Secondary Logo

Journal Logo

Contents: Abortion: Original Research

Opioid Analgesia for Medical Abortion

A Randomized Controlled Trial

Colwill, Alyssa Covelli MD, MCR; Bayer, Lisa L. MD, MPH; Bednarek, Paula MD, MPH; Garg, Bharti MBBS, MPH; Jensen, Jeffery T. MD, MPH; Edelman, Alison B. MD, MPH

Author Information
doi: 10.1097/AOG.0000000000003576

More than a quarter of a million medical abortions are performed annually in the United States.1 Medical abortion with mifepristone and misoprostol is the recommended regimen by the American College of Obstetricians and Gynecologists and the World Health Organization.2–4 Medical abortion has been extensively studied and is safe.5 However, abortion-related pain has not been systematically studied in clinical trials, limiting our ability to recommend appropriate analgesia.6 Medical abortion has been described as painful in 36–86% of patients, with up to one-third reporting more pain than anticipated.4,7–9

Nonsteroidal antiinflammatory drugs (NSAIDs) are the cornerstone of analgesia regimens for abortion-related pain. Yet, NSAIDs do not provide sufficient pain relief for many women undergoing medical abortion.10 Clinicians often prescribe oral opioids as an adjunct to NSAIDs to aid in pain control for medical abortion but no consensus exists on the use of opioids in this setting.11,12 The one randomized controlled trial that exists found no difference in pain score among women who received either ibuprofen or acetaminophen with codeine during methotrexate and misoprostol medical abortion, but the conclusions are limited by underdosing of a weak opioid.13

Immediate release oxycodone is known to be effective for severe pain outside of abortion care.14,15 We hypothesized that a more potent opioid could decrease the overall pain experienced during medical abortion. Our primary objective of this study was to estimate whether ibuprofen plus a strong opioid (oxycodone) given at a dose recommended for severe pain decreases pain scores in women undergoing medical abortion as compared with ibuprofen alone.


We conducted a randomized, double-blind, placebo-controlled trial based at Planned Parenthood Columbia Willamette in Oregon from May 2017 to May 2018. The Institutional Review Board at Oregon Health & Science University approved the study protocol. Women were invited to participate in the study after choosing and consenting to a medical abortion. Our main eligibility criteria were women aged 18 years or older, with gestations up to 10 0/7 weeks of gestation with no diagnosis of an early pregnancy failure, able to receive text messages, literate in English, able to avoid driving or alcohol use while taking study medications, no past use of methadone or heroin, no marijuana use greater than four times per week, no opioid use in the past 30 days, and no use of other pain medications. All participants completed written informed consent before any study procedures.

We collected baseline information including demographic, medical, and pregnancy history; body mass index; estimated gestational age by ultrasound examination; and anticipated maximum pain using an 11-point numerical rating scale (0–10). Oregon Health & Science University research pharmacy staff generated, assigned, and maintained the computer-generated 1:1 randomization scheme with random blocks ranging from 5–16. We did not collect baseline demographics on participants who were not randomized. The study drugs, 10 mg oxycodone or an identical placebo, were labelled in identical sequentially numbered bottles. Study staff, participants, and health care providers were blinded to treatment assignments; the randomization sequence was only unblinded after completion of all data collection and entry.

Our primary outcome was to determine whether women undergoing medical abortion who receive oxycodone 10 mg and ibuprofen 800 mg will report maximum pain scores at least 2 points lower on a numerical rating scale as compared with women using ibuprofen 800 mg and placebo within 24 hours postmisoprostol. Our secondary outcomes included maximum reported pain score stratified by gestational age (less than 7 weeks of gestation, 7–10 weeks of gestation), duration of maximum pain, number of ibuprofen tablets used, and use of adjunctive oxycodone.

All participants received four 200-microgram misoprostol buccal tablets for use 24–48 hours after mifepristone, six 4-mg ondansetron oral dissolving tablets to take as needed for nausea, and nine 800-mg ibuprofen tablets to start up to 1 hour before misoprostol administration and every 8 hours as needed for cramping. Participants were given one study drug tablet (10 mg oxycodone or identical placebo) to take at the onset of uterine cramping. In addition, we provided participants a paper prescription for six oxycodone 5-mg oral tablets (1 tablet orally every 4 hours as needed for pain) to fill and use only if needed (“adjunctive” medication).

