End of life care for people with alcohol and drug problems: Findings from a Rapid Evidence Assessment

People who use alcohol and other drugs(hereafter “substances”) and who are over the age of 40 are now more likely to die of a non‐drug related cause than people who use substances under the age of 40. This population will therefore potentially need greater access to palliative and end of life care services. Initially, the purpose of this rapid evidence assessment (REA), conducted August 2016–August 2017, was to explore the peer‐reviewed evidence base in relation to end of life care for people with problematic substance use. The following databases were searched using date parameters of 1 January 2004–1 August 2016: Amed, Psycharticles, Ovid, Ageinfo, Medline, Ebscohost, ASSIA, Social Care Online, Web of Knowledge, Web of Science, SSCI, Samsha, NIAAA. Data were extracted using a predefined protocol incorporat‐ ing inclusion and exclusion criteria. Given the dearth of evidence emerging on inter‐ ventions and practice responses to problematic substance use, the inclusion criteria were broadened to include any peer‐reviewed literature focussing on substance use specifically and end of life care. There were 60 papers that met the inclusion criteria. These were quality assessed. Using a textual thematic approach to catego‐ rise findings, papers fell into three broad groups (a) pain management, (b) homeless and marginalised groups, and (c) alcohol‐related papers. In general, this small and diverse literature lacked depth and quality. The papers suggest there are challenges for health and social care professionals in meeting the end of life needs of people who use substances. Addressing issues like safe prescribing for pain management becomes more challenging in the presence of substance use and requires flexible ser‐ vice provision from both alcohol/drug services and end of life care providers. Work is needed to develop models of good practice in working with co‐existing substance use and end of life conditions as well as prevalence studies to provide a wider context for policy development.


| BACKG ROU N D
In the last 20 years there have been numerous changes evident among the population who use substances whether this be alcohol or other drugs. The most important changes appear to be an increase in the number of older drugs users, a subsequent rise in rates of death from non-drug related conditions (Beynon, 2010), an increase in alcohol-related morbidity among older users (Kaplan et al., 2012) and the burgeoning of New Psychoactive Substances (NPS) use. The increase in older drug users may be associated with changes to the treatment management for illicit opiate users in the 1990's (McKeganey, 2006) and, in particular, the expansion and greater availability of methadone prescription (Clausen, 2008;Clausen, Waal, Thoresen, & Gossop, 2009). This increased availability means that increasing numbers of people with current and previous substance use have better survival rates and are more likely to die from the same chronic conditions that affect the general population (Corkery, 2008;Beynon et al., 2010). It may be that some of this population are using substances for longer or commencing drug use later in life but, whatever the cause, there has been an increase in older drug users accessing drug treatment services in the UK (Beynon, 2010;Beynon et al., 2010).
Before methadone treatment programmes were introduced, people who used substances had a greater risk of dying from overdoses (deliberate or accidental), as well as accidents, violence and disease.
These could be related to the substance itself or its route of administration, such as liver disease, HIV (Human Immunodeficiency Virus) and bacterial infections (Gibson et al., 2008). The adoption of evidence-based interventions like supervised injecting facilities, needle exchange programmes and outreach programmes for illicit drug users is considered to be an effective overarching policy approach for reducing harm (Ritter & Cameron, 2006). Harm reduction policies and related treatment approaches have led to older substance users (over the age of 40) now being more likely to die of a non-drug related cause than people using substances under the age of 40 (Benyon et al., 2010;Stenbacka et al., 2008). However, people using substances are still more likely to die at an earlier age than the general population and have patterns of disease and morbidity that reflect the impact of substance use or the traumatic life experiences more frequently encountered in this group (Beynon, Roe, Duffy, & Pickering, 2009;Beynon et al., 2010). In addition, there are higher rates of alcoholrelated morbidity and mortality associated with chronic and acute alcohol problems (Chang, Kreis, Wong, Simpson, & Guymer, 2008;Shield, Parry, & Rehm, 2014;Taylor et al., 2010). This highlights a more nuanced picture of alcohol-related harm given that the highest consumption is among more affluent groups of the population (Office for National Statistics, 2017) but the highest rates of alcoholrelated problems are among the least affluent (Erskine, Maheswaran, Pearson, & Gleeson, 2010).
The final set of changes is the increase in the use of NPS. Some of the more immediate consequences of NPS use are evident in the prisons system and in admissions to hospital accident and emergency departments for acute intoxication (Liakoni, Dolder, Rentsch, & Liechti, 2016;Ralphs, Williams, Ashew, & Norton, 2017). These harmful consequences are particularly associated with people who are homeless (Henshall et al., 2018). The impact of NPS use on mortality and morbidity, both medium and long-term has yet to be evaluated.
While there has been an improvement in life expectancy within the general population, it has been associated with more years living with poor health or disability at the end of life (Bell & Marmot, 2017). This has subsequently increased the anticipated number of deaths that are likely to need palliative and end of life care (Etkind et al., 2017). There appears to be only fragmented evidence relating to (a) the extent and nature of the care needs by people using substances and (b) the challenges services will face in supporting people with problematic substance use at the end of life. This Rapid Evidence Assessment (REA) aimed to explore what is already known about responses to end of life care need for people using substance and identify gaps in the evidence base. It forms one part of a wider programme of research on end of life care for people with problematic substance use (Galvani, Tetley, et al., 2016). What is known about this topic • There is an increase in older substance users who will require end of life care.

