Deprescribing medications for older adults in the primary care context: A mixed studies review

Abstract Aims This review investigates the factors that influence deprescribing of medications in primary care from the perspective of general practitioners (GPs) and community‐living older adults. Methods A mixed studies review structure was adopted searching Scopus, CINAHL, PsychINFO, ProQuest, and PubMed from January 2000 to December 2017. A manual search of reference lists was also conducted. Studies were included if they were original research available in English and explored general deprescribing rather than deprescribing of a specific class of medications. The Mixed Methods Assessment Tool was used to assess the quality of studies, and content analysis generated common categories across studies. Results Thirty‐eight articles were included, and 7 key categories were identified. The review found that the factors that influence deprescribing are similar across and within health systems and mostly act as barriers. These factors remained unchanged across the review period. The structural organisation of health systems remains poorly suited to facilitate deprescribing. Individual knowledge gaps of both GPs and older adults influence practices and attitudes towards deprescribing, and significant communication gaps occur between GPs and specialists and between GPs and older adults. As a result, deprescribing decision making is characterised by uncertainty, and deprescribing is often considered only when medication problems have already arisen. Trust plays a complex role, acting as both a barrier and facilitator of deprescribing. Conclusions Deprescribing is influenced by many factors. Despite recent interest, little change has occurred. Multilevel strategies aimed at reforming aspects of the health system and managing uncertainty at the practice and individual level, notably reducing knowledge limitations and closing communications gaps, may achieve change.

or more concurrent medications, may be appropriate in some circumstances. 6 However, it also places older adults at a potentially higher risk of adverse drug reactions, 7 resulting in increased personal and health care costs, hospitalisations, poor health outcomes, 8 and/or increased mortality rates. 9 This level of risk is considerable 10 and as such, warrants further attention as a public health 11 and ethical issue. 12 Medications are rarely indicated for lifelong usage, as many factors may change during the course of treatment. 13 This is due, in part, to the complexity of managing multiple morbidities experienced by many older adults, the lack of evidence available to inform decisions and changes in health, frailty, and, often, autonomy. 14 As a result, older adults and their general practitioners (GPs) face particular challenges when deciding on the most appropriate treatment regimens.
Deprescribing has been suggested as one intervention to reduce inappropriate polypharmacy. 7 The term deprescribing was first used by Woodward 15 and is defined as a systematic process supervised by a medical professional to reduce or discontinue long-term medications. 16 Deprescribing is indicated where the existing or potential harms outweigh existing or potential benefits of a particular medication/medications. Deprescribing may be relevant at any point in the life course, although it is most often considered in the context of medication use for older adults who, as a group, are growing exponentially worldwide. 17 Ideally, the process of deprescribing takes into account changes in the context of an individual's treatment goals, their current level of functioning, life expectancy, values, and preferences. 18 However, it is not a process that is routinely considered in primary care. 7 This is despite evidence of the usefulness of deprescribing to address polypharmacy and reduce mortality, 19 and an increased awareness of deprescribing at both international 20 and at national levels. 6,21,22 Three previous literature reviews in this research area are available. [23][24][25] One examined the barriers and facilitators of deprescribing from the perspective of patients, using a narrative synthesis of qualitative, quantitative, and mixed methods studies. 25 Anderson et al, 23 on the other hand, sought to understand barriers and facilitators from the perspective of the prescriber, conducting a synthesis of qualitative studies. These initial reviews by Reeve et al 25  The purpose of the current review is to build on Bokhof and Junius-Walker 24 work by investigating the factors that influence deprescribing from the perspective of both GPs and adults aged 65 years or older.
Independent, community-living older adults are the focus of this review, as the majority of this group retain autonomy and are capable of being responsible for their own health decisions, 17 hence the importance of considering deprescribing from their perspective. Given the increasing evidence that is now available on this topic, a review that incorporates all study types is warranted. Interventions to promote deprescribing will require change on the part of both prescribers and older adults, so it is important to consider how their views interact to create the context where deprescribing discussions can take place.

| METHODS
A mixed studies review methodology was used as a guide to explore the factors that influence deprescribing. 27,28 2.1 | Eligibility criteria

| Inclusion criteria
Full text primary research articles were included that were available in English, published between January 2000 and December 2017, and that investigated deprescribing or medication cessation/discontinuation in the primary care context for older adults, living independently in the community. General deprescribing was targeted. Articles were also included where deprescribing was mentioned as a mechanism to reduce polypharmacy or potentially inappropriate medications or in the context of the management of multiple morbidities.

