Our work provides the novel insight that social prescriptions can have a powerful impact on whether patients adhere to non-clinical health services provided by the voluntary and community sector. In particular, we found that patients who had a social prescription had better adherence to a community-sector health advice and navigation service in Germany compared to patients who self-referred.
Our study expands upon previous research by explicitly comparing two distinct pathways that patients can take to use such services: social prescription versus self-referral. Understanding these pathways is essential to identifying the ways in which social prescriptions might be of greatest benefit and to designing effective social prescribing schemes [12, 13, 17].
In addition, our study adds important insights to the mixed evidence available on the effectiveness and efficiency of such schemes in reducing GP workload and healthcare demand more generally [14, 15, 35, 36]. In particular, the non-significant findings of many previous studies might be partially explained by the multicausal nature of their outcome indicators. By using a more proximal indicator, we provide evidence that social prescribing may indeed be beneficial in stimulating adherence to community and voluntary sector services. Investigating process indicators offers important insights into how social prescribing works and, in turn, enhances our understanding of the mechanisms through which outcomes can be achieved . Further research on social prescribing schemes could analyse other process indicators that, like adherence, are more proximate to the receipt of a social prescription and are expected to bring about change in more distal, but ultimately more meaningful indicators such as health and quality of life.
Qualitative studies in this area of enquiry may help explain the mechanisms behind the effect of social prescribing we observed in our quantitative analysis. In particular, trust in GPs [22, 25] and respect for a GPs’ authority [38, 39] have been found to increase patients’ adherence to social prescriptions. Patients with social prescriptions might be more confident in the necessity of service than patients who self-refer. Indeed, GPs may also function as a catalyst for behavioural change and have a “priming effect” on patients’ responses to health-related interventions .
Our interaction analyses suggest that patients who visit a community-based service because of psychological concerns are not more likely to adhere to services if they have a social prescription as opposed to self-referral. This finding is corroborated by previous research, which has found that patients with mental illnesses are less likely to adhere to referral schemes [25, 30] or to complete referral programmes . While the explanations for this are numerous, one important reason for non-compliance with social prescriptions appears to be a self-perceived lack of need for continuing treatment [42, 43]. At a more practical level, our findings from the interaction analysis suggest that health care practitioners should target subgroups of patients for whom social prescribing is especially effective, i.e., patients who do not have a psychological concern. In contrast, patients with a psychological concern might need additional sources of support such as a mentoring programmes to improve their adherence to non-clinical community and voluntary sector health services. Further research on social prescribing could analyse other individual-level characteristics that, like psychological concerns, might moderate the effect of a social prescription on adherence. This might be helpful to better understand how policy makers and health care decision makers can maximize the effectiveness of social prescriptions.
Our study has several important limitations, each of which offers opportunities for further research. First, our results may be subject to omitted variable bias because our dataset did not allow us to include individual patient characteristics at a more granular level. In particular, health-related factors may have directly or indirectly influenced the effect of social prescriptions. Although the results of our sensitivity analyses (see Appendix D) suggest that this bias does not substantially affect our results, we nevertheless encourage future researchers to replicate and expand upon our analyses by using a richer set of individual characteristics to identify more precisely which patients are most likely to benefit from social prescribing.
A second limitation of our study is its focus only on one community sector provider, which might oversimplify the phenomenon of social prescribing. In other countries, such as the UK, social prescribing schemes include prescriptions to a broad range of community and voluntary sector services. Indeed, 13  point out that social prescribing should not be seen as a single intervention, but rather a series of relationships between referrer, patient, link worker and activity, all of whom interact with one other. Further research is therefore needed to investigate whether our findings are transferable to social prescribing schemes that encompass multiple services and activities. Although this was not possible in the German context, in which social prescribing and, indeed, non-clinical community and voluntary health services themselves are still in their infancy, our focus on one service has the advantage that we could rule out the presence of unobservable social prescribing activities in our analysis. Also, it would be interesting to examine whether the social prescribing effects hold across different settings or whether the results differ across health systems (e.g., in more integrated health systems, such as the English NHS).
While not a limitation, another point for future research is related to the setting of our study, which was conducted in a socially deprived urban area. Such areas are the setting for which social prescribing was originally conceived and in which there is a strong need to address the social determinants of health and wellbeing in addition to providing medical care [12, 44]. However, 25  showed that GPs faced greater difficulties in deprived areas when attempt to connect patients with non-clinical community and voluntary sector health services (in this case, an exercise programme) through social prescribing. Other research from a different context (i.e., doctor referrals to specialty care) has found that living in a deprived urban area is negatively associated with adherence to doctor referrals . Bearing this in mind and assuming that, in the absence of a social prescription, adherence in urban areas that are deprived does not differ from that in urban areas that are wealthier, our main effect may even represent an underestimate. Put differently, the effect of a social prescription on adherence might be higher in wealthier urban areas. We leave this conjecture, however, to be investigated in future work.