5.4 Consumer information
5.4.3 Quality report cards
Consumers need health information not only to choose the right services in the appropriate quantity but also to choose the best providers. If quality information affects consumer decisions, the provision of such quality information can improve the welfare effects of competition in the market for medical care. A large body of literature addresses the effects of quality information in the form of quality report cards, which have become increasingly common in the U.S. The advantage of quality report cards for empirical research is that they create natural experiments, which can be exploited for the identification of causal effects. In this section, I focus on the effects of quality information on consumers’ choices of health care providers.5
The main body of literature on quality report cards focuses on the choice of hospitals. In one
5For empirical studies investigating the effects of quality report cards on health plan choice, see Chernew and Scanlon (1998), Scanlon et al. (2002), Wedig and Tai-Seale (2002), Dafny and Dranove (2008) and Jin and Sorensen (2006).
of the first studies in this research area, Mennemeyer et al. (1997) examined the publication of death rates in five hospitals by the Health Care Financing Administration (HCFA) between 1982 and 1992. They found only a small effect of the data releases on the annual number of discharges, while press reports of single events reduced the number of discharges by 9%. Cutler et al. (2004) examined the annual publication of hospital-specific risk-adjusted mortality rates in the state of New York since the late 1980s as a series of mini experiments. They estimated that hospitals that received a high mortality flag experienced a decline in the number of admissions for bypass surgery in the subsequent year and a reduction in the mortality rate. The authors conclude that the publication of quality information can guide patients in their choice of provider and improve the quality of care. Quality report cards have also been shown to affect the market share of fertility clinics. Bundorf et al. (2009) estimated that clinics with higher birth rates in previous years experienced increases in their market shares after the publication of success rates in 1997.
Holding the success rate constant, clinics with younger patients who have higher chances of success experienced a decrease in their market share. The results suggest not only that quality information affects consumer decisions in this therapeutic area but also that patients are able to adjust the quality information for the case-mix of the treated patient populations. However, assisted reproductive therapy differs from other therapeutic areas because it leaves the patients time to collect information and is only used by younger patients who either can afford to pay for it by themselves or have special insurance for it.
One problem of the studies by Mennemeyer et al. (1997), Cutler et al. (2004) and Bundorf et al. (2009) is that quality report cards convey information that is also available to the consumers through other sources. The inability to control for information obtained from other sources can lead to an overestimation of the effects of quality report cards. Mukamel et al. (2004) used the predicted mortality rates of surgeons based on publicly available information as a measure of the quality information that would have been available to patients in the absence of report cards. To identify the effect of new information delivered by report cards they estimated the effect of the difference between the predicted and published mortality rate on the probability that a patient chose a specific surgeon. The results suggested that after the publication of the report cards patients were more likely to choose a physician whose published mortality rates were lower than what patients could infer from observable characteristics. In addition, the report cards reduced the importance of surgeon experience and prices for the patients’ choices. A problem of the econometric model by Mukamel et al. (2004) is that it assumes that the patients’ prior
beliefs about quality remain constant. Dranove and Sfekas (2008) enhanced the specification proposed by Mukamel et al. (2004) and incorporated regular updating of CABG patients’ beliefs about hospital quality over time. The analysis confirmed that quality report cards can affect the patients’ choice of provider but only when quality scores really are news to the patients.
Howard and Kaplan (2006) evaluated the U.S. national program for reporting quality outcomes in transplantation medicine and used a patient-level panel dataset to estimate the effect of the new information provided by quality report cards by controlling for hospital and time fixed effects. They only found a significant effect among younger and better-educated patients but not in the general patient population.
Information about hospital quality is not only published by public organizations but also in media hospital rankings. Pope (2009) analyzed the effect of the U.S. News and World Report (USNWR) hospital rankings – first published in 1990. The rank of a hospital in the previous year positively affected the number of non-emergency Medicare patients discharged from the hospital and the probability that a patient chose the hospital. The authors estimate that approximately 6% of all non-emergency Medicare patients changed their decisions because of the rankings. The effect might be larger than in other studies because Pope (2009) focus on non-emergency cases and because the rankings were published in an easy-to-understand manner in the mass media.
