A. Pal1, E. Mahmood3, J. Nicastro2, M. Sfakianos2, T. Dinitto1, S. M. Cohn1 2North Shore University And Long Island Jewish Medical Center,Department Of Surgery,Manhasset, NY, USA 3Northwestern University,Feinberg School Of Medicine,Chicago, IL, USA 1Staten Island University Hospital, Northwell Health,Surgery,Staten Island, NY, USA
Introduction: There is a need for predictive models that can help surgeons identify patients at greatest risk for wound dehiscence in order to guide their management to avoid evisceration. We sought to use a large database in order to examine risk factors for developing this complication after midline laparotomy.
Methods: The American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) is a prospectively collected surgical outcomes database compiled by manual chart abstraction. Exploratory laparotomy cases were queried using the primary CPT code from 2005-2013. The independent factors associated with wound dehiscence were examined by multivariate analysis using SAS JMP Pro 11 (Cary, NC, US). The cohort was split into a training dataset of patients from 2005-2009 and a prospective validation dataset from 2010-2013. A backwards logistic regression analysis was performed to identify predictors of wound dehiscence in the training set. The model was then tested in the validation set to estimate the receiver operating curves (ROC) and goodness of fit.
Results: A total of 16,793 patients were included in our analysis. 248 (1.47%) of these patients had a wound dehiscence. Significant predictors of wound dehiscence: deep wound infection (AOR=5.98, 95% CI 3.06 to 10.9, P<0.0001), postoperative pneumonia (AOR=3.25, 95% CI 1.99 to 5.11, P<0.0001), preoperative weight loss (AOR=3.11, 95% CI 1.29 to 10.2, P<0.0083), preoperative sepsis (AOR=3.03, 95% CI 1.91 to 4.70, P<0.0001), superficial wound infection (AOR=2.97, 95% CI 1.63 to 5.05, P<0.0007), and previous operation in the last 30 days (AOR=1.82, 95% CI 1.19 to 2.73, P<0.0061), smoking (AOR=1.49, 95% CI 1.01 to 2.18, P<0.044). The c-statistic for our model was reasonable: 0.73 in the training set and 0.70 in the validation set. The Hosmer-Lemenshow goodness-of-fit statistic was 0.89.
Conclusion: We identified a number of independent risk factors for the development of wound dehiscence which may inform the clinician and lead to improved selection of patients for measures which could reduce the likelihood of evisceration (retention sutures) after laparotomy.
