Multimethod study of a large-scale programme to improve patient safety using a harm-free care approach

Power, M, Brewster, L, Parry, G, Brotherton, A, Minion, J, Ozieranski, P, McNicol, S, Harrison, A and Dixon-Woods, M (2016) Multimethod study of a large-scale programme to improve patient safety using a harm-free care approach. BMJ Open, 6 (9). ISSN 2044-6055

Preview

Available under License Creative Commons Attribution. Download (2MB) | Preview

Abstract

Objectives: We aimed to evaluate whether a largescale two-phase quality improvement programme achieved its aims and to characterise the influences on achievement. Setting: National Health Service (NHS) in England. Participants: NHS staff. Interventions: The programme sought to (1) develop a shared national, regional and locally aligned safety focus for 4 high-cost, high volume harms; (2) establish a new measurement system based on a composite measure of 'harm-free' care and (3) deliver improved outcomes. Phase I involved a quality improvement collaborative intended to involve 100 organisations; phase II used financial incentives for data collection. Measures: Multimethod evaluation of the programme. In phase I, analysis of regional plans and of rates of data submission and clinical outcomes reported to the programme. A concurrent process evaluation was conducted of phase I, but only data on submission rates and clinical outcomes were available for phase II. Results: A context of extreme policy-related structural turbulence impacted strongly on phase I. Most regions' plans did not demonstrate full alignment with the national programme; most fell short of recruitment targets and attrition in attendance at the collaborative meetings occurred over time. Though collaborative participants saw the principles underlying the programme as attractive, useful and innovative, they often struggled to convert enthusiasm into change. Developing the measurement system was arduous, yet continued to be met by controversy. Data submission rates remained patchy throughout phase I but improved in reach and consistency in phase II in response to financial incentives. Some evidence of improvement in clinical outcomes over time could be detected but was hard to interpret owing to variability in the denominators. Conclusions: These findings offer important lessons for large-scale improvement programmes, particularly when they seek to develop novel concepts and measures. External contexts may exert far-reaching influence. The challenges of developing measurement systems should not be underestimated.