Table 8

Cox regression results for breast cancer mortality*

Analysis, exposure variableModel likelihood, df, p value†Coefficient (SE)Odds ratios by category‡
*There were 103 cases of breast cancer (ICD 9th 174, 175). In Cox regression, cases and controls were matched on age, race (white/non-white), and date of birth within five years.
†Model likelihood is difference in −2 log likelihoods between model with and without covariates; the only covariate in these models was exposure, so the p value of the model serves as a test of significance of the exposure coefficient, and as a test of exposure-response trend.
‡Categories for cumulative exposure are the same as in table 6.
Cumulative exposure0.88, 1 df, p = 0.340.0000049 (0.0000048)na
Log cumulative exposure, 20 year lag5.69, 1df, p = .010.084 (0.035)na
Categorical cumulative exposure lagged 20 years (quartiles)8.69, 4 df, p = 0.07na1.00, 1.76 (0.91–3.43), 1.77 (0.88–3.56), 1.97 (0.94–4.06), 3.13 (1.42–6.92)