Quality information also seems to affect the patients’ choices of providers in the long-term care sector. An analysis of a patient-level panel dataset showed that Medicare and Medicaid beneficiaries were more likely to choose nursing homes with high post-acute care quality scores after the publication of this information in 2002 (Werner et al., 2012). However, the effect was rather small and only the sub-domain of pain scores had a significant effect on nursing home choice.
In a much noticed study, Kolstad (2013) tested a theoretical model in which physicians not only respond to quality information because high quality pays off in the market but also because they have an intrinsic motivation to do a good job. The quality improvements due to intrinsic motivation was measured by the effect of the unpublished absolute number of fatalities on the number of fatalities in the subsequent year. Quality improvements due to profit incentives were measured by the effect of the difference between the unpublished absolute number of fatalities and the published risk-adjusted mortality rates on the physicians’ risk-adjusted mortality rates in the subsequent year. The demand-side effect of good quality was measured by the effect of the published risk-adjusted mortality rates on the patients’ propensity to choose a surgeon. The
results of the analysis of discharge data supported the theoretical predictions. Patients were less likely to choose a surgeon with a high risk-adjusted mortality rate after the publication of the report cards, and surgeons who learned from report cards that they performed worse than their colleagues later reduced their mortality rates. Most importantly, physicians also showed an intrinsic motivation to reduce their absolute mortality rate, and the intrinsic response was approximately four times as large as the response to profit incentives.
5.4.4 Discussion
While the Grossman model of health care demand suggests that health information affects the demand for medical care negatively, the theoretical predictions about the effects of health information on the supply of medical care are ambiguous because it is unclear whether physicians tend to over- or under-treat poorly informed patients.
Both the theoretical and empirical results on the effects of health information are mixed.
While better-informed patients appear to make better choices about the use of preventive care, the correlation between health information and physician visits seems to depend on the type and source of health information. Being generally well-informed does not affect the use of outpatient services in a uniform way, while the public provision of health information in a controlled way seems to reduce the need to see a physician. However, people who seek health information on the internet exhibit higher demand for physician visits. A possible explanation for this difference is that people use information from the internet more often when they have particular symptoms and become confused by all the possible causes of these symptoms. A limitation of the studies investigating information seeking on the internet is that they do not control for personality traits that are associated with both elevated medical service use and the propensity to seek health information on the internet.
The provision of quality information has consistently been shown to affect patients’ choices of providers, especially when they have time to make an informed decision. Quality report cards can thus provide incentives for physicians to provide good quality care and have the potential to increase the welfare effects of competition among hospitals and surgeons. However, the welfare effects of quality information also depend on the accuracy of the information. If the quality outcomes are not properly adjusted for the severity of the treated cases, hospitals might have incentives to refuse more severe cases, which can lead to limited access to care for sicker patients.
Cutler et al. (2004) cite several studies showing detrimental effects of quality report cards on
the quality of care and on the access to care of high-risk patients.
In general, the available evidence does not support the prediction of the Grossman model that better-informed patients use less care because they use medical services more efficiently.
Rather, it appears that better-informed patients use the same amount of care but navigate in the health care system more effectively and choose services more wisely.
Some studies use the effects of health information on service use as a test of the demand-inducement hypothesis assuming that better-informed patients are less prone to be influenced by their physicians. The available evidence, however, does not support this hypothesis. A possible explanation for the absence of a negative effect is that the effects of consumer information on the physician-patient interaction are obscured by a positive demand-side effect.
6 The Swiss health care system
The Swiss health care system is best described as a model of regulated competition (Enthoven, 1988). The regulation of the health care system is established on three levels of government.
The federal government is responsible for the national health insurance system; the registration and reimbursement of drugs and medical devices; the oversight of medical training, health pro-motion, and prevention; and the maintenance of health statistics. The cantons are responsible for regulating and overseeing medical practices and hospitals, organizing and financing med-ical education, regulating the prices of medmed-ical services and operating public hospitals. The municipalities mainly organize and finance the provision of nursing homes and home care. The federal government can mandate cantons to implement federal laws, and the cantons can delegate certain responsibilities to municipalities.