Do Hospitals Really Reward Glitz but Not Quality?

Abstract

The triple aims of American health care—better health, higher quality of care, and lower cost—remain as elusive as they are desirable.1 Since nearly all medical care for the severely ill and about a third of all US health care costs2 occur in hospitals, reaching these goals will presumably require aligning hospitals with them. One reasonable approach is to expect hospital chief executive officers (CEOs), who influence strategy and operations, to bear some accountability for meeting those aims: “pay for performance” should not apply only to physicians. So,howarewedoing?DoesCEOcompensation reflect success, if any, in reaching the triple aims? (I used thewordwebecause I have been a hospital CEO in San Francisco for the past 4years.) Those are thequestions that Joynt et al3 set out to answer by looking at the correlates of CEO remuneration at nonprofit hospitals. The authors reasoned that if, for example, quality of carematters to hospital boards of directors, which setCEOpay, thenCEOs shouldearnmoremoneyathospitals with better quality. Given that nonprofit hospitals have acommunitymission, Joyntetal alsowondered,perhapsskeptically,whetherCEOsathospitals thatprovidegreateramounts of charity carewouldearnmore.Generally, theygatheredpublicly available data and built linear regression models to determine which factors correlated with CEO income. The authors found several attributes that correlated positively with CEO pay (eg, teaching status and bed size), others that correlatednegatively (rural location andpoor payermix), andmanythathadnocorrelation (qualityandcommunitybenefit). So far, so good. But when they interpreted their findings, I think they took somemissteps that illustrate the difficultyofmakingconclusionsabout causality fromthepresence or absence of correlations. Let me try to explain. To begin, it is important to understand how most nonprofit hospital CEOs are paid. We have a base compensation, as well as incentive pay if we (meaning the hospital or perhaps the hospital system) meet certain targets; for me, these incentives include financial performance, several qualitymetrics, patient satisfaction, and community benefit. Presumably, a CEO’s base pay is related mostly to a hospital’s structural attributes,whereas incentive compensationdepends on performance. It is not unreasonable to expect that both types of factors would correlate with total compensation. Not surprisingly, the authors found that bigger, glitzier, moreprestigioushospitals—theYankeesandDodgersofhealth care—pay their CEOs a lot more money compared with other hospitals. Their conclusion that advanced technology drives CEO pay might be right, but an observational design cannot rule out alternatives, such as CEOs at fancier hospitals earn more because they are worth more, or because the members of the board compensation committees at glitzy hospitals are more accustomed to higher incomes. Nonetheless, their results suggest that the principles behindMoneyball4 have not had much of an effect on hospital C-suites. Joynt et al also found that the provision of charity care and community benefit did not correlate with CEO pay. Hospitals that provide large amounts of uncompensated charity care often struggle financially, making it less likely that their CEOs could be well compensated, even if their boards wanted to do so. On the surface, theirmostdisturbing findingwas thatCEO pay correlated with patient satisfaction, but not with quality. Theyseethisasamissedopportunityandrecommendthathospital boardsprovide incentives forCEOs tomeet quality goals. That advice seems strange, since every hospital CEO I know who receives incentive compensation already has qualityrelatedgoals, asdidapproximately three-quartersofCEOswho were surveyed recently.5 Amore likely explanation for the absenceof a correlationbetweenCEOpayandquality is thathospitalsareconcernedabout,measure,andrewarddifferentqualitymetrics—and Joynt et al looked at only a few of the several hundred possibilities. Indeed, the profusion of quality metrics challenges the ability of hospital quality departments to keep up withmeasuring and reporting them.6-10 We all know hospitals that havebeenhighly rankedby somestandards (eg, theUS News andWorld Report rankings or HealthGrades) but receivedawful gradesonothers (eg, LeapfrogorHospitalCompare). None of those rankings indicates how the hospital performedonmoremeaningful andobjectivemetrics, suchas severity-adjusted sepsis survival rates. Indeed, some hospital CEOshave likely figuredouthowtogame the systembyworking to align their incentive compensationwith themetrics that their hospital usually achieves—and then trumpeting those results. Thus, the opportunity that the authors have identified is to standardize how quality is measured, rather than allowing the indiscriminate proliferation of metrics to continue. By contrast, patient satisfaction correlated with CEO pay, likely because the subjective experience called patient satisfaction is easy to measure, even though what it actually means is unclear. Moreover, most measures of patient satisfaction correlate with one another, so it matters less what you ask about. Indeed, in our hospital, overall patient satisfaction is markedly higher among those who were in private rooms. So, too, is their satisfaction with the physicians, nurses, tests and treatments, and even the food! Talk about a halo effect. Finally, if you happen to practice in a large, high-tech urbanmajor teaching hospital in the South that has fewMedicaid patients and excellent patient satisfaction, do not assume that your CEO is at the top of the financial ladder, even if that is what Joynt et al’s regression model implies. Not only does correlationnot equal causation, it also applies only to the general, not the specific, case. Related article page 61 Research Original Investigation Compensation of CEOs at Nonprofit Hospitals