Advertisement

Perioperative opioid-prescribing practices of resident trainees compared with staff surgeons

Published:February 12, 2022DOI:https://doi.org/10.1016/j.ajog.2022.02.011

      Objective

      Little is known about the opioid-prescribing practices of surgical trainees. Our objective was to evaluate the opioid prescriptions of resident trainees compared with academic and community staff surgeons following elective hysterectomy.

      Study Design

      We performed a population-based cohort study using linked administrative data in Ontario, Canada, where all dispensed prescription opioids are recorded, regardless of insurance status.
      • Chan W.V.
      • Le B.
      • Lam M.
      • et al.
      Opioid prescribing practices for women undergoing elective gynecologic surgery.
      We included opioid-naïve women (age ≥18 years) who underwent elective hysterectomy between January 1, 2013 and March 31, 2019 and filled at least 1 opioid prescription in the perioperative period (day of hysterectomy to 7 days after). We excluded emergency surgeries, patients with malignancy, history of opioid toxicity, those who received opioids in the previous year, and those who underwent additional surgery in the 30 days preceding hysterectomy. The exposure was the opioid prescriber (trainee vs academic or community surgeon) for the opioid prescription filled on or closest to the date of surgery. The primary outcome was high-dosage opioid prescription, defined as >225 mg total oral morphine equivalents (OME), equivalent to >30 5 mg oxycodone tablets, well above the recommended total dosage following hysterectomy.
      Opioid Prescribing Engagement Network (OPEN)
      Prescribing recommendations.
      We conducted 3 pairwise comparisons, in which the trainees were compared with all the staff and each of the 2 staff groups. To account for confounding, we used inverse probability of treatment weighting where we generated a propensity score using multiple logistic regression for being in a given prescriber group regressed on the baseline covariates.
      • Austin P.C.
      • Stuart E.A.
      Moving towards best practice when using inverse probability of treatment weighting (IPTW) using the propensity score to estimate causal treatment effects in observational studies.
      Observations were weighted according to the inverse of the calculated probability of being in the given prescriber group. In the weighted sample, we reported the risk ratios and mean differences (MDs). We estimated the 95% confidence intervals (CIs) using nonparametric percentiles from 2000 bootstrapped samples using sampling that accounted for clustering within prescribers.
      • Austin P.C.
      • Leckie G.
      Bootstrapped inference for variance parameters, measures of heterogeneity and random effects in multilevel logistic regression models.
      Use of the study data was authorized by Ontario’s Personal Health Information Protection Act and did not require Research Ethics Board review.

      Results

      We included 20,352 patients who filled a perioperative opioid prescription (4362 [21.4%] written by trainees, 1727 [8.5%] by academic surgeons, and 14,263 [70.1%] by community surgeons). After weighting, the baseline covariates were balanced between the prescriber groups (Supplemental Table). High-dosage opioid prescriptions were filled by 2091 (10.3%) patients. There was no difference in the risk of receiving a high-dosage opioid prescription written by a staff surgeon vs that of receiving a prescription by a trainee [risk ratio (RR) 1.35; 95% CI, 0.93–2.00); there was also no difference between academic surgeon vs trainee (RR 1.22; 95% CI, 0.79–1.93) nor between community surgeon vs trainee (RR 1.36; 95% CI, 0.91–2.02). The findings were consistent in the subgroup analyses for minimally invasive and open hysterectomy (Table). Staff surgeons, from both the academic and community sites, prescribed more opioid quantities (mean difference in OMEs) than the trainees: MD 16.11 mg; 95% CI, 7.53–24.95.
      TableMain outcomes in weighted sample
      PrescriberHigh-dosage opioid prescriptionTotal perioperative OME (mg)OME value

      Recommended OME
      Opioid Prescribing Engagement Network (OPEN)
      Prescribing recommendations.
      Event rate (%)Risk ratio (95% CI)Mean±SDMean difference (95% CI)
      All hysterectomies (N=20,352)
      Trainee
      The trainee referent group weights differ for the 2 pairwise comparisons. The data presented above are from the weighted comparison of the trainees vs academic staff.
      6.0Ref144.59±74.56Ref
      All staff11.31.35 (0.93–2.00)164.59±79.7016.11 (7.53–24.95)
      Academic surgeon7.41.22 (0.79–1.93)157.60±70.6413.01 (1.85–25.77)
      Community surgeon11.81.36 (0.91–2.02)165.53±80.8016.75 (6.99–26.21)
      Minimally invasive hysterectomy (N=12,188)
      Trainee
      The trainee referent group weights differ for the 2 pairwise comparisons. The data presented above are from the weighted comparison of the trainees vs academic staff.
      5.5Ref142.21±72.47Ref<113 mg
      All staff10.01.25 (0.77–2.03)160.81±80.0415.85 (5.43–25.84)
      Academic surgeon6.71.22 (0.67–2.06)158.91±71.6516.70 (3.93–28.94)
      Community surgeon10.51.26 (0.78–2.04)160.91±81.5915.66 (4.73–27.08)
      Open hysterectomy (N=8,164)
      Trainee
      The trainee referent group weights differ for the 2 pairwise comparisons. The data presented above are from the weighted comparison of the trainees vs academic staff.
      7.1Ref149.49±78.59ref<150 mg
      All staff13.21.46 (0.96–2.29)170.29±78.9115.21 (2.73–26.98)
      Academic surgeon8.91.25 (0.75–2.00)154.65±68.825.16 (−8.85 to 20.44)
      Community surgeon13.71.48 (0.94–2.38)172.35±79.2917.17 (5.01–29.17)
      CI, confidence interval; mg, milligrams; OME, oral morphine equivalent; ref, reference interval; SD, standard deviation.
      Murji. Perioperative opioid-prescribing practices of resident trainees compared with staff surgeons. Am J Obstet Gynecol 2022.
      a The trainee referent group weights differ for the 2 pairwise comparisons. The data presented above are from the weighted comparison of the trainees vs academic staff.

