDS:

[Maps to: Providing information or education, Acquiring skills and competencies, Supporting behaviour change, Support]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results

Any support or education vs usual care

 

Adherence1910 studies significant improvements; 9 studies non-significant changes
Virological or immunological outcomes12Conflicting findings depending on outcomes, time points, etc

Summary of results:

Approximately half (10 of 19) of the interventions examined were associated with statistically significant increases in adherence to highly active antiretroviral therapy (HAART). Results were mixed in the 12 studies that measured clinical outcomes. A majority of studies (10 of 15) providing individual interventions reported significant improvement in adherence; no studies (0 of 4) reported improvement in groups. A majority of studies (6 of 7) over 12 weeks long significantly improved adherence; no studies less than 12 weeks (0 of 8) reported improvement. A majority of studies of medicines management skills (6 of 8) significantly improved adherence; a minority of studies (1 of 7) of cognitive behavioural therapy and motivational interviewing significantly improved adherence. Studies with marginalized populations were not successful.

Effectiveness statements:

There is some evidence that supportive and educational interventions improve adherence to HAART and improve clinical outcomes - results were mixed. There is some evidence that interventions aimed at individuals rather than groups, delivered over at least 12 weeks, and providing practical medicines management strategies rather than more complex psychologically-based approaches improve adherence - they are generally effective.

Russell 2006

Older adult medication compliance: integrated review of randomized controlled trials

[Maps to: Providing information or education, Supporting behaviour change, Acquiring skills and competencies, Support]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results
Cues vs controlAdherence64 studies significant increase; 2 studies non-significant changes
Organisers vs controlAdherence31 study significant increase; 2 studies non-significant changes
Cues and organisers vs controlAdherence21 study significant increase; 1 study non-significant change
Self-medication management program vs controlAdherence22 studies significant increase
Dose simplification: low vs higher frequency dosesAdherence33 studies significant increase
Brief counselling and education (1-3 days) vs controlAdherence2312 studies significant increase; 11 studies non-significant changes
Extensive counselling and education (> 3 days) vs controlAdherence178 studies significant increase; 9 studies non-significant changes
Counselling and education (unknown length) vs controlAdherence1Non-significant changes

Summary of results:

Half (31 of 57) of the interventions significantly improved medicines adherence when compared with control. All three studies assessing simplified dose regimens (lowered dose frequency) reported significant effects, and both studies on self-medication management programs reported significant benefits for adherence. Results for other interventions were mixed. A majority (4 of 6) studies evaluating cue interventions reported improved adherence compared with controls. Only half (1 of 2) of studies assessing cues combined with organisers and a minority (1 of 3) assessing organizers alone reported significant effects on adherence. Effects of counselling and education were also mixed, with half (20 of 41) of studies reporting significant effects on adherence. No study reported negative effects of any evaluated intervention on adherence.

Effectiveness statements:

There is some evidence that self-medication management programs improve adherence - they are generally effective. There is some evidence that simplified dose regimens improve adherence - they are generally effective. There is some evidence that counselling and education, cues and/ or organiser interventions improve adherence - the results are mixed. 

Schedlbauer 2004

Interventions to improve adherence to lipid lowering medication

[Maps to: Providing information or education, Supporting behaviour change]

Intervention and comparisonOutcomeNo. of studies or interventions (int)*Results
Simplification of drug regimen vs usual regimenAdherence21 study ARI = 11 more people out of 100 (no CI); 1 study non-significant reduction
Patient preference1ARI = 59 more people out of 100 (no CI)
Patient information and education vs usual careAdherence21 study ARI = 13 more people out of 100 (no CI) (newly prescribed) but non-significant increase (repeat prescriptions); 1 study non-significant increase
Intensified patient care (reminding) vs usual careAdherence31 study ARI = 24 more people out of 100 (no CI); 2 studies non-significant increase
Complex behavioural approach vs usual careAdherence1Non-significant increase

Summary of results:

A minority of studies (3 of 8) improved adherence significantly: 1 of 2 for simplified drug regimens; 1 of 2 for patient information and education; 1 of 3 for intensified patient care/reminding. A minority of studies (1 of 4) found a significant reduction in cholesterol levels or adverse effects (simplified drug regimen). Discontinuation rates were assessed in 1 study, with no significant effect. The class of lipid lowering medicines used in studies (eg statins, resins/bile acid sequestrants) did not seem to affect adherence rates. There were no studies evaluating decision support or administrative improvements.

Effectiveness statements:

There is insufficient evidence to support the use of interventions such as simplifying drug regimens, informing, educating, reminding and motivating patients, to improve adherence to self-administered medicines - they are generally ineffective.

Schroeder 2004 

Interventions for improving adherence to treatment in patients with high blood pressure in ambulatory settings

[Maps to: Providing information or education, Supporting behaviour change, Support, Improving quality]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results
Simplification of medicines regimen vs usual regimenAdherence9 int2 int non-significant increase; 7 int RR increase 8 to 19.6%
Patient education vs usual careAdherence6 int3 int non-significant increase; 2 int non-significant decrease; 1 int ARI = 24%
Patient motivation, support and reminders vs usual careAdherence24 int10 int ARI up to 24%
Complex health and organisational interventions (combined interventions and structure hypertension management) vs usual careAdherence18 int8 int ARI up to 41%

Summary of results:

19 of the 38 studies showed significant increases in adherence. Some studies evaluated multiple types of adherence-enhancing interventions (therefore effects by number of interventions are reported here). Simplification of dosing regimens increased adherence in 7 out of 9 interventions. Patient education increased adherence in 1 out of 6 interventions. Patient motivation, support and reminders increased adherence in 10 out of 24 interventions (successful interventions included reminder charts, self-determination training, reminders and packaging, social support, nurse phone calls, family member support, electronic medication cap aid and telephone-linked computer counselling). Complex interventions increased adherence in 8 out of 18 interventions (successful interventions included work site care; combined home visits, education and special dosing devices; educational leaflet, reminders and educational newsletter; and pharmacist-led patient medicines management and advice interventions). The effects of interventions on adherence rates was variable and where significant ranged from 5% to 41% increase.

Effectiveness statements:

The overall results of all types of interventions to improve adherence to antihypertensive medicines were mixed. There is sufficient evidence that simplification of medicines regimens improves adherence - it is generally effective. There is insufficient evidence that patient education improves adherence - it is generally ineffective. There is some evidence that patient motivation, support and reminders or complex or combined interventions improve adherence - the results were mixed.

Spurling 2007

Delayed antibiotics for respiratory infections:

[Maps to: Facilitating communication and/or decision making, Minimising risks or harms]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results
Delayed vs immediate antibioticsAntibiotic use6ARR = 64 fewer people out of 100 used antibiotics with delayed antibiotics (95% CI 81 to 38 fewer)
Clinical: sore throat symptoms2Significant increase at day 3 in numbers with pain (1 study), malaise (1 study) and fever severity (2 studies) with delayed antibiotics; non-significant changes for severity of pain (1 study), malaise (1 study) or fever at day 1 (2 studies)
Clinical: otitis media symptoms2Significant increase at day 3 in pain severity with delayed antibiotics (1 study); non-significant changes at day 7 or for pain to day 7 (1 study); significant increase at days 3 to 7 for malaise and malaise severity with delayed antibiotics (1 study); non-significant change in fever severity (1 study)
Clinical: common cold symptoms1Non-significant changes at any time point for symptoms or severity
Clinical: cough symptoms2Non-significant changes
Supplementary medicines use21 study significant increase with delayed antibiotics MI = 0.59 (95% CI 0.25, 0.93); 1 study non-significant decrease with immediate antibiotics
Adverse effects: vomiting31 study significant increase with delayed antibiotics; 2 studies non-significant changes
Adverse effects: stomach ache1Non-significant changes
Adverse effects: diarrhoea42 studies significant decrease with delayed antibiotics; 1 study non-significant decrease with delayed antibiotics; 1 study non-significant increase with delayed antibiotics
Adverse effects: rash2Non-significant changes
Satisfaction5ARR = 6 fewer people out of 100 were satisfied with their treatment with delayed antibiotics (95% CI 12 to 3 fewer)
Delayed vs no antibioticsAntibiotic use2Non-significant increase
Clinical: signs and symptoms2Non-significant changes (sore throat symptoms, cough symptoms)
Adverse effects1Non-significant changes (vomiting, rash, stomach ache, diarrhoea)
Satisfaction2Non-significant increase

Summary of results:

For delayed versus immediate antibiotics: In meta-analysis, antibiotic use was significantly reduced with delayed antibiotics (6 studies but there was high heterogeneity), but patient satisfaction was also reduced (5 studies). One of 2 studies reported significantly higher supplementary medicines use with delayed prescribing. The effects were mixed for clinical outcomes for sore throat and otitis media, with both worse symptoms and no differences reported at different time points for delayed compared with immediate antibiotics; for cough or common cold there were no studies reporting significant differences in clinical outcomes between delayed and immediate antibiotics. Effects of delayed antibiotics were also mixed for adverse effects: a minority of studies (1 of 3) found significantly more vomiting, while half of studies (2 of 4) reported less diarrhoea, while for other adverse events there were no significant differences. For delayed versus no antibiotics: Two studies showed no significant changes in antibiotic use with delayed antibiotics, and no changes in symptom resolution, adverse events or patient satisfaction.

Effectiveness statements:

There is sufficient evidence that delayed antibiotics decrease antibiotic use in comparison to immediate antibiotics - they are generally effective. There is insufficient evidence of an effect of delayed antibiotics on antibiotic use in comparison to no antibiotics - they are generally ineffective. There is sufficient evidence that delayed antibiotics are associated with lower satisfaction - they are generally ineffective. There is some evidence that delayed antibiotics increase supplementary medicines use - results are mixed. There is insufficient evidence that delayed antibiotics improve clinical outcomes or adverse effects - results are mixed.

