Persistent Opioid Use After Hysterectomy in the United States, 2005-2015.

For this retrospective cohort study, we used MarketScan® Commercial Claims and Encounters database and Medicare Supplemental and Coordination of Benefits databases (copyright © 2015 IBM Watson Health. All rights reserved). 10 , 11 These data have been validated by IBM Watson Health and are population-based and de-identified healthcare claims drawing from approximately 150 payors in the U.S., representing employees, dependents, and retirees with employer-based insurance in the U.S. These databases include longitudinal inpatient and outpatient healthcare claims allowing researchers to observe dates of service, patient-level diagnosis codes and procedure codes, and all reimbursed outpatient prescription medications. Unique individuals can be followed over time using encrypted identification numbers, and detailed enrollment data were used to ensure that only individuals who could generate a claim were included in the population at risk at any given time. In 2014, these databases included approximately 50.9 million individuals. For additional context, in 2014, 55.4% of the U.S. population, or 174.4 million individuals, had employment-based insurance; thus, this database includes a significant proportion of those with employer-based insurance. 12 This study was determined to be exempt from further review by the Institutional Review Board at the University of North Carolina at Chapel Hill as only de-identified data were available. We included women aged 18 years and older who underwent a benign hysterectomy between January 1, 2005, and September 31, 2015, based on the following Current Procedural Terminology (CPT) codes: Abdominal hysterectomy (58150, 58152, 58180), Vaginal hysterectomy (58260, 5f8262, 58263, 58267, 58270, 58275, 58280, 58290, 58291, 58292, 58293, 58294), and Laparoscopic hysterectomy (58541, 58542, 58543, 58544, 58550, 58552, 58553, 58554, 58570, 58571, 58572, 58573), as described in prior literature. 13 If a patient had multiple CPT codes falling into more than one route, they were assigned to the most invasive category coded: abdominal (most invasive), followed by laparoscopic, vaginal (least invasive). 13 To characterize baseline health characteristics, we used MarketScan data from 2004 – 2015. Patients were required to have at least 180 days of prior continuous enrollment in the MarketScan data and were required to have evidence of at least one prescription fill during this time to indicate prescription coverage benefit. Patients were followed for up to one-year post surgery and were censored at occurrence of invasive surgery during follow-up 14 , discontinuation of insurance coverage, reaching the end of the study data (December 31, 2015), or reaching 1-year post surgery. To examine possible indications for hysterectomy, we examined diagnosis codes in the 30 days prior to hysterectomy, using Clinical Classification Software diagnosis groups: 46 – Benign neoplasm of uterus, 168 – Inflammatory disease of female pelvic organs, 169 – Endometriosis, 170 – Prolapse of female genital organs, and 171 – Menstrual disorders). 15 Because surgical complications may impact postoperative opioid use, we examined diagnosis and procedure codes indicative of surgical complications (lysis of adhesion, bowel injury, ureteral injury, acute renal failure, urinary tract infection, shock or sepsis, wound disruption, postoperative infection, CPR, unplanned intubation, and prolonged intubation) occurring on the day of surgery, and in the 60 days after surgery. We evaluated prescription claims for opioids which included codeine, dextropropoxyphene hydrocodone, hydromorphone, oxycodone, pentazocine, tramadol, fentanyl, meperidine, morphine, oxymorphone, pethidine, and tapentadol. We focused on opioid naïve patients undergoing hysterectomy, and excluded women with prevalent opioid use defined as an opioid prescription filled between 180 to 30 days prior to hysterectomy ( Figure 1 ). In addition, women who had ≥ 2 opioid prescriptions filled in the 30-days prior to surgery were also excluded. While we assumed that one prescription within 30-days prior to surgery represented opioids prescribed in a preoperative visit for use for postoperative pain, evidence of two or more opioid prescriptions was indicative of potential prevalent use for nonsurgical related pain. We defined perioperative opioid use as any opioid prescription filled within 30 days before or 7 days after surgery. The prescription claims data include National Drug Code allowing for exact identification of the specific opioid prescribed, the number of units dispensed, and the number of days supplied (length in days the prescription is intended to last). We describe the median quantity dispensed and days supply in the perioperative period by specific opioid. We report the proportion of women who received a perioperative opioid prescription. To compare those with and without a perioperative opioid prescription we used the absolute standardized mean difference (SMD). The absolute SMD is useful in large observational cohort studies when sample size is large, as it is not influenced by sample size. Differences in SMD less than 0.10 are considered negligible. 