Weng W, Lipner R. — American Board of Internal Medicine
Presented: AcademyHealth Annual Research Meeting, June 2012
Research Objective: Using patient-level chart audits, disparities in diagnosis and treatment of osteoporosis received by minority (non-Hispanic African American, Hispanic and Asian) versus white patients were examined. These differences were parsed by patient, physician and practice-level factors.
Study Design: The overall racial/ethnic differences were examined in eight osteoporosis care measures (pharmacologic therapy for patients with osteoporosis, DXA scan, screen for falls risk and complete falls risk assessment, weight-bearing exercise participation status and advice, calcium intake counseling, adequate calcium intake, vitamin D intake counseling, and adequate vitamin D intake). For each measure which had significant racial/ethnic differences, a series of hierarchical linear regression models were fitted to decompose the racial/ethnic differences by “within-physician” effect (i.e., the degree to which minorities received lower quality care than non-minorities within the same physician panel), and “between-physician” effect (i.e., the degree to which minorities were more likely to receive care from physicians providing overall lower quality care). Analyses assessed the impact on racial/ethnic differences of patient demographics and physician-level factors including physicians’ cognitive skills (equated ABIM Internal Medicine Certification Examination scores), subspecialty certification, and practice infrastructure (NCQA's Physician Practice Connections Readiness Survey scores), share of minority patients in the patient panel, and geographic location of practice.
Population Studied: 37,445 chart reviews of women age 65 and older, men age 70 and older, or patients with osteopenia, prior low-impact fracture, or osteoporosis (29,308 white, 2,966 African American, 2,560 Hispanic, and 1,900 Asian patients) from 1,481 board certified internists who completed the ABIM Osteoporosis Practice Improvement Module from January 2008 to December 2011.
Principal Findings: There were no or small disparities across all quality measures for Hispanic and Asian patients compared to white patients. African-American patients had similar performance compared to white patients in pharmacologic therapy (difference = -1.7%) and falls risk assessment (different = 2.8%), but poorer performance in other measures such as weight-bearing status and advice (-4.3%, P-value = 0.03), adequate vitamin D intake (-7.2%, P-value < 0.001), adequate calcium intake (-10.8%, P-value < 0.001) and DXA scan (-12.1%, P-value < 0.001). For DXA scan, “within physician” and “between-physician” effects accounted for 44% and 56% of the differences between African-American, respectively. Share of African-American patients in the patient panel was the dominant physician-level factor, explaining most of the differences (43%). Even though certification in rheumatology was associated with higher quality of care, it did not help explain the racial/ethnic differences nor did practice infrastructure. Calcium and vitamin D measures followed similar patterns.
Conclusions: There were no significant disparities observed for Hispanic or Asian patients. African-American patients received a similar level of pharmacologic therapy, but large racial/ethnic differences were found for other measures, which were due to both “within-physician” effect and physician-level factors. Share of African-American patients in the panel is the dominant factor. However, practice infrastructure does not matter.
Implications for Policy, Delivery or Practice: The study findings suggest that initiatives targeting physicians who have a large share of African-American patients would more likely mitigate the racial/ethnic disparity than policies designed to improve practice infrastructure. A large “within-physician” effect also implies that interventions targeted at meeting different needs of minority patients may also help mitigate the disparity.
For more information about this presentation, please contact Research@abim.org