Table 13. Results, by individual review: Part E
  1. * Numbers of individual studies contributing to results for each outcome are reported unless otherwise indicated as numbers of interventions (int) per outcome.

    RR = Relative Risk; OR = Odds Ratio; ARR = Absolute Risk Reduction; ARI = Absolute Risk Increase; MR = Mean Reduction; MI = Mean Increase; AMR = Absolute Mean Reduction; AMI = Absolute Mean Increase; 95% CI = 95% Confidence Interval

Stevenson 2004

A systematic review of the research on communication between patients and healthcare professionals about medicines:

[Maps to: Providing information or education, Facilitating communication and/or decision making, Improving quality, Support, Minimising risks or harms]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results

Interventions promoting doctor-patient communication: training seminars for doctors vs no seminar

 

Repeated patient complaint2Increase with intervention (significance unclear)
Asked patient to repeat instructions or demonstrate use32 studies increase (significance unclear); 1 study increase at follow up
Patient medicines information recall1Increase with intervention (significance unclear)
Addressed patient fears about new medicines1Significant increase

Interventions promoting doctor-patient communication: patient communication skills training vs medicines education

 

Medicines question asking skill1Significant increase with communication skills training
Acquisition of medicines knowledge1Significant increase with communication skills training
Patient problems and symptoms1Non-significant change
Number of medicines questions asked1Significant increase with communication skills training
Interventions promoting doctor-patient communication: medicines fact sheet plus doctor counselling vs fact sheetMedicines knowledge1Significant increase with combined intervention
Interventions promoting doctor-patient communication: medicines fact sheet plus doctor counselling vs no interventionMedicines knowledge1Significant increase with intervention

Interventions promoting pharmacist-patient communication: modified pharmacy services and medicines review vs usual care

 

Adherence (self report)32 studies significant increase; 1 study increase and decrease
Adherence - prescription refill21 study significant increase; 1 study decrease and no change
Clinical outcomes2Significant improvement and no change
Patient satisfaction3Significant increase
Cost of medicines1Significant decrease
Medicines-related problems1Significant decrease
Interventions promoting pharmacist-patient communication: advertising campaign promoting question asking (no control)Number of medicines questions asked1Non-significant change
Information tailored to patient1Increase with intervention (significance unclear)
Interventions promoting pharmacist-patient communication: written questions for pharmacist plus counselling vs usual careNumber of medicines questions asked1Significant increase
Patient recall of medicines information1Non-significant change
Adherence1Non-significant change

Interventions promoting pharmacist-patient communication: patient prompt for question asking plus counselling vs usual care

 

Patient recall of medicines information1Non-significant change
Adherence1Non-significant change
Patient recall of medicines information1Non-significant change
Adherence1Non-significant change
Number of medicines questions asked1Non-significant change
Interventions promoting pharmacist-patient communication: pharmacist questioning protocol for adherence problems vs usual careAdherence1Significant increase
Satisfaction with answers to medicines questions1Significant increase

Interventions promoting nurse/ assistant-patient communication: telephone follow-up vs no call

 

Number reporting adverse effects1Non-significant change
Adherence (self report)1Non-significant change
Adherence - pharmacy records1Non-significant change
Number stopping due to adverse events1Non-significant change
Usefulness of service1Majority felt intervention useful (significance unclear; no control)

Interventions promoting nurse/ assistant-patient communication: face-to-face consultation vs usual care

 

Adherence1Significant increase
Perceived barriers to adherence1Non-significant change
Discussions with doctor about medicines issues1Significant increase
Patient analgesia use1Increased following intervention (significance unclear; no control)

Summary of results:

Doctor patient communication (5 studies): There were 4 studies on communication skills training. One study targeted patients and compared it to medicines education and found it improved medicines knowledge, question asking, and question asking skill but not clinical outcomes. Three studies targeted doctors: 1 found it increased the number of times doctors addressed patients’ fears about new medicines; the majority (2 of 3) of studies found it increased how often doctors asked patients to repeat instructions about use; 1 study showed it improved patient medicines recall, and the times doctors repeated patient complaints (2 of 2 studies) but significance was unclear. In another study, fact sheets with counselling by doctors increased patient medicines knowledge compared to fact sheets alone.
Pharmacist patient communication (6 studies): 1 study evaluated communication skills training targeted to pharmacists and found patients were more satisfied with pharmacist time and answering  their questions; 1 evaluated a mass media campaign targeting patients in which the number of questions asked did not increase, but information was more tailored by pharmacists; written prompts used by patients in 1 study did not increase questions asked, but prompts to patients to write questions for pharmacist did increase questions asked, but not adherence or patient recall; 3 studies changed pharmacist visits (clinic or home) which improved satisfaction and medicines problems and decreased costs, but effects were mixed for adherence and clinical outcomes.
Nurses or medical assistants and patient communication (5 studies): 3 studies in which face-to-face education/counselling was provided found, in individual studies, significantly increased adherence and increased discussions with doctors about medicines, but no change to barriers to adherence. Two studies evaluated telephone contact to discuss medical problems: 1 study found no difference in reporting of adverse effects or in adherence; the other study found more discussed issues on the call and found the calls useful.

Effectiveness statements:

There is insufficient evidence to determine whether interventions to improve two-way communication between patients and healthcare professionals improve outcomes related to communication, adherence and medicines use or clinical outcomes.

Stone 2002

Interventions that increase use of adult immunization and cancer screening services: a meta-analysis

[Maps to: Providing information or education, Supporting behaviour change, Improving quality, Minimising risks or harms]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results
Organisational change vs usual care/controlImmunisation uptake10Significant increase; OR = 16.0 (95% CI 11.2 to 22.8)
Provider reminder vs usual care/controlImmunisation uptake22Significant increase; OR = 3.80 (95% CI 3.31 to 4.37)
Patient financial incentive vs usual care/controlImmunisation uptake8Significant increase; OR = 3.42 (95% CI 2.89 to 4.06)
Provider education vs usual care/controlImmunisation uptake13Significant increase; OR = 3.21 (95% CI 2.24 to 4.61)
Patient reminder vs usual care/controlImmunisation uptake23Significant increase; OR 2.52 (95% CI 2.24 to 2.82)
Patient education vs usual care/controlImmunisation uptake22Significant increase; OR = 1.29 (95% CI 1.14 to 1.45)
Provider financial incentive vs usual care/controlImmunisation uptake4Non-significant increase
Feedback vs usual care/controlImmunisation uptake2Non-significant increase

Summary of results:

Many interventions significantly increased use of adult immunisation. Relative effectiveness of interventions: organisational change was the most effective; provider reminder, patient financial incentives, provider education were effective; patient reminder and education were less effective. Provider financial incentives and feedback non-significantly increased uptake. 

Effectiveness statements:

There is some evidence that many interventions increase uptake of adult immunisation - they are generally effective. Relative effectiveness of interventions: organisational change was the most effective; provider reminder, patient financial incentives, provider education were effective; patient reminder and education were less effective. Provider financial incentives and feedback non-significantly increased uptake. There was limited information for mass media interventions and regulatory or legislative actions. 

van Eijken 2003

Interventions to improve medication compliance in older patients living in the community: a systematic review of the literature

[Maps to: Supporting behaviour change, Improving quality]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results
Single generalised intervention vs controlAdherence13 int3 int significant increase; 2 int non-significant increase; 5 int non-significant difference; 3 int increase (2) or no difference (significance unknown)
Multifaceted generalised intervention vs controlAdherence3 int1 int significant increase; 2 int non-significant changes
Multifaceted tailored intervention vs controlAdherence7 int3 int significant increase; 2 int non-significant increase; 1 int non-significant change; 1 int non-significant decrease

Summary of results:

A minority of single interventions (5 of 13) showed improved adherence compared to control, and 3 were significant. A minority (1 of 3) of multifaceted generalised interventions significantly improved adherence. Almost half (3 of 7) multifaceted tailored interventions found significant improvements in adherence. Proportionately more multifaceted interventions improved adherence compared to single interventions; and proportionately more tailored interventions improved adherence compared to generalised interventions.

Effectiveness statements:

There is some evidence that multifaceted tailored interventions increase medicines adherence among older people in the community - results are mixed. There is insufficient evidence to support the use of generalised multifaceted interventions and single interventions among older people to increase adherence - they are generally ineffective. There is some evidence that multifaceted interventions improve adherence more than single interventions and tailored more than generalised interventions.

Van Wijk  2005

Effectiveness of interventions by community pharmacists to improve patient adherence to chronic medication: a systematic review

[Maps to: Providing information or education, Support]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results
Community pharmacist delivered education, monitoring, medicines/chart review and/or counselling vs usual careAdherence (self report)75 studies non-significant changes; 2 studies significant increase at follow-up (6 months and longer) with intervention
Community pharmacist delivered education, monitoring, medicines review and/or counselling vs usual careAdherence - pill counts54 studies non-significant changes; 1 study significant increase with intervention
Community pharmacist delivered education, monitoring, and/or counselling vs usual careAdherence - pharmacy records42 studies non-significant changes; 2 studies significant increase with intervention
Community pharmacist delivered monitoring and counselling vs usual careAdherence - medication event monitoring system1Significant increase with intervention

Summary of results:

A minority of studies (6 of 17) reported significant improvements in adherence to chronic medicines with interventions delivered by community pharmacists, compared with usual care. Effects of the range of different interventions assessed were mixed overall: both positive and no effects on adherence were found for interventions delivered individually or in combination, and including patient education and counselling at each prescription refill, monthly counselling and monitoring, encouragement and reward for adherence, incorporation of patients' questions into counselling, or chart review or problem identification.

Effectiveness statements:

There is insufficient evidence that community pharmacist interventions improve patient adherence to chronic medicines - they are generally ineffective.  

