A tool for prediction of chronic postoperative pain in cardiac surgery

Objective. To develop a tool for prediction of chronic postoperative pain based on identified predictors and decision-making algorithm for its prevention in patients undergoing cardiac and/or aortic surgery. Material and methods. An iterative logistic regression analysis was performed to identify optimal prognostic model for chronic postoperative pain. The baseline model included intense pain in early postoperative period (OR=6.3; p=0.0001). The following variables were sequentially added: central sensitization (CSI), anxiety (HADS), and pain catastrophizing (PCS). Models were compared using the AIC, BIC and likelihood ratio test (LRT). The quality of the final model was assessed by ROC analysis. Results. Inclusion of CSI (Model #2) significantly improved the model compared to baseline one (LRT: p=0.019). Inclusion of anxiety (p=0.24) or pain catastrophizing (p=0.69) did not improve prognosis. The final model included two predictors: intense pain in early postoperative period (β=1.69; p<0.001) and central sensitization (β=0.051; p=0.022). ROC analysis showed AUC 0.78 (95% CI 0.68–0.88). Sensitivity was 66.7%, and specificity was 80.0% at a threshold of 0.36. Conclusion. Original predictive model enables effective identification of patients with high risk of chronic postoperative pain after cardiac surgery and facilitates transition from reactive treatment to proactive prevention. This improves quality of life in late postoperative period.