16 We used log-binomial regression to identify independent predictors of perioperative opioid receipt. We estimate risk ratios and 95% confidence intervals adjusted for demographic and clinical factors at baseline. Age, year of hysterectomy, US region of residence, route of hysterectomy, and surgical setting were included as categorical variables, using the category with the most patients as the referent. Length of hospitalization and Charlson Comorbidity Index were included as linear terms. Surgical complication, prior diagnoses, and receipt of non-opioid analgesics during baseline were included as binary variables, 1 indicating presence of the listed condition, with 0 as the referent. To evaluate the risk of prolonged opioid use, we examined the proportion of women who filled at least one prescription in monthly intervals of follow-up. Among women with complete follow-up at each time point, we reported the proportion with continued monthly fills at various intervals through one-year after hysterectomy. In addition, we report the proportion with any opioid fill during intervals of follow-up, not requiring any fills in prior months (no continuous use requirement). To facilitate comparisons with prior research in this area, we also conducted a secondary analysis examining the proportion of patients with any opioid prescription filled between 90- and 180-days after hysterectomy, a common marker of prolonged opioid use in prior literature. 5 , 17 We conducted stratified analyses of continued monthly use stratified by the initial opioid prescribed. We additionally compared those with a complication during surgery to those with no complication.

A total of 569,634 women meeting our enrollment criteria underwent a hysterectomy during the study period, among which 176,537 (30.9%) were prevalent opioid users and were thus excluded. Among the 393,097 remaining women in our study population, the average follow-up time was 273.6 days. Laparoscopic hysterectomy was the most common route accounting for 48.8% of all patients, followed by abdominal hysterectomy (34.0%), and vaginal hysterectomy (17.2%) ( Table 1 ). Throughout the study period, the proportion of abdominal and vaginal hysterectomies declined from 57.3% (abdominal) and 23.0% (vaginal) in 2005, to 23.3% (abdominal) and 11.5% (vaginal) in 2015. Meanwhile, the proportion of hysterectomies conducted laparoscopically increased from 19.7% in 2005 to 65.2% in 2015 (data not shown). During the month prior to hysterectomy, 60% of all women in the study population were diagnosed with benign neoplasm, 56% had a diagnosis code indicating menstrual disorder, 31% were diagnosed with endometriosis, and 30% were diagnosed with inflammatory pelvic disease ( Table 1 ). Bowel injury was the most common intraoperative complication, occurring in 4.2% of surgeries, followed by lysis of adhesion, occurring in 3.8% of surgeries ( Appendix 1 , available online at http://links.lww.com/AOG/B663 ). We found that 331,322 (84.3%) patients undergoing benign hysterectomy filled a perioperative opioid prescription ( Table 1 ). Those who filled an opioid prescription were younger (46.6±9.2 vs 50.0±11.5), more likely to reside in the South (49.6% vs 44.0%), less likely to reside in the Northeast (9.9% vs 14.4%), had a lower Charlson Comorbidity Index (mean 0.3 vs 0.5), and were more likely to have a diagnosis of menstrual disorder (57.6% vs 48.2%). After multivariate adjustment, we found that the youngest (adjRR 18–24 =0.91, 95%CI: 0.87, 0.95) and oldest (adjRR 65–74 =0.84, 95%CI: 0.83, 0.85 ; adjRR 75+ =0.70, 95%CI: 0.69, 0.72) patients were less likely to receive perioperative opioids compared to those aged 45–54 years of age ( Figure 2 ). We did not observe meaningful differences in year of procedure, geographic region, surgical route, hysterectomy indication, surgical setting (inpatient vs. outpatient), intraoperative surgical complications, or Charlson Comorbidity Index. For perioperative opioids associated with hysterectomy, the median number of pills dispensed was 30 (IQR 25, 40), and the median days’ supply was 4 (IQR 3, 5) ( Table 2 ). Oxycodone was the most common opioid prescribed in the perioperative period (45.0% of patients), followed by hydrocodone (38.9% of patients). Among patients receiving a perioperative opioid prescription, 14.2% had an additional fill in the month (0 – 30 days) after hysterectomy. Those undergoing abdominal surgery were the most likely to have additional fills in the first month of follow-up (20.4%) followed by vaginal (11.6%) and laparoscopic hysterectomies (10.8%). The proportion of women with continuing monthly fills of opioids through 2, 3, 6, and 12 months post hysterectomy was 1.40%, 0.34%, 