Vergouwen 2003

Improving adherence to antidepressants: a systematic review of interventions

[Maps to: Providing information or education, Supporting behaviour change, Support, Improving quality]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results

Education vs usual care (outpatient)

 

Adherence43 studies non-significant changes; 1 study significant increase compared to verbal information only
Depression1Significant increase; adherence not measured
Dosage and frequency vs usual careAdherence1Significant increase with choice of frequency
Collaborative (primary care) vs usual careAdherence119 studies significant increase; 2 studies non-significant changes
Depression1110 studies significant reduction; 1 study non-significant changes
Education (primary care) vs usual careAdherence33 studies non-significant changes
Depression32 studies significant reduction; 1 study non-significant change 

Summary of results:

Outpatient setting: A minority of studies (1 of 4) comparing education with usual care found significant increases in antidepressant medicines adherence - symptoms of depression were not measured.  Another study comparing education to verbal information only significantly reduced depression, but adherence was not measured. Significantly improved adherence was found when patients actively chose their dosage regimen. Primary care setting: There was no significant difference in adherence in 3 of 3 studies evaluating education; and the majority of studies (9 of 11) evaluating collaborative care significantly improved adherence when compared with usual care. Symptoms of depression were improved in the majority of primary care studies (2 of 3 for education, and 10 of 11 for collaborative care).

Effectiveness statements:

There is some evidence that collaborative care interventions in primary care settings improve both adherence and depression - they are generally effective. There is insufficient evidence to support the use of educational interventions in primary care or outpatient settings - they are generally ineffective. There is some evidence that educational interventions in an outpatient setting using combined written and verbal information, or involving patient choice of dose regimen, improves adherence; but insufficient evidence to reduce depression. 

Vermeire 2005

Interventions for improving adherence to treatment recommendations in people with type 2 diabetes mellitus

[Maps to: Providing information or education, Acquiring skills and competencies, Supporting behaviour change]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results
Education/facilitation vs usual careClinical outcome14Significant reduction

Nurse led interventions vs usual care

 

Adherence1Non-significant changes
Clinical outcome21 study significant reduction; 1 study significant reduction and non-significant changes
Home aides versus usual careClinical outcome1Reduction (significance unknown)

Diabetes education campaigns vs usual care/other intervention

 

Adherence1Increase (significance unknown)
Clinical outcomes42 studies reduction (significance unknown); 2 studies non-significant changes

Pharmacy-based interventions vs usual care

 

Adherence1Significant increase in medication possession ratio
Clinical outcomes31 study significant reduction; 1 study significant reduction and non-significant changes; 1 study reduction unknown

Dosing and frequency interventions

 

Adherence21 study significant increase with only once daily; 1 study increase with once daily (significance unknown)
Clinical outcome1Significant reduction and non-significant changes
Patient participation vs routine counsellingClinical outcome1Reductions (significance unknown)
Oral vs injectable insulinAdherence1Non-significant changes

Summary of results:

Meta-analysis for education/facilitation interventions from 3 to 48 months showed a significant decrease in glycosylated haemoglobin (clinical outcome). Separate meta-analysis for nurse-led, pharmacy-based and diabetes educator-led interventions also showed significant decreases. A minority of studies (3 of 8) reported significant increases in adherence: 2 of 2 studies that decreased dosing from 3 to 1 or 2 times daily and 1 of 2 pharmacy-based interventions. This latter pharmacy-based intervention showed improvement in adherence and clinical outcomes. One study of oral versus injectable therapy reported an increase in patient satisfaction but no effect on adherence. Another study of diabetes education reported increased knowledge but no effect on glycosylated haemoglobin. None of the included studies assessed major outcomes such as mortality or morbidity, and only 1 study reported on economic outcomes and quality of life.

Effectiveness statements:

There is insufficient evidence to support the use of interventions to improve adherence to treatment in people with type 2 diabetes - they are generally ineffective. There is some evidence that these interventions improve clinical outcomes - results are mixed - nurse-led interventions, home aides and diabetes education are generally effective in improving clinical outcomes.

Volmink 2007

Directly observed therapy for treating tuberculosis

[Maps to: Supporting behaviour change, Minimising risks or harms]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results

DOT vs self-administration of treatment

 

Cure4Non-significant increase
Cure or completion of treatment4Non-significant increase 
Completion of treatment1Non-significant increase 
DOT (home) vs self-administration of treatmentCure3ARI = 6 more people out of 100 (95% CI 1 to 11 more)
Cure or completion of treatment3ARI = 6 more people out of 100 (95% CI 1 to 11 more) 

DOT (clinic) vs self-administration of treatment

 

Cure2Non-significant decrease
Cure or completion of treatment2Non-significant decrease 
DOT home vs DOT clinicCure or completion of treatment1Non-significant increase when at home
DOT (home) family member vs DOT (home) community health workerCure or completion of treatment1Non-significant decrease with community health worker
DOT for prophylaxis in IV drug users DOT vs IV drug users self-administrationCompletion of treatment1Non-significant increase
DOT for prophylaxis where IV drug users choose own location vs IV drug users treatment centreCompletion of treatment1Non-significant decrease when attending centre

Summary of results:

There were no significant differences in cure, cure/completion of treatment, or completion of treatment alone between directly observed therapy (DOT) and self-administration. There was a small but significant difference between DOT (home) versus self-administration, on cure and completion rates favouring DOT at home. There were no significant differences in cure or completion of treatment whether DOT was provided by a family member or a health worker. There were no significant differences in cure or completion of treatment between DOT for prophylaxis and self-administration. No trials measured the effect of DOT on patients keeping their outpatient appointments while taking treatment.

Effectiveness statements:

There is insufficient evidence that DOT improves completion of treatment in people with tuberculosis or latent tuberculosis - it is generally ineffective. Although there may be a small benefit of DOT provided at home, compared with self-administration, there is insufficient evidence to determine if one form of DOT (eg provided at home or in clinics, or provided by family members or healthcare workers) is more effective than another.

Zygmunt 2002

Interventions to improve medication adherence in schizophrenia

[Maps to: Providing information or education, Facilitating communication and/or decision making, Supporting behaviour change, Support]

Intervention & comparisonOutcomeNo. of studies or interventions (int)*Results
Individual interventions vs standard care (or non-specific counselling)Adherence42 studies significant increase; 2 studies non-significant changes
Group interventions vs standard care (or social skills training)Adherence41 study significant increase; 3 studies non-significant changes
Family interventions vs standard care (or other intervention)Adherence123 studies significant increase; 9 studies non-significant changes
Community-based interventions vs standard care (or other intervention)Adherence104 studies significant increases; 6 studies non-significant changes
Multimodal psychosocial interventions vs standard careAdherence62 studies significant increase; 2 studies increase (significance unknown); 2 studies non-significant changes
Multimodal psychosocial interventions vs other interventionAdherence31 study significant increase; 2 studies non-significant changes

Summary of results:

Only a minority of single or multimodal psychosocial interventions in schizophrenia, such as individual interventions (2 of 4), group interventions (2 of 4) and family therapy (3 of 12), community based interventions (4 of 10), mixed interventions and comparisons (5 of 9) improved adherence to antipsychotics. Little relationship was found between intensity of intervention and improvement in adherence. Five of the 9 studies that had a specific goal to improve adherence improved it.

Effectiveness statements:

There is insufficient evidence to support the use of psychosocial interventions, delivered either as single or multicomponent interventions, to improve adherence to antipsychotic medicines when compared with standard care or with other interventions - they are generally ineffective.

Discussion

Summary of main results

This overview included 37 systematic reviews assessing the effects of a range of interventions with diverse aims in relation to consumers' use of medicines. These aims included: support for behaviour change; promotion of communication and informed decision making; risk minimisation; skills acquisition; and education or information provision. We identified no reviews aiming to promote consumer participation in medicines-related activities at the systems level.

Some effective interventions were simple, while others were complex. Looking collectively across reviews, promising interventions to improve adherence and other key medicines use outcomes (eg adverse events, knowledge) included: medicines self-monitoring and self-management; simplified dosing; and interventions directly involving pharmacists in medicines management. Other strategies such as: reminders; education delivered together with self-management skills training, counselling or support; financial incentives; and lay health worker interventions, also showed promise in relation to adherence but the effects of these interventions was less consistent.

No included intervention was effective to improve all medicines use outcomes across all diseases, populations or settings. For some interventions, such as providing information or education as single interventions, the accumulated evidence suggests they are ineffective. For many other interventions there is insufficient evidence to determine how effective or ineffective they might be to change medicines use outcomes with any degree of certainty. Assembling and assessing the review-level evidence on consumer's medicines use across diseases and settings has also identified major intervention, outcome and population gaps in the research that should be addressed as priorities.

Overall completeness and applicability of evidence

Interventions and outcomes

Although this overview includes 37 reviews of interventions, there were few strategies for which there was sufficient evidence to make conclusive statements about their effects on medicines use. Even for categories with proportionately more reviews, there was not necessarily sufficient evidence to draw conclusions. It is clear, however, that strategies to improve medicines use has focussed on educating consumers and supporting and changing those behaviours of consumers most practically related to taking their medicines, for example, providing reminders to patients or information pamphlets about how to take medicines. There were proportionately fewer reviews for facilitating communication and decision making, and minimising risks or harms, and, notably, no reviews addressing the effectiveness of strategies to increase consumer participation at a system level for better medicines use. The focus of existing research on education and changing practical behaviours is in contrast to emerging research suggesting that there are numerous other factors which determine how and why people take medicines (Britten 2004; Coulter 2006; Munro 2007; Pound 2005; Soumerai 2006; Townsend 2003).

