Opioid Prescription After Carpal Tunnel Release Is Declining Independent of State Laws

Daniel J. Cunningham, MD, MHSc; Eliana B. Saltzman, MD; Daniel J. Lorenzana, MD; Christopher S. Klifto, MD; Marc J. Richard, MD; Tyler S. Pidgeon, MD


J Am Acad Orthop Surg. 2021;29(11):486-497. 

In This Article


Study Design

This is a retrospective, observational study of perioperative opioid filling in patients undergoing primary CTR between 2010 and 2018 using a large, national database. This study was designed and reported in accordance with the Strengthening Reporting of Observational Studies in Epidemiology (STROBE) statement on observational studies.[20]

Variables and Data Sources

The PearlDiver Mariner data set was used to evaluate opioid filling trends in all patients undergoing primary open and endoscopic CTR (see Supplemental Digital Content, Appendix for CPT and ICD procedural codes used to identify primary CTR, http://links.lww.com/JAAOS/A587). In line with recent recommendations on opioid-related database studies in orthopaedic surgery, this data set was selected because of the granular, patient-level information that it provides on opioid filling.[21] PearlDiver is a payer database covering 122 million distinct patients between Q1 2010 and Q2 2018 that includes information on patients in all US states and territories. This database is collated and updated by PearlDiver, Inc. Importantly, PearlDiver facilitates longitudinal, patient-specific analysis of International Classification of Diseases (ICD)-9 diagnosis codes, ICD-10 diagnosis and procedure codes, Current Procedural Terminology (CPT) codes, and National Drug Code (NDC) for the duration of their insured status. This database tracks insured patients across states, prescribers, and care settings including inpatient and outpatient settings. This database includes information on any prescription that was filled at all pharmacies within the United States except for inpatient pharmacies. Because most patients are not kept as inpatients for CTR, this limitation is not likely to alter study inferences. A variety of payer types (commercial, Medicare, Medicaid, and cash pay) are included. To select a cohort of patients with broad applicability to patients undergoing primary elective CTR; exclusion criteria included patients less than 18 years of age; same-day distal radius fracture fixation (see Supplemental Digital Content, Appendix for codes used to exclude fracture, http://links.lww.com/JAAOS/A587); patients without active insurance status from 6 months preoperative to 30 days, 90 days, and 1 year postoperative; and patients with exceptionally high perioperative opioid demand (>8,000 morphine milliequivalents or 1,067 oxycodone 5-mg pills filled from 1 month preoperative to 1 year postoperative). Three cohorts were identified to analyze 30-day, 90-day, and 1-year perioperative opioid filling patterns: (1) active insurance status from 6 months preoperative (pre-op) to 1 month postoperative (post-op), n = 75,933; (2) active insurance status from 6 months pre-op to 3 months post-op, n = 74,915; and (3) active insurance status from 6 months pre-op to 1 year post-op, n = 58,290. These groupings were selected to maximize short-term and long-term follow-up cohort sizes. Descriptions of the 30-day and 1-year cohorts are included in the Supplemental Digital Content, Appendix, http://links.lww.com/JAAOS/A587. Patients could be in all three groups depending on their perioperative insurance status. Outcomes included opioid volume filled among patients who filled at least one prescription, the rate of opioid filling and no filling, and opioid refills over the study time frames. Cumulative oral morphine milliequivalents were tabulated over these time frames and were converted to oxycodone 5-mg pills for ease of interpretation using conversion factors proposed by the Centers for Disease Control (CDC).[22] Rates of one and two or more opioid prescriptions during each time frame were also calculated.

Baseline patient and surgical factors were recorded including age, sex, obesity, Charlson comorbidity index (CCI), preoperative (6 months to 1 month preoperative) opioid filling, year of surgery, and state of opioid prescription. Opioid prescribing within the 6 month to 1 month preoperative window was evaluated as an indicator of chronic opioid use, similar to definitions used by the CDC, but modified to exclude the 1-month preoperative period due to the fact that patients may have filled an opioid intended for postoperative usage before their surgery.[23]

State-specific legislation or governing-body opioid prescribing rule changes were reviewed to determine dates of opioid-limiting laws (see Supplemental Digital Content, Appendix Table 1, http://links.lww.com/JAAOS/A587).[4] As some legislation only pertained to specific subpopulations of state residents (eg, patients on Medicaid), only rules ertaining to all state residents were considered for the purposes of these analyses. Of note, some state rules had gradual roll-out of various portions of opioid limitations. Only the rule's passage date or effective date was considered. Furthermore, only legislation passed and/or enacted before April 1, 2018, was considered because this database only includes patients through the second quarter of 2018. For comparison, opioid prescribing in states without opioid legislation were also evaluated. March 1, 2017, was used as the before/after date for these states without opioid legislation because it was the mean date of legislation passage/effectiveness in the cohort of states in which opioid legislation was passed. States with fewer than 11 patients (most commonly in comparing postact outcomes in low cohort states) were excluded from analyses because PearlDiver does not permit statistical calculations on populations of patients less than 11. Once these before/after dates were identified, opioid prescribing in specific states was analyzed before and after the passage of these laws. Furthermore, yearly state-level mean opioid prescribing was also recorded.

Statistical Analysis

Descriptive statistics including means (standard deviations) or proportions (percentages) were calculated and displayed as appropriate for baseline characteristics. Unadjusted outcomes were calculated for each time frame including mean oxycodone 5-mg pills per patient who filled at least one opioid prescription and rates of one or more or two or more opioid prescriptions. A Student t-test was used to assess changes in volume of oxycodone 5-mg pills filled per filler before and after legislation within each state. Multivariable linear regression using ordinary least squares regression was used to assess the adjusted associations of age, sex, obesity, CCI, preoperative opioid filling, year of surgery, and state of opioid prescription on the volume of opioid prescribed and the rate of opioid filling for each perioperative time frame. Where possible, the impact of orthopaedic versus nonorthopaedic prescriber was assessed. Adjusted mean estimate, 95% confidence interval, and P-value were displayed. Multivariable logistic regression was used to assess the adjusted associations of age, sex, obesity, CCI, preoperative opioid filling, year of surgery, and state of opioid prescription on the rate of two or more opioid prescriptions in the early and cumulative perioperative time frames, and the rate of one or more opioid prescriptions in the late perioperative time frame. Adjusted odds ratio, 95% confidence interval, and P-value were displayed. CCI of 0, year 2010, state of Florida, and no preoperative opioid prescribing were used as reference controls in multilevel variables. P-values less than 0.05 were considered statistically significant. Because of the large sample size and the likelihood of achieving exceptionally small P-values, the clinical relevance was also considered. In the absence of evidence to guide researchers on clinically important differences in opioid prescribing, statistically significant factors altering filling by 5 oxycodone 5-mg pills or 0.2 change in odds were considered to be clinically relevant. Analyses in the body of the study focus on the 90 day post-op cohort, whereas additional groups are included in the Supplemental Digital Content, Appendix, http://links.lww.com/JAAOS/A587.


No sources of funding were present for this study.