Abstract
We study the long-term effects on hospital activity of a three-year national integration programme. We use administrative data spanning from 24 months before to 22 months after the programme, to estimate the effect of programme discontinuation using difference-in-differences method. Our results show that after programme discontinuation, emergency admissions were slower to increase in Vanguard compared to non-Vanguard sites. These effects were heterogeneous across sites, with greater reductions in care home Vanguard sites and concentrated among the older population. Care home Vanguards showed significant reductions beginning early in the programme but falling away more rapidly after programme discontinuation. Moreover, there were greater reductions for sites performing poorly before the programme. Overall, this suggests the effects of the integration programme might have been lagged but transitory, and more reliant on continued programme support.