International Health Care Systems P4P Quality

P4P in the UK

General Practitioners (GPs) in the UK NHS are paid a mixture of capitation, lump sum allowances, and a pay-for-performance bonus. The P4P element, the Quality and Outcomes Framework (QOF), rewards GPs according to their performance on a large number of indicators. QOF payments represented up to 20% of GPs ’ income in the first year of its introduction.

In the 2006-2007 fiscal year, the NHS updated their quality indicator set, adding new quality metrics and retiring other ones.  The QOF is a piece-wise linear payment schedule with minimum and maximum thresholds.  In this year, minimum thresholds for payments were raised from 25% to 40% for all indicators. The maximum thresholds were raised for nine clinical indicators, whereas they were left unchanged for 25 other indicators whose definitions remained consistent. This change provides a unique quasi-experiment to evaluate how GPs respond to changes in payment thresholds by comparing the changes in their performance on indicators with increased thresholds against those with thresholds kept the same.

How did these changes effect physician behavior?  Did they improve quality for metrics with higher thresholds or those added to the QOF program?  Did quality decrease for metrics removed from the QOF program?

A paper by Feng et al. (2013) attempts to answer this question.  They use data from Scotland and apply a difference-in-differences methodology.  The find the following results.

“We find no ‘ discouragement ’ effect for GPs with a very low performance index (i.e. below 40% in 2005/2006) as the result of the increased maximum payment threshold… the effect of the increased maximum payment threshold under the QOF scheme in 2006 had a positive effect on GPs ’ QOF performance. We also conclude that the Under Performers improved the most in 2006/2007, followed by Competent Performers and Excellent Performers.”

This could mean that either that the QOF is working well, or that this is a natural phenomenon of regression  to the mean.

Leave a Reply

Your email address will not be published. Required fields are marked *