Low-Value Service Variation and Physician Characteristics
SCHWARTZ-DOCTOROFMEDICINETHESIS-2017.pdf (423.4Kb)(embargoed until: 2019-05-01)
MetadataShow full item record
CitationSchwartz, Aaron. 2017. Low-Value Service Variation and Physician Characteristics. Doctoral dissertation, Harvard Medical School.
AbstractImportance: Reducing spending on unnecessary medical procedures is a global priority. Understanding patterns of low-value service use across physicians can inform efforts to reduce wasteful care.
Objective: To quantify the extent of physician-level variation within region and within provider organization in the rates of primary care related low-value health care services, and to assess for associations between low-value service rates and physician characteristics.
Design, Setting, and Participants: Retrospective analysis of low-value service use in 2008‒2013 using Medicare fee-for-service claims and enrollment data for 4,797,293 beneficiaries served by 66,675 generalist physicians. We employed multilevel models to quantify the magnitude of service use across physicians, adjusted for patient clinical and sociodemographic characteristics, within region and within provider organizations. We examined associations between rates of low-value services and physician characteristics related to education, demography, academic status, pharmaceutical/device payment, and patient panel size.
Main Outcomes and Measures Annual per beneficiary count of 17 primary care related services that provide minimal clinical benefit.
Results: The average rate of low-value services among attributed beneficiaries was 35.6 services per 100 beneficiaries per year, with considerable variation across physicians (within region 90th/10th percentile ratio, 2.14; 95% CI, 2.12‒2.16; within organization 90th/10th percentile ratio, 1.57; 95% CI, 1.56‒1.58). Greater low-value service rates were associated with educational characteristics (DO credential, foreign medical graduate status), demographic characteristics (older age, female gender), academic status (lack of professorship), pharmaceutical/device company payment (both any payment and greater size of payment), and larger patient panel size. However, association magnitudes were generally minimal.
Conclusions and Relevance: Although variation in low-value service use is substantial even within the same organization, physician characteristics are associated with only small differences in service use. Therefore, direct measurement of service use is likely to be superior to use of proxy measures in attempts to target physicians for waste-reduction interventions.
Citable link to this pagehttp://nrs.harvard.edu/urn-3:HUL.InstRepos:32676133