Effect of UK Quality and Outcomes Framework pay-for-performance programme on quality of primary care: systematic review with quantitative synthesis.

Abstract

To systematically review the one and three year impact on quality of care of the introduction and withdrawal of financial incentives in the UK Quality and Outcomes Framework pay-for-performance programme. Systematic review with quantitative synthesis. MEDLINE, Embase, CINAHL, PsycINFO, and Scopus databases were searched from 1 January 2004 to 3 September 2024. Eligible studies used repeated cross sectional or cohort designs with consistent measurement before and after incentive introduction or withdrawal. Studies with a minimum of three time points before and after intervention were included in quantitative synthesis. Analysis used both reported impact if available and de novo interrupted time series analysis of extracted raw data if not reported by the original study. Meta-analysis was not appropriate; findings were quantitatively synthesised by reporting medians and interquartile ranges of changes in quality or reported narratively. Risk of bias was assessed using the Mixed Methods Assessment Tool. 30 studies were included, with 11 providing data for quantitative synthesis. Across all indicators, evidence was found of improvement in recorded quality at one year after incentive introduction (83 indicators; median change compared with that predicted by pre-incentivisation trends 6.1 (interquartile range (IQR) 1.9 to 14.6) percentage points) but less consistently at three years (72 indicators; median change 0.7 (-2.1 to 8.9) percentage points). Impact was higher for process indicators with lower performance in the year before incentivisation. Incentive withdrawal was associated with reduction in recorded quality compared with predicted at both one year (31 indicators; median change -10.7 (IQR -17.9 to -3.8) percentage points) and three years (31 indicators; median change -12.8 (-21.0 to -4.4) percentage points). The largest changes with both incentive introduction and withdrawal were for complex process indicators such as diabetes foot screening, with smaller changes in simple processes such as blood pressure measurement, intermediate outcomes, and treatment indicators. For all types of indicators, the reduction in quality following incentive withdrawal generally matched or exceeded the gains observed after incentive introduction (for example, for 14 indicators with data for both, median change at three years for incentive introduction was a 1.4 (IQR -0.9 to 4.6) percentage point increase versus a 3.9 (IQR 2.2 to 11.6) decrease for incentive withdrawal). Quality and Outcomes Framework incentives consistently improved quality of care at one year beyond that predicted by pre-incentivisation trends, but by three years the impact was inconsistent and not clearly better than trend. Gains from incentivisation seemed to reverse after incentive withdrawal. Prospero CRD42023467627.