Impact of a social prescribing intervention in North East England on adults with type 2 diabetes: the SPRING_NE multimethod study

Background Link worker social prescribing enables health-care professionals to address patients’ non-medical needs by linking patients into various services. Evidence for its effectiveness and how it is experienced by link workers and clients is lacking.

Objectives To evaluate the impact and costs of a link worker social prescribing intervention on health and health-care costs and utilisation and to observe link worker delivery and patient engagement.

Data sources Quality Outcomes Framework and Secondary Services Use data.

Design Multimethods comprising (1) quasi-experimental evaluation of effects of social prescribing on health and health-care use, (2) cost-effectiveness analysis, (3) ethnographic methods to explore intervention delivery and receipt, and (4) a supplementary interview study examining intervention impact during the first UK COVID-19 lockdown (April–July 2020).

Study population and setting Community-dwelling adults aged 40–74 years with type 2 diabetes and link workers in a socioeconomically deprived locality of North East England, UK.

Intervention Link worker social prescribing to improve health and well-being-related outcomes among people with long-term conditions.

Participants (1) Health outcomes study, approximately n = 8400 patients; EuroQol-5 Dimensions, five-level version (EQ-5D-5L), study, n = 694 (baseline) and n = 474 (follow-up); (2) ethnography, n = 20 link workers and n = 19 clients; and COVID-19 interviews, n = 14 staff and n = 44 clients.

Main outcome measures The main outcome measures were glycated haemoglobin level (HbA1c; primary outcome), body mass index, blood pressure, cholesterol level, smoking status, health-care costs and utilisation, and EQ-5D-5L score.

Results Intention-to-treat analysis of approximately 8400 patients in 13 intervention and 11 control general practices demonstrated a statistically significant, although not clinically significant, difference in HbA1c level (–1.11 mmol/mol) and a non-statistically significant 1.5-percentage-point reduction in the probability of having high blood pressure, but no statistically significant effects on other outcomes. Health-care cost estimates ranged from £18.22 (individuals with one extra comorbidity) to –£50.35 (individuals with no extra comorbidity). A statistically non-significant shift from unplanned (non-elective and accident and emergency admissions) to planned care (elective and outpatient care) was observed. Subgroup analysis showed more benefit for individuals living in more deprived areas, for the ethnically white and those with fewer comorbidities. The mean cost of the intervention itself was £1345 per participant; the incremental mean health gain was 0.004 quality-adjusted life-years (95% confidence interval –0.022 to 0.029 quality-adjusted life-years); and the incremental cost-effectiveness ratio was £327,250 per quality-adjusted life-year gained. Ethnographic data showed that successfully embedded, holistic social prescribing providing supported linking to navigate social determinants of health was challenging to deliver, but could offer opportunities for improving health and well-being. However, the intervention was heterogeneous and was shaped in unanticipated ways by the delivery context. Pressures to generate referrals and meet targets detracted from face-to-face contact and capacity to address setbacks among those with complex health and social problems.

Limitations The limitations of the study include (1) a reduced sample size because of non-participation of seven general practices; (2) incompleteness and unreliability of some of the Quality and Outcomes Framework data; (3) unavailability of accurate data on intervention intensity and patient comorbidity; (4) reliance on an exploratory analysis with significant sensitivity analysis; and (5) limited perspectives from voluntary, community and social enterprise.

Conclusions This social prescribing model resulted in a small improvement in glycaemic control. Outcome effects varied across different groups and the experience of social prescribing differed depending on client circumstances.