Share this article on:

Oral Compared With Intravenous Sedation for First-Trimester Surgical Abortion: A Randomized Controlled Trial

Allen, Rebecca H. MD, MPH; Fitzmaurice, Garrett ScD; Lifford, Karen L MD, ScD; Lasic, Morana MD; Goldberg, Alisa B. MD, MPH

doi: 10.1097/AOG.0b013e3181938758
Original Research

OBJECTIVE: To test the equivalency of oral sedation and intravenous sedation for pain control in first-trimester surgical abortion.

METHODS: Women undergoing suction curettage at less than 13 weeks of gestation were randomly assigned to oral sedation, 10 mg of oxycodone and 1 mg of lorazepam, or intravenous sedation, 100 micrograms fentanyl and 2 mg midazolam. All patients received 800 mg of preoperative ibuprofen and a 20-mL paracervical block with 1% lidocaine. The primary outcome was intraoperative pain as measured on a 21-point verbal rating scale that had a range from 0 to 100 (0=no pain and 100=worst pain ever) with an equivalence margin for the treatment group comparison of ±10.

RESULTS: Of 130 women, 65 were randomly assigned to oral sedation and 65 to intravenous sedation. The groups differed at baseline by age and preoperative ratings of depression, stress, and anxiety; however, when adjusted for these differences, the primary results were unaffected. Mean intraoperative pain scores, controlling for age and preoperative depression, stress, and anxiety, were 61.2 for oral sedation and 36.3 for intravenous sedation (mean difference 24.9, 95% confidence interval 15.9–33.9). Other findings included no difference in postoperative adverse effects and less satisfaction with pain control with oral sedation compared with intravenous sedation.

CONCLUSION: Oral sedation, as studied, is not equivalent to intravenous sedation for pain control during first-trimester surgical abortion.



Oral sedation is not equivalent to intravenous sedation for pain control during first-trimester surgical abortion.

From the Departments of 1Obstetrics and Gynecology, Women and Infants’ Hospital, Providence, Rhode Island; 2Biostatistics, Harvard School of Public Health, Boston, Massachusetts; 3Obstetrics and Gynecology, Boston University Medical Center, Boston, Massachusetts; 4Anesthesiology, Perioperative and Pain Medicine, Brigham and Women’s Hospital, Boston, Massachusetts; and 5Obstetrics, Gynecology, and Reproductive Biology, Brigham and Women’s Hospital and Planned Parenthood League of Massachusetts, Boston, Massachusetts.

Financial support for this study was provided by an anonymous foundation.

The opinions expressed in this article do not necessarily reflect those of Planned Parenthood Federation of America, Inc.

Presented at the Association of Reproductive Health Professionals, Society of Family Planning, and Planned Parenthood Federation of America Annual Meeting, Washington, DC, September 17–20, 2008.

Corresponding author: Rebecca H. Allen, MD MPH, 101 Dudley Street, Providence, RI 02905; e-mail:

Financial Disclosure The authors did not report any potential conflicts of interest.

In the United States, an estimated 1.2 million abortions were performed in 2005.1 Of these procedures, 88% occurred during the first 12 weeks of pregnancy.2 Since the legalization of abortion in the United States, many different types of analgesia have been used during first-trimester abortion procedures, ranging from a local paracervical block alone to general anesthesia.3 Most abortions in North America are performed under local anesthesia combined with preoperative nonsteroidal antiinflammatory drugs (NSAIDs).4 Some providers also use nonpharmacologic pain control techniques during the operative procedure such as focused breathing, visualization, and localized massage.4 To provide additional pain control, many clinics offer oral or intravenous (IV) anxiolytics and narcotics. Previous studies of IV narcotics and anxiolytics (moderate sedation) have shown minimal to no decrease in pain score compared with local anesthesia alone.5–8 However, one study demonstrated increased patient satisfaction with additional medications despite no large effect on pain scores.7 Given the modest effect of intravenous moderate sedation on pain during first-trimester abortion, we hypothesized that a similar effect could be achieved with oral sedation.

One previous study has shown a small reduction in anxiety with the use of oral anxiolytics for abortion, but no reduction in pain.9 There is a need to explore the efficacy of oral agents in surgical abortion for several reasons. First, administering IV sedation is not feasible in many private offices and clinics where women receive general gynecologic care. Improving pain control for early abortion with oral agents may enable more physicians to offer abortions in their offices and general gynecology clinics, thus expanding access to care. Additionally, moderate IV sedation requires close monitoring that can significantly increase the cost of providing abortion services. Given that many patients pay directly for abortions, increasing the cost of services can be prohibitive for many women. Oral medications may also be more appealing to patients seeking sedation who are fearful of needles.