0.07%, and 0.02%, respectively ( Table 3 ). Those undergoing abdominal surgery consistently had the higher proportion of patients continuing treatment. Table 4 displays the proportion of patients with any opioid prescriptions in the months after surgery, not requiring fills in any prior months (no continuous use requirement). The proportion of women with any opioid fill occurring 2, 3, 6, and 12 months post hysterectomy was 3.4%, 2.7%, 2.7%, and 2.6%, respectively. In stratified analyses, we found that patients whose initial perioperative prescription was for oxycodone had lower proportions with prolonged use compared to those initiating on hydrocodone ( Appendix 2 , available online at http://links.lww.com/AOG/B663 ). We also found that those with surgical complications during surgery or in the 60 days post-surgery were more likely to have prolonged opioid use compared to those without complications. ( Appendix 3 , available online at http://links.lww.com/AOG/B663 ). In order to provide results comparable to those reported in prior literature, we also estimated the proportion of patients with any opioid prescription between 90–180 days after hysterectomy. For this analysis, we limited the population to the 293,491 women who had at least 180 days of follow-up enrollment in the data source. Among the 248,822 patients with a perioperative opioid prescription, 6.8% had an opioid prescription in the 90–180 days post-surgery. Using this definition, patients undergoing abdominal surgery were most likely to have prolonged use (7.0%), followed by laparoscopic (6.7%) and vaginal (6.6%). Meanwhile, among the 44,669 patients with no perioperative opioid prescription, 4.1% received an opioid fill in the 90–180-day period (abdominal:4.7%; laparoscopic:3.7%; vaginal:4.1%), suggesting that 2.7% (6.8% - 4.1%) had prolonged opioid use related to postoperative pain.

Overall, 84% of opioid-naïve patients undergoing a hysterectomy filled a perioperative opioid prescription. Our analysis of predictors of opioid receipt found that younger and older women were less likely to receive perioperative opioids. This is consistent with other studies that have found that older age is associated with less opioid use in patients undergoing hysterectomies for benign indications as well as with other surgeries including breast cancer and orthopedic procedures. 8 , 18 – 20 These analyses additionally identified many statistically significant predictors, however our large sample size led to very precise estimates with narrow confidence intervals, and while statistically significant (alpha=0.05), these results were not likely clinically meaningful. Notably, while we observed a shift towards less invasive (laparoscopic procedures), we did not observe differences in opioid receipt by year or route of surgery. A recent study by Thompson et. al. also reported a trend towards minimally invasive surgical techniques, however they found that the amount of opioids prescribed upon discharge increased during this same time period. 13 Among patients receiving perioperative opioids, 0.34% had at least one opioid prescription monthly through three months and 0.07% had continued use through six months, suggesting that rate of continued use of opioids after hysterectomy is quite low (1 in 1,538). The percentage of patients with any opioid fill in the third (2.7%) or sixth (2.7%) month after hysterectomy were substantially higher, highlighting the importance of how prolonged opioid use is defined and the impact continuous use measures have on overall estimates. Of note, when we mirrored prior studies by defining prolonged use as any opioid prescription occurring between 90–180 days post-surgery, we found that 6.8% of patients who received perioperative opioids had “prolonged use”. This definition is much broader and less stringent as it does not require regular, recurring receipt of opioids. We also found that among patients who did not receive any perioperative opioid prescription, 4.1% of patients had an opioid prescription in the 90–180 days after hysterectomy, suggesting that a substantial portion of the prolonged use may not be stemming from exposure at the time of hysterectomy. We saw differences by surgical route when examining postoperative opioid use, with patients undergoing abdominal surgery being more likely to have continued opioid use. This is consistent with expectations, as abdominal hysterectomies are associated with long recovery times compared to minimally invasive (laparoscopic, vaginal) routes. 