Clinical Knowledge and Trends in Physicians' Prescribing of Opioids for New Onset Back Pain, 2009-2017

IMPORTANCE Opioid musculoskeletal pain overprescribing was widespread in the mid-2000s. The degree to which prescribing changed as awareness of the danger grew among physicians with different levels of clinical knowledge remains unstudied. OBJECTIVE To compare the association of clinical knowledge with opioid prescribing from 2009 to 2011 when prescribing peaked nationally with 2015 to 2017 when guidelines shifted away from opioid prescribing. DESIGN, SETTING, AND PARTICIPANTS This cross-sectional study included 10 246 midcareer general internal medicine physicians in the United States who saw patients who were Medicare beneficiaries with Part D enrollment from 2009 to 2017. MAIN OUTCOMES AND MEASURES Any opioid prescription and high dosage or long duration (HDLD) (>7 days or >50 daily morphine milligram equivalents) opioid prescriptions filled within 7 days of applicable visits for new low back pain concerns. Associations between opioid prescribing for new low back pain concerns during outpatient visits and clinical knowledge measured by prior year American Board of Internal Medicine (ABIM) Maintenance of Certification examination performance were estimated using serial cross-sectional logit regressions. Regression covariates included yearly examination quartile (ie, knowledge quartile) interacted with 3-year group dummies (ie, early: 20092011; middle: 2012-2014; late: 2015-2017), state and year dummies, physician, practice, patient characteristics, and state opioid regulations. RESULTS Of the 55 387 low back pain visits included in this study, 37 185 (67.1%) were visits with female patients, 41 978 (75.8%) were with White patients, and the mean (SE) age of patients was 76.2 (<0.01) years. The rate of opioid prescribing was 21.6%(11 978) for any opioid prescription and 17.6%(9759) for HDLD prescriptions. From 2009 to 2011, visits with physicians in the highest and lowest knowledge quartiles had similar adjusted opioid prescribing rates with a 0.5 (95% CI, -1.9 to 3.0) percentage point difference. By 2015 to 2017, visits with physicians in the highest knowledge quartile prescribed opioids less frequently that physicians in the lowest knowledge quartile (4.6 percentage point difference; 95% CI, -7.5 to -1.8 percentage points). Visits in which HDLD opioids were prescribed showed no difference in the early period but showed a difference in the late period when comparing physicians in the highest and lowest knowledge quartiles (early period: difference -0.1; 95% CI, -2.4 to 2.2 percentage points; late period difference: 4.8; 95% CI, -7.4 to -2.1 percentage points). CONCLUSIONS AND RELEVANCE In this cross-sectional study, when the standard of care shifted away from routine opioid prescribing, physicians who performed well on an ABIM examination were less likely to prescribe opioids for back pain than physicians who performed less well on the examination.