Would you like it if your doctor was paid more if your blood pressure was under control, and less if it wasn’t? Do you think it would make you healthier?
The issue of whether, and how, to incentivise doctors is highly controversial. Traditionally, doctors worldwide have been paid in three ways: with a regular salary, with a fee for each consultation or procedure they perform, or with a lump sum based on the number of patients under their care. Until recently, pay has not reflected whether the care provided by doctors actually benefits patients. As a result, healthcare systems around the world often provide poor value, with mediocre care being delivered at high cost. The idea of paying doctors more if they provide better care, known as pay-for-performance, has been repeatedly touted as a solution and adopted worldwide.
But does pay-for-performance actually improve patients’ health?
The world’s largest and most studied pay-for-performance programme is the UK’s Quality and Outcomes Framework (QOF). Introduced in 2004, this scheme allows family doctors to increase their incomes based on performance on more than 100 indicators of quality related to several major diseases, including heart disease, stroke and diabetes. QOF covers almost all family practices in the UK and nearly £6 billion (US$9 billion) was invested in incentive payments over the first seven years, with additional billions invested in administration and information technology. Much of the research suggests that this led to improvements in quality indicators, particularly in the early years, and in the process increased family doctors’ incomes by up to a quarter.
Given that QOF targeted some of the leading causes of death, it might be expected to have extended the lives of people with these conditions. We evaluated whether mortality rates declined more in the UK after QOF was initiated than in similar countries that didn’t introduce pay-for-performance, and found that the scheme was associated with a reduction in mortality of just 3.7 deaths per 100,000 in its first seven years. This effect was not statistically distinguishable from zero.
Clinical trials vs the real world
In some ways this was surprising. Projections based on the apparent improvements for incentivised activities in the early years of the scheme, combined with evidence from clinical trials on the patient benefit of these activities, estimated that the scheme should have reduced mortality by 11 deaths per 100,000 in the first year alone.
What can explain the apparent contradiction that quality of care under pay-for-performance has seemingly improved, yet patient health has not?
One reason is the potential disparity between the effectiveness of treatments in clinical trials and the outcomes for patients using the same treatments in the “real-world”. Clinical trials aim to create ideal testing conditions for treatments, often excluding patients on the basis of age, illness or infirmity. These excluded groups – in whom the treatment is effectively untested – typically make up the majority of patients who receive treatment after the trial is concluded. For example, for the major trials of blood sugar control in diabetes, between 49% and 97% of diabetic patients would not have been eligible to participate. As a result, evidence from randomised trials may not translate to the wider population, particularly for older and sicker patients.
Interventions in clinical trials may also be significantly altered when put into practice. We know from clinical trials that keeping blood pressure and cholesterol at optimum levels and minimising variation over time is beneficial. But QOF indicators only require the most recent measurement to be under the relevant threshold, even if numerous previous measurements were too high. These performance measures bear little resemblance to the interventions in clinical trials that established the benefits of blood pressure and cholesterol control.
To complicate matters further, for patients with chronic conditions, very low levels of some biological parameters like blood glucose are not necessarily an indication of good health and can be associated with higher risk of death and other complications. The necessity of creating straightforward, universal performance metrics in pay-for-performance schemes has perhaps oversimplified more intricate clinical relationships.
Medicine is complicated. The behaviour of patients and doctors is complicated. To date, the simple incentives used by national programmes haven’t been effective in improving patient health. Yet these schemes are now widespread. It is critical that we find a way to make pay-for-performance work for patients, or else look for better ways to fix healthcare.