Search

Objective Assess the relative importance of proximity and other hospital characteristics in the choice of hospital for breast cancer surgery by race/ethnicity. Data SEER-Medicare data. Study design Observational study of women aged >65 years receiving surgery for stage I/II/III breast cancer diagnosed in 1992-2007 in Detroit (N = 10,746 white/black), Atlanta (N = 4,018 white/black), Los Angeles (N = 9,433 white/black/Asian/Hispanic), and San Francisco (N = 4,856 white/black/Asian). We calculated the distance from each patient's census tract of residence to each area hospital. We estimated discrete choice models for the probability of receiving surgery at each hospital based on distance and assessed whether deviations from these predictions entailed interactions of hospital characteristics with the patient's race/ethnicity. We identified high-quality hospitals by rates of adjuvant radiation therapy and by survey measures of patient experiences, and we assessed how observed surgery rates at high-quality hospitals deviated from those predicted based on distance alone. Principal findings Proximity was significantly associated with hospital choice in all areas. Minority more often than white breast cancer patients had surgery at hospitals with more minority patients, those treating more Medicaid patients, and in some areas, lower quality hospitals. Conclusions Residential location alone does not explain concentration of racial/ethnic-minority breast cancer surgery patients in certain hospitals that are sometimes of lower quality.

Objective Assess validity of the retrospective Dartmouth hospital referral region (HRR) end-of-life spending measures by comparing with health care expenditures from diagnosis to death for prospectively identified advanced lung cancer patients. Data/Setting/Design We calculated health care spending from diagnosis (2003–2005) to death or through 2011 for 885 patients aged ≥65 years with advanced lung cancer using Medicare claims. We assessed the association between Dartmouth HRR-level spending in the last 2 years of life and patient-level spending using linear regression with random HRR effects, adjusting for patient characteristics. Findings For each $1 increase in the Dartmouth metric, spending for our cohort increased by $0.74 (p