Table 3.
Variable | p value | OR (95% CI) |
---|---|---|
Age per year | 0.10 | 1.006 (0.999–1.014) |
Sex, female v. male | 0.049 | 1.242 (1.001–1.541) |
Primary cancer type | ||
Lung v. other | < 0.001 | 2.139 (1.702–2.687) |
Breast v. other | 0.85 | 0.974 (0.743–1.277) |
Kidney v. other | 0.84 | 0.959 (0.647–1.422) |
Prostate v. other | 0.08 | 0.685 (0.447–1.050) |
Type of surgery | ||
Fracture fixation v. prophylactic | < 0.001 | 1.478 (1.225–1.785) |
CI = confidence interval; OR = odds ratio.
We computed the risk of death given the presence of each variable using Cox regression. We adjusted each variable for the presence of all others. Cox regression is appropriate for evaluating and adjusting for the effect of multiple factors (analogous to multiple regressions) when the outcome involves censored data such as survival.