21 , 22 When clinically possible, minimally invasive routes are preferred, however abdominal surgery may be performed in patients who have complicated clinical cases. These patients may also have indications that result in more postoperative pain. 21 In stratified analyses, we also found that patients who received hydrocodone and those with surgical complications were more likely to have prolonged opioid use. The increased proportion with prolonged use among patients receiving hydrocodone may be due to varying factors, such as changes in hydrocodone scheduling and temporal prescribing trends which were not accounted for in the current analysis. Our results are similar to prior research examining prolonged opioid use after surgery. In a study using U.S. commercial claims data from 2013–2014, Brummett et al. found that approximately 6% of patients undergoing hysterectomy who received a perioperative opioid had use in the 90–180 days after surgery. 5 Clarke et al. used a slightly more conservative definition of prolonged use, requiring at least 1 opioid prescription in the 1–90 days after surgery, and at least 1 opioid prescription in the 91–180 days after surgery. When using this definition they found that 2.5% of patients undergoing open hysterectomy had prolonged use. 9 Our definition is more conservative than those used by Brummett and Clark. Our estimate of 0.34% of patients with prolonged use through 3 months required at least 1 prescription in 3 consecutive windows during follow-up: 1–30 days, 31–60 days, and 61–90 days of follow-up. While this definition is more conservative than those used in prior studies, we believe that it more accurately represents continued opioid exposure that may be related to postoperative pain. Overall, prior literature defines prolonged opioid use in varying ways, which paint different conclusions about the magnitude of iatrogenic use after hysterectomy. Opioids are an important part of the healthcare system and evidence supports their efficacy in managing short-term or acute pain. 23 Striking the balance between adequate and safe pain management is clinically challenging. Inadequate management of postoperative pain can lead to increased surgical complications, increased risk of chronic pain, and decreased quality of living. 24 , 25 Over prescribing for post-operative pain has been widely reported to lead to nonmedical use and diversion. Bartels found that among women undergoing cesarean sections, 83% reported that they consumed less than half of their prescribed opioids, and 77% of patients stored these drugs in unlocked locations. 26 These trends have been seen in other populations, with a large systematic review finding that across a broad range of 10 surgeries, the reported proportion of consumed opioids ranged from 6%−59%, with 70% of patients keeping excess opioids. 27 Looking specifically at hysterectomies, the tendency to overprescribe opioids persists. As-Sanie et al. conducted a prospective study of patients undergoing laparoscopic, vaginal, or abdominal hysterectomy for benign indications at one institution and found that the median number of opioids prescribed was equivalent to 40 tablets of 5 mg hydrocodone. 28 On average, patients used less than half of the prescribed opioids, with 22 tablets left over, and forty percent felt they received too much opioid medication after surgery. Johnson and Makai conducted a systematic review of perioperative opioid management for minimally invasive hysterectomy (laparoscopic and vaginal). 29 Using information from twenty-one studies, the authors reported opioid consumption of 50 to 100 oral morphine equivalents (10–20 tablets of 5 mg hydrocodone) over the first two postoperative weeks while physicians prescribed between 125 to 300 oral morphine equivalents (25–60 tablets of 5 mg hydrocodone). A more recent study by Wong et al. assessing opioid use after laparoscopic hysterectomy also found excessive prescribing of opioids with the equivalent of 20 tablets of 5mg oxycodone prescribed with only 5 tablets of oxycodone consumed. 30 Another factor that may impact opioid use after hysterectomy is the integration of enhanced recovery after surgery (ERAS) protocols. 31 As these standardized pathways become further used for hysterectomies, the use of postoperative opioids may decrease. 32 As more recent data become available, further research evaluating the impact of ERAS on opioid prescribing habits and prolonged opioid use would be an important contribution to the field. Overall, the increase in physician awareness of overprescribing habits, the use of ERAS protocols, integration of perioperative multimodal analgesia, and a more personalized approach to prescribing opioids using individual patient characteristics will hopefully allow for more optimal prescribing practices in the gynecology population. 