| Summary: | Abstract Background Rapid evaluation was at the heart of National Health Service England’s evaluation strategy of the new models of care vanguard programme. This was to facilitate the scale and spread of successful models of care throughout the health & social care system. The aim of this paper is to compare the findings of the two evaluations of the Enhanced health in Care Homes (EHCH) vanguard in Gateshead, one using a smaller data set for rapidity and one using a larger longitudinal data set and to investigate the implications of the use of rapid evaluations using interrupted time series (ITS) methods. Methods A quasi-experimental design study in the form of an ITS was used to evaluate the impact of the vanguard on secondary care use. Two different models are presented differing by timeframes only. The short-term model consisted of data for 11 months data pre and 20 months post vanguard. The long-term model consisted of data for 23 months pre and 34 months post vanguard. Results The cost consequences, including the cost of running the EHCH vanguard, were estimated using both a single tariff non-elective admissions methodology and a tariff per bed day methodology. The short-term model estimated a monthly cost increase of £73,408 using a single tariff methodology. When using a tariff per bed day, there was an estimated monthly cost increase of £14,315. The long-term model had, using a single tariff for non-elective admissions, an overall cost increase of £7576 per month. However, when using a tariff per bed-days, there was an estimated monthly cost reduction of £57,168. Conclusions Although it is acknowledged that there is often a need for rapid evaluations in order to identify “quick wins” and to expedite learning within health and social care systems, we conclude that this may not be appropriate for quasi-experimental designs estimating effect using ITS for complex interventions. Our analyses suggests that care must be taken when conducting and interpreting the results of short-term evaluations using ITS methods, as they may produce misleading results and may lead to a misallocation of resources.
|
|---|