Study staff launched the text-message platform when participants notified them of misoprostol ingestion (0 hours). At 6 and 24 hours, participants responded to an automated survey using text message (TextIt), which included questions on maximum pain (11-point numerical rating scale), duration of maximum pain, time of study drug administration, number of ibuprofen tablets used, whether the oxycodone adjunctive prescription was filled (yes or no) and used (yes or no; if yes, number of tablets taken), nausea or vomiting (yes or no), satisfaction with pain medications (yes or no), adequacy of blinding (do you believe the study drug was oxycodone or placebo?), whether other medical and nonmedical therapies were used for pain (free text), and belief of passing the pregnancy (yes or no).

Participants returned for a standard follow up visit within 21 days for confirmation of medical abortion success.16 We reviewed medical records to verify completion of medical abortion as well as to check for unanticipated phone calls or visits. If a participant did not return to this scheduled visit, we sent a text message to check whether the participant was “feeling back to normal” and had seen another health care provider. We considered a participant lost to follow-up if we received no notification of misoprostol ingestion, no response to text or phone contacts during data collection, or they did not return to their follow-up visit (or no response to text or phone contact).

Previous studies suggest that the 11-point numeric pain scale is as sensitive to changes in clinical pain as the visual analog scale and a change of 2 points or more is needed to determine a difference in pain intensity.17,18 We based our sample on the assumptions of nonnormally distributed data, and used a Wilcoxon rank-sum test data simulation using specified parameters (delta=2, sigma=2.6, alpha=0.05) for a moderate effect. A sample size of 34 participants per group provided 80% probability of detecting the 2-point difference in the simulated data. To allow equal power for stratification by the two gestational age groups, we doubled the sample and to allow for up to 10% drop-out, we planned to enroll 152 participants (less than 7 weeks of gestation: 38 placebo, 38 oxycodone; 7–10 weeks of gestation: 38 placebo, 38 oxycodone). Before completion of the study, without breaking the randomization schema or further evaluating outcomes, we found that 30.8% of enrolled participants had not taken the study drug. Therefore, we increased the total enrollment by 25% in the less than 7 weeks of gestation group to a total of 172 participants to have sufficient power to determine our primary outcome (less than 7 weeks of gestation: 48 placebo, 48 oxycodone; 7–10 weeks of gestation: 38 placebo, 38 oxycodone).

We used the REDCap electronic data capture tool at Oregon Health & Science University for data management. We exported data directly from REDCap into Stata 15.1 for statistical analysis. We used independent two-sample t-tests to compare continuous variables, χ2 tests to compare categorical variables and Wilcoxon rank-sum tests to compare medians. Odds ratios (ORs) were computed using simple logistic regression. We analyzed our primary cohort using an intent-to-treat approach. Additionally, we performed a per-protocol analysis.


We screened a total of 512 women and randomized 172 (86 per treatment group; Fig. 1). Regarding gestational age stratification, we recruited 96 women with gestations of less than 7 weeks and 76 women with gestations between 7 and 10 weeks. Text-message surveys were completed by 170 of 172 (98.8%) of participants. Treatment groups did not differ in demographic characteristics (Table 1) but when stratified by gestational age (Appendix 1, available online at, more Hispanic women were allocated to oxycodone in the later gestational age groups (P=.01).

Fig. 1.
Fig. 1.:
CONSORT (Consolidated Standards of Reporting Trials) flow diagram.Colwill. Opioid Analgesia for Medical Abortion. Obstet Gynecol 2019 .
Table 1.
Table 1.:

Overall, women reported that their maximum reported pain level was a median of eight. We found no difference in the maximum reported pain level between the treatment (P=.92, Table 2) or the stratified gestational age treatment (7 weeks of gestation or less P=.75; greater than 7 weeks of gestation P=.62, Table 3) groups. We also performed noninferiority testing and the maximum pain scores did not differ between groups by more than 1 point on the numerical rating scale (P=.001). Oxycodone treatment did not reduce the proportion of women with reported moderate-severe pain (numerical rating scale greater than 7) or the maximum median duration of pain, which was about 1 hour in length (P=.39, Tables 2 and 3). The onset of maximum mean pain occurred approximately 2 hours postmisoprostol in both the treatment (Table 2) and the stratified gestational age groups (Table 3). Baseline anticipated pain correlated with the maximum pain that participants experienced (r=0.43, P<.01).