      Conclusion

      Trainees wrote >20% of the hysterectomy-associated opioid prescriptions. The high-dosage opioid prescriptions were similar between the trainees and staff. Although trainees prescribed statistically lower opioid dosages, the clinical significance of the approximately 2 fewer 5 mg oxycodone pills per patient is uncertain. A limitation of evaluating the trainee prescribing practices is the unknown influence of supervising attendings on trainee prescriptions. Nonetheless, prudent perioperative opioid prescribing is a shared responsibility of staff and trainees alike and should be integrated into medical education.

      Appendix

      Supplemental TableBaseline characteristics of patients after inverse probability of treatment weighting
      CharacteristicsTrainee vs all staff surgeonsTrainee vs academic surgeonTrainee vs community surgeon
      Trainee (N=4362)All staff (N=15,990)St DiffTrainee (N=4362)Academic surgeon (N=1727)St DiffTrainee (N=4,362)Community surgeon (N=14,263)St diff
      Age (y), mean±SD47.97±10.5548.04±10.340.0149.39±10.8749.46±11.800.0147.61±10.8747.87±11.800.03
      Charlson index (%)
       06.2%5.8%0.026.6%6.7%0.005.9%5.7%0.01
       10.5%0.6%0.000.8%0.7%0.010.5%0.5%0.01
       20.2%0.1%0.020.2%0.3%0.010.2%0.1%0.01
       ≥30.1%0.1%0.010.1%0.1%0.000.0%0.1%0.02
       No hospitalizations93.0%93.4%0.0292.3%92.3%0.0093.5%93.6%0.01
      Comorbidities (%)
       Any mental health diagnosis29.6%30.7%0.0232.7%33.0%0.0128.9%30.3%0.03
       Substance use disorder1.1%1.0%0.011.3%1.2%0.001.0%1.0%0.00
       Psychotic disorder1.8%1.9%0.002.1%2.3%0.011.9%1.9%0.00
       Mood disorder6.3%6.6%0.017.3%7.6%0.016.0%6.4%0.02
       Anxiety disorder24.4%24.9%0.0126.3%26.3%0.0023.9%24.6%0.02
      Income quintile (%)
       1st17.5%16.4%0.0315.9%15.7%0.0017.9%16.4%0.04
       2nd20.2%19.8%0.0119.5%19.6%0.0020.3%19.9%0.01
       3rd19.0%20.9%0.0520.6%20.6%0.0018.7%21.0%0.06
       4th22.2%22.1%0.0021.7%22.0%0.0122.3%22.1%0.00
       5th21.0%20.8%0.0022.3%22.0%0.0120.8%20.5%0.01
      Rural residence (%)13.9%13.3%0.029.3%9.2%0.0014.6%13.5%0.03
      Minimally invasive hysterectomy (%)65.3%60.2%0.1167.2%69.2%0.0464.4%59.6%0.10
      Surgery duration quintile (%)
       1st20.1%19.6%0.018.3%8.1%0.0121.9%20.5%0.04
       2nd20.5%20.4%0.0011.7%11.4%0.0121.1%20.7%0.01
       3rd19.0%19.2%0.0116.0%15.4%0.0218.9%19.3%0.01
       4th19.3%19.6%0.0125.1%24.2%0.0218.6%19.5%0.02
       5th20.1%20.1%0.0037.5%39.3%0.0418.4%19.0%0.02
      Diagnosis (%)
       Abnormal uterine bleeding52.1%50.8%0.0241.2%40.5%0.0154.6%52.2%0.05
       Prolapse27.1%26.0%0.0232.4%32.4%0.0025.3%25.0%0.01
       Pelvic inflammatory disease6.4%7.1%0.0325.7%25.3%0.0128.4%27.1%0.03
       Pain or endometriosis27.8%27.2%0.011.3%1.4%0.011.9%1.7%0.01
       Postmenopausal issues1.8%1.7%0.0142.2%43.8%0.0343.7%45.1%0.03
       Fibroids43.2%44.3%0.0211.2%11.6%0.0113.5%13.3%0.01
       Ovarian mass or cysts12.8%12.9%0.0041.2%40.5%0.0154.6%52.2%0.05
      Length of hospital stay (d), mean±SD1.81±1.011.78±0.970.031.55±1.021.51±0.880.041.88±1.051.81±0.990.07
      SD, standard deviation; St Diff, standardized difference.
      Murji. Perioperative opioid-prescribing practices of resident trainees compared with staff surgeons. Am J Obstet Gynecol 2022.

      References

        • Chan W.V.
        • Le B.
        • Lam M.
        • et al.
        Opioid prescribing practices for women undergoing elective gynecologic surgery.
        J Minim Invasive Gynecol. 2021; 28: 1325-1333.e3
        • Opioid Prescribing Engagement Network (OPEN)
        Prescribing recommendations.
        (Available at:)
        • Austin P.C.
        • Stuart E.A.
        Moving towards best practice when using inverse probability of treatment weighting (IPTW) using the propensity score to estimate causal treatment effects in observational studies.
        Stat Med. 2015; 34: 3661-3679
        • Austin P.C.
        • Leckie G.
        Bootstrapped inference for variance parameters, measures of heterogeneity and random effects in multilevel logistic regression models.
        J Stat Comput Simul. 2020; 90: 3175-3199