Adherence

The other major focus of the included reviews was on adherence. There was often insufficient evidence to make any conclusions about the effects of interventions on consumer wellbeing and health status, or effects on a system level. The narrow range of outcomes reported reflects the included reviews' primary focus on medicines adherence, and even where reviews did not focus solely on promoting adherence there was often only a limited range of outcomes reported. It is not clear whether reviews and their included studies are selectively reporting adherence as a primary outcome. There may be a few reasons for this lack of reporting: a general lack of awareness of review authors or study authors of the complexity of factors likely to affect adherence and medicines use (relating to reporting of relevant outcomes by included studies and/or reviews); or a lack of awareness of the many factors that can affect the healthcare behaviours of consumers more generally.

Other outcomes including adverse events

It appears that reviews, and probably the research on medicines use, are still centred around adherence, which is undoubtedly an essential component of prescribing and medicine taking. However, other outcomes such as safe use or risk minimisation, informed decision making or overall satisfaction with medicines use and effects, are arguably as important as adherence but seem to be regarded as less so, or may be used as surrogates for adherence. Few reviews reported on possible adverse events related to better adherence. This may be due to the lack of adverse event reporting in primary research; nonetheless before implementing strategies to improve adherence the potential harms of those strategies should be adequately evaluated in trials. Monitoring for harms associated with interventions should also be done once interventions are implemented, to provide a more complete picture of medicines use and the effects of interventions.

Many reviews did not report outcomes which would seem intuitive and important for understanding consumers' medicines use. For example, reviews of educational and informational interventions did not routinely report outcomes such as knowledge, understanding or recall of information; and reviews of more behaviourally-based interventions, such as reminder packaging aids, did not always assess participants' skill level to use packaging, or to correctly fill aids. The reviews tended to focus on endpoint outcomes such as adherence or treatment (clinical) outcomes, which limits inferences about how and why interventions may or may not be effective. Assessing a broader range of outcomes should therefore be a priority for future research in this area, and should be guided by the direct and indirect aims of interventions under assessment.

Interventions

The intervention taxonomy developed alongside this overview represents a conceptual framework for organising the evidence (Lowe 2010; Ryan 2010). It provides a broad definition of consumers' medicines use that extends beyond adherence to envisage a more complex and interactive role for consumers in relation to decision making and management. We hope that this overview, and the taxonomic structures developed alongside it, may encourage other researchers to consider and report a wider range of outcomes than those typically captured in existing research on consumers' medicines use.

One limitation of this current overview is that it did not have a primary focus on assessing the evidence on strategies provided at a system level, such as changing the role of pharmacists or financial structures, that may indirectly influence consumers' medicines use. This overview did include some evidence about such 'indirect' to consumer interventions, and some conclusions about their effects have been made because many reviews which took a broad approach to medicines use in a particular clinical area often included such interventions at a system level. However, we cannot draw strong conclusions about these interventions.

Populations

We identified significant gaps in the populations assessed by included reviews. Many reviews included cross-disease populations, and there was a spread of acute and chronic conditions represented, as well as interventions specifically addressing immunisation uptake and contraceptive use. However, there are also clear gaps in the evidence for a range of populations. Very few reviews included studies involving children and adolescents, their parents or other carers. There has been more recognition in the consumer health literature of carers' issues, which may eventually lead to more reviews in this area. For now, however, there is insufficient evidence to draw conclusions about the effects of interventions targeting carers with respect to medicines use.

A major gap in the review literature relates to polypharmacy; in particular, what are the effects of interventions for people taking medicines for more than one concurrent health problem (multimorbidity)? This gap is likely to arise for several reasons, including that people with more than one co-existing condition are often directly excluded from trials; and even where they are not directly excluded, information about comorbid conditions often fails to be adequately reported or implications considered (Boyd 2005; Trumble 2006). While a number of included reviews did report some details about co-occurring conditions, there was little exploration of the issues apart from those affecting very old and/or frail populations (eg polypharmacy in elderly patients). Rising chronic disease rates internationally means that multimorbidity rates are also rising, particularly among younger populations (Fortin 2005; Fortin 2007; Smith 2007; Starfield 2003). Multimorbidity is associated with substantially poorer outcomes than among others in the community, including higher rates of hospitalisation, depression, and premature death, as well as higher rates of medicines use (polypharmacy) and medicines adverse events (Fortin 2005; Boyd 2005; Smith 2007). Research is needed as a priority to develop an evidence base applicable to people with more than one concurrent condition, and for medicines use in these populations, in order to improve health and other outcomes. Systematic reviews could usefully contribute to this area by reporting data on co-occurring conditions wherever it is available from primary research and reporting it as a gap where it is not (May 2009; Ryan 2009c).

Quality of the evidence

The quality of the reviews and of included studies within this overview was variable. We excluded reviews rated as low quality using the AMSTAR tool (Shea 2007), and while all Cochrane reviews were rated as high quality, most non-Cochrane reviews achieved only a rating of moderate quality. Only one review met all 11 AMSTAR criteria adequately. This suggests that the majority of the reviews included in this overview have limitations in design and/ or execution that may influence the results when considered both individually and collectively.

Similarly, at the study level, methodological quality ranged from studies that were well designed and conducted to those with serious methodological limitations. Around half of included reviews included only RCTs. While this ensured rigorous design for assessing intervention effects, it did not guarantee that the included studies were of high quality. In fact almost all reviews of RCTs reported methodological limitations in included trials, even in the small number of reviews that were highly restrictive about study inclusion based on rigorous methodological design and quality (eg Haynes 2008; Heneghan 2006a). The other half of reviews in this overview included studies of other designs, most commonly quasi-randomised trials, but sometimes designs such as controlled before-and-after studies. These reviews, too, noted that included studies generally had methodological limitations.

We have reflected limitations in the quality of the evidence by interpreting the results and formulating the statements of intervention effectiveness in light of the quality of included studies. However, methodological limitations at both the included study and review levels mean that the results of this overview should be interpreted with caution, as there is the possibility of bias arising from different sources within and across reviews.

Potential biases in the overview process

This overview developed and adopted rigorous methods with the aim of reducing the impact of bias contributed by the overview process itself.

A major strength of this overview, compared with other previous overviews in the area, is the comprehensiveness of our searches. We used handsearching, performed by two researchers working independently, to identify all reviews potentially relevant to consumers' use of medicines. Since the research evidence on interventions directed to consumers is typically not well indexed, database searches do not reliably identify all relevant reviews. We believe that handsearching is therefore essential to ensure identification of a comprehensive set of reviews in this area.

We selected reviews for inclusion in this overview based on relevance. The dataset we have presented is, therefore, only a selection of the total available data. However, we developed and piloted criteria to enable us to select reviews for inclusion in a consistent manner, based on their focus and content; we systematically assessed the quality of reviews using an established and validated tool (Shea 2007); and all review selection and ranking steps were performed by two researchers to maximise consistency of judgement.

A strength of this overview is that we developed an evidence rating scheme to enable consistent judgements to be made and statements formulated about the effects of interventions, across diseases, populations and settings.

We also sought to reduce double counting of the evidence, by excluding from this overview non-Cochrane reviews which overlapped with included Cochrane reviews. However, some individual studies have contributed evidence to multiple included reviews, and this remains a limitation.

Agreements and disagreements with other studies or reviews

This overview is the first synthesis of reviews to take a broad perspective on evidence-based prescribing and medicines use; broader than adherence, but focussed on consumers' medicines use. To our knowledge the overview by Van Dulmen 2007, which focussed on adherence to medical treatments, including appointment keeping, dietary and exercise recommendations, is the most comparable piece of research on review-level evidence in this area. In contrast to Van Dulmen, this overview excluded duplicative reviews and took a more inclusive approach to medicines use interventions. Despite the differences in inclusion and exclusion criteria, both overviews found some evidence for the effectiveness of interventions to improve adherence primarily for supporting behaviour change, such as simplified dosing regimens and reminders. However, the effects of such interventions on other outcomes were not addressed in Van Dulmen's overview. This overview is also more current. For example, we identified and included a further four Cochrane reviews that have been published since 2005 (Halpern 2006; Heneghan 2006a; McIntosh 2006; Rueda 2006) that explicitly focus on adherence to medicines and which would potentially fit Van Dulmen's inclusion criteria; as well as several other DARE reviews focussing on adherence which may also be relevant (Amico 2006; Olthoff 2005; Russell 2006; Van Wijk 2005).

The World Health Organization (WHO) (Chetley 2007) has produced the manual How to improve the use of medicines by consumers. This manual emphasises planning a strategy after determining the reasons for irrational medicines use, using social marketing. In this overview, there were no reviews assessing this process for improving medicines use - which in itself is a strategy. The key messages, however, are similar to this overview. The focus is on empowerment at both an individual and system level, and includes issues such as self-medication and self-management, safe use and misuse. The other focus of the WHO review is on communication as a principal tool by which other strategies are implemented at an individual or system level. Strategies that are used in combination with other strategies, such as by providing information and supporting behaviour change, are promoted as being more effective than single strategies.

The UK's National Institute for Health and Clinical Excellence (NICE) has also recently completed guidance for medicines adherence (NICE 2009). The overarching principles for adherence to medicines in its guidance are patient involvement in decision making, and supporting adherence. Interventions targeted at both professionals and patients were included in the NICE review, in contrast to our overview which focussed on consumer interventions. In addition, reviews and trials not specific to medicines use were also included in their review. For this reason, it is difficult to compare findings. However, the evidence and the recommendations are similar to the underlying concept in this overview, in that the goal of an intervention for consumers is not just adherence to any medicines prescribed by a physician.

Authors' conclusions

Implications for practice

This overview presents current evidence from reviews that have synthesised trials and other studies evaluating the effects of consumers' medicines use interventions. Thirty-seven included reviews contributed evidence about a wide range of interventions, but relatively fewer outcomes were reported, with the exception of adherence. Decision makers who are faced with implementing interventions to improve medicines use can use this overview to inform these decisions, and also to consider the range of different approaches available to improve medicines use by using our taxonomy. Researchers and funders can also use this overview to determine where more research is needed. The limitations of the literature due to the lack of evidence for important outcomes related to medicines use and for important populations, such as people with multimorbidity, should also be considered in decisions for practice and policy.