| Conceptual framework
• Meeting end of life needs for this group of people will require flexible service provision.
• Problematic substance users often present with complex social and medical problems that make accessing formalised end of life care services more difficult to navigate than other populations. However, the dearth of literature resulted in the definition changing to include any alcohol or drug use, rather than problematic use, except we retained problematic prescription drug use to ensure manageability of the evidence.

| Design
A Rapid Evidence Assessment (REA) methodology was identified as the most appropriate research tool to use to enable a speedy identification of key pieces of evidence. This would inform our wider study and provide a reference document to underpin further work on this topic in policy or practice. REAs can be defined as providing.
'… a more structured and rigorous search and quality assessment of the evidence than a literature review' but one critique is that it has narrower parameters and is not "as exhaustive as a systematic review" (Department for International Development (DFID), 2017: online). REAs, therefore, stem from Systematic Review methodology that, historically, focus on interventions and their effectiveness using experimental or quasi-experimental research design. By contrast, REAs are used to gain an overview of the prevalence and quality of evidence focusing on topic areas to support commissioning or programming decisions and identifying evidence gaps requiring further research (DFID, 2017: online). An REA is conducted within a shorter timeframe than a Systematic Review but retains the key characteristics of systematic review; transparency, replicability and comprehensiveness (Government Social Research (GSR) and EPPI Centre, 2009;Galvani & Forrester, 2011;GSR, 2013).
Our initial goal was establishing if there were any interventions for this group. As the review proceeded, the lack of a cohesive body of evidence to answer the research questions indicated became clear. Therefore, a combination of a REA and systematic mapping methodology (Clapton, Rutter, & Sharif, 2009) was adopted.
Systematic maps aim to describe the existing literature and gaps in the literature, in a broad topic area and the literature quality and content can be analysed in depth or more superficially as appropriate to individual projects (Clapton et al., 2009, p. 11). The review, therefore, was conducted with the rigour of planning and approach of an REA. However, as a result of our experience conducting the REA and the diversity of the literature found, it also encompassed systematic mapping of the review's findings.

| Search methods
This REA was an iterative review, the findings of which sought to support the wider programme of research of which it was part. To ensure it did so, five separate protocols were developed for the ini- These original protocols for the review were adapted as the REA proceeded due to the very limited relevant data generated by them.
This strategy allowed us to refine the questions and the focus of the research once the review was being conducted. Subsequently, the search terms were broadened to capture all the literature within this field within our search parameters (see Table 1 Table 2).

| Data extraction and synthesis
Given the dearth of intervention studies and the limited literature found through the initial searching, a systematic mapping approach was adopted. This facilitates a visual map of the existing literature on the broad topic of substance use and end of life care. It also allowed categorisation of the existing evidence in a number of ways, including its methodology, focus and country of origin. Textual narrative synthesis (Barnett-Page & Thomas, 2009) was used to explore any similarities, differences and relationships between papers. The textual narrative approach involves a commentary describing study characteristics, context, quality and findings. This was facilitated by a three-step process (Lucas, Baird, Arai, Law, & Roberts, 2007) including, (a) study grouping; in which studies belonging to each of the sub-groups, for example pain management or alcohol were identified with two researchers independently categorising and theming the papers. (b) Themes produced by each researcher were compared and a consolidated list produced. (c) sub-group synthesis were then developed. If there were any disagreement, a third researcher would review the paper. Study commentaries were produced in an excel file to summarise key aspects of the papers in relation to the sub-group within which they were included. This included both key findings and/or recommendations that speak to the aims of this REA. In terms of quality control, each commentary was reviewed by a second researcher independently to assess the summary in relation to the original paper. The empirical studies in the evidence base were scored on all six criteria on a three-point scale reflecting the extent to which the studies followed good research practice: 3 = no concerns; 2 = some minor concerns; 1 = major concerns. This resulted in a score ranging from 6 to 18 for each study. Studies were then assigned a quality category of high, moderate or low, based on their score. It is important to note that a low or moderate "quality" rating does not imply that a study was poorly designed or executed and does not suggest that its conclusions are incorrect or unreliable. It can simply mean that the report of the study did not fully explain its design or methods.

