Start spreading the news: A structural estimate of the effects of New York hospital report cards
Introduction
One of the key strategies for improving health care quality is the release of “report cards” on everything from enrollee satisfaction with their health plans to cardiovascular surgery mortality. Report cards may improve the performance of health care markets in at least two ways. First, they may enable enrollees and patients to identify and patronize higher quality payers and providers. Second, they may encourage payers and providers to further improve quality so as to increase demand.1 The latter is predicated on the former, because improvements in quality will not increase demand if consumers are unaware of them. Thus, if report cards fail to influence demand, both of these benefits may be lost.
As we describe below, several researchers have studied whether report cards affect demand, with mixed results. Dafny and Dranove (2005) suggest that one reason for the mixed findings may be that report card rankings could comport with prior beliefs about quality. For example, it seems doubtful that any positive report card could elevate the reputation of the Mayo Clinic above its current stature. Thus, even a glowing report might not increase Mayo's market share. Report cards will likely have the largest impact on market shares when the results are contrary to prior beliefs.
In this study we propose and implement a methodology to assess the effectiveness of the “news” that report cards provide to the market. Studying the immediate aftermath of the introduction of New York's cardiovascular surgery report cards in December 1990, we find that higher-ranking hospitals did not appear to gain market share. However, once we control for prior beliefs about quality, we find that when report card scores differ from prior beliefs, patients do respond by changing their choice of hospital. We conclude that report cards may be valuable in precisely those situations where they are needed most, when the facts about quality differ from preconceptions. This effect is not symmetric, however—hospitals whose report card rankings were lower than prior beliefs experienced a statistically significant decrease in demand, but hospitals with higher-than-expected scores appear to reap no benefits from the positive news. This is consistent with an earlier finding by Scanlon et al. (2002) who found a similar asymmetric response to health insurance report cards. Based on these findings as well as anecdotal conversations with health care executives, we speculate that this asymmetric response might prove to be a consistent expression of health consumer behavior.
Section snippets
Brief background on report cards
Detailed discussions of health care provider report cards appear elsewhere, so we will summarize some of the key facts, especially those pertaining to the New York report cards that we study.2 The Health Care Financing Administration (now the Center for Medicare and Medicaid Services) released the first publicly disseminated provider report cards in 1984. These reported hospital mortality rates for a wide range of conditions and procedures. The HCFA report
Previous literature
Numerous researchers have attempted to study the effects of report cards on market shares. Unless otherwise mentioned, none of these studies account for patient prior beliefs.
Mennemeyer et al. (1997) examined the effects of the report cards published by HCFA between 1986 and 1992. They found that hospitals with higher than expected mortality rates (where the expectations were based on patient characteristics, not market perceptions of quality) experienced almost no effect on market share, but
Why report cards?
The additional information afforded by report cards might provide two critical benefits4:
(B1) Report cards could enable patients
Model
In this section we present a highly stylized model that shows how to incorporate prior beliefs into estimates of the impact of report cards. This approach is similar to that of Chernew et al. (in press), although the empirical implementation is somewhat different. Suppose that an individual i who visits hospital j expects to obtain utility Uij based on their expectation of the hospital's quality Qj, their travel time, Tij, and a random component ηij distributed according to the type-I extreme
Data and methods
Our data come primarily from New York state hospital inpatient records for the years 1989 to 1991. Data from 1990 to 1991 were collected as part of the Healthcare Cost and Utilization Project State Inpatient Databases (SID). Data from 1989 were collected as part of New York State's SPARCS program. The two data sources are equivalent: they are collected from the same hospital records and contain the same information. The data include diagnoses, procedures performed, hospital identifiers, and
Results
Table 3 contains the estimate of Eq. (4). For comparison sake, we also include an estimate of “naïve” Eq. (5). Both models include interactions between age, race, sex, and private insurance and the hospital fixed effects. We also include the travel time to the hospital, the travel time interacted with patient characteristics, and the estimated fixed effect by hospital and year. Because these hospital-year effects were estimated in a previous step, we adjusted the standard errors of the
What is an improved report card score worth?
The results of model (5) indicate that a hospital receiving negative news can expect to gain approximately 7% in market share for each one standard deviation increase in report card score. To get a feel for what a good report card is worth, consider a hospital that currently performs 500 surgeries, was perceived to be two standard deviations below average in quality, but receives a mean report card score. That hospital would perform an additional 70 surgeries as a result of the report card.
Conclusions
In this study, we showed that when hospital report cards provide information that differs from patients’ prior beliefs, patients respond to this information by moving to higher-quality hospitals. We also showed that this effect is primarily due to shifting away from hospitals with negative news, rather than shifting towards hospitals with positive news. This finding may explain why hospital executives with whom we have spoken indicate that their hospital governing boards pay considerable
References (17)
- et al.
Information and consumer choice: the value of publicized health plan ratings
Journal of Health Economics
(2006) - et al.
The impact of health plan report cards on managed care enrollment
Journal of Health Economics
(2002) - Altman, L., 1990. Heart-surgery Death Rates Decline in New York. The New York Times, December 5, section...
- Altman, L., 1992. Surgical Scorecards: Can Doctors be Rated Just Like Ballplayers? The New York Times, January 14,...
- Chernew, M., Gowrisankaran, G., Scanlon, D., in press. Learning and the value of information: the case of hospital...
- et al.
The role of information in medical markets: an analysis of publicly reported outcomes in cardiac surgery
American Economic Review
(2004) - Dafny, D., Dranove, D., 2005. Do Report Cards Tell Patients Anything They Don’t Already Know? The Case of Medicare HMO...
- Davis, H., 1992. Heart surgery mortality declines 36% across state; improvement in WNY matches the average. Buffalo...
Cited by (99)
Effects of disclosing inspection scores of health facilities
2022, Socio-Economic Planning SciencesMeasuring quality effects in equilibrium
2022, Journal of Health EconomicsHospital report cards: Quality competition and patient selection
2021, Journal of Health EconomicsWhat do consumers learn from regulator ratings? Evidence from restaurant hygiene quality disclosures
2021, Journal of Economic Behavior and OrganizationThe industrial organization of health care markets
2021, Handbook of Industrial Organization