Some interventions hold promise for improving consumers' medicines use outcomes, including adherence and other key outcomes such as adverse events, knowledge, and clinical outcomes. These include self-monitoring and self-management, simplified dosing and pharmacist-based interventions. Other interventions show promise in relation to adherence but effects were less consistent; these included reminders, education delivered together with self-management skills training, counselling or support, financial incentives and lay health worker interventions. The evidence suggests that some interventions, such as information or education provided alone, are ineffective, while for many others there is insufficient evidence to determine effects on medicines outcomes related to health consumers, providers or systems.

What is clear from this accumulated evidence is that there is not one single approach that appears effective across all clinical situations or for all outcomes. Similarly, not all complex interventions are more effective than simple strategies to improve medicines use.

The evidence assembled within this overview can be used in different ways to inform decision making. It enables identification of interventions that are effective or promising, and which may be suitable for uptake to achieve a range of medicines use goals. It also clearly identifies those interventions for which the evidence indicates ineffectiveness, or uncertainty. This too can inform decision making: for example, if a decision needs to be made from among a range of intervention options, knowing which interventions are ineffective or of uncertain effectiveness may preclude them being implemented. When selecting interventions for an evidence-based approach to medicines practice and policy, attention should therefore be given both to those interventions that appear to work but also those where evidence is lacking, thus narrowing choices to a smaller range of (potentially effective) options. Researchers and funders may use this overview to identify areas where more research is needed.

Implications for research

Despite the large body of evidence assembled in this overview, there are still many areas in which the evidence is uncertain, and a large number of interventions on consumers' medicines use still require rigorous assessment. Gaps in the medicines intervention and outcome taxonomies, and the evidence accumulated within both, indicate a clear need for further high-quality research.

In general terms, outcomes could be better reported in future research on consumers' medicines use, and should be guided by the aims of the interventions under investigation, as well as by the medicines outcome taxonomy presented here. Several populations deserve particular attention in research, including children and young people, carers, and those with multiple co-existent conditions. Further research is also needed on a range of additional interventions to improve safe and effective medicines use by and for consumers. These interventions are comprehensively described in our intervention taxonomy, which focusses on the aim of the intervention (eg supporting behaviour change) as opposed to the type of intervention (eg reminders). This change in perspective could be used to focus systematic reviews and objectives of individual studies in the future.

Acknowledgements

The Canadian Agency for Drugs and Technologies in Health (CADTH) contracted the Cochrane Effective Practice and Organisation of Care (EPOC) Group and the Cochrane Consumers and Communication Review Group to summarise the evidence on interventions for healthcare professionals and healthcare consumers in relation to evidence-based prescribing. This overview presents results from part of the project that focusses on interventions targeted to consumers.

We thank the Cochrane Consumers and Communication Review Group Managing Editor Dr. Megan Prictor, and Editor Prof. Sandy Oliver for their assistance with this review.

Appendices

Appendix 1. Medicines outcome taxonomy

MAJOR OUTCOME CATEGORYEXAMPLES OF OUTCOMES
Consumer outcomes
Consultation and communication by consumerCommunication aides (eg summaries, recordings, internet), communication enhancement (eg improved communication with provider)
Knowledge and understandingInformation access and use, knowledge acquisition (level, change in levels, family members’/ carers’ knowledge, knowledge about expected and undesired effects of treatment, knowledge of risk/ accurate knowledge of risk, changes to beliefs about disease/treatment), knowledge retention
Consumer involvement in care processDecision making (decision making process, decision support provided, decisional conflict, decisions made, patient and carer preferences, agreement between personal values and choices/ outcomes), availability of patient-held information
Consumer evaluation of careConsumer-professional interactions (experience of), perceptions and ratings of care/ interventions/ treatment, satisfaction (with information provided, with decision made, with care, carer satisfaction, sense of control)
Support and consumer skills acquisitionPractical support (eg technical aids), psychosocial support (eg self-help groups, peer or family support), self care skills, communication skills, activities of daily living skills
Health status and wellbeingClinical and physiological outcomes, ie physical health (patient or carer), psychological health (patient or carer), psychosocial outcomes (quality of life, personal cost of illness, personal cost of medicines)
Health behaviourRelated to attitudes towards the condition/ treatment
Consumer adverse eventsComplications, morbidity/ mortality, relapse, side effects of medicines
System benefitsHospital and specific service use, adverse events (system – complaints and litigation, reporting of adverse events), costs
Provider outcomes
Consultation and communication by providerPractice style – level of patient-centred care
Knowledge and understandingAttitudes towards treatments
Evaluation of careSatisfaction, anxiety of professional

Appendix 2. Evidence rating scheme

Summary statement: Sufficient evidence from studies

Translation:Evidence to make a decision about the effect of the intervention(s) in relation to a specific outcome(s). This includes evidence of an effect in terms of (i) benefit or (ii) harm. Statistically significant results are considered to represent sufficient evidence on which to base decisions, but a judgement of 'sufficient evidence' is also made based on the number of studies/ participants included in the analysis for a particular outcome. A rating of 'sufficient evidence' is often based on meta-analysis producing a statistically significant pooled result that is based on a large number of included studies/ participants. This judgement may also be made based on the number of studies and/or study participants showing a statistically significant result - for example (in a narrative synthesis) a result where 12 studies of a total of 14 for a specific outcome showed a statistically significant effect of an intervention would be considered to represent 'sufficient evidence.'

Summary statement: Some evidence from studies

Translation: Less conclusive evidence to make a decision about the effects of a particular intervention(s) in relation to a specific outcome(s).This may be based on narrative syntheses of review results. In this case, the result is qualified according to the findings of the review - for example, 'some evidence (5 studies of 9) reported a positive effect of .....' {This would be based on a more equivocal set of results than those obtained for 'sufficient evidence' above. For example, while 12/14 statistically significant studies would be classed as 'sufficient evidence', 5/9 statistically significant studies is more equivocal and would be classed as 'some evidence.'} This may also be based on a statistically significant result obtained in a small number of studies; a statistically significant result obtained from studies with a small number of participants; or a statistically significant result obtained from studies of low quality.

Summary statement: Insufficient evidence from studies

Translation: Not enough evidence to support decisions about the effects of the intervention(s) on the basis of the included studies. This should be interpreted as 'no evidence of effect', rather than 'evidence of no effect'. Statistically non-significant results are considered to represent insufficient evidence. Where the number of studies is small, and/or the number of participants included in the studies is small, 'insufficient evidence' might reflect underpowering of the included studies to be able to detect an effect of the intervention. Where the number of studies is large, and/or the number of participants included in these studies is large, 'insufficient evidence' may reflect underlying ineffectiveness of the intervention to affect the outcomes being examined.

Summary statement: Insufficient evidence to determine from studies

Translation: Not enough evidence to be able to determine whether an intervention is effective or not on the basis of the included studies. This statement is about reporting gaps in the evidence (ie where there are too few studies to be able to determine effects), rather than the situation of the summary statement above, which is about ineffectiveness (eg several studies reporting a statistically non-significant result). It is likely to arise when the numbers of included studies is very small.

Appendix 3. Results of reviews - Summary of effects of interventions

The effects of interventions are presented below by intervention category, based on the aims of the interventions and our taxonomy's organisation. Many reviews evaluated interventions which fall into multiple categories and so contributed evidence to determining the effects of different interventions.

Refer to Additional Table 9 for the results of each review presented individually as: quantitative results; a narrative summary of the results; and effectiveness statements.

Providing information or education

Several reviews evaluated interventions to enable people to know about their treatments and/or health, by providing information or by educating consumers about medicines. Few of these reviews separated out patient information or education interventions, as many interventions were multi-faceted and included an information or education component. Most reviews also did not pool results over studies but instead vote counted positive and negative results across studies.

Most reviews examining education showed no differences in adherence or in clinical outcomes: 1 of 2 studies improved adherence in the review by Schedlbauer 2004; 1 of 6 studies in Schroeder 2004; 1 of 4 studies in Vergouwen 2003; and 3 of 8 studies in Vermeire 2005. A review on written medicines information (Nicolson 2009) also found adherence to medicines instructions improved in only a minority of studies (2 of 6 studies), although half of studies showed improvements in knowledge (6 of 12 studies) and recall of side effects (3 of 6 interventions).

A review of 39 studies (Zygmunt 2002) found that psycho-education interventions (including dissemination of knowledge about disease, treatment and medicines) delivered in different ways were ineffective when compared with usual care. However, one review (Vermeire 2005) of education interventions for people with type II diabetes did show an improvement in glycosylated haemoglobin levels (an indicator of long-term glucose control), although the review authors questioned the clinical importance of this improvement because it was small in size. Patient education and information also resulted in increased immunisation rates, in 3 of 5 studies of mass mailings (Maglione 2002), and an odds ratio (OR) of 1.29 (95% confidence Interval (CI): 1.14 to 1.45) from 22 studies (Stone 2002). Few reviews reported other outcomes.  

In reviews which included interventions with an information and education component, results were mixed. In a large review of many interventions across health conditions, less than half of the studies (41 of 93 interventions) improved adherence, or clinical outcomes (29 of 93 interventions) (Haynes 2008). In a review of complex interventions in older adults, about half of studies (20 of 41) improved adherence (Russell 2006). A minority of studies in some reviews improved adherence: 8 of 18 studies for hypertension (Schroeder 2004); and 1 of 6 studies for counselling and education for contraceptive use (Halpern 2006). This is in contrast to a review in which education combined with self-management skills training for antiretroviral therapy improved adherence in the majority (6 of 8) of studies (Rueda 2006), and a meta-analysis of 26 interventions including an education component that showed a small improvement in adherence to antiretroviral therapy (standardised mean increase (standardised MI) = 0.35 (95% CI 0.20 to 0.51)) (Amico 2006). Lewin 2005 also showed better immunisation rates in three studies with lay health workers providing interventions which included education ((relative risk (RR) = 1.30 (95% CI 1.14 to 1.48)). A single study in Olthoff 2005 reported improved adherence with education and training for glaucoma. In addition, single studies in a review by Stevenson 2004 of interventions combining education and communication training, counselling or tailoring of medicines found improved knowledge and adherence, but not clinical outcomes.  