| Exclusion criteria
Deprescribing of specific medications or medication classes were not included, as unique factors related to individual medication use, such as specific withdrawal issues, might have influenced deprescribing practices and attitudes in these instances. Articles discussing deprescribing within long-term residential aged care facilities were excluded, as the nature of such care changes the relationship between primary care providers and older adults, limiting the ability of older adults to act autonomously. 29 Similarly, articles discussing deprescribing within acute care hospital settings were excluded, as it is less likely that there is an established, ongoing relationship between the prescriber and older adult in these settings. Articles were also excluded if they only related to discontinuing medications during palliative stages of care, as the context of medication deprescribing is markedly different in these scenarios. 30 • deprescribe* OR discontin* OR cessation OR cease OR withdraw*OR stop* OR reduc* OR optim* • polypharmacy OR medication OR medicines OR prescribing OR "prescription drug" An example search strategy is detailed in Appendix S1.
Search terms were applied to abstracts, keywords, and titles. In addition, a secondary search of reference lists of relevant articles was conducted, to check for other potentially eligible articles. The search was conducted by the main author, following refinement of the search terms under the guidance of a university librarian.

| Identifying and selecting relevant articles
After duplicate citations were removed, one reviewer (RG) screened titles for relevant articles. Abstracts and/or full texts were reviewed if the article met the inclusion criteria. Key data from potential articles were extracted into a table. This was reviewed by all authors to confirm the relevance and appropriateness of each article. As the focus of the review was on autonomous community-living adults, any articles that included community-living older adults with cognitive impairment were excluded.

| Quality assessment of included articles
The quality of the included articles was assessed using the Mixed Methods Appraisal Tool (MMAT). 31 This checklist is suited for use when studies with a range of methodological techniques need to be assessed. The first 2 authors screened the studies to determine their quality rating, applying the scoring system described in Pluye et al, 32 and scores were confirmed by the third author.

| Analytic approach
The findings sections of included studies were analysed using content analysis. 33 Findings were explored across all articles to identify factors influencing deprescribing. The first author became familiar with the content by reading each study multiple times. Unique codes were identified and tabled. The relationship between codes was considered in order to organise the codes into groups. These groupings were discussed by all researchers in order to refine the final categories.
Seven key categories were generated from the data.     Abbreviations: ADEs, adverse drug reactions; CP, consultant pharmacists; EBM, evidence-based medicine; GPs, general practitioners (this includes primary care family physicians); N/A, not applicable; NP, nurse practitioners; QoL, quality of life mix of primary care prescribers including GPs, nurse practitioners, GPs' assistants, consulting pharmacists, and geriatricians (n = 7).

| Quality of the included studies
The quality of the included studies varied (see Table 1). All but one study 44 reached 50% or more in the MMAT appraisal score. The 2 researchers who applied the MMAT came to a consensus, prior to the use of the tool, on the meaning of the criteria applied to the qualitative studies. This is recommended by the MMAT developers. 31 This may mean that we applied the tool differently than others. Few qualitative studies included researcher reflexivity as required by the MMAT. Response rates for the quantitative surveys of GPs were generally low, with no study reaching the required MMAT response rate (≥60%). However, it should be noted that a low response rate is typical when surveying this population 72 and that the responses were adequate to address the descriptive, exploratory aims of the studies. The 2 studies that used a modified Delphi technique were not scored, as the criteria used in the tool were found to be not applicable.

| Findings
Seven key categories were identified that influence deprescribing.
These were the health care system, older adult and GP characteristics, knowledge limitations, beliefs about medication use, GPs' perceptions of older adults, older adults' perception of GPs, and fears regarding deprescribing.