28 , 33 This study provides important pharmacoepidemiologic data regarding different methods to assess prolonged opioid use. The current findings describe the postoperative opioid prescriptions in a ten-year period during which the US opioid prescribing rate peaked and will be able to serve as a benchmark for future work as the clinical landscape changes, with advances in surgical techniques and implementation of ERAs pathways. This study should be interpreted with several limitations. While the insurance claims data provide a detailed longitudinal picture of a patient’s healthcare, there are certain clinical details that are not captured. We only observe outpatient opioid prescriptions which are reimbursed through insurance and cannot observe inpatient opioid administration. However, our main focus was outpatient opioid use, as this is the setting where patients are under less medical supervision and may be at higher risk of nonmedical use. In addition, we cannot see prescriptions that are paid for by cash or received from other sources. Furthermore, we do not have a direct link between opioid prescriptions and indications for prescribing. We assume that patients who received opioids near surgery were receiving this for post-surgical pain management. Because these patients had no evidence of opioids in the six months prior, this assumption is clinically reasonable, but may still result in misclassifying opioids prescribed for other indications in certain cases. The study was conducted among commercially insured patients, which may not be generalizable to those with public insurance (Medicare, Medicaid) or who are uninsured. However the commercially insured population represents the largest group of Americans (55.4% of the U.S. population in 2014). 12 We also found that 30% of patients undergoing benign hysterectomy received opioid prescriptions in the 182 days prior to surgery, suggesting that a meaningful proportion of women undergoing these procedures may be on opioid treatment at the time of procedure, or had recent use. This study focuses on population-based data regarding perioperative opioid prescriptions and long-term use. We are unable to assess individual factors such as indication for surgery and other comorbidities/factors that may impact the surgery. Given the lack of clinical detail, it is difficult to make specific recommendations on opioid prescriptions for surgery. Our findings highlight that 84% of hysterectomy patients filled an opioid prescription for surgical pain, and that rates of continuous use of opioids was low, suggesting that persistent opioid use after hysterectomy may not be as high as previously reported. Further research examining postoperative pain management in these women are needed to add to the research surrounding opioid use for post-hysterectomy pain. In efforts to address the US opioid crisis, there has been heightened attention to pain management practices for surgical pain, and the risks of iatrogenic opioid addiction stemming from opioids prescribed after surgery. We were not able to estimate how much of prolonged postsurgical use was related to persistent pain, inappropriate use, or for pain unrelated to original surgery. 1 Overall we found that a very small proportion of patients who received perioperative opioids for hysterectomy had prolonged use. Of those who received a perioperative opioid prescription, fewer than 4 per 1,000 patients had at least one monthly fill for 3 months after surgery. When continuing fills were not required, we found that fewer than 7 per 100 patients filled a prescription in the 90–180 days after surgery. In comparison, 4 per 100 patients who did not receive a perioperative opioid filled a prescription in the 90–180 days after hysterectomy, suggesting that a substantial proportion of use during this time may be unrelated to perioperative opioid exposure.

Perioperative opioid use for postsurgical pain is a critical issue to assess given the contribution of prescription opioids to the opioid epidemic in the US. 1 , 2 The Center for Disease Control and Prevention states that “we now know that overdoses from prescription opioids are a driving factor in the 15-year increase in opioid overdose deaths”. 3 , 4 A few studies have shown that there is a risk of persistent opioid use among patients who were opioid naïve prior to surgery, with rates varying from 0.12% after cesarean delivery to as high as 6.5% after major surgery. 5 , 6 Hysterectomy is the second most common surgery in women in the US after cesarean section. Prolonged opioid use after a hysterectomy may be particularly important to assess given the risk of persistent postsurgical pain after hysterectomy 7 , 8 which may result in persistent narcotic use. Brummett et al 5 found a ~6% rate of persistent opioid use after a hysterectomy while Clarke et al 9 reported rates of 1.5% and 2.5% for opioid use after minimally-invasive and abdominal hysterectomies, respectively. While these prior studies have explored persistent opioid use after major surgeries overall, they did not focus specifically on hysterectomy; thus, details regarding this patient population and those at risk for persistent opioid use are not available. Given limitations in existing literature and the fact that persistent postsurgical pain after hysterectomy may increase the use of prolonged opioid use, we sought to conduct a larger-scale, population-based study to examine 1) variables associated with opioid prescriptions filled perioperatively and 2) the risk of long-term opioid use after hysterectomy.