Table 2.
Table 2.:
Primary and Secondary Outcomes of Entire Cohort (Intention-to-Treat Analysis)
Table 3.
Table 3.:
Stratification by Gestational Age in Intent-to-Treat Cohort

Participants used a similar median number of ibuprofen tablets [oxycodone 2 (range 0–9), placebo 2 (range 0–7), P=.59]. At least half of all participants filled their adjunctive medication prescription but there was no difference between groups (oxycodone 49%, placebo 62%, P=.09; less than 7 weeks of gestation: oxycodone 50%, placebo 66%, P=.11; 7–10 weeks of gestation: oxycodone 49%, placebo 58%, P=.42). Less than half of the participants in either group used the adjunctive medication (oxycodone 40%, placebo 48%, P=.28) and the median number of adjunctive oxycodone tablets used was two (total 10 mg) (Table 2, P=.59). In total, we prescribed 1,032 tablets of oxycodone for adjunctive medication (six tablets per participant). Our research population only used 15% of what we prescribed—leaving 85% of prescribed tablets unused. Only four women (2% of the cohort) used all six tablets of the adjunctive oxycodone medication. No one reported using more than the six tablets prescribed.

We performed a per-protocol analysis to compare our main outcomes between participants that adhered to the study protocol by taking the study drug as prescribed (Table 4). We found no differences in baseline characteristics, maximum reported pain score, duration of pain, number of ibuprofen tablets used, proportion of participants who filled and used oxycodone, or satisfaction (Appendices 2 and 3, available online at A total of 49 participants (placebo group n=21; oxycodone group n=28) did not follow the protocol by: not taking study drug but taking the baseline ibuprofen (placebo group n=16; oxycodone group n=24); taking ibuprofen and adjunctive drug without study drug (placebo group n=2; oxycodone group n=2); taking study drug and not taking ibuprofen or adjunctive medication (placebo group n=1; oxycodone group n=1); or who took no medications at all (placebo group n=2; oxycodone group n=1) (Appendix 4, available online at We performed additional analyses to compare our main outcomes between participants that did and did not take study drugs. We found no differences in baseline characteristics, however nonprotocol followers were more likely to report lower maximum pain (P<.01), a decreased duration of maximum pain (P<.01), and use less ibuprofen and adjunctive pain medication (P<.01) (Appendices 5 and 6, available online at

Table 4.
Table 4.:
Results of Protocol Followers (Per-Protocol Analysis)

Women's satisfaction with the prescribed pain medication did not differ between treatment groups (OR 1.16; 95% CI 0.62–2.18). The incidence of nausea or vomiting was similar between groups (OR 1.28; 95% CI 0.69–2.38). Blinding to allocation group was adequate (placebo 70%, oxycodone 47% P=.09).

The majority of women completed their follow up (85%, 150/172; in-person visit n=135 participants, text message: n=15) and this did not differ by treatment group (P=.85). Thirteen percent (22/170) of participants made an extra phone call to Planned Parenthood Columbia Willamette or to the study phone but only six (4%) total participants called with complaints of pain (four placebo, two oxycodone). Ten women had unscheduled visits with Planned Parenthood Columbia Willamette or another health care provider (oxycodone four, placebo six, P=.81); one for pain, one for bleeding, and eight for nonurgent issues.

Two women had incomplete abortions or an ongoing pregnancy requiring surgical aspiration (oxycodone one, placebo one) as confirmed at their follow-up appointment, which is consistent with the typical completion rate for medical abortion of 95–99%.2 One additional woman required an extra dose of misoprostol to complete, and one participant completed with expectant management (oxycodone two, placebo zero). All of these participants reported by text survey that they believed they had passed the pregnancy. No participant reported a serious adverse event due to the study drugs. One participant in the oxycodone group presented to the emergency department with dizziness and was diagnosed with an inner ear infection. One participant in the placebo group presented with heavy bleeding, which was treated with expectant management.


