The number of opioid overdose deaths in the United States has quadrupled in 15 years, a dramatic manifestation of the current opioid abuse epidemic.1 This rise parallels a sharp increase in the amount of legal prescription opioids dispensed.2 The abundance of prescription opioids available is a pathway for nonmedical use, defined by the National Institute on Drug Abuse as the “use of a medication without a prescription, in a way other than as prescribed, or for the experience or feelings elicited.”3 Most individuals using prescription opioids for nonmedical purposes have obtained them from family members or friends3 with unused opioids prescribed for other purposes.4
Although prescription opioids are a primary method for postoperative pain management, studies suggests that more than half of prescribed opioids go unused5,6 or are not disposed of properly.7 Despite cesarean delivery being the most common major surgical procedure performed in the United States,8 little is known about opioid prescribing practices or use after hospital discharge in women who undergo this operation.9 Although development of chronic opioid dependence after cesarean delivery is relatively low, the large number of procedures performed annually in the United States makes this an important public health issue10 Adjusting postcesarean delivery opioid prescribing practices to better match actual patient need may lead to reductions in unused opioids available for diversion, nonmedical use, and development of chronic dependence as well as reduce wasted resources. This study aims to characterize postdischarge opioid use after cesarean delivery and to identify factors associated with variations in prescribing practices and patient opioid consumption.
MATERIALS AND METHODS
We conducted a prospective observational cohort study of all women undergoing cesarean delivery between May 25, 2016, and August 30, 2016, at an academic tertiary medical center. The study was conducted in two 4-week blocks separated by a gap of 4 weeks as a result of availability of the research staff. Women with major perioperative morbidities such as cesarean hysterectomy or intensive care unit admission and women with chronic opioid use during pregnancy were excluded.
Eligible women were asked to participate in the study on postoperative day 1 or 2. Participants were asked questions about expectations regarding their delivery, analgesic medication use, and adequacy of analgesia (Survey 1; Appendix 1, available online at https://links.lww.com/AOG/A958). Follow-up data were obtained during a second survey after hospital discharge, beginning on postoperative day 14. Participants were contacted by phone or email, per their preference, and asked whether they had received and filled a prescription for opioid medication (Survey 2; Appendix 2, available online at https://links.lww.com/AOG/A959). They were queried regarding the number of pills used and remaining, their pain experience (worse than, better than, or as expected), and the number of opioid pills prescribed relative to pain experienced (too few, appropriate, or too many). Women who were still taking opioid medication at the time of initial contact were recontacted weekly until they had stopped opioid medication. Attempts were made to contact women at least three times before they were deemed lost to follow-up. Non–English-speaking women were contacted with the assistance of a phone interpreter and no women were excluded as a result of language barriers.
Demographic and delivery information as well as in-hospital opioid and nonopioid analgesic use was extracted from the electronic medical record. The Tennessee Controlled Substance Monitoring Database was accessed to confirm whether an opioid prescription was filled and for what amount and to ascertain whether additional prescriptions were obtained during the pregnancy and postpartum period. This database also includes prescriptions obtained in the states immediately surrounding Tennessee. Morphine milligram equivalents were calculated for all in-hospital and postdischarge opioids to allow for comparisons between opioids of different strengths (ie, oxycodone compared with hydrocodone). Where appropriate, we describe opioid strength in both morphine milligram equivalents and the number of 5-mg oxycodone tablets. We also classified postdischarge prescriptions by health care provider type. Two health care provider groups deliver care at our institution—a faculty provider group composed of physicians, certified nurse-midwives, and advanced practice nurses who practice out of one location (faculty practice) and a certified nurse-midwife group overseen by two physicians who practice out of a different location (certified nurse-midwife practice). The certified nurse-midwife practice frequently cares for low-risk women who usually seek nonmedicated vaginal deliveries.
