Should We Set Goals?

The Boston Globe had an interesting article yesterday on academic research in the business literature on goal-setting. Academics at Harvard, Northwestern and University of Pennsylvania recently published a paper called “Goals Gone Wild.” (note that free signup is required to download 28 page research paper, which is entertaining and informative reading)

“Goals Gone Wild” recounts many instances where clear goals tripped companies up, causing lost reputation, lost profits, corporate bankruptcy, and even deaths.   At Sears Auto Repair, stretch goals for revenue led mechanics to bill for work not needed or performed. At Enron, the goal of maximizing revenue led to managers ignoring profitability.  Ford set a goal of a new car that was under 2000 lbs, cost less than $2000, and was available for 1970 model year, and the company produced the self-immolating Ford Pinto which cost 53 lives.  Goals can encourage cheating (Chicago teachers penciled in student answers on a high stakes test in Freakonomics), and might also inhibit learning (air traffic controllers with straightforward goals were less likely to develop new ways to address safety concerns). 


The authors (Ordoñez, Schweitzer, Galinsky and Bazerman) point out that goal setting can be harmful because goals can be

·        Too specific

·        Too narrow

·        Too numerous

·        Too challenging, and

·         Have inappropriate time horizons

What does this have to do with managing health care costs?


I spent a number of years designing and negotiating “pay for performance” contracts, and I’ve also designed and implemented physician compensation systems. This article made me think critically about the work we’ve done to create the right goals and the right incentives in health care.


First, let’s examine physician compensation. On a system-wide level, we have a fee for service payment methodology which accomplishes exactly what we would expect – the system pays for more units of service, and we deliver more units of service.   In physician practices, compensation is generally on the spectrum between fixed salary and entirely productivity – and we see higher productivity when this leads to higher physician income.  That’s good, to the extent that patients value access, and we don’t have enough of many types of doctors.  That’s bad, though, when a patient has to return to a rheumatologist to review a (normal) x-ray since the physician insists on a billable office visit when there is no incremental value to the patient.  


I have heard from many physicians that any effort to develop a “perfect” compensation system is fruitless, since physicians should be driven by professionalism, not financial incentives.  The answer, as is often the case, is the middle ground.  Financial incentives alone don’t drive performance – and  the combination of appropriate pay, the right resources, and professional satisfaction are all critical to delivering the right care to our patients. 


The authors of “Goals Gone Wild” note that we tend to understand intrinsic motivation (“I want to provide good care to my patient because that’s why I became a physician”) but we exaggerate the impact of external motivation. (If it was all about the money, absolutely no one would go into child psychiatry.) 


I remember heated discussions about whether “pay for performance” goals should be outcomes (such as diabetic control) or process measures (such as whether or not patients had screening for diabetic kidney disease).   Ordoñez, et al, would have recommended “learning goals” (process measures) rather than “performance goals” (outcome measures).   I believe they would be especially critical of outcome goals if they were possible to manipulate, such as the UK NHS pay for performance system, where physicians can exclude individual patients from their measurement). 


I also remember heated discussions about whether we should have just a few measures to allow us to concentrate on those, or whether we should have a broad range of measures to be sure enough patients were benefiting from our efforts.  I argued for a limited number of measures a few years ago, and have more recently migrated toward preferring enough different measures in pay for performance to encourage system redesign.  “Goals Gone Wild” reminds me of the hazard of having too many goals. 


Goals are clearly a good idea in some circumstances;  Ordoñez et al use a medical metaphor, and suggest we should  “conceptualize goal setting as a prescription-strength medication that requires careful dosing, consideration of harmful side effects, and close supervision.”  This is sage advice for those designing physician compensation and pay for performance metrics.