The purpose of this study was to test the equivalency of oral sedation (oxycodone 10 mg orally plus lorazepam 1 mg sublingually) and intravenous sedation (fentanyl 100 micrograms plus midazolam 2 mg) for first-trimester surgical abortion. We chose the oral regimen based on accepted starting doses for these medications in opioid-naïve individuals and consideration of opioid equianalgesic tables. We hypothesized that oral sedation would be equivalent to intravenous moderate sedation as defined by ±10 points on a 21-point verbal rating scale that had a range from 0 to 100 (0=no pain and 100=worst pain ever). We also sought to describe patient satisfaction with each method of pain control and characteristics of each method, such as adverse effect profile, recovery room time, and postoperative pain.

Back to Top | Article Outline


We conducted a randomized, double-blind, placebo-controlled trial at Planned Parenthood League of Massachusetts in Boston, Massachusetts from July 2006 to July 2007. The Partners Human Research Committee and the Planned Parenthood Federation of America granted approval for the study protocol, and all patients provided informed written consent. Women at least 18 years old requesting surgical abortion with an intrauterine pregnancy between 5 0/7 weeks and 12 6/7 weeks confirmed by pelvic ultrasonography were recruited only after a decision to proceed with the abortion had been made. The participants spoke English or Spanish or had staff or a professional translator available to translate for all study procedures. Participants had to be eligible for intravenous sedation per clinic protocol (no food or nonclear liquids for 6 hours before the procedure, no clear liquids for 2 hours before the procedure, and has an escort or ride home). Exclusion criteria included a known allergy or intolerance to any medications used in the study (ibuprofen, lidocaine, fentanyl, midazolam, oxycodone, and lorazepam), chronic narcotic, benzodiazepine, or barbiturate use within the past year, history of IV drug use within the past year, and weight less than 120 lb. The standard of care at Planned Parenthood League of Massachusetts for pain control was the choice between intravenous sedation with local anesthesia and ibuprofen or local anesthesia and ibuprofen alone. Potential participants were told that the trial was being conducted to learn whether oral sedation was equivalent to intravenous sedation, but that both were used for pain control for first-trimester surgical abortion.

After consent and study enrollment, participants were randomly assigned to their treatment group in an allocation ratio of 1:1. Computerized randomization lists were generated in blocks of four and six without stratification by a researcher not involved in participant recruitment or patient care. At the point of randomization, a sealed, opaque envelope containing the randomization assignment was retrieved by study staff and opened by a research nurse. The research nurse prepared the study medications and placebos based on the randomization scheme, thus maintaining blinding to study allocation for participants, study staff, and clinicians. The medications and participant data sheets were labeled with, and participants enrolled by, consecutive numbers. No identifiers of treatment group were placed on participant data sheets or medications, only the number in order of enrollment.

After randomization, we surveyed participants with a preoperative questionnaire measuring self-assessment of pain tolerance, preoperative nausea and vomiting levels, expected pain levels, acceptable pain levels, alcohol use, and self-reported race, ethnicity, and primary language. We also administered the Depression Anxiety Stress Scales (DASS), a set of three self-report scales designed to measure the negative emotional states of depression, anxiety, and stress.10 In addition, participant descriptors, including age, height, weight, gestational age, obstetric history, uterine position, and type of health insurance were abstracted from the medical record.

Once the preoperative questionnaire was completed, all participants received 800 mg of oral ibuprofen. Participants then ingested the oral pills they were assigned and waited 60 minutes to allow for onset of action. Women assigned to the oral group ingested two oral 5-mg oxycodone tablets and one sublingual 1-mg lorazepam tablet, whereas women assigned to the IV group ingested three placebo tablets, two oral tablets (1,000 mg Vitamin C) and one sublingual tablet (1,000 micrograms vitamin B-12). The active tablets and vitamin placebos were similar in color and size but not identical. After 60 minutes, the participant was asked to guess her group assignment (to assess the efficacy of blinding), and then the participant was taken to the procedure room.

Women assigned to the IV group received 2 mL of fentanyl (100 micrograms) and 2 mL of midazolam (2 mg), whereas women assigned to the oral group received one 2-mL syringe of normal saline followed by another. The syringes for active medication and placebo were identical in appearance and labeling. Approximately 2 minutes after the administration of the IV medications, the procedure nurse assessed the participant’s level of sedation by using standard criteria, the Observer’s Assessment of Alertness/Sedation (OAA/S).11 Participants received supplemental oxygen during the procedure. An additional 25–50 micrograms of fentanyl was allowed if participants requested additional pain medication during the procedure.

