Health Care Program Meeting
March 1, 2013
Robert Kaestner, University of Illinois and NBER, and Anthony Lo Sasso, University of Illinois at Chicago
By exploiting a unique health insurance benefit design, Kaestner and Lo Sasso provide novel evidence on the causal association between outpatient and inpatient care. Their results indicate that greater outpatient spending was associated with more hospital admissions: a $100 increase in outpatient spending was associated with a 2.7 percent increase in the probability of having an inpatient event and a 4.6 percent increase in inpatient spending among the enrollees in their sample. Moreover, the researchers show that the increase in hospital admissions associated with greater outpatient spending was for conditions in which it is plausible to argue that the physician and patient could exercise discretion.
Jason Abaluck, Yale University and NBER, and Leila Agha, Boston University
There is enormous variation in medical treatment across physicians, hospitals, and regions, but designing reforms to lower costs and maintain quality requires identifying specific instances of inefficient spending. Abaluck and Agha develop a measure of the efficiency of health care delivery based on the frequency of negative CT scans for pulmonary embolism. Their model shows how to transform the fraction of negative tests into a measure of medical care efficiency that links directly to welfare. They apply their model using a 20 percent sample of Medicare claims data from 2000-2009. The empirical assignment of testing outcomes is validated using chart and billing data from two large hospitals. The researchrs find that 80 percent of doctors are performing too many tests, in the sense that on the margin they perform tests even if the costs exceed the benefits. If all doctors tested only when the benefits exceeded the costs, the proportion of patients given a chest CT in this sample would fall by 15 percent, from 3.63 to 3.08 percent. The financial savings would be about $66 per person tested, while the medical benefits from reduced mortality risk from treatment of false positives would be $242 per person tested. Together, these factors would roughly double the welfare increase from testing over a world with no over treatment. The researchers also find that more experienced doctors and doctors in regions with lower spending overall are less likely to over test.
David Cutler, Harvard University and NBER; Jonathan Skinner, Dartmouth College and NBER; Ariel Stern, Harvard University; and David Wennberg, Dartmouth Medical School
There is considerable controversy about the causes of risk-adjusted regional variation in healthcare expenditures. Cutler, Skinner, Stern, and Wennberg use three surveys to test whether demand-side factors such as patient preferences, and supply-side factors such as financial incentives, organizational structure, and professional beliefs, can potentially explain observed regional variations in Medicare expenditures. The first survey asked elderly Medicare enrollees about preferences for end-of-life care. The second and third asked cardiologists and primary care physicians about organizational and financial pressures, and presented them with vignettes on how they would treat specific patients. The surveys were linked to Medicare end-of-life utilization data at the hospital-referral region level. The results show that physician beliefs about appropriate care have a large impact on end-of-life spending and matter much more than do patient preferences. While organizational factors influence physician beliefs, financial considerations do not appear to matter. The best explanation for large observed variation in views of appropriate treatment is physicians' beliefs about their own productivity. Finally, many physicians' beliefs are more aggressive than warranted by professional guidelines, highlighting the likelihood for waste in U.S. health care.
Mireille Jacobson, RAND Corporation and NBER; Joseph Newhouse, Harvard University and NBER; and Craig Earle, Harvard University
In the economics literature, physician-induced demand is a phenomenon whereby physicians respond to negative income shocks by exploiting their agency relationship to provide excessive care. In principle, however, increasing provision could remedy under-provision of care. Jacobson, Newhouse, and Earle demonstrate such a case in the context of a 2005 reform to Medicare's reimbursement policy which significantly reduced payments to physicians for many chemotherapy drugs. They demonstrate a sharp change in the treatment patterns of Medicare beneficiaries diagnosed with lung cancer in the months just after, relative to just before, that payment change: a roughly 10 percent increase in the likelihood of chemotherapy treatment. The treatment change was specific to the physician-office setting, not the outpatient-hospital setting in which there was no change in reimbursement until a full year later. In addition, the types of drugs used by physicians changed: drugs that lost the most margin were used less intensively and expensive drugs, favored by a uniform 6 percent average margin built into the new payment system, were used more intensively than prior to the reform. Surprisingly perhaps, the change in treatment improved patient outcomes: the likelihood of death at 3, 6, and 9 months after diagnosis declined by 2-3 percent. The improvement in survival was even larger for the oldest patients, who had been under-treated according to the clinical literature.
Benjamin Handel, University of California at Berkeley and NBER, and Jonathan Kolstad, University of Pennsylvania and NBER
Traditional models of insurance choice are predicated on rational choice and risk protection. When these models are taken to data, it is typical to use the choices that consumers make from menus of health insurance options to estimate their risk preferences. A key empirical assumption is that, conditioning on observed health risk, risk preferences represent the primary component of persistent unobserved preferences. If other factors, such as information about plan options or perceived plan hassle costs, also affect choices, then risk preference estimates will be biased. In addition to having positive implications for choice predictions, omitting such unobserved choice factors can have normative implications for welfare analysis. Handel and Kolstad combine administrative data on health plan choices with unique survey data on consumer beliefs and other unobserved preference factors in order to separately identify risk preferences, information frictions, and plan hassle costs. These datasets are linked at the individual level, allowing the researchers to develop a simple empirical framework. They show that including additional factors in choice affects standard preference parameters. Then they develop a welfare framework that integrates both information frictions and hassle costs, and assess the welfare impact of a counterfactual menu design with only a high-deductible health plan option. The welfare loss from the restricted menu of plans is 46 percent lower after accounting for information frictions and hassle costs. This illustrates that welfare implications, and subsequent policy decisions, differ substantially after accounting for these additional choice factors.