Demographic, delivery, anesthetic, and in-hospital opioid use patterns were examined to identify variables associated with higher morphine milligram equivalent use including body mass index, race, type of insurance, number of prior cesarean deliveries, depression, smoking status, other comorbidities, type of skin incision, whether the patient labored, and whether a vaginal delivery was desired. In-hospital opioid use was adjusted to account for variable durations of inpatient admission by dividing the total morphine milligram equivalent used by the number of hours admitted postcesarean delivery (morphine milligram equivalents per hour). We also determined total morphine milligram equivalent use during four 12-hour epochs (less than 12 hours, 12–23 hours, 24–35 hours, 36–47 hours) and from 48 hours until discharge. Per routine, all women undergoing cesarean delivery with neuraxial (spinal or epidural) anesthesia receive one dose of neuraxial morphine at the time of surgery, which confers analgesia for up to 24 hours after delivery. The four women who did not receive neuraxial morphine were excluded from this portion of the analysis.
Study data were collected and managed using REDCap,10 a password-protected, secure online electronic database hosted at Vanderbilt University Medical Center.10 All statistical analyses were performed using (SAS 9.4). Chi-square or Fisher exact tests were used to analyze categorical variables as appropriate. Student t test, Mann Whitney U, and Pearson correlation coefficient were used as appropriate to analyze continuous variables. Multivariable logistic regression was used to evaluate associations between higher morphine milligram equivalent use and variables found to be significant in univariate analysis. Manual backward elimination was used to select a model that retained only those variables significant at the .05 significance level. Because the purpose of this project was quality improvement, the internal review board at Vanderbilt University Medical Center waived the requirement for written informed consent.
Of the 286 women who underwent cesarean delivery during the study period, 251 met eligibility criteria. Of these, 246 (98%) agreed to participate, and 188 (75%) completed Survey 2 (Appendix 2, https://links.lww.com/AOG/A959). The final sample size included 179 women (71% of eligible patients) with complete data (Fig. 1). There were no significant differences in demographic or obstetric characteristics nor opioid prescriptions rendered between respondents and nonrespondents.
Postdischarge, most participants (83%) used at least one opioid and the median duration of use was 8 days (interquartile range 6–13 days). Thirty-one women (17%) used opioids for more than 2 weeks postpartum, the maximum being 39 days. Of the 165 women who received a prescription, the total prescribed opioid ranged from 60 (eight tablets) to 630 (84 tablets) morphine milligram equivalents (median 225 morphine milligram equivalents [30 tablets], interquartile range 210–262). Most of these women (75% [123/165]) had unused tablets. The median morphine milligram equivalent unused for the entire population was 75 (interquartile range 0–187), corresponding to a median of 10 unused oxycodone tablets per patient. Every additional 10 tablets of oxycodone tablets prescribed were associated with approximately two tablets leftover (P<.001). Actual unused morphine milligram equivalents during this study period totaled 19,047, which is equivalent to 2,540 unused 5-mg oxycodone tablets or a median of 10 tablets per person. Of the 119 women with unused opioids, most (63%) stored them in an unlocked location within their home and only seven women (6%) disposed of them. Almost one third of women (n=46 [28%]) responded that they received too many opioid tablets.
Participants who used some opioids after discharge were not prescribed fewer opioids than women who used all opioids (P=.09); however, the actual opioid morphine milligram equivalents used was significantly less (Fig. 1; P<.001). Women who used all prescribed opioids were also more likely to report that they received too few tablets than women who used some or none of the prescribed opioids (33% [14/42] compared with 4% [5/123], P<.001). Women who reported receiving too few opioids were more likely to have public health insurance (73.7% [14/19] compared with 48.3% [70/145], relative risk 1.5, 95% confidence interval [CI] 1.1–2.1) and to have used opioids for more than 2 weeks before pregnancy (31.6% [6/19] compared with 12.3% [18/146], relative risk 2.6, 95% CI 1.2–5.6). Only four women were found to have received an additional prescription postdischarge according to our examination of the Tennessee Controlled Substance Monitoring Database. All received prescriptions from health care providers outside our system and two were from dental providers.