The surgical abortion was performed in the standard fashion using Pratt dilators and electric or manual suction curettage. The paracervical block was standardized and consisted of 20 mL of 1% plain-buffered lidocaine distributed in the following fashion: 2 mL into the anterior lip of the cervix before tenaculum placement and 8 mL into the 4 o’clock and 8 o’clock positions around the cervix, respectively. Sharp curettage was avoided unless deemed clinically necessary by the operating physician. Six attending-level surgeons performed the procedures. Within 3 minutes after removal of the speculum, study personnel asked participants to report the level of pain using the 21-point 0 to 100 verbal pain rating scale. The participant was asked, “On this scale where 0 is no pain and 100 is the worst pain ever, what was that like for you?” We also measured the participant’s pain rating with IV insertion in a similar manner to determine baseline pain tolerance.

After the abortion procedure, participants were transferred to the recovery room for standard postoperative care. Upon arrival in the recovery room, participants were again asked to guess their assigned treatment to assess the efficacy of blinding. Approximately 30 minutes after arriving in the recovery room, study staff administered a postoperative questionnaire, which measured secondary outcomes such as postoperative pain and adverse effects of the regimen, including levels of dizziness, drowsiness, nausea, and vomiting. Participant satisfaction with the pain control regimen was also measured with a series of questions. Additional information was collected from the medical record, including the type of aspiration performed (electric or manual), the use of sharp curettage, the name of the operating physician, acute complications, operative time, and recovery room time. Participants were discharged after having met standard clinic discharge criteria.

We compared the equivalency of two groups: 1) local anesthesia with ibuprofen and intravenous sedation using fentanyl and midazolam and 2) local anesthesia with ibuprofen and oral sedation using oxycodone and lorazepam. The primary outcome for this trial was the intraoperative pain score as measured by a verbal rating pain scale containing 21 points for pain discrimination that ranged from 0 to 100. The 11-point, 0 to 10 verbal rating pain scale has been validated by various studies to be an effective way to measure acute pain intensity and is used widely in the abortion literature.6,7,9,12,13 However, research has shown that many participants treat the 11-point scale as a 21-point scale by responding with numbers between two whole numbers on the 0 to 10 measure.14 For example, in our pilot study, we found that 6% of participants responded between two whole numbers on the 0 to 10 scale, and we analyzed the scale as a 21-point scale.5 Therefore, for this trial, we chose to use a 21-point, 0 to 100 verbal rating scale to measure pain.

For this equivalency trial, we set the equivalence bounds at ±10, based on clinical knowledge of the 0 to 100 pain scale.15,16 Our expectation for the “true” mean difference between oral sedation and intravenous sedation was 2.5 points on the pain scale. We assumed a standard deviation of 22, derived from prior work at our institution.5 We planned to perform one interim analysis at 50% enrollment, and using the O’Brien-Fleming criterion, the interim analysis had an alpha of 0.005; the final analysis had a planned alpha of 0.048.17 Based on these assumptions, we determined that a total sample size of 234 participants (117 per group) was required to have a power of 82% to test equivalence. To account for a potential 10% participant dropout rate or incomplete data, we added 26 extra participants for 260 total (130 in each arm).

The statistical software package SPSS 16.0 (SPSS, Inc., Chicago, IL) was used for all data analyses. Descriptive statistics (means and percentages) for the baseline characteristics of the two study groups were compared using χ2 tests for categorical data, χ2 tests for trend for ordinal data,18,19 and t tests for continuous data. The comparison of treatment groups used the “intention-to-treat” principle, and pain scores were analyzed as both continuous and ordinal; for the latter, 0 to less than 40 was labeled mild pain, 40 to less than 70 was labeled moderate pain, and 70 or greater was labeled severe pain. In the analyses of our primary and secondary outcomes, we controlled for factors that differed between groups, despite randomization, at a significance level of alpha=0.10; these factors included age and depression, stress, and anxiety scores on the DASS. Adjusted analyses were performed using both linear and ordinal regression where appropriate for the type of outcome.