| FINDING S
The initial search of peer-reviewed articles resulted in a large number of papers. The inclusion and exclusion criteria were applied at abstract and full text reading stages (see Figure 1 below).
There were 60 papers generated from our search after applying the inclusion and exclusion criteria. They were, however, extremely diverse in methodology, focus and audience. The systematic map ( Figure 2) illustrates the range of evidence found among the 60 papers.
All empirical studies (n = 32) referred to in this REA were scored according to the DFID (2014) guidance (Table 3). Table 3 Table 4.

| Thematic groups
As the systematic map shows, the topic focus of the evidence could be grouped around three main themes: pain management (n = 25), homeless and miscellaneous populations (n = 24) and alcohol-related papers (n = 6). In addition, five remaining papers were diverse in focus and methodology.

| Homeless and marginalised groups
This group of papers focussed on people who are homeless or pre-

| Mental Health, HIV and miscellaneous groups
Mental ill health was highlighted in many of the papers identified in this review. Depression is known to be associated with chronic and terminal illness and with pain, both as a factor in causation and arising as a consequence of illness (

| Miscellaneous
There were five papers that fell outside of the thematic groups identified. Beynon, McVeigh, Hurst, and Marr (2010)

| D ISCUSS I ON
There is a lack of diversity, quality, breadth and depth to the litera- needs. It appears policy and practice have yet to respond to the emerging needs of people with problematic substance use at or near, their end of life in a significant way. This is a concern given the trends around substance use among older age groups and the increasing longevity of the general population. This absence of practice response is reflected in the gaps in the evidence base and can reflect a lack of patient involvement in order to drive through relevant and effective health improvements that reflect the complexity and need of this population (Luxford & Sutton, 2014 (Wilson et al., 2015). Research should also include the experiences, views and attitudes of social and healthcare professionals in responding to the overlapping issues of substance use and palliative and end of life conditions. This could include concerns about potential safety risks for outreach or community-based practice in particular, as highlighted by Galvani, Dance, and Wright (2018) in their study of hospice and specialist substance use staff. Lastly, there was limited evidence found in relation to alcohol (for example, problematic alcohol consumption associated with liver cirrhosis specifically at the end of life). Further work needs to be done in this area given alcohol remains the most commonly used substance. The full REA report for this work can be accessed for further information (Witham, Galvani, & Peacock, 2018).

| Limitations
REAs have a number of limitations including the breadth and depth of the searching. The number and type of databases used for searching are often limited to allow for a more rapid result and usually do not include the comprehensive searching involved in systematic reviews (, undated). These limitations apply to this study too. The decision to set the cut-off date at 2004 and to access only English language literature meant that some studies may have been missed and it is not possible to know how many would have met our inclusion criteria. Nevertheless, agreeing and adhering to a date limiter is an accepted feature of REA's as is the need to set limits on the retrieval of full texts (Government Social Research & EPPI Centres, 2009). Thirteen databases across health and social care sources were accessed for this review. It is possible that more databases may have produced additional material.
Similarly, expanding our search terms to include specific health conditions and specific types of licit and illicit drugs may have produced further hits. Data extraction and critique is often limited with REAs resulting in a focus on methodology and only key elements of data. While this review went slighter deeper in our reading and grouping of content within the evidence, a less rapid review could have included this greater depth.

| CON CLUS ION
This REA set out to explore current responses to and models of practice for, people living with problematic substance use and palliative and end of life care conditions. There is a dearth of research available on this topic and thus the focus of this review was broadened to identify what evidence was available, its focus, quality and the gaps in the research evidence base. The resulting body of work comprised 60 papers, primarily from peer-reviewed journals. It was quite disparate in This is clearly an area of work where far more research is needed. It is a new area of work and research focus and that has to be considered in considering our findings. However, the gaps identified are considerable and need to be filled in order to provide an evidence base on which to build future good policy and practice, both in the UK and internationally. Ultimately, this work is needed to ensure that this growing group of people have good quality care and equal access to service provision.