We found that oxycodone does not reduce the amount or duration of maximum pain experienced or the duration of overall pain in women undergoing medical abortion up to 10 0/7 weeks of gestation. Our findings did not differ by gestational age. Overall, we found that the use of opioids provided no overall benefit for pain control for women undergoing medical abortion. Our study also helps to further characterize the pain experienced by women undergoing medical abortion as it has not been well-described. Women reported a relatively high peak pain level of 8 out of 10 on a numerical rating scale, which occurred 2.5 to 4 hours after misoprostol use and lasted for about 1 hour.

Interestingly, we had a relatively high proportion of women who did not adhere to the study protocol. The majority of these women used ibuprofen alone and had lower reported pain scores and shorter duration of pain than protocol followers, implying that their pain was not severe enough to need additional opioids and that ibuprofen was sufficient. Because we did not initially account for so many women deviating from protocol, we increased our sample size in the lower gestational age cohort to be powered to analyze our primary outcome. However, analyzing our outcomes using a per-protocol analysis also failed to show a difference in maximum pain scores, strengthening our conclusions that the use of opioids does not reduce maximum pain induced by medical abortion.

Women were instructed to take oxycodone at the onset of uterine cramping in an attempt to provide pain relief when needed most. We feel it is unlikely that the timing of administration was too late to provide participants’ coverage of their peak pain, because oxycodone provides analgesic relief in as soon as 15 minutes. Even if women missed the peak coverage with administration, we would anticipate a shorter duration of pain, but we did not find this. We did find that a significant subset of women still experience severe pain, despite use of ibuprofen and oxycodone, indicating an area of focus for future research.

The major strength of this study is its design as a placebo-controlled, double-blind, randomized controlled trial, which reduced confounding. Our study is the only one to date that examines the effect of opioid use for medical abortion using the current recommended regimen with mifepristone and misoprostol. We used an oxycodone dosage of 10 mg, which is appropriate to treat severe pain. In addition, oxycodone was only available at this clinic to eligible women through study participation to increase internal validity. Our data collection rate was near 100%, indicating text messaging is a reliable form of data collection in this population. Text-message use also increased follow-up data for women who were unable to return to their scheduled clinic visit.

Limitations include that the adjunctive medication was prescribed and not dispensed, potentially a limiting factor for women who may not have had access to it owing to financial or logistical issues, thus potentially lowering its use if it were otherwise dispensed. This clinic site did not prescribe routine opioids. Providing a prescription mimics most clinical scenarios, so the usage rate in this study likely simulates real-world usage. Women who enrolled had to be accepting of the possibility of using an opioid and be opioid naive, which may skew the population towards women who are fearful of pain and more likely to use opioids. Our exclusion criteria limit the generalizability of our results for women who may experience pain differently, such as those who have or are using illicit drugs or chronic opioids or women with chronic pain disorders. A participant's anticipated pain did strongly correlate with experienced pain. Therefore, it would be reasonable to use this information to determine whether to offer additional therapies to aid in pain control on an individual patient basis. Although we did not study pain management for medical management of early pregnancy loss, we believe, given the similarities in the pain experience, that it is reasonable to extrapolate our study findings to this patient population as well.

As health care providers, we have been the largest contributor to the opioid epidemic, but changing prescribing patterns can have a significant effect on exposure and availability of opioids to the public. Even with prescribing few tablets, 85% (874 tablets) of the prescribed oxycodone tablets for adjunctive medication went unused. Therefore, all opioid prescribing patterns should be evaluated and individualized to prevent overprescribing. We can conclude that routinely prescribing opioids for medical abortion up to 10 0/7 weeks of gestation is unnecessary, but, if opioids are requested, we would recommend providing four tablets or fewer.

Authors' Data Sharing Statement

  • Will individual participant data be available (including data dictionaries)? Yes.
  • What data in particular will be shared? All de-identified individual participant data collected during the trial.
  • What other documents will be available? Study protocol.
  • When will data be available (start and end dates)? Immediately following publication and ending 3 years after article publication.
  • By what access criteria will data be shared (including with whom, for what types of analyses, and by what mechanism)? Researchers who provide a methodologically sound proposal and rationale for use of the data set, their proposed analyses and results through academically established means. Oregon Health & Science University maintains a high community standard for the free release of data and materials. Transfer of resources is subject to the acceptance of a Materials Transfer Agreement as required by policy at Oregon Health & Science University. Oregon Health & Science University understands and agrees to comply with the NIH policy on Sharing Research Data and on Sharing Model Organisms.


