Table 4.

Race/ethnicity, insurance, and the relative odds of an advanced-stage diagnosis of breast cancer, across prediction modelsa

Race/ethnicity (white is reference category)Insurance (private is reference category)
BlackHispanicAPIAI/ANUninsuredMedicaidMedicare
#1. Bivariate logistic (race/ethnicity)1.81 (<0.001)1.23 (0.15)1.13 (0.49)1.38 (0.47)
#2. Bivariate logistic (insurance)2.96 (<0.001)2.43 (<0.001)1.46 (<0.001)
#3. Two-variable logistic (race/ethnicity and insurance)1.56 (<0.001)1.14 (0.37)1.09 (0.64)1.11 (0.82)2.71 (<0.001)2.16 (<0.001)1.46 (<0.001)
#4. Single-equation w/o bio variables (Supplementary Table S1)1.31 (0.01)0.97 (0.83)0.98 (0.99)1.14 (0.76)2.06 (<0.001)1.65 (<0.001)1.32 (0.02)
#5. Single-equation with bio variables (Table 3)1.19 (0.01)0.97 (0.85)1.01 (0.96)1.25 (0.61)1.85 (0.001)1.63 (<0.001)1.33 (0.02)
#6. 2SRI w/o bio variables (Supplementary Table S1)1.26 (0.03)0.89 (0.52)0.92 (0.72)0.95 (0.93)1.73 (0.06)1.49 (0.02)1.29 (0.06)
#7. 2SRI with bio variables (base-case model; Table 3)1.16 (0.07)0.89 (0.40)0.94 (0.76)1.02 (0.98)1.60 (0.07)1.48 (0.01)1.29 (0.04)
Additional sensitivity analyses
#8. 2SRI base-case w/o missing value imputation (Supplementary Table S2)1.28 (0.10)1.00 (0.99)1.14 (0.64)0.79 (0.71)1.62 (0.10)1.30 (0.12)1.27 (0.22)
#9. 2SRI base-case with sample weights (Supplementary Table S3)1.23 (0.10)1.04 (0.84)1.01 (0.96)0.96 (0.94)1.87 (0.02)1.35 (0.11)1.22 (0.24)
#10. 2SRI model excluding mammography-capacity (Supplementary Table S4)1.16 (0.09)0.89 (0.44)0.96 (0.81)1.44 (0.36)1.72 (0.003)1.57 (<0.001)1.36 (0.02)
#11. Propensity score–adjusted model via IPTW (Supplementary Table S5)1.22 (0.07)1.05 (0.79)0.94 (0.77)1.09 (0.84)1.93 (0.002)1.68 (<0.001)1.38 (0.02)
#12. Propensity score–adjusted model via SMRW (Supplementary Table S5)1.20 (0.12)1.001 (0.99)0.88 (0.55)0.98 (0.97)2.06 (0.002)1.66 (<0.001)1.38 (0.02)
  • aTable entries are logistic model-derived ORs, with corresponding P values in parentheses. ORs in rows #1–3 are derived from logistic models with either one (race/ethnicity or insurance) or two predictor variables (both race/ethnicity and insurance) estimated with the 7,503 observations used in the base-case multivariable models from Table 3; ORs in rows #4–12 are taken directly from the indicated tables.