Education and/or counselling delivered by pharmacists, often as part of a more comprehensive package of care, similarly showed mixed results for adherence and clinical outcomes . Adherence to chronic medicines for different diseases improved in a minority of studies (6 of 17) with community pharmacists (Van Wijk 2005), but effects were mixed in populations with heart failure alone (Koshman 2008). In contrast, adherence improved in a majority of studies (4 of 6) involving both community and hospital pharmacists (Morrison 2001). In Beney 2000, looking at the effects of outpatient pharmacists, the five studies in which interventions were primarily focussed on duration showed mixed results for adherence. Overall, however, the studies including an educational component in Beney 2000 showed improvement in clinical outcomes (6 of 8 studies) such as blood glucose levels and costs (3 of 3 studies). They showed no change in quality of life (1 study) or adverse effects (1 study).

Facilitating communication and/or decision making

Few reviews focussed on improving communication skills and/or decision making about medicines. Therefore this section reports the results from one key review, however noting that this review included several before and after studies.

Stevenson 2004 assessed interventions promoting communication between patients and healthcare professionals. The review divided studies into three groups according to whether the intervention was designed to improve communication between patients and doctors, patients and pharmacists, or patients and nurses or medical assistants. 

  • Doctor-patient communication (5 studies): There were four studies in communication skills training. One study targeted patients and compared it to medicines education and found it improved medicines knowledge, question asking, and question asking skill but not clinical outcomes. Three studies targeted doctors in which physician communication outcomes, such as addressing patient fears (1 study of 1) and asking patients to repeat instructions (2 of 3 studies) significantly increased, and patient medical recall was increased (1 study of 1). In another study, fact sheets with counselling by doctors increased patient medicines knowledge compared to fact sheets alone.

  • Pharmacist-patient communication (6 studies): One study evaluated communication skills training targeted to pharmacists and found patients were more satisfied with pharmacist time and answering  their questions. One study evaluated a mass media campaign targeting patients in which the number of questions asked did not increase, but information was more tailored by pharmacists. Written prompts used by patients in one study did not increase questions asked. Prompts to patients to write questions for pharmacists did increase questions asked, but not adherence or patient recall. Three studies changed pharmacist visits (clinic or home) which improved satisfaction and, in single studies, decreased medicines problems and costs, but effects were mixed for adherence and clinical outcomes.

  • Nurses or medical assistants-patient communication (5 studies): Three studies assessed provision of face-to-face patient education/counselling, with single studies reporting significantly increased adherence and increased discussions with doctors about medicines, but no change to barriers to adherence. Two studies evaluated telephone contact to discuss medical problems: one study found no difference in reporting of adverse effects or in adherence; the other study found patients discussed more issues on the call and found the calls useful.

There were also other reviews which included numerous strategies that may have promoted communication and decision making. General results from those reviews are as follows.

Cross-disease reviews of interventions without a specific focus on communication and/or facilitating decision making reported mixed effects for various outcomes. Similar to Stevenson 2004, Roughead 2005 evaluated studies which changed the interactions between pharmacists and patients, and most studies improved knowledge (4 studies of 6) and medicines use (6 studies of 9), but only a minority of studies improved adherence (2 studies of 8). In addition, one large review of 93 interventions aiming specifically to improve adherence (Haynes 2008) reported mixed effects on both adherence and clinical outcomes.

Disease-specific reviews, where interventions included a communication or decision-making support component, were largely inconclusive. Reviews on compliance therapy in schizophrenia (McIntosh 2006) and written action plans for asthma in children (Bhogal 2006) showed no significant or consistent effects on adherence or clinical outcomes. In other reviews, adherence was improved in a minority of studies on structured counselling in contraceptive use (Halpern 2006) or in single or multimodal psychosocial interventions in schizophrenia (Zygmunt 2002). Delaying antibiotic prescriptions (leaving the decision to initiate therapy up to patients) reduced antibiotic use, but also significantly reduced patient satisfaction. Delaying prescriptions also resulted in unchanged or significantly worse clinical outcomes, and mixed adverse effects (Spurling 2007).

Acquiring skills and competencies

Few reviews focussed on interventions to train or assist consumers to develop skills around medicines monitoring, management and/or use. Most of these reviews showed mixed results for adherence and clinical outcomes, but overall improvements in other outcomes related to medicines use, such as knowledge. A review of highly active antiretroviral therapy showed improved adherence improved in most studies (6 studies of 8) and mixed results in clinical outcomes (Rueda 2006). Another review of antiretroviral therapy (Amico 2006) found a small effect from a meta-analysis of 26 interventions to improve adherence (SMD = 0.35 (95% CI 0.20 to 0.51)). In this review, high intensity interventions (eg skills training) were no more effective than low intensity interventions (eg education). In older adults (Russell 2006), self-management of medicines also showed improvement in adherence (2 studies of 2). 

In contrast, in a review of 22 studies where pharmacists provided care in medicines use and management, few studies (2 studies of 8) improved adherence and there were mixed results for clinical outcomes in 16 studies (Roughead 2005). Morrison 2001 found interventions provided by pharmacists improved adherence in most studies (5 studies of 7), but improved clinical outcomes in only a minority (1 study of 5). However, counselling of patients and physicians together improved both adherence and clinical outcomes in half of studies (2 studies of 4). There were also mixed improvements in clinical outcomes in a diabetes education programme, with non-significant decreases in glycosylated haemoglobin (4 studies of 4), and mixed results for other clinical outcomes (Vermeire 2005). In a review of self-monitoring for oral anticoagulation therapy (Heneghan 2006b) clinical outcomes improved in most studies (6 of 11), and adverse events were reduced (such as thromboembolic events: absolute risk reduction (ARR) of 2 out of 100 people from meta-analysis of 13 studies (95% CI 4 to 2 fewer)).

Three reviews also reported other outcomes related to drug use. Roughead 2005 found that a majority of studies showed significant improvements in knowledge (4 of 6 studies) and medicines use (6 of 9 studies). These included improvements following education on techniques for using medicines (eg inhaler use) (2 of 2) and improved risk management (2 of 2). However, there were no differences in quality of life (11 of 16 studies) or adverse effects (3 of 4 studies).  Morrison 2001 on pharmacist counselling and education reported individual studies with significantly improved medicines error identification, possible improvements in knowledge, and correct use of inhaler, and decreased adverse experiences. Heneghan 2006b reported that a significant proportion of people undertaking self-monitoring for oral anticoagulation (mean 22%) were unable to complete treatment and dropped out.

Bhogal 2006's review compared asthma management in children who used written action plans based on symptoms, or based on clinical measures (Bhogal 2006). Improvements in clinical outcomes and other outcomes were inconsistent between the two types of action plans, but significantly more children (but not parents) intended to use the symptom-based written action plans.

Supporting behaviour change

Several reviews assessed strategies to promote or support medicine-related behaviour change, including changes to address under-use, overuse or misuse of medicines. One large review (93 interventions) across diseases reported mixed effects of interventions to support behaviour change for adherence and clinical outcomes (Haynes 2008). Most of the effective interventions for short-term treatment were simple, while most of the effective interventions for long-term treatment were complex.

Disease-related reviews

When considering all reviews across diseases using simple or complex interventions, the results were mixed. Various interventions to support behaviour change showed improvements in medicines use, with few clear overall patterns.

  • Simplified dosing regimens (eg decreasing frequency of doses from four to twice daily; changing formulation from tablet to transdermal form) improved adherence in the majority of studies in several reviews (Schedlbauer 2004; Schroeder 2004; Vermeire 2005), including adherence to chronic medicines in older adults (Russell 2006).

  • Self-monitoring and/or management interventions to support behaviour change in people using oral anticoagulation medicines improved clinical outcomes in the majority of studies and decreased treatment-related adverse events (Heneghan 2006b). However, a significant proportion of people self-monitoring was unable to complete treatment, and dropped out. Both studies in Russell 2006 on self-medication management programs for older people improved adherence. Bhogal 2006 found mixed results for clinical outcomes, when comparing different written action plans for asthma. The review found that more children intended to use the plan, and had a lower risk of exacerbations, when symptoms rather than clinical measures were used to guide treatment decisions.

  • Support and education, alone or as part of multifaceted interventions, improved adherence in approximately half of studies (Rueda 2006 and Schroeder 2004), but effects on clinical outcomes were mixed.

In contrast, many reviews reported better adherence and other outcomes in only a minority of studies. These included interventions including patient motivation and support to promote behavior change in hypertension (Schroeder 2004); counselling to support behaviour change in a range of conditions including glaucoma, schizophrenia, or for oral contraception (Halpern 2006, McIntosh 2006, Olthoff 2005, van Eijken 2003 and Zygmunt 2002); and directly observed therapy for tuberculosis in Volmink 2007.

Interventions including education or information to support behaviour change

Results were also mixed for reviews of interventions which included an education or information component to support behaviour change . Adherence improved in a minority of those studies in Schedlbauer 2004, Schroeder 2004, and Vergouwen 2003. There were few consistent effects on knowledge, clinical or other outcomes. In another review, education and facilitation in diabetes significantly improved metabolic control (glycosylated haemoglobin levels), but only a minority of studies improved adherence (Vermeire 2005). Nicolson 2009 also showed that only a minority of studies (2 of 6 studies) providing written information improved adherence with medicines instructions. Half of studies (20 of 41 studies) using a combination of education and counselling among older people improved adherence (Russell 2006). In a single study on glaucoma, education and medicines usage training improved adherence (Olthoff 2005).