1. Jones RK, Jerman J. Abortion incidence and service availability in the United States, 2011. Perspect Sex Reprod Health 2014;46:3–14.
2. Medical management of first-trimester abortion. Practice Bulletin No. 143. American College of Obstetricians and Gynecologists. Obstet Gynecol 2014;123:676–92.
3. Kulier R, Kapp N, Gülmezoglu AM, Hofmeyr GJ, Cheng L, Campana A. Medical methods for first trimester abortion. The Cochrane Database of Systematic Reviews 2011. Art. No.: CD002855. doi: 10.1002/14651858.CD002855.pub4.
4. World Health Organisation Task Force on Post-ovulatory Methods of Fertility Regulation; Special Programme of Research, Development and Research Training; World Health Organization. Comparison of two doses of mifepristone in combination with misoprostol for early medical abortion: a randomised trial. BJOG 2000;107:524–30.
5. The safety and quality of abortion care in the United States. Washington, DC: The National Academies Press; 2018.
6. Fiala C, Cameron S, Bombas T, Parachini M, Saya L, Gemzell-Danielsson K. Pain during medical abortion, the impact of the regimen: a neglected issue? A review. Eur J Contracept Reprod Health Care 2014;19:404–19.
7. Raghavan S, Comendant R, Digol I, Ungureanu S, Friptu V, Bracken H, et al. Two-pill regimens of misoprostol after mifepristone medical abortion through 63 days' gestational age: a randomized controlled trial of sublingual and oral misoprostol. Contraception 2009;79:84–90.
8. el-Refaey H, Templeton A. Early abortion induction by a combination of mifepristone and oral misoprostol: a comparison between two dose regimens of misoprostol and their effect on blood pressure. Br J Obstet Gynaecol 1994;101:792–6.
9. Winikoff B, Dzuba IG, Creinin MD, Crowden WA, Goldberg AB, Gonzales J, et al. Two distinct oral routes of misoprostol in mifepristone medical abortion: a randomized controlled trial. Obstet Gynecol 2008;112:1303–10.
10. Raymond EG, Weaver MA, Louie KS, Dean G, Porsch L, Lichtenberg ES, et al. Prophylactic compared with therapeutic ibuprofen analgesia in first-trimester medical abortion: a randomized controlled trial. Obstet Gynecol 2013;122:558–64.
11. Fiala C, Cameron S, Bombas T, Parachini M, Agostini A, Lertxundi R, et al. Pain management for up to 9 weeks medical abortion—an international survey among abortion providers. Eur J Obstet Gynecol Reprod Biol 2018;225:181–4.
12. Guilbert ER, Hayden AS, Jones HE, White KO, Steven Lichtenberg E, Paul M, et al. First-trimester medical abortion practices in Canada: National survey. Can Fam Physician 2016;62:e201–8.
13. Wiebe E. Pain control in medical abortion. Int J Gynaecol Obstet 2001;74:275–80.
14. Lugo RA, Kern SE. The pharmacokinetics of oxycodone. J Pain Palliat Care Pharmacother 2009;18:17–30.
15. Litkowski LJ, Christensen SE, Adamson DN, Van Dyke T, Han SH, Newman KB. Analgesic efficacy and tolerability of oxycodone 5 mg/ibuprofen 400 mg compared with those of oxycodone 5 mg/acetaminophen 325 mg and hydrocodone 7.5 mg/acetaminophen 500 mg in patients with moderate to severe postoperative pain: a randomized, double-blind, placebo-controlled, single-dose, parallel-group study in a dental pain model. Clin Ther 27;2005:418–29.
16. PPFA manual of medical standards and guidelines. 2018.
17. Jensen MP, Chen C, Brugger AM. Interpretation of visual analog scale ratings and change scores: a reanalysis of two clinical trials of postoperative pain. J Pain 2003;4:407–14.
18. Farrar JT, Berlin JA, Strom BL. Clinically important changes in acute pain outcome measures. J Pain Symptom Manage 2003;25:406–11.

Supplemental Digital Content

© 2019 by the American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.