Table 3. Cox Proportional Hazards Model Including Interaction Term Between Prediagnosis BMI and Postdiagnosis Aspirin Use.
| Characteristic | HR (95% CI) | P value |
|---|---|---|
| Prediagnosis BMI | ||
| Normal (<25) | 1 [Reference] | [Reference] |
| Overweight BMI (25-30) | 1.11 (0.85-1.45) | .45 |
| Obese BMI (≥30) | 1.45 (1.07-1.96) | .02 |
| Current regular aspirin use | ||
| No aspirin use | 1 [Reference] | [Reference] |
| Aspirin use | 0.59 (0.48-0.90) | .01 |
| Interaction effect | ||
| Current regular aspirin use with prediagnosis BMI | ||
| Aspirin use with normal BMI (<25) | 1 [Reference] | [Reference] |
| Aspirin use with overweight BMI (25-30) | 1.77 (1.03-3.07) | .04 |
| Aspirin use with obese BMI (≥30) | 1.22 (0.68-2.18) | .51 |
| Simple main interactionsa | ||
| Aspirin use vs not at normal weight | 0.59 (0.39-0.90) | .01 |
| Aspirin use vs not at overweight | 1.05 (0.74-1.49) | .80 |
| Aspirin use vs not at obese | 0.72 (0.48-1.08) | .11 |
| Weight change | ||
| No change | 1 [Reference] | [Reference] |
| Weight loss (≥10%) | 1.11 (0.85-1.44) | .45 |
| Weight gain (≥10%) | 0.79 (0.62-1.00) | .05 |
| Age at initial diagnosis, y | ||
| 45-65 | 1 [Reference] | [Reference] |
| <45 | 1.14 (0.89-1.48) | .30 |
| ≥65 | 1.37 (1.07-1.77) | .01 |
| Sex | ||
| Men | 1 [Reference] | [Reference] |
| Women | 1.07 (0.87-1.32) | .50 |
| Race and ethnicity | ||
| Non-Hispanic White | 1 [Reference] | [Reference] |
| Non-Hispanic Black or Hispanic | 0.90 (0.69-1.17) | .43 |
| Otherb | 0.80 (0.54-1.18) | .26 |
| Stage at initial diagnosis | ||
| I/II | 1 [Reference] | [Reference] |
| III | 1.25 (0.91-1.71) | .18 |
| IV | 1.43 (1.07-1.92) | .02 |
Abbreviation: BMI, body mass index.
a
The reference for each HR in this section is no aspirin use for the corresponding weight.
b
Other race includes Asian or Pacific Islander, American Indian or Alaska Native, and other or not specified.