Reminders to support behaviour change

Reviews including studies evaluating the use of reminders to support behaviour change also showed mixed effects. Adherence improved by a small amount in a minority of studies (1 of 3 studies) in Schedlbauer 2004 and in a review of 24 studies (Amico 2006) (Mean Increase (MI) = 0.35 (95% CI 0.20 to 0.51)). Packaging showed mixed results for adherence and ease of use in long-term medications in Heneghan 2006a, but a majority of studies (3 of 4 studies) increased adherence for malaria in Orton 2005. Alarms (1 study) and compliance devices (2 studies) improved adherence in another review (Olthoff 2005).

Among older people, a majority of cue interventions for behaviour change, half of the studies of cues plus organizers, and a minority of studies of organizers alone, improved adherence (Russell 2006). Reminders were generally successful in improving adherence to immunisation uptake (Jacobson 2005; Stone 2002), however small the effect (Maglione 2002). However in another review on older adults, practical reminder devices such as reminder packaging, pillboxes, organizers, charts or schedules, improved adherence in only 3 of 13 assessed interventions (van Eijken 2003).

Interventions for organisational change

The few reviews which evaluated changing care organisation and delivery to support behaviour change show mixed effects. Most collaborative care studies in primary care, involving multimodal interventions to support behaviour change, improved both adherence (9 of 11 studies) and depressive symptoms (10 of 11 studies; Vergouwen 2003). Lay health workers also improved immunisation uptake (Lewin 2005). However, effects of pharmacy-based education and facilitation (Vermeire 2005) or services provided by community or hospital pharmacists (Beney 2000; Koshman 2008; Morrison 2001) were mixed for adherence and clinical outcomes, but improved for appropriate medicines use and dose (Beney 2000). Few studies (1 of 3) improved adherence with multifaceted generalised interventions including telephone-linked computer system and instructions with reminders for older people (van Eijken 2003).

Support

A number of reviews assessed strategies to assist and encourage consumers to manage their health and medicines use. No reviews included studies of support provided as a single intervention to improve adherence or medicines use. However, two key reviews focussing on support and psychosocial interventions had conflicting results (Vergouwen 2003; Zygmunt 2002); we highlight them here.

In Vergouwen 2003, 9 of 11 studies evaluating collaborative care involving psychological and psychiatric care for people with depression showed improved adherence, and 10 of 11 studies showed improved depression. In comparison Zygmunt 2002 evaluated single or multimodal psychosocial interventions for people with schizophrenia, and found that only a minority of the interventions improved adherence to antipsychotic medicines. Specifically, adherence improved after: individual interventions (2 of 4 studies), group interventions (2 of 4 studies) and family therapy (3 of 12 studies), community based interventions (4 of 10 studies), and mixed interventions and comparisons (5 of 9 studies). However, 5 of the 9 studies which had a specific goal to improve adherence led to improvements. 

The little or no difference in adherence with interventions focussing on support alone is similar to the results of several reviews in which support was provided in combination with other strategies. A meta-analysis of 26 interventions of adherence in antiretroviral therapy found a small positive effect (Amico 2006). One of 6 studies including support in counselling improved adherence to hormonal contraceptives (Halpern 2006). No studies including motivational interviewing improved adherence to antiretroviral therapy (Rueda 2006). One study in McIntosh 2006 providing compliance therapy specifically to improve adherence, found no significant difference in adherence or clinical outcomes, attitudes towards medicines, quality of life, or health services use. 

Mixed results for adherence and other outcomes were found in other reviews. A minority of studies (10 of 24 studies) of interventions including social and family support and counselling for people with hypertension, increased adherence (Schroeder 2004). Van Wijk 2005 found mixed effects for adherence in support interventions which included encouragement, counselling, and problem identification. Russell 2006 found that half of studies (20 of 41 studies) of education and counselling interventions improved adherence. Stevenson 2004, a review that included before-and-after studies, also found mixed results for adherence in studies which promoted contact (face-to-face or by telephone) between patients and nurses or medical assistants, but found increased discussion of medicine issues with doctors. Stevenson also reported a study involving medicines counselling visits between pharmacist and patients, which found mixed results, including a tendency for improved quality of life and knowledge in the intervention group. Before-and-after studies of interactions between pharmacists and patients were also reviewed by Lummis 2006. This review found that overall support and encouragement by pharmacists for the use of patients’ own medicines in hospital improved the identification of medicines errors and reduced costs to hospitals and patients.

More broadly, a large review of adherence-promoting interventions by Haynes 2008, found that most of the effective interventions in long-term treatments were complex. These included interventions such as counselling, reinforcement, family therapy, psychological therapy and crisis intervention. However, both Zygmunt 2002 and Amico 2006 found no relationship between the complexity of an intervention (which included support) and better adherence.

Minimising risks or harms

In a number of reviews, strategies focussed on preventing or managing adverse events or complications; whether for individuals experiencing an emergency event or needing ongoing treatment, or for reducing population-level risks.

There were few specific strategies to minimise risks and harms of treatment or disease at an individual level. Of note are two reviews of interventions which place responsibility for care with the patient. Heneghan 2006b evaluated self-monitoring of clinical measures with oral anticoagulation medicines, with or without self-adjustment of medicines. The majority of studies found fewer adverse events (eg thromboembolic events: absolute risk reduction of 2 fewer people out of 100; 95% CI 2 to 4 people fewer) and improved therapeutic outcomes (6 of 11 studies). However, a significant number of people self-monitoring (on average 22%) were unable to complete treatment, and dropped out. Another review comparing specific monitoring plans showed that risk of asthma exacerbations requiring acute care was lower for symptom monitoring, compared with peak flow monitoring action plans (Bhogal 2006).

A review of 24 studies, including interventions of education or counselling about adverse effects of anti-retroviral medicines, found small improvements to medicines adherence (MI = 0.35 (95% CI 0.20 to 0.51); Amico 2006). The interventions with education or counselling were considered medium-level intensity, and level of intensity was not related to effect size, nor was duration of the intervention. In one large, high-quality review, several studies that examined the effects of telling patients about adverse effects of medicines showed no significant decrease in adherence (Haynes 2008). 

Interventions involving healthcare professional and patient interactions found mixed results for various outcomes. Overall, a review of 22 studies of pharmaceutical care interventions for managing health or medicines-related problems (Roughead 2005) showed that a minority of studies (2 of 8 studies) significantly improved adherence, but a majority showed significant improvements in knowledge (4 of 6 studies), medicines use (6 of 9 studies) and risk management (2 of 2 studies). A minority of included studies showed an improvement in adverse events (1 of 4 studies); and health resource use (2 of 8 studies), but effects on clinical outcomes were mixed. In a review of interventions to improve communication between healthcare professionals and patients (Stevenson 2004), modified pharmacy services and medicines review decreased medicines problems (1 study of 1), but telephone contact by nurse or assistant did not change the number of people reporting adverse events or stopping treatment because of them. Lummis 2006 primarily included before-and-after studies to evaluate the use of patients’ own medicines (POMs) in hospital with pharmacist assessment. Results from included studies showed that the identification of patient medicines errors was increased when POMs were used together with pharmacist assessment. There was no difference in medicine administration errors on wards when POMs were used without pharmacist review. Royal 2006 also found mixed results for pharmacist-led interventions to decrease adverse events and mortality; a minority of studies (4 of 15 studies) reduced hospital admissions, but there were no significant changes to emergency department visits. Finally, Koshman 2008 assessed the effects of community pharmacist-led interventions for people with heart failure, and reported mixed results. There were significantly fewer hospitalisations with pharmacist collaborative care interventions (ARR = 12 fewer people per 100 (95% CI 22 to 1 people fewer)), but not with pharmacist-directed care. Neither intervention type had a significant effect on mortality.

Public health interventions to minimise risks or harms

Reviews showed a range of interventions improved immunisation uptake. A broad overview of interventions including meta-analyses (Stone 2002) showed that organisational change interventions were most effective (10 studies, OR = 16.0 (95% CI 11.2 to 22.8). Also effective were reminders (23 studies, OR = 2.52 (95% CI 2.24 to 2.82), patient financial incentives (8 studies, OR = 3.42 (95% CI 2.89 to 4.06) and education (22 studies, OR = 1.29 (95% CI 1.14 to 1.45). Maglione 2002 reviewed the effect of mass mailings, showing that most of the mass mailing studies found improved immunisation uptake, but effects were small and not clinically significant. More recent reviews showed increased immunisation uptake with lay health worker interventions (Lewin 2005); and with reminder and recall interventions which included person-to-person calls, letters, autodialer computer reminders and patient plus provider reminders, which resulted overall in an 11% absolute increase in immunisation rates (Jacobson 2005). Person-to-person calls were the most effective single intervention. In particular, however, one study showed no effect of reminders in adolescents, and reminders with outreach did not significantly increase immunisation.

One strategy to aid in the eradication of tuberculosis was reviewed. Volmink 2007 found no significant differences in treatment completion or cure of tuberculosis with directly observed therapy (DOT) overall, although there was a small difference in favour of DOT delivered at home compared with self-administration.

Spurling 2007 reviewed delayed prescribing as a strategy to reduce widespread antibiotic resistance. Meta-analysis of 6 studies showed that delayed prescribing reduced antibiotic use (ARR = 64 fewer people out of 100 used antibiotics with delayed antibiotic interventions (95% CI 81 to 38 fewer)), but heterogeneity was high; and 1 of 2 studies showed increased supplementary medicines use. Delayed prescribing also had mixed effects on clinical outcomes and adverse effects.

Improving quality

Several reviews focussed on strategies to improve care coordination or integration, such as substitution or expansion of care or aiming to address barriers to medicines use. These reviews primarily examined changing healthcare professionals' roles and their interactions with consumers (eg pharmacists and nurses), and, to a lesser extent, financial incentives and medicines pricing policies.

