AAPS Main Site  |  Past & Future Meetings  |  Centennial 2022 Meeting
American Association of Plastic Surgeons

Back to 2021 Abstracts


A Model To Predict Chronic Nerve Pain After Burn Injury
Kevin M. Klifto, PharmD, Pooja S. Yesantharao, MS, A. Lee Dellon, MD, PhD, C. Scott Hultman, MD, MBA.
Johns Hopkins University School of Medicine, Baltimore, MD, USA.

PURPOSE:A model that predicts a patientís risk of developing chronic, neuropathic pain(CNP) after burn injury may facilitate selective medical and/or surgical management starting at initial care and continuing throughout reconstruction to avoid chronic opioids. We constructed a mathematical model predicting a patientís risk of developing CNP after burn injury.
METHODS:A retrospective analysis was conducted from 1880 adults admitted to a single institutionís Burn Unit from 2014-2019. Of the 1880 patients, 113 developed CNP. CNP was defined as patient-described pain evaluated by ≥two clinicians for ≥six months after burn injury, unrelated to a pre-existing illness/medications. The modified Delphi process was used for selection of 78 potential risk variables. Multivariate regression techniques were used to derive the model, Brier scores assessed model performance, Area-under-the-curve(AUC) assessed model discrimination, Hosmer-Lemeshow goodness-of-fit test assessed model calibration, and stratified K-fold cross-validation assessed model accuracy and generalizability. Follow-up was set to six months.
RESULTS:Prevalence rates of CNP were similar in the development(6%) and validation(5.4%) cohorts. CNP risk score=-4.1+0.34(substance use)+0.45(tobacco use)+0.93(surgical treatment)+1.02(%TBSA). Algorithm=1-1/[1+exp(risk score)] for six months post-burn CNP risk score. The model was calibrated to accurately predict the probability of developing CNP(Brier score=0.14). Stratified K-fold cross-validation(k=5): (AUC=0.72; 95%CI 0.64-0.83) A good fit was indicated using the Hosmer-Lemeshow goodness-of-fit test(p=0.21; 10 groups, each with varying %TBSA). As the number of risk factors increase, the probability of CNP increases.
CONCLUSIONS:We present a novel model that accurately predicts a patientís risk of developing CNP after burn injury, with substance use, tobacco use, surgical treatment, and %TBSA representing the greatest predictors.


Back to 2021 Abstracts