| The health care system
The structures and practices within the health care systems repre- Older adults often need to consult with more than one prescriber in the course of managing multiple morbidities; this may result in an increased risk of poor communication regarding medications between prescribers and confusion about responsibilities. 34,35,38,40,46,50,56,61,62,66,67 General practitioners view this as a potential barrier to enacting deprescribing. They felt that specialist prescribers regarded their treatment as a priority 62,64 and focused on a specific area, resulting in no one taking responsibility overall. 66,67 Also, patients sometimes believed that specialists had more authority than GPs 34,49,59 and faced confusion about which prescriber was responsible for, and authorised to deprescribe. 49 Conversely, GPs and older adults valued the long-term relationships that they developed in the primary care context, and this worked to facilitate deprescribing. Familiarity allowed GPs to build trust and gain an overview of their older patients' preferences, health concerns, and medications. 35,50,53,70 General practitioners generally felt that this positioned them as the gatekeepers or coordinators of their older patients' care. 53,60,66,67 However, this perception was contradicted by another recurrent dialogue where GPs noted that the hierarchy between themselves and their specialist counterparts prevented them from questioning medication prescribing decisions, even if it meant continuing medications with no clear indications or that were potentially inappropriate. 35,38,40,46,50,54,56,62,63 Both older adults and their GPs thought that there was not enough time during consultations to review medications, consider patient preferences, and determine the priorities patients valued. 34,[37][38][39]48,50,53,55,56,60,61,63,64,66,70 Furthermore, GPs were not adequately reimbursed for more complex consultations, especially with patients with multiple morbidities, 53,67 including undertaking medication reviews, and deprescribing discussions and follow-up. 34,35,46,56 Some GPs avoided addressing complex issues because of their awareness of the lack of time. 60,66 Access to support alongside or within individual practice organisations may influence deprescribing. Changes in practice were suggested, including the ability to consult with pharmacists, 50,51,60,62,63 inviting fellow GPs to conduct independent prescribing reviews, 38 phone consults with geriatricians and/or specialists, 66 and referrals to services able to provide nonpharmacological options. 70 However, it is notable that the Delegation of tasks to other team members was mentioned, 51,70 such as medication reconciliation to nurses 50 or follow-up to pharmacists. 57,58 In practice, this may not happen, as some GPs were uncomfortable with delegation, fearing it would undermine their accountability and authority as a prescriber. 50 Disease-specific treatment guideline use generated much concern among GPs, who thought that they promoted ongoing prescribing and encouraged polypharmacy. 35,37,56,62,66,70 General practitioners noted that guidelines did not take into account treatment decisions in the context of multiple morbidities or provide guidance on the appropriate circumstances in which to discontinue medications. 37,46,47,53,56,62,64,67 Additionally, GPs felt unsure of the usefulness of guidelines, because they were based on clinical trials that rarely included older adults, especially those with multiple morbidities, 37,46,56 and did not take into account outcomes valued by the patient. 46 The way guidelines were used varied among GPs, with those in the study of Sinnott et al 64 prioritising the management of one disease over another, whereas those in another study 62 seemed to apply guidelines one after another, without ranking which treatment was most essential. Some GPs were less concerned about following guidelines when prescribing for older patients, preferring to prioritise quality of life, 53,63 while others thought that the use of guidelines would ensure best practice. 46 General practitioners were hoping for more useful tools to help them rank the treatment of various diseases, 67 recognise potentially inappropriate medications, 39 or guide deprescribing of specific classes of medication. 44

| Older adult and GP characteristics
General practitioners' approaches to the management of their older adults' medications varied. 46 Concerns about polypharmacy use in their older adult patients were frequently raised by health care providers. 34,35,40,50,53,56,62,66,67 However, their views of what constituted polypharmacy varied, 50,56 and this was often assessed in an individual, case-by-case manner. 34,50,56 Increased pill burden or the number of medications was a factor in triggering a medication review with a view to deprescribing. 48 Some prescribers in another study thought that polypharmacy should be framed as a risk to patients who are aging in the same way as they would frame a discussion about the risk of other health problems such as stroke. 34 Older adults varied in their interest for more information about their medications and involvement in decision making. 61,71 For example, raising the topic of discontinuing a medication was initiated by only 18% of older adults in a Norwegian study, 68 while 55% had done so in a US study 51 and 42% in a Canadian study. 69 The Canadian study found that initiating a deprescribing conversation was more likely in older adults who searched for information about and had an awareness of medication harms. 69 In contrast, some older adults felt fearful about expressing their medication preferences and did not share their experience of trialling deprescribing themselves until after the fact. 42,61,71

