Does prevention-focused integration lead to the triple aim? An evaluation of two new care models in England.

OBJECTIVES

To examine the effectiveness of two integrated care models ('vanguards') in Salford and South Somerset in England, United Kingdom, in relation to patient experience, health outcomes and costs of care (the 'triple aim').

METHODS

We used difference-in-differences analysis combined with propensity score weighting to compare the two care model sites with control ('usual care') areas in the rest of England. We estimated combined and separate annual effects in the three years following introduction of the new care model, using the national General Practice Patient Survey (GPPS) to measure patient experience (inter-organisational support with chronic condition management) and generic health status (EQ-5D); and hospital episode statistics (HES) data to measure total costs of secondary care. As secondary outcomes we measured proxies for improved prevention: cost per user of secondary care (severity); avoidable emergency admissions; and primary care utilisation.

RESULTS

Both intervention sites showed an increase in total costs of secondary care (approximately £74 per registered patient per year in Salford, £45 in South Somerset) and cost per user of secondary care (£130-138 per person per year). There were no statistically significant effects on health status or patient experience of care. There was a more apparent short-term negative effect on measured outcomes in South Somerset, in terms of increased costs and avoidable emergency admissions, but these reduced over time.

CONCLUSION

New care models such as those implemented within the Vanguard programme in England might lead to unintended secondary care cost increases in the short to medium term. Cost increases appeared to be driven by average patient severity increases in hospital. Prevention-focused population health management models of integrated care, like previous more targeted models, do not immediately improve the health system's triple aim.