Blogs by the online BMJ diabetes champions and guests
ADA Annual meeting 9 June 2012 - Should We Pay People to Change Their Behavior? Behavioral Economics and Diabetes
The use of immediate reinforcement to change adverse behaviors has been most studied in addiction. The essential characteristics of reinforcement are: frequent monitoring of the target behavior, reward each time that desired behavior occurs and withhold reinforcement when the desired behavior does not occur.
An example study was at the VA with alcoholic subjects. The setting was group therapy. The monitoring was frequent breath analysis for alcohol. The reinforcement was fishbowl random rewards as follows: 50% a statement “good job” written on the card; the remaining cards providing an award ranging from $1-2 with 1/13 worth $20 and 1/500 worth $100. The average per patient cost was $200 for the program. Alcohol free breath analysis was 20% in the control group and 86% in the reinforcement group. Two factors are important. One is the amount of reward. When a program with a per patient reward of $80 was compared to one with an average reward o $240, the $80 was not effective. The second factor is the value of the reward to the participants. The best approach is to ask them. Some prefer cash; other store gift cards and others still token gifts. Group interaction is also important. An individual approach is not as effective. The rewards have to be immediate. An analysis of various addiction programs has shown this to be the most effective intervention; the VA has just adopted it nationwide.
Using an reward averaging $50/week resulted in an 0.77 increase in medication adherence (Petry Am. J. Med in press). In hypertensive subjects, medication adherence increased from 75% to 99.5% (Petry Am J Med 2011). When rewards are added to the LEARN weight loss program, 5% weight loss increased to 66% of female college students. The approach has also been shown to be effective in seniors who wish to increase their exercise.
Dr. Volpp began his presentation by discussing why we engage in health adverse activities. We like immediate gratification. We are loss adverse. We are overly optimistic. We have a problem with self-control. He used the example of the growth of payday lenders. In 1990 there were 55 Starbucks and 500 payday lenders. In 2005 there were 8,600 Starbucks and 22,000 payday lenders. This is true because the immediate reward of $50 before payday is of more value than the $60 needed to pay back next week when your paycheck arrives. He then pointed out the value of opt-out versus opt-in programs. He used the example of four European countries in which you have opt-in to an organ donation program versus the several which require an opt-out. Generally the organ donor rate in opt-in countries is less than 10% and greater than 80% in opt-out countries. This has been applied successfully to pharmacy refill programs. When the customer has an active choice to have automatic prescription refills, automatic refills increase markedly. In the US over 80% of private insurance plans have reward incentives and disincentives. An example is co-payments that are known to change drug use. A typical program might have a 0$ co-pay for generic cardiovascular drugs and a higher one for patented cardiovascular drugs. Long term rewards like annual premium deductions for health insurance do not work neither do those having a high threshold such a a large amount of weight loss. Finally he pointed out two other successful approaches to improve adherence. On is the use of peer mentoring, which appears to be more effective than financial incentives and the other is hovering by which he means using devices that automatically report behavior. One such device wirelessly transmits weight, blood pressure and SMBG values and has been shown to improve desired self-monitoring.
Currently most private insurance plans in the US have provider and patient incentives, but governmental programs are far behind. The programs tend to be in silos, effecting individual diseases and not practices. Pay for performance for diabetes that would require major changes in a practice for 10% of patients and required by only one insurance plan are just not effective. Cost sharing with patients is also not effective because essential medications or treatment end up be equally affective as non-essential ones. The approach that seems most effective currently is to adjust the patient co-payments to the cost/benefit of the drug. For example, beta-blockers are cheap and effective and have zero co-payments. Other examples are lipid, diabetes and hypertension drugs. A scheme might be 0% co-pay for generic drugs, 50% co-pay for preferred drugs and 75% copay for non-preferred drugs. Such a plan increased use by 5-10% and adherence improved as well.
Dr. Parchman discussed the difference in the RCT setting and clinical practice. In the US, 51% of physicians’ offices are solo physician with one or more nurse practitioner or physician’s assistants; 35% have 2 to 5 physicians; 9% have 6 to 10 physicians; 4% have 10 to 25 physicians and 1% have more than 25 physicians. In a study in which the primary care physicians were directly observed, the patients had, on average, 4.8 comorbidities and were taking 6.4 medications. The visit lasted 17 minutes and 17 issues were addressed. 27% of the time was spent on medications and 13% on patient education. The speaker felt that the current fee for service could not result in comprehensive care of these complex patients. He discussed the Patient-Center medical home concept that has be promoted by the NCQA. There are currently over 3,300 recognized programs with average incentive payments of between $10,000-$20,000. The theory is that you will need to reorganize care to treat complex patients. The concept is team-based care (see Bodenheimer at chcf.org for more details).
COMMENT: This was a fascinating session touching on area that I know little about. The presentations by Drs. Petry and Volpp were scholarly and convincing. Behavioral economic theory would seem to have a place in our armenarium.