| Knowledge limitations
Knowledge limitations were commonly cited as a major barrier. General practitioners reported that knowledge gaps, particularly in their pharmacological knowledge, negatively impacted their confidence and willingness to deprescribe. 34,35,50,60,62,66 They were unsure of the potential interactions between multiple medications, 35,56,62 unsure of the ongoing benefit of long-term medication use (eg, preventives), 43,56,62 and unsure of the outcome of deprescribing. 34,37 In practice, then, GPs make assessments of the potential harms or benefits of medication based on the individual needs of their patients. 53,54 Additionally, GPs lacked information about nonpharmacological options 39 and how to develop, implement, and monitor a deprescribing plan. 44 Furthermore, GPs in the study by Riordan et al 60 noted that they were still influenced by pharmaceutical company salespeople, even though they perceive their information to be biased.
An incomplete clinical picture regarding their older patients was another knowledge limitation that negatively impacted GPs' capability to deprescribe. This may result from patients not communicating important information regarding their medications. 35,37,50,56,61 Furthermore, other prescribers, such as medical specialists, may not effectively communicate all their treatment plans (eg, medication indications and/ or duration of treatment) to the GP. 34,35,38,40,46,48,50,51,70 Older adults had limited knowledge about their medications. The study by Weir et al 71 found that this affected their confidence to initiate deprescribing discussions. The reasons for and the potential harms of their medications were not always known, 71 or the ongoing need for some medications was not clear. 55 Patients assumed, as their doctor had more medical knowledge than they did, that it must be appropriate to continue their medication, otherwise, their doctor would stop writing repeat prescriptions. 59 Poor health literacy may also be a factor that influences deprescribing decision making. 41 General practitioners in Clyne et al 39 65 and 55% in an Australian study 58 having experienced deprescribing. Knowledge of the process of deprescribing was limited, for example, with regard to tapering dosages or trialling deprescribing 59,68 or with regard to the potential reduction in risks that deprescribing offers. 45

| Beliefs about medication use
General practitioners generally held positive beliefs about medications.
Their previous experiences of the clinical usefulness of medications and their belief that they generally cause few serious side effects resulted in them favouring continued prescribing over deprescribing. 37,43 This bias towards prescribing was further promoted by the uncertain outcome of deprescribing. 34,70 In comparison, older adults hold conflicting ideas around medication use, often expressing concurrent positive and negative beliefs. 36,41 These conflicting beliefs are clearly illustrated in the 3 quantitative studies that used the Patient's Attitudes Toward Deprescribing survey. Reeve et al, 58 Sirois et al, 65  Older adult responses in many of the qualitative studies elaborate on the reasons for their conflicting beliefs. Medications were valued, as they were perceived to extend life and improve well-being, and their use was to be expected during older age. 71 Older adults were more likely to believe that their medications were necessary following the testimony of their GP, 59 recall of the usefulness of medications for family members and friends, 71 ongoing symptom relief, 36,41,71 or avoidance of preventable health issues. 55 Concurrently, older adults expressed a strong dislike of using medications long-term. They prefer to take as few as possible. 42,49,55,68,71 Sometimes, nonpharmacological options were followed in order to stop medications. 42 From a practical point of view, taking multiple medications was perceived to be a burden by some, 36,71 and notably, in the US health system, was costly. 42 When taking into account competing outcomes, older adults valued ongoing quality of life more than extending life expectancy, suggesting that, if the side effects from a particular medication were too significant, they may consider discontinuing that medication. 45,59

| GP perceptions of older adults
General practitioner perceptions of their older patients influenced their willingness to consider raising the topic of deprescribing. General practitioners perceived their older patients to be generally resistant to change and that they would be unlikely to accept their advice to deprescribe, 35,37,43,50,62 especially if they suggested stopping a medication which the older adult perceived was giving them symptom relief. 37 Alongside this, some believed that older adults themselves had no problem with polypharmacy. 62 Explanation of potential risks and uncertainties was seen as being particularly hard. 47,64 General practitioners in the study by Schuling et al 62 noted that explaining deprescribing to their older patients was made more difficult because of their patients' age and sometimes their poor education. Another study, however, also of Dutch GPs, noted that some believed that even very old people were capable of entering into a shared decision-making process. 53 General practitioners felt pressured to meet their older adults' (and/or family members') expectations to prescribe medications, [37][38][39]54,56,60,70 although some noted that with careful explanation, they may accept the suggestion of alternative treatments. [37][38][39] General practitioners in the study by Wallis et al 70 observed that it is important to remember that patients are not coming to an appointment expecting a discussion about deprescribing.

