INTRODUCTION: Multi-modal pain regimens within enhanced recovery after surgery pathways have markedly decreased postoperative opioid use. However, 5–10% of gynecologic surgery patients still become opioid-dependent postoperatively, and there is paucity of data on outcomes in patients with complex pain disorders. OBJECTIVE: To construct a predictive model based on perioperative patient and surgical characteristics, validated survey scores, and postoperative pain score trajectories to assess differences in postoperative opioid use. METHODS: Subjects undergoing benign gynecologic surgery, including hysterectomy or apical prolapse surgery, were enrolled, completed preoperative brief pain inventory (BPI) and validated questionnaires on pain catastrophizing (PCS), pain behavior, anxiety, and depression. Patients undergoing surgery for malignancy or concomitant non-gynecologic surgery were excluded. Subjects completed daily BPI and pain medication diary starting 6 hours postoperatively and daily until postoperative day seven. Demographic variables, perioperative characteristics and postoperative office visits, office communication, emergency department visits, opioid refills through the state prescription drug monitoring program, and adverse events were collected until 30 days postoperatively. A composite pain measure was created using postoperative day 1 BPI scores, and clustering models were evaluated to group subjects into similar postoperative pain score trajectory. Total postoperative opioid consumption in morphine milliequivalents (MME) was compared between groups, in addition to demographic and perioperative variables. A prediction model was created using binary logistic regression and adjusted for age, BMI, hysterectomy, intraoperative opioid use, PCS score, anxiety, depression, and pain behavior to identify significant predictors of group allocation. A sample size of 66 was needed to meet 90% power. Significance was defined as p < 0.05. RESULTS: A total of 81 subjects were enrolled and 66 with complete data were analyzed. Postoperative day 1 scores identified 3 clinically meaningful phenotypes with different opioid requirements: low pain, moderate pain, and high pain. Groups were similar in age, BMI, comorbid conditions, preoperative psychometric scores, and perioperative characteristics (Tables 1 and 2). However, compared to the low and moderate groups, the high group had more subjects, respectively, with diabetes mellitus (0% vs 7.7% vs 30.8%, p=0.024), generalized anxiety disorder (28.6% vs 12.8% vs 46.2%, p = 0.03), and primary diagnosis of abnormal uterine bleeding (0% vs 11.5% vs 45.5%, p = 0.035). Total postoperative opioid use differed significantly between groups (low 12.18 ± 12.01, moderate 25.77 ± 53.62, high 63.27 ± 45.89 MME, p = 0.015). On logistic regression, preoperative PCS scores (aOR 1.11, p = 0.04) and anxiety (aOR 1.17, p = 0.017) were significantly predictive of higher postoperative pain trajectories, with model accuracy at 83.3% and sensitivity of 94.2%. CONCLUSIONS: Using our model, we can predict with 83.3% accuracy which patients undergoing benign gynecologic surgery will have higher pain trajectories based on postoperative day one pain scores and baseline psychometrics. Subjects in the high pain group used 5–6 times more opioids than those in the low group. Preoperative pain catastrophizing and anxiety are significant predictors of postoperative pain and opioid use. Future studies with larger sample sizes are needed to test and validate this model for generalizability and reliability.Table 1Table 2
Boyd et al. (Fri,) studied this question.