A novel preoperative model to predict 90-day surgical mortality in patients considered for renal cell carcinoma surgery

Abstract

Introduction Surgical benefits for renal cell carcinoma must be weighed against competing causes of mortality, especially in the elderly patient population. We used a large cancer registry to evaluate the impact of patient and cancer-specific factors on 90-day mortality (90DM). A nomogram to predict the odds of short-term mortality was created.

Materials and Methods The National Cancer Database was queried to identify all patients with clinically localized, nonmetastatic disease treated with partial or radical nephrectomy. Using a random sample of 60%, multiple logistic regression with 90DM outcomes were performed to identify preoperative variables associated with mortality. Variables included age, sex, race, co-morbidity score, tumor size, and presence of a thrombus. A nomogram was created and tested on the remaining 40% of patients to predict 90DM.

Results 183,407 patients met inclusion criteria. Overall 90DM for the cohort was 1.9%. All preoperative variables significantly influenced the risk of 90DM. Patient age was by far the strongest predictor. Nomogram scores ranged from 0 to 12. Compared to patients with 0 to 1 points, those with 2 to 3 (odds ratio [OR] 2.89, 2.42–3.46; P < 0.001), 4 to 5 (OR 6.25, 5.26–7.43; P < 0.001), and >6 (OR 12.86, 10.83–15.27; P < 0.001) were at incrementally significantly higher odds of 90DM. Being >80 years of age alone placed patients into the highest risk of surgical mortality.

Conclusions Management of localized kidney cancer must consider competing causes of mortality, especially in elderly patients with multiple co-morbidities. We present a preoperative tool to calculate risk of surgical short-term mortality to aid surgeon–patient counseling.