Objective: The ability to accurately predict postoperative mortality is expected to improve preoperative decisions for elderly patients considered for colorectal surgery.
Methods: Patients undergoing colorectal surgery were identified from the National Surgical Quality Improvement Program database (2005–2007) and stratified as elderly (>70 years) and nonelderly (<70 years). Univariate analysis of preoperative risk factors and 30-day mortality and morbidity were analyzed on 70% of the population. A nomogram for mortality was created and tested on the remaining 30%.
Results: Of 30,900 colorectal cases, 10,750 were elderly (>70 years). Mortality increased steadily with age (0.5% every 5 years) and at a faster rate (1.2% every 5 years) after 70 years, which defined “elderly” in this study. Elderly (mean age: 78.4 years) and nonelderly patients (52.8 years) had mortality of 7.6% versus 2.0% and a morbidity of 32.8% versus 25.7%, respectively. Elderly patients had greater preoperative comorbidities including chronic obstructive pulmonary disease (10.5% vs 3.8%), diabetes (18.7% vs 11.1%), and renal insufficiency (1.7% vs 1.3%). A multivariate model for 30-day mortality and nomogram were created. Increasing age was associated with mortality [age >70 years: odds ratio (OR) = 2.0 (95% confidence interval (CI): 1.7–2.4); >85 years: OR = 4.3 (95% CI: 3.3–5.5)]. The nomogram accurately predicted mortality, including very high-risk (>50% mortality) with a concordant index for this model of 0.89.
Conclusions: Colorectal surgery in elderly patients is associated with significantly higher mortality. This novel nomogram that predicts postoperative mortality may facilitate preoperative treatment decisions.