Risk score for nonhome discharge after lower extremity bypass.

OBJECTIVE

Patients undergoing lower extremity bypass (LEB) for peripheral artery disease require intensive health care resource utilization including rehabilitation and skilled nursing facilities. However, few studies have evaluated factors that lead to nonhome discharge (NHD) in this population of patients. This study sought to predict NHD by preoperative risk factors in patients undergoing LEB for peripheral artery disease using a novel risk score.

METHODS

The Vascular Study Group of New England database was queried for elective LEB for peripheral artery disease including claudication and critical limb ischemia from 2003 to 2017. Patients were excluded if the procedure was not elective, if they were not admitted from home, if they were bedridden, or if they died during the index admission. Only preoperative factors were considered in the analysis. The primary end point was NHD including rehabilitation and skilled nursing facilities. Data were split two-thirds for model derivation and one-third for validation. In the derivation cohort, bivariate analysis assessed the association of preoperative factors with NHD. A parsimonious manual stepwise binary logistic regression for NHD aimed at maximizing the C statistic while maintaining model simplicity was performed. A risk score was developed using the β coefficients and applied to the validation data set. The risk score performance was assessed using a C statistic and Hosmer-Lemeshow test for model fit.

RESULTS

There were 10,145 cases included with an overall NHD rate of 26.4% (n = 2676). Mean age was 66 years (range, 41-90 years). NHD patients were older (72 years vs 64 years; P < .01) and more frequently male (57.2% vs 42.8%; P < .01) and nonwhite (16.1% vs 9.9%; P < .01); they more frequently had tissue loss (54.2% vs 23.0%; P < .01), anemia (16.0% vs 5.3%; P < .01), severe cardiac comorbidity (21.8% vs 10.5%; P < .01), and insulin-dependent diabetes (33.3% vs 18.2%; P < .01). On multivariable analysis, factors associated with NHD included age, sex, nonwhite race, tissue loss, cardiac comorbidity, partial ambulatory deficit, and insulin-dependent diabetes. The C statistic was 0.78 in the derivation group and 0.79 in the validation group, with Hosmer-Lemeshow P > .999. The risk score ranged from 0 to 18, with a mean score of 4 (standard deviation ±3.5). The risk score was divided into low risk (0-4 points; n = 5272 [52%]; NHD = 10.1%]), moderate risk (5-9 points; n = 3663 [36.7%]; NHD = 36.7%), and high risk (≥10 points; n = 1210 [11.9%]; NHD = 66.1%).

CONCLUSIONS

This novel risk score was highly predictive for NHD after LEB for peripheral artery disease using only preoperative comorbidities. High-risk patients account for 12% of LEB but nearly a third of all patients requiring NHD. This risk score can be used preoperatively to determine high-risk patients for NHD, which may help improve preoperative counseling and hospital efficiency by allocating resources appropriately.