Development of a Clinically Actionable Incisional Hernia Risk Model after Colectomy Using the Healthcare Cost and Utilization Project

Abstract

Background: Incisional hernia remains a persistent and burdensome complication after colectomy. Through individualized risk-assessment and prediction models, we aimed to improve preoperative risk counseling for patients undergoing colectomy; identify modifiable preoperative risk factors; and encourage the use of evidence-based risk-prediction instruments in the clinical setting.

Study design: A retrospective review of the Healthcare Cost and Utilization Project data was conducted for all patients undergoing either open or laparoscopic colectomy as identified through the state inpatient databases of California, Florida, and New York in 2009. Incidence of incisional hernia repair was collected from both the state inpatient databases and the state ambulatory surgery and services databases in the 3 states between index surgery and 2011. Hernia risk was calculated with multivariable hierarchical logistic regression modeling and validated using bootstrapping techniques. Exclusion criteria included concurrent hernia, metastasis, mortality, and age younger than 18 years. Inflation-adjusted expenditure estimates were calculated.

Results: Overall, 30,741 patients underwent colectomy, one-third of these procedures performed laparoscopically. Incisional hernia repair was performed in 2,563 patients (8.3%) (27-month follow-up). Fourteen significant risk factors were identified, including open surgery (odds ratio = 1.49; p < 0.0001), obesity (odds ratio = 1.49; p < 0.0001), and alcohol abuse (odds ratio = 1.39; p = 0.010). Extreme-risk patients experienced the highest incidence of incisional hernia (19.8%) vs low-risk patients (3.9%) (C-statistic = 0.67).

Conclusions: We present a clinically actionable model of incisional hernia using all-payer claims after colectomy. The data presented can structure preoperative risk counseling, identify modifiable patient-specific risk factors, and advance the field of risk prediction using claims data.