| Older adults' perceptions of their GP
Trust was an important factor mentioned in multiple studies. Older adults' hypothetical interest in deprescribing was associated with a higher physician trust score. 57,58 However, in practice, those who reported higher trust were less likely to have experienced deprescribing. 52 Older adults reported that their trust in the prescribing practices of their GP was based on the perceived clinical knowledge of their GP, a belief that their GP would make decisions with their best interests in mind, and on the strength of the relationship established between themselves and their GP based on mutual respect, good communication, and knowledge of their preferences. 36,41,49,55,59,61,68,71 Sometimes, this level of trust meant that they did not ask for important medication information. 61 Some older adults did qualify their trust, suggesting that they needed to find information for themselves in order to maintain responsibility for their own health. 55,49,71 Others remarked on the paternalistic nature of their relationship with their GP, although they were generally accepting of this, that it was wise not to argue, and preferred to follow their doctor's advice. 41,61,71 Weir et al 71 compared attitudes of older adults across 3 groups and noted that those who were frailer and/or lacked an understanding of their medications were happy to abdicate decision making about their medications, including deprescribing, to their doctor.
Both GPs and older adults recognised that trust could be undermined when different prescribers gave conflicting advice about deprescribing. 38,49,55,62,68 Finally, others, in the study by Moen et al, 55 noted a general distrust of the health system rather than of a specific prescriber and were sceptical about the ongoing influence of pharmaceutical companies on prescribers.

| Fears
Both older adults and GPs feared the potential for unfavourable outcomes from deprescribing, such as a return of symptoms, withdrawal effects, or previously avoided serious events such as stroke, occur-

| Knowledge limitations
General practitioners noted that they lacked pharmacological knowledge in the context of treating older, multimorbid adults. This suggests that medical curricula should be revised to include more specific geriatric pharmacology and deprescribing education, 75 together with ongoing professional development after graduation. Pharmacists, located either in the community or within GP practices, can also supply information.
Those located within GP practices may be better placed to collaborate and meet the immediate information needs of GPs. 76 The review found that the evidence available to guide GPs when

| Communication gaps
The and Junius-Walker 24 identified trust as being an important characteristic that was necessary to facilitate deprescribing, but the role of trust may be more complex than this suggests. In some studies, trust is associated with a willingness to consider deprescribing, suggesting that it may facilitate acceptance of a prescriber's recommendation to deprescribe, even in the face of uncertainty. 57,58 However, other research demonstrates that trust can also act as a barrier to deprescribing. For example, trust was used by patients to explain why they unquestioningly accepted their medication regimen, resulting in lower reports of deprescribing. 52  Currently, GPs may receive little information about specialist consultation outcome/s. Furthermore, GPs sometimes continued medications they considered inappropriate because they were not willing to challenge their specialist counterparts, and some older adults believed that only their specialists had the authority to deprescribe. Despite these challenges, GPs remain well placed to act as the overall coordinators of their older adults' medication regimens as they may be the only prescriber who can see the full picture of what is being taken. This suggests that a collaborative approach with improved communication between specialist and GP prescribers is needed, along with clarification of lines of responsibility.

| Areas for further investigation
Future deprescribing research needs to explore aspects of diversity among older adults. Age was used to define the samples in the reviewed studies; however, chronological age is only loosely related to physiological changes that occur during the aging process, suggesting that this group should not be thought of as homogenous. The majority of older adults continue to live in the community and are a diverse group that contribute to society as mentors, consumers, members of the workforce, caregivers, and innovators, despite the challenges of the aging process. 17 The review highlights that several other areas also require further exploration. Work is required to understand the extent to which health literacy, which is known to be lower in this age group, 81 and socio-economic characteristics influence older adults' decisions. Additionally, research is required to further understand the complex role of trust within long-term doctor-patient relationships, as it applies to deprescribing.

| Limitations
Our review only included studies available in English. Despite the search methodology including multiple databases and a variety of relevant search terms, a significant number of the included studies were identified via hand searches of reference lists of related articles and using citation tracking. This reduces the reproducibility of our methods. Finally, the MMAT used to assess the quality of the included studies is still in development, so the quality scores given should be treated cautiously.

| CONCLUSION
This review investigated the factors that influence deprescribing from the perspective of both GPs and older adults aged 65 years or older and living independently in the community. The review found that these factors mostly act as barriers to prevent deprescribing from entering into discussions during consultations. They have remained static across the review period and are similar across health systems.
To achieve change, multilevel strategies should be prioritised to address structural constraints within health systems and to manage uncertainty at the practice and individual level, reducing knowledge limitations and closing communication gaps.