M. A. Chaudhary1, N. Bhulani1, D. Sturgeon1, N. K. Kwon1, E. D. Jager1, T. P. Koehlmoos2, A. J. Schoenfeld1, A. H. Haider1 1Brigham And Women’s Hospital,Center For Surgery And Public Health,Boston, MA, USA 2Uniformed Services University Of The Health Sciences,Bethesda, MD, USA
Introduction:
Post-surgical opioid prescription is widely maintained to be associated with the burgeoning opioid epidemic. Literature on appropriate post-surgical opioid prescription practices are now making way into policy. Stratification tools for identifying patients at risk of sustained opioid use are not available. In this context, we sought to generate a robust risk-score for identifying post-surgical patients susceptible to sustained opioid use.
Methods:
The Military Health System Data Repository (MDR) was queried for TRICARE claims data (2006-2014). Adult (18-64 years) patients who underwent one of 10 common general surgical, cardiovascular, orthopedic or urological procedures were included in the study. Opioid use 6 months prior to the index procedure and up to 6 months following discharge was assessed. A 75% random sample was generated from the study cohort for model generation and the remaining 25% sample was used for internal validation. Multivariable logistic regression models were utilized determine the predictors of sustained opioid use and a 100-point risk-score was generated using variables available at hospital discharge. C statistic and calibration curves were used to determine model performance.
Results:
A total of 86,356 patients records were included in the study. Among these, 64,767 randomly selected records were used for risk-score generation and 21,589 for internal validation. A total of 7.4% (n=6,365) of patients met criteria for sustained use after the index surgical procedure. The logistic regression model using only variables available at discharge provided robust identification of sustained post-surgical opioid use (area under the receiver operator curve=0.73). The risk-score included the following variables: age [{25-34, score (s):4}{35-44, s:6}{45-54, s:5}{55-64, s:4}], sex (female, s:4), race (non-white, s:-2), lower socio-economic status (s:7), prior opioid use (s:26), comorbid diagnosis [{diabetes, s:5}{depression, s:5}{anxiety, s:7}], procedure type [{urological, s:4}{orthopedic, s:7}], ICU admission (s:5), length of stay (>3 days, s:4) and discharge disposition (non-home discharge, s:13). The risk-score was further categorized based on the likelihood of sustained opioid use (Table). The risk-score performance in the validation sample was at par with the test sample.
Conclusion:
This study developed and validated a risk-score for sustained post-surgical opioid use that may have significant utility for clinical practice in discharge planning and engagement of pain management services for at-risk patients. Timely identification and appropriate outpatient care planning may reduce the incidence of sustained post-surgical opioid use in these patients.
