Purpose: There is no consensus on the best comorbidity measure in candidates for radical cystectomy. The aim of this study was to identify tool best suited to identify patients at risk for 90-day or premature long-term non-bladder cancer mortality.
Methods: We studied 1268 patients who underwent radical cystectomy to identify patients at risk for 90-day and later-than-90-day mortality, respectively. Six classifications were investigated as possible predictors of both types of mortality. Multivariable models including age as continuous variable and each classification separately were calculated. A heuristic ranking was based on the evaluation of the hazard ratios, p values, Akaike's information criteria, and concerning the logit models also the areas under the curve.
Results: The median follow-up was 5.7 years. Within 90 days after surgery, the mortality rate was 4.2%. The greatest independent contribution concerning the prediction of 90-day mortality was seen with the American Society of Anesthesiologists (ASA) physical status classification (classes 3-4 versus 1-2: hazard ratio 7.98, 95% confidence interval 3.54-18.01, p < 0.0001). In the longer term, countable diseases (Canadian Cardiovascular Society classification of angina pectoris, conditions contributing the Charlson score) were of greater importance. The results of heuristic ranking were confirmed by multivariate analyses including age and all classifications together.
Conclusions: Besides to chronological age, clinicians should pay particular attention to the ASA classification to identify patients at risk for 90-day mortality after radical cystectomy, whereas long-term mortality is more determined by countable comorbid diseases.
Keywords: 90-day mortality; Bladder; Comorbidity; Competing mortality; Competing risk analysis; Cystectomy; Logit model; Proportional hazards model.