An independent Data Safety and Monitoring Board, chosen and convened by the study investigators, was scheduled to meet once at 25% enrollment to assess safety and once at 50% enrollment to evaluate the results of the interim analysis. Prespecified stopping rules included lack of safety, futility, or efficacy. Statistical monitoring criteria used the O’Brien-Fleming stopping boundaries to preserve the overall type-1 error rate at 5%. Based on the results of the interim analysis, which was done at 50% enrollment and presented here, the Data Safety and Monitoring Board recommended stopping the study. The study was halted due to futility, ie, it was considered highly improbable that further data collection could yield evidence of equivalence.

Back to Top | Article Outline


A total of 2,193 patients seeking surgical abortion services between 5 and 13 weeks of gestation were screened for study eligibility between July 2006 and July 2007. Of these, 891 were ineligible, 1,170 declined to participate, and 132 were enrolled. The first two participants were excluded from the analysis for protocol violations, having received oxycodone combined with acetaminophen instead of oxycodone alone, leaving 130 total, or 65 in each arm (Fig. 1). No participants withdrew from the study.



Overall, participants reported a mean age of 25.0 years, and those who declined to participate had a mean age of 26.4 years (P=.02). Those who enrolled in the trial were of greater gestational age (mean 60 days) than those who declined to enroll (mean 57 days) (P=.02). Forty-five percent of participants and 46% of nonparticipants were nulliparous (P=.82). More than one half of participants (62%) and nonparticipants (55%) had at least one prior termination of pregnancy (P=.16). A total of 40% of nonparticipants were enrolled in Medicaid, 37% in private insurance plans, and 23% were self-pay compared with participants, who reported 56% Medicaid, 20% private insurance, and 24% self-pay (P=.001).

Of those who enrolled in the study and were assigned to oral or intravenous sedation, both groups had similar sociodemographic and medical characteristics except that participants in the intravenous group were older and had less severe scores on the DASS instrument (Table 1). There were no differences between the groups in preoperative nausea and vomiting levels or in alcohol consumption in the past 6 months. When asked what an acceptable level of pain would be for the abortion procedure, participants in the oral group selected a mean pain score of 39.2 (median 40) and the intravenous group reported a mean of 37.6 (median 40) on a scale from 0 to 100, with no difference between the groups (P=.65).

Table 1

Table 1

The groups had similar baseline pain scores with IV insertion (Table 2). For intraoperative pain, the mean difference in pain scores between the groups was 24.9 (95% confidence interval [CI] 15.9–33.9), with the IV group having significantly lower pain scores than the oral sedation group (Table 2). After adjusting for age and the DASS depression, anxiety, and stress scores, there was no discernible change in the mean difference in intraoperative pain scores between groups. Because the 95% CI for the mean difference does not fall within the predetermined equivalency bounds, ±10, the two groups are not equivalent. Study physicians performed manual uterine aspiration in five (7.7%) women in the oral arm and six (9.2%) women in the IV arm (P=.58). Only one woman underwent sharp curettage in the oral arm compared with none in the IV arm. No uterotonic agents were used in either group postoperatively. No participants received additional pain medication intraoperatively. Adjustment for physician effects on pain scores had no discernible effect on the treatment arm comparison (data not shown). Although postoperative pain scores differed significantly between groups in the unadjusted analysis (mean difference 7.4, 95% CI 1.0–13.7), the difference was reduced and nonsignificant after adjusting for age and the DASS depression, anxiety, and stress scores (adjusted mean difference 6.2, 95% CI –0.6 to 13.0).

Table 2

Table 2

There was no statistically significant difference between the two groups in operative time or recovery time (Table 3). In the unadjusted analysis, the oral group experienced significantly more postoperative nausea than the IV group, with an odds ratio of 2.4 (95% CI 1.1–5.5). However, when adjusted for age and DASS depression, anxiety, and stress scores, the adjusted odds ratio declines to 2.3 and is no longer significant (95% CI 0.94–5.5). In both adjusted and unadjusted analyses, there was no difference between the groups in postoperative vomiting, dizziness, or drowsiness. Analyzing participants’ expectations of pain compared with their actual experience revealed that the oral group was more likely to say their pain was the same or greater than expected than the IV group (Table 3). The intravenous sedation group found their pain control to be more acceptable than the oral group on several measures (Table 3). Controlling for age, and depression, stress, and anxiety DASS scores did not qualitatively alter these results. There were no adverse events in the trial. At no time during the trial did the oxygen saturation level in participants fall below 90%, and reversal agents were not needed.