Postdischarge opioid use exhibited a bimodal distribution with peaks of zero and 200 total morphine milligram equivalents (27 tablets). The latter also demarcated the top quartile of postdischarge opioid consumption, which we designated as the “top opioid quartile” group. The remaining women comprised the “average opioid use” group. More women in the top opioid quartile smoked during pregnancy and had public health insurance, but these groups did not differ in terms of other demographic or delivery characteristics (Table 1). The major difference between top opioid quartile and average opioid use groups was in-hospital opioid use. The top opioid quartile group had greater in-hospital opioid use during the three 12-hour epochs between 12 and 47 hours postdelivery (Table 2). Compared with the average use group, the top opioid quartile group had a 2.6-fold greater opioid morphine milligram equivalent use per hour of stay, even after adjusting for smoking and insurance during pregnancy (odds ratio 2.59, 95% CI 1.61–4.17) and the mean in-hospital morphine milligram equivalent use per inpatient hour was 67% greater in the higher opioid use group. More than 90% of women were prescribed either oxycodone or oxycodone–acetaminophen while an inpatient.
The total morphine milligram equivalents prescribed at discharge were weakly correlated with in-hospital opioid morphine milligram equivalent use per hour (r=0.35). However, there were significant differences in opioid morphine milligram equivalents prescribed according to health care provider type with the certified nurse-midwife practice prescribing 50% less opioid morphine milligram equivalents at discharge than the faculty practice (161±121 compared with 227±93, P=.02). Oxycodone and oxycodone–acetaminophen accounted for 87% of prescriptions followed by hydrocodone–acetaminophen (9%). There were no differences between women with and without unused opioids in terms of the total morphine milligram equivalents prescribed at discharge (210morphine milligram equivalents [28 tablets], interquartile range 90–225 compared with 225 morphine milligram equivalents [30 tablets], interquartile range 210–281, P=.12).
We found that, once discharged after cesarean delivery, more than three fourths of women did not use all of their prescribed opioids corresponding to a median of 10 unused tablets of 5 mg oxycodone per patient. Aside from the cost associated with unused and wasted opioids, the majority of these opioids are stored in unlocked locations, risking potential nonmedical use and diversion. On the other hand, a smaller subset of women used all prescribed opioids and often reported not receiving enough pain medication after discharge. The top quartile of opioid use had significantly greater in-hospital opioid use compared with the average use group. The association found between smoking and higher opioid use is consistent with other studies examining chronic opioid use11 or opioid use after vaginal delivery.12
There are several studies characterizing opioid use after hospital discharge.5–7,9,13 One study of 250 patients undergoing outpatient wrist and hand surgery reported an average of 19 unused opioid tablets per person 2 weeks after surgery, amounting to an estimated 4,600 leftover tablets.13 Another examined opioid use after cesarean delivery and reported findings similar to those of our study: 53% of women used no or very few opioids after discharge, 77% stored opioids in an unlocked location, and high in-patient opioid use was associated with higher postdischarge opioid use. That study's small sample size (n=30) and low survey response rate (26%) introduced risk of sampling bias, potentially limiting validity.9 Of note, national organizations provide few guidelines for prescribing opioids postdischarge. The Centers for Disease Control and Prevention recommends that “3 days or less will often be sufficient [for acute pain].” They state, “more than seven days will rarely be needed.”14 Importantly, our data do not support this recommendation.
A stratified prescribing scheme based on reliable predictors of individuals' postdischarge needs such as in-hospital use is a potential strategy for reducing current overprescription practices. Any proposed scheme should carefully consider the risks and inconveniences of underprescription, including strict state restrictions on phoning in prescriptions. Our data suggest that health care providers are not currently considering in-hospital opioid use in determining the amount of opioid prescribed at discharge. Instead, prescriptions appear to be personalized to health care provider background (certified nurse-midwife compared with faculty practice), not patients.
Our study has several limitations. Regarding our method of determining the number of used and leftover pills, we chose to rely on patient report during postdischarge surveys, introducing the possibility of recall bias. We opted against asking patients to keep a log of opioid use to avoid patient attention to opioid use that might alter consumption. We accessed the Tennessee Controlled Substance Monitoring Database to ascertain whether participants received prescribed opioids from health care providers outside our system. Even so, it remains possible that participants obtained opioids from friends, family, or illicit sources that were not disclosed to study personnel. In addition, although our sample size provides robust characterization of opioid use in our population, we were not powered to examine specific demographic and obstetric characteristics associated with higher opioid use. It is possible that differences exist that our sample was underpowered to detect. Finally, we acknowledge the influence of heightened awareness surrounding opioid overprescribing in Tennessee. Although we did not encounter direct evidence of this, implicit bias could have led to differential prescribing practices based on health care providers' assumptions concerning the sociodemographic and geographic characteristics within our patient population.