Four reviews evaluated the roles of healthcare professionals in various diseases/conditions. Roughead 2005 reviewed 22 studies of pharmaceutical care interventions (consultations between pharmacist and patient), and found that a minority of studies (2 of 8 studies) showed significant improvements in adherence; and there were mixed results for clinical outcomes in 16 studies; and mortality and morbidity in 25 studies. However, a majority showed significant improvements in knowledge (4 of 6 studies) and medicines use (6 of 9 studies), better techniques for using medicines (eg inhaler use) (2 of 2 studies), quality of life (11 of 16 studies), and improved risk management (2 of 2 studies). Beney 2000, assessing the effects of additional outpatient pharmacist services for patients, reported similar findings. This review reported mixed effects for adherence and clinical outcomes. However, rates of appropriate medicines use and appropriate medicines dose were both significantly improved, as were costs of medicines (3 of 3 studies), numbers of medicines (2 of 3 studies), and knowledge of medicines (2 of 3 studies). Adverse effects were unchanged in a single study. Another review by Stevenson 2004 focussing on interventions to improve communication between providers and patients, found mixed results for adherence in studies which promoted contact between patients and nurses or medical assistants either face to face or by telephone. It did find better discussion of medicines issues and a tendency for better quality of life and knowledge. Two studies changed pharmacist visits (clinic or home) which improved satisfaction and decreased medicines problems as well as costs. Effects were mixed for adherence and clinical outcomes. One further review looking across diseases/conditions assessed interventions involving different health professionals that aimed to decrease medicines-related adverse events (Royal 2006). In this review, pharmacist-led interventions had mixed effects. Mortality decreased in half of studies (2 of 4 studies), and in a minority of studies there were fewer adverse events (1 of 3 studies) and hospital admissions (4 of 15 studies). There were no changes to emergency department visits, however. Interventions led by primary healthcare professionals or nurses did not consistently change any reported outcomes.

Other reviews evaluated the changing of roles in relation to specific diseases/conditions. Two reviews (Bower 2006; Vergouwen 2003) reported positive effects of collaborative care interventions delivered by various healthcare professionals. Vergouwen 2003 found that the majority of studies improved adherence to antidepressant medicines (9 of 11 studies) and depression (10 of 11 studies). Bower 2006 similarly reported significantly improved adherence (OR = 1.92 (95% CI 1.54 to 2.39)) and significantly lower depressive symptoms (OR = 0.24 (95% CI 0.17 to 0.32). van Eijken 2003 reviewed seven multifaceted interventions which included new services provided by pharmacists and nurses to older adults with various conditions. They found mixed results with 3 of the 7 interventions increasing adherence in older adults. In hypertension (Schroeder 2004), 8 of 18 studies evaluating complex interventions improved adherence (from 5% to 41%); one study of work site care showed better adherence and clinical outcomes; home visits and education improved adherence in 2 studies; a pharmaceutical care model in 2 studies improved adherence with 1 improving hypertension; and rewards in 1 study improved adherence with no effects on clinical outcomes.

In heart failure (Koshman 2008), interventions involving pharmacist-led services had mixed effects overall. Pharmacist-directed care interventions had mixed effects on adherence and improved health-related quality of life in the minority of studies (1 of 6 studies) but had no significant effects on mortality or hospital admissions. However pharmacist collaborative care interventions significantly decreased hospital admissions with an ARR of 12 people fewer per 100 (95% CI 22 to 1 fewer) hospitalised. Mortality was not significantly changed and effects on adherence and health-related quality of life were mixed in single studies.

One review (Lummis 2006), which included before-and-after studies, evaluated the use of patients’ own medicines in hospital (POM). Individual studies showed that the identification of patient medicines errors was increased with POM but there was no difference in medicines administration errors on the wards. Studies also showed costs to hospitals and patients after discharge were reduced with POM (3 of 3 studies), but workload for pharmacists was increased (1 of 1 study) while dispensary staff workload was decreased (2 of 2 studies).

To improve immunisation uptake, Stone 2002 reviewed and meta-analysed the results of various interventions in adults: organisational changes improved adherence the most (OR = 16.0 (95% CI 11.2 to 22.8) from 10 studies). Lewin 2005 investigated the effect of lay health workers to improve knowledge, attitudes or behaviour to increase uptake. From the 3 studies found, there was a significant increase in immunisation uptake in adults and in children (RR = 1.3 (95% CI 1.14 to 1.48)).  Stone 2002 also meta-analysed the effect of patient financial incentives from 8 studies. Immunisation uptake improved (OR = 3.42 (95% CI 2.89 to 4.06)) which was a relatively less effective intervention to improve immunisation uptake than organisational change interventions. Financial incentives, in general, were reviewed by Giuffrida 1997. They also found overall increases in adherence, but most studies did not show significant changes: studies comparing 5 financial interventions, such as rewards, compared to no intervention found increased adherence (2 non-significant); while studies comparing 8 financial interventions to another intervention (eg telephone prompts) showed increases (2 non-significant). 

Three reviews assessed the effects of different pharmaceutical pricing policies (Aaserud 2006; Austvoll-Dahlgren 2008; Maio 2005) which indirectly influence consumers’ medicines use through different pricing structures and/or by altering the financial impact of medicines use. Although we included indirect-to-consumer interventions in this overview, we did not seek to include all reviews which used financial interventions or other indirect strategies to change consumer medicines use. There are likely to be other reviews of these interventions not included in this overview (eg if they were classified as too 'indirect to consumer') and which might contribute evidence on the effects of these interventions. We therefore provide only an overview of results from these reviews on financial interventions here. 

Aaserud 2006 assessed pharmaceutical pricing and purchasing policies, and reported primarily on reference pricing policies (where one reference medicine is chosen from a therapeutically similar group, the price of which is covered; if people opt for a more expensive option they have to pay the price difference). Reference pricing increased the use of specific reference medicines, and decreased the use of cost share medicines, without affecting total reference medicines use, or use of medicines other than those in the reference group. Reference pricing also decreased total medicines expenditure but had mixed effects on healthcare use, including increased emergency visits and hospital admissions through emergency in a minority (1 of 10 interventions) of cases, and significantly increased non-emergency hospital admissions and physician visits (5 of 10 interventions). Austvoll-Dahlgren 2008 assessed cap and co-payment strategies where patients share payment in different ways for their prescription medicines and reported similar effects. Caps, co-payments, co-insurance with ceiling strategies and changes in tiered co-payments, alone or in different combinations, significantly decreased insurance plan overall medicines expenditure and overall prescription medicines use. However, these policies had mixed effects on discretionary and essential medicines use, patterns of healthcare use and patient medicines expenditure. The effects of fixed co-payments, fixed co-payments plus co-insurance with or without ceilings, or of index pricing policies were unable to be determined. Pharmacy Utilisation Management (PUM) strategies (cost-sharing and administrative restriction processes) were assessed in another review in older adult populations (Maio 2005). Although PUM strategies involving caps or co-payments reduced prescription medicines use and medicines costs, they also both increased health service use and may reduce health status among older adults. PUM strategies involving formularies may reduce costs without significant effects on health system use or health status.

Consumer system participation

We found no reviews addressing interventions for consumer system participation in medicines-related activities, such as in research planning, or formulary or policy decisions.

What's new

Last assessed as up-to-date: 3 March 2010.

DateEventDescription
13 December 2011AmendedDue to continuing problems with display of this overview in PDF format, we have now split the Characteristics of Included Studies, and Results by Individual Review, into 5 tables each (Parts A to E).

History

Protocol first published: Issue 2, 2009
Review first published: Issue 5, 2011

DateEventDescription
21 September 2011AmendedWe split each of the large tables on Characteristics of included studies, and Results by individual review, into three parts, to improve their presentation.

Contributions of authors

Rebecca Ryan contributed to: developing the project plan, the overview methods and evidence rating scheme; developing the selection criteria, the taxonomies for interventions and outcomes; development and writing of the protocol; and conduct and writing of the overview.

Nancy Santesso contributed to: developing the project plan, the overview methods and evidence rating scheme; developing the selection criteria, the taxonomies for interventions and outcomes; development and writing of the protocol; and conduct and writing of the overview.

Sophie Hill contributed to: developing the project plan, the overview methods and evidence rating scheme; developing the selection criteria, the taxonomies for interventions and outcomes; development and writing of the protocol; and drafts of the overview.

Caroline Kaufman contributed to: drafts and writing of the protocol.

Dianne Lowe contributed to: conduct and writing of the overview.

Jeremy Grimshaw contributed to: conceived the project, contributed to the project planning and development, and development and writing of the protocol and drafts of the overview.

Declarations of interest

None known

Sources of support

Internal sources

  • No sources of support supplied

External sources

  • Canadian Agency for Drugs and Technologies in Health, Canada.

    Supported the development and updating of the Rx for Change database, the consumer arm of which contributed to this overview.

Notes

The protocol for this overview was originally published in the standard format of a Cochrane review protocol, with the following citation:

  • Ryan R, Santesso N, Hill S, Kaufman C, Grimshaw J. Consumer-oriented interventions for evidence-based prescribing and medicine use: an overview of Cochrane reviews (Protocol). Cochrane Database of Systematic Reviews 2008, Issue 4. Art. No.: CD007508. DOI: 10.1002/14651858.CD007508.

On issue 2 2009 of The Cochrane Library the Cochrane Consumers and Communication Review Group withdrew the above protocol and republished it in 'overview' format with the citation below. The contents of the protocol were unchanged.

  • Ryan R, Santesso N, Hill S, Kaufman C, Grimshaw J. Consumer-oriented interventions for evidence-based prescribing and medicine use: an overview of Cochrane reviews (Protocol). Cochrane Database of Systematic Reviews 2009, Issue 2. Art. No.: CD007768. DOI: 10.1002/14651858.CD007768.

Ancillary