Table 2A.
Primary cox regression analysis of DII and incident prostate cancer (N = 30,972; events = 1,546).
| DII | Model 0 | Model 1 | Model 2 | Model 3 |
|---|---|---|---|---|
| Per 1-unit increase | 0.951 (0.927–0.977) P <0.001 |
0.983 (0.953–1.014) P = 0.287 |
0.987 (0.956–1.019) P = 0.417 |
0.986 (0.956–1.018) P = 0.389 |
| Q1 | 1.000 (reference) | 1.000 (reference) | 1.000 (reference) | 1.000 (reference) |
| Q2 vs Q1 | 0.893 (0.779–1.024) P = 0.105 |
0.956 (0.832–1.098) P = 0.523 |
0.957 (0.833–1.100) P = 0.538 |
0.955 (0.831–1.097) P = 0.514 |
| Q3 vs Q1 | 0.856 (0.746–0.983) P = 0.028 |
0.966 (0.834–1.118) P = 0.640 |
0.971 (0.838–1.124) P = 0.689 |
0.966 (0.835–1.119) P = 0.648 |
| Q4 vs Q1 | 0.781 (0.678–0.899) P <0.001 |
0.913 (0.776–1.073) P = 0.270 |
0.931 (0.791–1.096) P = 0.392 |
0.928 (0.788–1.093) P = 0.370 |
| P for trend | <0.001 | 0.323 | 0.453 | 0.426 |
Values are HRs (95% CIs), with two-sided Wald P values. Model 0: unadjusted. Model 1: age and total energy intake. Model 2: additionally ethnicity, education, household income, smoking, and alcohol. Model 3: additionally SBP, DBP, and diabetes. Time scale: years since baseline; Efron ties. Trend uses quartile number as a continuous term. HR, hazard ratio; CI, confidence interval. The corresponding secondary logistic models are retained in Table 2B.