Table 3

Table 3

To explore whether the difference in pain control was mediated by sedation level, as measured by the OAA/S, we analyzed the difference in pain scores between groups, controlling for sedation score. Not surprisingly, the IV group had significantly higher sedation scores on the OAA/S than the oral group (P<.001), with 49.2% in the oral group scoring in the alert (not sedated) category compared with 23.1% in the IV group. After appropriate adjustment for the OAA/S score, we found that only a relatively small proportion (16%) of the differences in pain score could be explained by differences in sedation.

Finally, to assess the efficacy of blinding, we asked participants to guess their treatment arm after oral agent ingestion and after IV agent administration. A total of 52 of 61 participants (85.2% with four missing values) in the oral arm correctly guessed their treatment assignment and 53 of 61 participants (86.9% with 4 missing values) in the IV arm correctly guessed their treatment assignment. If blinding were perfect, such guesses would reflect chance alone, and each group would guess correctly approximately 50% of the time. Exact tests for a binomial probability of 0.5 lead to rejection of this null hypothesis (P<.001). Therefore, it seems highly likely that blinding was imperfect in our study.

Back to Top | Article Outline


The ideal method of pain control for first-trimester surgical abortion should be safe, effective, and affordable. We studied oral sedation (oxycodone 10 mg orally plus lorazepam 1 mg sublingually) as an alternative to moderate intravenous sedation (fentanyl 100 micrograms plus midazolam 2 mg) and found that oral sedation was not equivalent. Furthermore, women in the IV group were more satisfied with their pain control than women in the oral group. There were no differences between the groups in terms of postoperative adverse effects; however, the study was underpowered for this assessment.

Paracervical blocks using procaine-derived anesthetics combined with NSAIDs are widely used for pain control during first-trimester surgical abortion.3,4 However, on average, women receiving this regimen report experiencing moderate pain, ranging from 5 to 7 on a 10-point scale.5,9,13 In our study, the oral arm reported pain levels (mean 61.2, median 65) comparable with these published results for local anesthesia and NSAIDs. While this is suggestive, we cannot conclude that oral sedation is no better than local anesthesia alone, because we did not have a local anesthesia alone arm of the trial to make a direct comparison. Furthermore, it is not known whether higher doses of lorazepam or oxycodone or longer preoperative waiting periods might have improved efficacy.

Women in the IV arm reported pain levels (mean 36.3, median 30) that are lower than pain levels reported in previously published trials of moderate intravenous sedation for first-trimester surgical abortion. This is likely due to dosage variations. One randomized trial of 368 women comparing local anesthesia alone with local anesthesia with the addition of intravenous fentanyl (50–100 micrograms) found that IV fentanyl reduced the pain of first-trimester abortion by only 1.0 point on a 0 to 10 point verbal rating scale, from 5.3 to 4.3.6 This pain reduction was less than what study participants desired (2.0 points), and the authors concluded that the pain reduction provided by intravenous fentanyl was of questionable clinical significance. Another randomized study of 100 participants compared local anesthesia alone with local anesthesia in combination with intravenous fentanyl and midazolam.7 Patients randomly assigned to the treatment arm received 25 micrograms of fentanyl and 2 mg of midazolam. The treatment arm reported a mean score of 5.5, whereas those in the placebo arm reported a mean score of 5.0, which was not significantly different. However, women who received intravenous sedation reported increased satisfaction with their abortion procedure. Using 100 micrograms of fentanyl and 2 mg of midazolam, our participants reported intraoperative pain scores that matched their preoperative assessment of an acceptable level and reported high satisfaction with the pain control method. Therefore, it seems that moderate intravenous sedation with this regimen is effective for pain control in most patients.

The strengths of this study include random assignment and enrolling a diverse population in an actual practice setting. However, we did exclude women weighing less than 120 lb because we used standard medication dosing and did not want to risk overdosing this population. Our study does have limitations, because pain, a biopsychosocial experience influenced by many factors, is difficult to measure. We attempted to control for this with randomization, double-blinding, measuring baseline pain scores with IV insertion, and adjusting for the DASS instrument results. However, blinding seems to have been imperfect in our study and this may have influenced the results. In addition, all the women in our study agreed to have their analgesia regimen randomly assigned and thus were willing to give up control over their choice of pain management. Although not a significant difference, it is notable that more women who opted not to participate in the study were undergoing their first termination of pregnancy than women who enrolled. Therefore, we likely recruited women who were not as anxious about their pain control compared with other women who declined to participate in the study. Finally, of 1,302 women, we were only able to recruit 132, or 10%. We adequately remunerated participants for their time and effort and we had dedicated study staff for recruitment. However, difficulties in enrolling potential participants likely stemmed from stress surrounding the abortion procedure, the fact that all participants had to be IV eligible, and the 1 hour wait time for the oral medications to take effect. For many women, the additional wait time was too long for them to justify participation in the study. This selection bias may make our results less generalizable to the population of women seeking abortions.

