Social prescribing: where is the evidence?

SOCIAL PRESCRIBING PROLIFERATION

Social prescribing is the topic of the moment. Many national organisations and individuals from policy, practice, and academia (such as NHS England, the RCGP, the Mayor of London, and National Institute for Health Research) are rightly advocating social prescriptions as an important way to expand the options available for GPs and other community-based practitioners to provide individualised care for people’s physical and mental health through social interventions. No robust figures exist but it is thought that around 20% of patients consult their GP for primarily social issues, given this and the driving forces of an ageing population, increased complex health and social needs, and increasing demand on services, social prescribing is rapidly gaining popularity.

As a concept and a model for delivering health and social interventions, social prescribing has proliferated without a concomitant evidence base.¹ This is partly due to resource limitations on evaluators and partly due to difficulties in conceptualising what social prescribing is and what good evidence for a complex service might look like. Here, we briefly outline different models of social prescribing, the current evidence base and its limitations, explore problems relating to what constitutes good evidence, and discuss some potential ways forward.

An immediate difficulty is the range of activity that the term ‘social prescribing’ embraces. Such heterogeneity is a function of social prescribing being the demand-driven formalisation of referrals to existing community services and organisations, which is necessarily locally different. More generally, at one extreme there are narrow interventions that focus on one clinical area and aim to prevent or reduce progression to chronic disease. Such interventions tend to include targeted life-style interventions (for example physical activity, healthy eating or cooking), medicines management or group mentoring, and are typically accessed through the healthcare system. At the other extreme, a large number of schemes are based on an understanding of the social determinants of health, recognising the interconnections between activity levels, social connectivity,² and mental health.³ These schemes use a range of activity types (such as income support, leisure, or social support), and intensity (in terms of number and length of sessions) of ‘link-worker’ support (that is, professionals who can inform or use goal-setting and health coaching).

Distinctions also need to be made with respect to approaches for different populations. Social prescribing originally operated in deprived areas, managing populations suffering complex physical and mental health problems, financial difficulties, social and emotional problems, substance abuse, and chaotic lifestyles. These populations often needed social alongside medical solutions.⁴ In time, social prescribing principles have been increasingly applied to older populations, with a view to addressing loneliness, improving physical activity levels, and mental wellbeing.⁵

Understanding such diversity is important, because relevant outcomes are likely to vary according to different social prescribing schemes. Generally, programmes are aimed at addressing the social determinants of ill health, supporting patients with non-clinical root causes of poor health, or preventing disease. For the most part, interventions that can be categorised in narrow clinical terms offer a more methodologically robust evidence base,⁶ but a recent systematic review⁷ detailing social prescribing schemes for diabetes treatment noted the variety of models.

EVALUATING SOCIAL PRESCRIBING

Social prescribing schemes located within the health care system are more likely to be subject to formal evaluation, because of a need to demonstrate effectiveness and value for money, a more established research culture, closer ties to universities, and the availability of validated tools by which to measure health outcomes. A current review has identified >60 reported outcomes in use and though some are in domains beyond those usually reported in the biomedical literature, identifying relevant, validated outcomes in a system perhaps less attuned to measurement and evaluation, is a challenge.⁸ The draft Common Outcomes Framework for Social Prescribing produced by NHS England could be helpful in this respect. This identifies impacts on the person, community groups, and on the health and care system, and should facilitate comparison and synthesis of study results. Less helpfully, the choice of specific indicators is left to local discretion.

