Predictive modeling for chronic pain after ventral hernia repair

Background: Few studies predict which patients have dissolution of their postoperative discomfort or develop chronic pain after ventral hernia repair (VHR). This study develops a predictive model to determine which patients are at the greatest risk of chronic pain after VHR.

Methods: A prospective study of VHR patients was performed via the International Hernia Mesh Registry. Anonymous, self-reported, quality of life data using the Carolinas Comfort Scale (CCS) was recorded preoperatively, and 1,6, and 12 months postoperatively. Pain was identified as a score of 2 or more (mild but bothersome to severe) for any Carolinas Comfort Scale pain-specific questions. Logistic regression analyses were performed to determine statistically significant predictors of chronic pain. Univariate analysis selected potential predictors with a P value less than .15, and a subsequent multivariable model was built using backward elimination setting retention criterion at P < .15. Goodness-of-fit of the model was tested using Hosmer-Lemeshow test. A value of greater than 70% for the area under the curve (AUC) was considered most accurate diagnostically. The final model was then internally validated with bootstrap analysis.

Results: A total of 887 patients underwent VHR between 2007 and 2014. The patients had an average age of 57.2 ± 12.8 years, 52.4% were female, 17.0% were active smokers, and 13.2% used narcotics preoperatively. With 74% follow-up at 1 year, 26.0% of the patients reported chronic discomfort. After logistic regression model, independent predictors of pain at 6 months were preoperative pain score 2 or more (P < .0001), preoperative narcotic use (P = .06), and 1-month postoperative pain score 2 or more (P < .0001), AUC = .74. Baseline, 1-month, and 6-month predictors determined the final multivariate regression model for prediction of chronic pain at 1 year, AUC = .73. Older age was protective against chronic pain (odds ratio [OR] .98, 95%confidence interval [CI] = .96 to .998, P = .03), female sex increased risk with an OR of 1.7(CI = 1.1 to 2.7, P = .02); preoperative pain, and recurrent hernia repair nearly doubled the risk of developing chronic pain postoperatively (OR = 3.0, CI = 1.8 to 4.8, P < .0001 and 1.6, CI = .98 to 2.6, P = .06, respectively). Importantly, presence of pain at 1 month was a strong predictor of chronic pain at 1-year follow-up (OR = 2.6, CI = 1.7 to 4.2, P < .0001).

Conclusions: Patients who have preoperative pain and at 1 month postoperatively are significantly more likely to have chronic pain. Both short- and long-term pain can be predicted from female sex, younger age, and repair of recurrent hernias. This predictive model may aid in preoperative counseling and when considering postoperative intervention for pain management in VHR patients.