Further studies of oral options for pain control during first-trimester surgical abortion are needed. We attempted to design a safe and practical regimen that would be equivalent to moderate intravenous sedation and failed. Our previous investigation found that the addition of lorazepam alone to local anesthesia did not improve pain scores and was associated with more dissatisfaction.5 In this study, adding oxycodone to the oral premedications did not decrease pain compared with moderate intravenous sedation. To date, nonintravenous methods of pain control for first-trimester surgical abortion have been disappointing, but other investigators may have more success with different regimens. For now, the majority of women who wish to minimize pain during surgical abortion will likely have acceptable pain control and be satisfied with the IV regimen used in this study.

Back to Top | Article Outline


1. Jones RK, Zolna MR, Henshaw SK, Finer LB. Abortion in the United States: incidence and access to services, 2005. Perspect Sex Reprod Health 2008;40:6–16.
2. Strauss LT, Gamble SB, Parker WY, Cook DA, Zane SB, Hamdan S, et al. Abortion surveillance—United States, 2004. MMWR Surveill Summ 2007;56:1–33.
3. Paul M, Lichtenberg ES, Borgatta L, Grimes DA, Stubblefield PG, editors. A clinician’s guide to medical and surgical abortion. New York (NY): Churchill Livingstone; 1999.
4. Lichtenberg ES, Paul M, Jones H. First trimester surgical abortion practices: a survey of National Abortion Federation members. Contraception 2001;64:345–52.
5. Allen RH, Kumar D, Fitzmaurice G, Lifford KL, Goldberg AB. Pain management of first-trimester surgical abortion: effects of selection of local anesthesia with and without lorazepam or intravenous sedation. Contraception 2006;74:407–13.
6. Rawling MJ, Wiebe ER. A randomized controlled trial of fentanyl for abortion pain. Am J Obstet Gynecol 2001;185:103–7.
7. Wong CY, Ng EH, Ngai SW, Ho PC. A randomized, double blind, placebo-controlled study to investigate the use of conscious sedation in conjunction with paracervical block for reducing pain in termination of first trimester pregnancy by suction evacuation. Hum Reprod 2002;17:1222–5.
8. Rawling MJ, Wiebe ER. Pain control in abortion clinics. Int J Gynaecol Obstet 1998;60:293–5.
9. Wiebe E, Podhradsky L, Dijak V. The effect of lorazepam on pain and anxiety in abortion. Contraception 2003;67:219–21.
10. Lovibond PF, Lovibond SH. The structure of negative emotional states: comparison of the Depression Anxiety Stress Scales (DASS) with the Beck Depression and Anxiety Inventories. Behav Res Ther 1995;33:335–43.
11. Chernik DA, Gillings D, Laine H, Hendler J, Silver JM, Davidson AB, et al. Validity and reliability of the Observer’s Assessment of Alertness/Sedation Scale: study with intravenous midazolam. J Clin Psychopharmacol 1990;10:244–51.
12. Wiebe ER. Comparison of the efficacy of different local anesthetics and techniques of local anesthesia in therapeutic abortions. Am J Obstet Gynecol 1992;167:131–4.
13. Romero I, Turok D, Gilliam M. A randomized trial of tramadol versus ibuprofen as an adjunct to pain control during vacuum aspiration abortion. Contraception 2008;77:56–9.
14. Jensen MP, Turner JA, Romano JM. What is the maximum number of levels needed in pain intensity measurement? Pain 1994;58:387–92.
15. Schuirmann DJ. A comparison of the two one-sided tests procedure and the power approach for assessing the equivalence of average bioavailability. J Pharmacokinet Biopharm 1987;15:657–80.
16. Phillips KF. Power of the two one-sided tests procedure in bioequivalence. J Pharmacokinet Biopharm 1990;18:137–44.
17. O’Brien PC, Fleming TR. A multiple testing procedure for clinical trials. Biometrics 1979;35:549–56.
18. Cochran WG. Some methods for strengthening the common χ2 tests. Biometrics 1954;10:417–51.
19. Armitage P. Tests for linear trends in proportions and frequencies. Biometrics 1955;11:375–86.
© 2009 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.