Our study has several strengths. To limit selection bias, we recruited all women undergoing cesarean delivery regardless of language or other factors. In fact, 30% of our population was identified as non–English-speaking. A single prior study of women undergoing cesarean delivery combined data from patients undergoing thoracic surgery.9 Our study limited inclusion to women undergoing a single type of surgery, uncomplicated cesarean delivery, which is a relatively uniform procedure with relatively uniform intraoperative anesthetic management. In addition, we excluded women with perioperative morbidities that could significantly alter pain and opioid use. Finally, the prospective design allowed for collection of detailed information about our institution's current variations in health care provider opioid prescribing patterns, patient expectations for delivery as well as patient experience with opioid use, which helped to inform differences in opioid use patterns and which will inform potential improvements.
In conclusion, most women are prescribed opioids after cesarean delivery well in excess of the amount they use. However, a small subset of women use all opioids prescribed. Homogenous or “one-size-fits-all” opioid prescribing risks underprescribing to an important subset of the population. Further study is needed to determine whether in-hospital opioid use can guide postdischarge opioid prescribing to reduce unused opioids without undertreating pain.
1. Rudd RA, Seth P, David F, Scholl L. Increases in drug and opioid-involved overdose deaths—United States, 2010–2015. MMWR Morb Mortal Wkly Rep 2016;65:1445–1452.
2. Jones CM. Heroin use and heroin use risk behaviors among nonmedical users of prescription opioid pain relievers—United States, 2002–2004 and 2008–2010. Drug Alcohol Depend 2013;132:95–100.
3. National Institute on Drug Abuse. Prescription drugs: abuse and addiction. Bethesda (MD): National Institute on Drug Abuse; 2011:1–16.
4. SAMHSA Center for Behavioral Health Statistics and Quality. Results from the 2013 national survey on drug use and health: summary of national findings. Rockville (MD): SAMHSA Center for Behavioral Health Statistics and Quality; 2014:1–184.
5. Kim N, Matzon JL, Abboudi J, Jones C, Kirkpatrick W, Leinberry CF, et al. A prospective evaluation of opioid utilization after upper-extremity surgical procedures: identifying consumption patterns and determining prescribing guidelines. J Bone Joint Surg Am 2016;98:e89.
6. Harris K, Curtis J, Larsen B, Calder S, Duffy K, Bowen G, et al. Opioid pain medication use after dermatologic surgery: a prospective observational study of 212 dermatologic surgery patients. JAMA Dermatol 2013;149:317–21.
7. Maughan BC, Hersh EV, Shofer FS, Wanner KJ, Archer E, Carrasco LR, et al. Unused opioid analgesics and drug disposal following outpatient dental surgery: a randomized controlled trial. Drug Alcohol Depend 2016;168:328–334.
8. National Center for Health Statistics. National vital statistics reports. Hyattsville (MD): National Center for Health Statistics; 2015:1–68.
9. Bartels K, Mayes LM, Dingmann C, Bullard KJ, Hopfer CJ, Binswanger IA. Opioid use and storage patterns by patients after hospital discharge following surgery. PLoS One 2016;11:e0147972.
10. Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research electronic data capture (REDCap)—a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform 2009;42:377–81.
11. Bateman BT, Franklin JM, Bykov K, Avorn J, Shrank WH, Brennan TA, et al. Persistent opioid use following cesarean delivery: patterns and predictors among opioid-naive women. Am J Obstet Gynecol 2016;215:353.e1–18.
12. Jarlenski M, Bodnar LM, Kim JY, Donohue J, Krans EE, Bogen DL. Filled prescriptions for opioids after vaginal delivery. Obstet Gynecol 2017;129:431–437.
13. Rodgers J, Cunningham K, Fitzgerald K, Finnerty E. Opioid consumption following outpatient upper extremity surgery. J Hand Surg Am 2012;37:645–50.
14. Dowell D, Haegerich TM, Chou R. CDC guideline for prescribing opioids for chronic pain—United States, 2016 [published erratum appears in MMWR Recomm Rep 2016;65:295]. MMWR Recomm Rep 2016;65:1–49.