Testing of a novel easy-to-use mortality index in a radical prostatectomy cohort

To test a novel and easy-to-use mortality index developed and validated in a nationally representative US sample of older adults in patients selected for radical prostatectomy.

The sample comprised 2205 consecutive patients who underwent radical prostatectomy in the years 1992-2005. The median age was 64 years and the median follow-up was 9.7 years. A slightly modified Lee mortality index was calculated based on information obtained by chart review. The 10-year competing mortality rates were compared with overall mortality rates previously reported in a US population-based cohort using the given confidence intervals with the 2-tailed Wald test. Cox proportional hazard models were calculated to analyze the combined effects of variables using overall mortality as study endpoint.

The modified Lee mortality index provided a clear dose-response pattern in the radical prostatectomy cohort with 10-year competing mortality rates between 2.5% and 50%. Overall, the competing mortality rates reached only narrowly two-thirds of the values predicted by the modified Lee mortality index, reflecting selection for healthy operative candidates (chi-square homogeneity test; P = .0043). In the multivariate analysis considering the modified Lee mortality index, age, and tumor-related variables, the Lee mortality index was identified as an independent predictor of overall survival. Comorbidity measures delivering similar prognostic information as the modified Lee mortality index, however, largely replaced this index in the multivariate models.

The modified Lee mortality index could be suited as comorbidity measure in men with prostate cancer. Practical use would require adjustment for selection effects.