In response to a recent article on the topic of economic motivation theory, Michael Kirsch sent me information about a very interesting study (May 2010 issue of the British Medical Journal) done to evaluate the effects of monetary incentives on clinic, physician, and staff work performance.
From 1999 to 2007, 35 medical facilities of Kaiser Permanente in NorthernCalifornia, were given financial incentives for ensuring that their patients got regular screening for diabetic retinopathy and screening for cervical cancer – eye exams and PAP smears*.
The results were less than stellar. In eligible patients (i.e. diabetics and sexually active women without hysterectomies) over 4 years, the rate of screening for diabetic retinopathy increased a little over 3 percentage points from 84.9 to 88.1% and over one year the rate for screening for cervical cancer increased by a paltry 0.6 percentage points. And then it got worse.
After these financial incentives were stopped, the screening rates for these tests fell dramatically to levels that were significantly lower than they were before the bonuses were started. After the incentives were stopped, screening rates for diabetic retinopathy dropped to 80.5% over 4 years and the screening rates for cervical cancer dropped to 74.3% over 5 years. What is going on?
This is yet another example of how economic motivation theory can be very counter-intuitive. One would assume that performance would increase linearly with increased rewards but in many contexts researchers have found the opposite effect. The mechanism is thought to work this way; an external reward or punishment (apart from base compensation) has the effect of decreasing internal motivators (based on autonomy, mastery, and purpose) so much so that this negates or even reverses the positive effects of a person’s external motivators (the drive to earn more) such that their total motivational drive and hence, their work performance, is decreased. Economists say that the internal motivators are “crowded out” in these cases.
The financial incentives in this study were given to be distributed throughout each facility rather than to each physician and as part of this program regular reminders for screening exams were sent out to staff. The modest but statistically significant increases in the rates for these two screening tests were likely as a results of the regular reminders rather than any financial incentives (unfortunately there was no control group without incentives to test this).
But the negative effects of monetary incentives on performance can clearly be seen after the incentives are discontinued. This is what economists call “motivational spillover.” This is what happens when you start giving someone an external or financial motivation to do something that they were already doing as part of the internal motivators of their job (mastery and purpose). Take, for example, the economic parable of the man and his lawn.
The story goes that a man was upset that his neighbor kids would always play on his lawn and damage it. So he decided to pay each child to play on his lawn. The surprised kids gladly accepted. After a few days the man told them that he could only afford to pay them half of the initial rate. The kids accepted this reduced rate but were less then enthusiastic. After a few more days the man cut his pay to almost nothing and the children were so upset that they left, vowing never to play on his lawn again unless he increased their pay. Problem solved.
In this case, the man’s pay “crowded out” the kid’s internal motivators (autonomy, and fun as the purpose) for playing on his lawn and the dominance of the external motivator spilled over into further activity. In the case of the medical incentives, regular screening exams are supposed to be part of what the staff at the clinic does and involves internal motivators as part of their autonomy, mastery, and purpose (taking care of patients) and it is these internal motivators that were impaired by the incentives. Clearly the clinic lost far more than they gained by instituting incentives and then discontinuing them. Interestingly the screening rates increased slightly after incentives were reinstated but did not get back to the original levels. Thus there was a net loss in performance even after restarting the incentives.
This is likely the reason why small monetary or other incentives for performance rarely work in socialized medical systems.
This study is consistent with a growing body of evidence that pay-for-performance does not work and can reduce overall care, continuity of care, and impair further efforts to improve care. So what is to be done?
The first thing is to try and get policy makers to understand that efforts to increase overall compensation by relatively small incremental increases tied to performance are very unlikely to work and as in the case above, will lead to minimal gain for money spent and may lead to a net loss in performance.
To properly compensate primary care practitioner’s level of education, effort, and time, a significant net increase in base pay should be provided and performance should be enhanced or maintained by efforts that maintain or improve the staff’s perceptions of autonomy, mastery, and purpose. For example, money is probably much better spent on regular educational activities for the staff that enhances their intrinsic motivators — education for staff members about the importance of and new methods of preventative care is much more likely to be effective than rewarding and/or punishing them for specific outcome indicators.
*The BMJ study did find that diabetic control and blood pressure control did improve significantly over the time span of the study however, there were no internal or external controls for these measures nor any way to differentiate them from other variables such as notifications and increased staff awareness of these measures that may had significant influence and so these measures were not included in the results of this study.
Chris Rangel is an internal medicine physician who blogs at RangelMD.com.
Submit a guest post and be heard.