Pay for performance doesn’t work in difficult patient populations

Pay for performance.  It’s a lovely sounding concept. If you’re a good doctor, defined by having healthy patients who meet predetermined quality indicators, then you get paid more.

What could be simpler, right?


Not all patients are created equal.  Some are highly educated, highly literate, highly motivated to prioritize health.  They have good jobs with health insurance, so critical medical care isn’t prohibitively expensive. They don’t need to choose between paying out of pocket for $400 worth of necessary blood pressure medications or paying the rent.

And then there are my patients. Many are unemployed. Only a handful have private insurance plans.  Half are underinsured, with Medicaid or Medicare. Half are completely uninsured.  Only half speak English as a primary language. For the majority of my patients, housing is uncertain, jobs are not to be found, their neighborhood schools are struggling, and still my patients are human, living in bodies subject to human frailties.  They develop high blood pressure and diabetes and coughs and colds and prostate cancer and neurofibromatosis the same as anyone else.

A terrifying article (for those who take care of socioeconomically challenged patient panels) appeared in the Journal of the American Medical Association in September 2010.  A Harvard group led by Clemens S. Hong examined, as the title of the article spells out, the “Relationship between patient panel characteristics and primary care physician clinical performance rankings.”

They used data from 125,303 adult patients who visited one of 13 clinic sites linked by a common electronic medical system.  9 were hospital-affiliated practices, 4 were community health centers. 162 primary care physicians were ranked based on clinical performance scores.

Here’s the upshot: physicians whose clinical performance measurements ranked in the top tertile of the physicians, compared to the bottom tertile, had fewer minority patients, fewer non-English speaking patients, and fewer patients on Medicaid or without insurance.

This suggests to me that if your patients are English speaking, white, and well-insured, they’re going to understand what’s going on with them medically, easily access what they need to stay well, and be able to pay for the care you recommend.  You’re going to look good, and make more money for taking care of an easier-to-care-for population.

It doesn’t matter how good of a doctor you are if your patients are unable to follow your suggested care.  My patients skip specialty appointments because they don’t have a ride to the doctor.  They don’t take their insulin because our subsidized clinic pharmacy ran out and they can’t afford to pay full price in the drug store.  They take two of the same med because they can’t read the bottles and leaving the hospital they were given the brand name of the generic they take at home.

Adjusting for patient panel characteristics led to a relative mean change of 7.6 percentiles in physician rankings.  More than one-third of the primary care physicians (59/162) ended up in a different quality tertile, and would have earned different quality payments.

As a physician, I am rewarded on results, not effort. My patients, through no fault of my own, and really no fault of their own either, do not always take the steps I may recommend for them to be healthy and for me to attain all my quality indicators.  Luckily, I am meeting my pay-for-performance quality indicators (hooray for being rewarded for prescribing generic drugs!). But caring for difficult patient populations is a handicap for physicians in pay-for-performance incentive systems, potentially directing physicians away from caring for the populations who need our care the most.

Kohar Jones is a family physician who blogs at Progress Notes.

Submit a guest post and be heard on social media’s leading physician voice.

View 11 Comments >

Most Popular

Get KevinMD's 5 most popular stories.