Presently, there is more evidence of impact at the individual level than other areas, though, due to methodological reasons, this evidence is likely to be biased (lacking control groups, regression to the mean particularly). Reviewing the effectiveness of 15 programmes, Bickerdike et al,⁹ concluded there was insufficient methodologically robust evidence to assess success or value for money. Chatterjee et al,¹⁰ mapped outcomes for 86 projects in the UK and highlighted evidence gaps relating specific populations, and the types of pathways in use. Subsequent work by Loftus et al,¹¹ reported that while social prescribing was linked with better patient outcomes, GP workload overall was not reduced. Two recent reports have assessed the impact of social prescribing on healthcare demand, with mixed findings.^12,13 Moffatt et al,¹⁴ recently demonstrated the beneficial capacity that ‘link worker’ models of social prescribing offer, allowing more extensive approaches not possible in routine care. Thus, the extent to which potential impacts of social prescribing vary according to pathway and between cohorts is currently unknown. One of our authors is leading a review seeking to redress this evidence gap (Husk et al, unpublished data, 2018).

DIFFICULTIES IN GENERATING EVIDENCE

The evidence base relating to social prescribing is clearly problematic. There are three main reasons why generating robust studies of social prescribing are difficult: the methodological, the issue of generalisability, and the practical.

Methodological problems with generating a robust evidence base are considerable. Given the fact that social prescriptions are local context dependent and necessarily heterogeneous, there is confusion about the nature of what constitutes social prescribing. Linked to this, the multiple components that constitute a social prescription mean that evaluations are likely to be difficult to manage, compare, and assess for quality. Importantly, many of these components are rooted in contexts where, for example, local activity options may be shaped by local advocates and programme impacts are affected by available activities. There is also the challenge of selecting, using, and reporting relevant validated outcomes; in short, it is very difficult to agree on what constitutes ‘success’ or ‘effectiveness’ for these systems and, furthermore, deprived communities could find it harder to demonstrate impact — potentially increasing health inequalities.

The ability to make generalisable claims from any results is limited. With the reliance on local contextual factors, there are difficulties in designing evidence which is useful outside of the area under study. Further, attributing any change to social prescriptions is hard given these broader influencing factors. Attribution aside, there are differential regional and local interests which impact on the outcomes that are selected; that is, what is important for one area may well not be prioritised in others. Timing is also central; for example, when is the best time to evaluate services which take significant time to set up and embed?

Last, there are practical challenges to generating evidence around social prescriptions. Initially developing a collaborative relationship while maintaining researcher independence is challenging. What might be considered a suitable control group? How, given the complexity, is it possible to track impacts on health and social care use? Given resourcing constraints, can small organisations be expected to engage in data collection? Consenting and information governance can also be complex and extremely difficult to navigate. Perhaps most centrally is the fact that, not only are outcomes difficult to select, but their measurement can alter practice because services are naturally sensitive to positive or negative measures.

Each of these issues is complicated, and much of the discussion in the field focuses on tackling these areas. We think there are some key ways in which evidence might be generated which address such difficulties.

First, it is important to conceptualise social prescribing not as an intervention, but as a system. Each element of this system requires a robust and relevant evidence base. For activities, this might include randomised controlled trials or reviews of effectiveness, but for pathway features (like a ‘link worker’ element) this might include qualitative descriptions of patient experience, or realist evaluations of pathway sections (for example, enrolment, engagement, and adherence).¹

Second, reporting contextual factors and their impact is central to robust evidence. A good example is that much of the current evidence rightly relates to health, with little consideration given to the broader system in which social prescriptions happen and what the impact will be on social care services. Which raises the question of whether primary-care located models are the best possible model.

Third, being realistic about what outcomes are relevant and useful is important. While there is merit in assessing the impact on physiological outcomes (such as HbA1c), it is also important to capture the impact on the wider determinants of health. Finally, robustly recording the pathways individuals take through services is important so as to assess reach, scope, and acceptability. Along with coherent baseline data this enables assertions to be made around who, and in what way, social prescriptions might be of maximum benefit.

In summary, we would argue that social prescriptions have the potential to greatly benefit individuals with complex health and social care needs. However, it is important that interest, investment and innovation are supported and informed by a high-quality concomitant research programme that addresses the points raised, if this potential is to be fully realised.

Provenance

Commissioned; externally peer reviewed.

Competing interests

The author has declared no competing interests.