Better Health Greater Cleveland President Dr. Randall Cebul, M.D., responds to New York Times on Measuring Outcomes for Disadvantaged Populations


Robert Pear’s April 27th article in the Times ‘Health Law’s Pay Policy Is Skewed, Panel Finds’ highlights an important and difficult dilemma in devising new provider payment methods under the Affordable Care Act. These methods seek to reward physicians for the quality of health care outcomes experienced by disadvantaged patients using the same standards as those used for more advantaged patients, despite their being at predictably higher risk of poor outcomes. As a result, providers seeking to “score well” on quality measures, and thereby receive favorable recognition for the care they deliver – and enhanced payments – have an incentive to avoid caring for vulnerable populations.  How to avoid this unintended consequence of methods that are designed to reward better quality for all Americans is the dilemma Pear describes.

One proposed approach to avoiding this problem, according to an expert panel of the National Quality Forum, is to “level the playing field” across providers by using statistical methods that account for patient attributes that predict poor outcomes, such as poverty, low literacy or educational attainment, minority race, or lack of health insurance – factors largely outside of the doctor’s control.  Others appropriately point out that this approach merely disguises predictable disparities, providing doctors with no incentive to improve the care of their most vulnerable patients, since their disadvantaged status is hidden in a statistical black box. Indeed, since the methods would give extra credit to those providers who disproportionately care for vulnerable populations, this approach risks a “recruit and neglect” strategy if the extra credit is overly generous.

As the President of Better Health Greater Cleveland, a regional health improvement collaborative in northeast Ohio, I oversee twice yearly public reporting of patient achievement on quality standards across a large and diverse group of primary care practices, including those who provide care mostly to homeless patients as well as those who provide care for the most affluent in our region.  Since 2008, in addition to quality measures on a variety of important chronic medical conditions, we require all of our clinical partners to submit patient-level data on insurance status, race and ethnicity, language preference, and household income and educational attainment.  We report results without risk-adjustment methods, so nothing is hidden, but we also report our results in patient sub-groups by these “disparities” measures, to highlight the challenges to our patients, providers, and our region.  In addition to reporting results on nationally endorsed but locally vetted standards, to identify the best in class, we also report improvement in achievement over time, to recognize effectiveness in improving quality.

Our results have been remarkably consistent over the past seven years.  Patients who achieve best are more likely to be higher educated affluent older white males who live in the suburbs, while those who achieve least well are more likely to be poor minority patients with lower educational attainment who live in Cleveland’s inner city. And, while we have documented improvements in most of these disadvantaged sub-groups over time, we also find that more advantaged patients both achieve better and improve faster – paradoxically leading to larger gaps in outcomes even as the health outcomes for all patient groups get better.  These results are consistent with Thomas Piketty’s description of the “inevitability” of increasing inequality critiqued in a recent Times magazine story by David Leonhardt (“All for the 1 percent, 1 percent for all”, May 4, 2014).

There are no easy answers to this payment dilemma, but there are principles and approaches that should be considered.  First, as expected of all providers who seek to be rewarded for “meaningful use” of electronic health records, we should expect our peers to collect and report information about their patients’ preferred language, gender, race, and ethnicity.  Additional data about patients’ insurance status, income, and education also should be provided.  Second, rewards for better clinical performance should recognize improvement over time and not just achievement, which is more easily obtained among patients who are more advantaged.  And finally, rewards for performance among disparities populations should be transparent, with the playing field leveled as fairly as possible for other related factors that are outside of the doctor’s control.