Variation in Quality and Costs of Care Across Physicians and Its Determinants

DSpace/Manakin Repository

Variation in Quality and Costs of Care Across Physicians and Its Determinants

Citable link to this page


Title: Variation in Quality and Costs of Care Across Physicians and Its Determinants
Author: Tsugawa, Yusuke ORCID  0000-0002-1937-4833
Citation: Tsugawa, Yusuke. 2016. Variation in Quality and Costs of Care Across Physicians and Its Determinants. Doctoral dissertation, Harvard University, Graduate School of Arts & Sciences.
Full Text & Related Files:
Abstract: This dissertation evaluates one of the key determinants of health care quality and costs – practice patterns of physicians. For decades, rapid health care spending growth and suboptimal quality of care have been fundamental issues of the U.S. health care system. A large body of literature has demonstrated substantial geographic variation in health care utilization without concomitant improvement in patient outcomes in higher spending regions. This literature has spurred debate about whether current levels of health care utilization are indeed socially wasteful – i.e., generate higher spending with no improvement in patient outcomes. While informative, however, this literature has not investigated variation due to individual physician behaviors, despite the central role of physicians as key decision makers in health care. In fact, surprisingly little is known about how individual physicians vary in their care, the determinants of that care, and the implications of that variation for patient outcomes. This dissertation attempts to shed light on these questions by analyzing the productivity of health care spending at the physician level.

Chapter 1 investigates the proportion of service use variation that can be explained by variation in individual physician practice patterns, and examines the impact of that variation on patient outcomes. I analyze data on Medicare beneficiaries hospitalized with medical conditions treated by general internists. Using a cross-classified multilevel model, I find that variation in spending across physicians exceeds variation across hospitals (10.9% and 6.2% of overall spending, respectively). As for evaluating the impact of between-physician variation in spending on patient outcomes, I exploit a natural experiment of physicians who specialize in hospital-based care – hospitalist physicians. Hospitalists routinely work in shifts and therefore patients are plausibly quasi-randomized to these physicians within a hospital based on physician work schedule. Among 272,979 hospitalizations treated by 8,489 hospitalists, hospitalists in the highest-spending quartile had lower 30-day patient mortality than hospitalists in the lowest quartile within the same hospital, despite similar patient characteristics (adjusted mortality rate 10.7% vs. 11.2%; adjusted odds ratio 0.94, 95%CI: 0.90 to 0.98, p=0.002). I observe no relationship between physician spending and patients’ readmission rates. Given larger variation in spending across physicians than across hospitals, our findings suggest that policies focused on individual physicians may be as or more effective than those targeted toward hospitals or regions. Moreover, interventions targeted at high-spending physicians to reduce spending, without accounting for their quality of care, may have the unintended consequence of negatively impacting patients’ health.

Chapter 2 begins my evaluation of the upstream determinants of variation in quality of care across physicians with a special focus on physicians’ years in practice. Physicians with longer years in practice may accumulate, or conversely exhibit obsolescence of, knowledge and skills. However, the association between physicians’ years in practice and patient outcomes is poorly understood. Using data on Medicare beneficiaries aged 65 years or older hospitalized during 2011-2013 with a medical condition, I investigate the association between hospitalist physicians’ years since residency completion and patient outcomes, adjusting for patient and physician characteristics and hospital-specific fixed effects. I again rely on quasi-randomization of patients to hospitalists to circumvent the possibility that physicians with greater years in practice may treat patients that are sicker on unobserved dimensions. Of 386,159 hospitalizations treated by 14,650 hospitalists, hospitalists in practice longer had higher patient mortality than hospitalists in practice for fewer years, despite similar patient characteristics. Each additional 10 years in practice was associated with 0.5% increase (95% CI: 0.3% to 0.7%, p<0.001) in patients’ mortality. Significant effects were present for low- and medium-volume physicians, but not high-volume physicians. Readmissions and costs of care were not meaningfully associated with physician years in practice. This study has implications for recent debates in the medical community regarding how best to ensure maintenance of clinical skills over a physician’s career. Our findings suggest that evaluating patient outcomes, particularly among older physicians with low-to-medium patient volume, may be necessary to guarantee that quality care provided by physicians is high throughout their careers.

Chapter 3 assesses another upstream determinant of between-physician variation in quality of care – physician sex. Studies have found differences in practice patterns between male and female physicians, with female physicians more likely to adhere to clinical guidelines and evidence-based practice. However, whether patient outcomes differ between male and female physicians is largely unknown. While physician sex is not a modifiable factor, understanding whether quality of care differs between male and female physicians is critically important, as it allows us to further investigate which aspects of practice patterns that vary between male and female physicians lead to better patient outcomes. Using nationally representative data on Medicare beneficiaries in 2012-2013, I examine the association between physician sex and patient outcomes among general internists. Despite similar observed illness severity of patients, female physicians have lower 30-day patient mortality (adjusted mortality rate 10.9% vs 11.4%; adjusted risk difference -0.5%, 95%CI: -0.7% to -0.4%, p<0.001) and lower 30-day readmissions (adjusted readmission rate 15.1% vs 15.8%; adjusted risk difference -0.7%, 95%CI: -0.8% to -0.5, p<0.001) within same hospital. These findings are unaffected when restricting analyses to hospitalist physicians for whom patients are plausibly randomized. Although the exact mechanism underlying these differences remains unclear, understanding why these differences in care quality exist, and what we might do to alleviate them, is critical to ensuring that all patients get high quality care.
Terms of Use: This article is made available under the terms and conditions applicable to Other Posted Material, as set forth at
Citable link to this page:
Downloads of this work:

Show full Dublin Core record

This item appears in the following Collection(s)


Search DASH

Advanced Search