Risk prediction of 30-day mortality after lower extremity major amputation.

OBJECTIVE: Universal risk calculators may underestimate mortality risk, whereas purely observational administrative data may lack appropriate granularity to individualize risk. The purpose of this study was to create a simple risk prediction model to identify the factors associated with 30-day morality after lower extremity major amputation for ischemic vascular disease.
METHODS: The Veterans Affairs Surgical Quality Improvement Program national data set was queried from 2005 to 2015 to identify 14,890 patients undergoing elective above-knee or below-knee amputation for rest pain, tissue loss, or gangrene. The data set was divided into a two-thirds derivation set and one-third validation set for the purpose of creating a risk prediction model. The primary end point was 30-day mortality. Eight independent risk factors for mortality resulted from the model and were assigned whole number integer risk scores. Summary risk scores were collapsed into categories and defined as low (0-3 points), moderate (4-7 points), high (8-10), and very high (>10).
RESULTS: Mortality in the derivation data set was 4.6% (n = 453). By multivariable backward elimination, predictors of 30-day mortality (odds ratio [95% confidence limits]) included preoperative do not resuscitate order (3.1 [2.3-4.0]), congestive heart failure (2.8 [2.1-3.6]), age >80 years (1.8 [1.4-2.2]), chronic renal insufficiency (2.1 [1.7-2.5]), above-knee amputation (1.8 [1.4-2.2]), dependent functional status (2.0 [1.6-2.5]), coronary artery disease (1.3 [1.1-1.6]), and chronic obstructive pulmonary disease (1.3 [1.0-1.6]); the final model held a C statistic of 0.74. In both the derivation and validation sets, 30-day mortality correlated with risk category. Among the defined categories in the derivation set, 30-day mortality rates were 2.3% for low-risk patients, 4.3% for moderate-risk patients, 7.5% for high-risk patients, and 17.5% for very-high-risk patients, with similar results for the validation data set.
CONCLUSIONS: This risk prediction model uses eight easily obtainable clinical metrics that allow early assessment of 30-day mortality risk of patients undergoing major lower extremity amputation for ischemic indications. The internal validation of the risk score demonstrates the increased mortality with increasing risk category. Reliable expected mortality prediction is critically important for surgeons to make recommendations in accordance with a patient's or family's goals of care. These data may also be used to set realistic expectations for hospital-based quality initiatives and to provide guidance in preoperative medical optimization.