Introduction Pay for performance is increasingly used as a way of improving the quality of medical care. We previously showed that a pay for performance scheme targeting a range of processes measures in hospitals in the North West of England was associated with a substantial reduction in mortality for pneumonia, myocardial infarction and heart failure equivalent to 890 fewer deaths (Sutton et al. Reduced Mortality with Hospital Pay for Performance in England. New England Journal of Medicine 2012;367:1821–28). This analysis only assessed mortality in the first 18 months after introduction of the scheme. We now report mortality outcomes at 42 months to see whether the effect was sustained.

Methods Difference-in-differences regression analysis based on mortality for 230,985 patients admitted with pneumonia, myocardial infarction and heart failure to incentivised hospitals 18 months before and 42 months after the introduction of the program. These were compared with mortality in the following control groups: 1,260,545 patients admitted for the same three conditions to all 132 other hospitals in England, 50,400 patients admitted for six non-incentivised conditions to the incentivised group of hospitals and 285,301 patients admitted for non-incentivised conditions to all other hospitals in England. Analyses were adjusted for differences in age, gender, primary diagnosis, co-morbidities, type of admission, and location from which the patient was admitted.

Results Preliminary analyses suggest that the gains in mortality seen 18 months after the introduction of the pay for performance programme were not sustained at 42 months.

Discussion Pay for performance schemes remain controversial, and there are many unanswered questions about how and when they work. Our previous analyses were important because the incentive scheme that was introduced (Premier HQID) had no impact on mortality when introduced in the US, but appeared to have a substantial impact on mortality when introduced in the UK. However, our long term analyses suggest that these improvements were not sustained. We will comment on a number of possible reasons for the observed effects. One factor is that during the study period, the financial incentives changed from being bonuses to penalties for hospitals in the scheme.

Declaration of competing interests None.