In a 2018 analysis of induced abortion care in the United States, the National Academies of Sciences, Engineering, and Medicine concluded that legal induced abortion is safe and effective, yet the quality of care varies geographically owing to state-level regulations.1 Timeliness and equity, two dimensions of quality, greatly depend on the distribution of services, requiring care to be readily available locally to all women when needed. Obstetrician–gynecologists (ob-gyns) play an important role in ensuring timely access to abortion care because they may be the first clinicians a patient with an unintended pregnancy encounters.
The most recent representative survey of U.S. ob-gyns performed in 2008–2009 found that 14% reported providing abortion care.2 Female physicians and physicians living in an urban area or reporting they were Jewish were significantly more likely to provide abortions, whereas those older than 35 years of age, living in the South and Midwest (compared with the Northeast), or reporting high religious motivation were less likely to provide. The survey did not ask about the abortion methods provided. Medication abortion has become increasingly popular in recent years, accounting for 31% of nonhospital abortions in 2014.3 However, a study from 2007 suggested that uptake of medication abortion among those not already providing surgical abortion was limited.4
The objective of this study was to provide an updated estimate of the proportion of ob-gyns who provide induced abortion, disaggregated by method, as well as physician and practice characteristics associated with provision. We also aimed to explore the barriers to provision of medication abortion, including the requirement to stock mifepristone in one's office.
This cross-sectional survey included Fellows and Junior Fellows of the American College of Obstetricians and Gynecologists (ACOG) who were currently in practice. The American College of Obstetricians and Gynecologists is a professional society that represents approximately 90% of practicing U.S. ob-gyns and has more than 58,000 members. The study population included 1,000 members of ACOG's Collaborative Ambulatory Research Network, which is a demographically representative group of practicing ACOG members who voluntarily participate in surveys conducted by the ACOG Research Department.5 We randomly selected 1,000 participants from a list of more than 1,400 current members of the Collaborative Ambulatory Research Network. With a minimum response rate of 60%, the maximum margin of error in the estimation of proportions with a 95% CI in a random sample of 600 participants is ±4.0%.
Data collection took place between August 2016 and March 2017. We initially invited participants via email with a link to access the survey, which was administered using Qualtrics software. The email invited participants to complete a survey on “selected ob-gyn practices” and said the survey included “questions about how you manage patients with early pregnancy loss and unintended pregnancy and those seeking contraception.” Those who were retired or did not wish to participate were given a link to opt out. We sent email reminders weekly for 5 weeks to those who had not completed the survey, and we mailed one postcard reminder. Participants who had not completed the electronic survey after these reminders were mailed a paper version of the questionnaire and a prepaid return envelope. We sent a second mailing containing a shortened version of the questionnaire to those who completed neither the electronic nor the first mailed survey.
We pretested the survey among practicing ob-gyns for meaning and respondent interest. The final version included 14 demographic questions and 19 questions about induced abortion provision (the shortened version contained eight abortion questions), in addition to questions on other family planning topics. Participants who reported that at least some of their patients were of reproductive age (15–49 years old) were asked whether they “had any patients in the last 12 months who wanted or needed an abortion or termination of pregnancy.” If they answered affirmatively, we then asked whether, in the prior 12 months, they had provided a surgical abortion (dilation and sharp curettage, electric or manual vacuum aspiration) or a medical (medication) abortion. If they did provide any abortion care, we asked them to estimate the number of procedures and the locations where the procedures took place (ambulatory surgical center, hospital operating room, outpatient office setting of primary practice, Planned Parenthood or other specialized clinic, or other location).
Because medication abortion may be easier to provide in one's office than surgical abortion, we specifically explored the barriers to providing the former. We asked those who did not provide medication abortion in the prior year their reason. Participants could select from a list of 10 responses or write in a response, and multiple responses were allowed.
The U.S. Food and Drug Administration requires Mifeprex (mifepristone 200 mg) to be dispensed by the clinician in an office, clinic, or hospital; it may not be dispensed by prescription at a pharmacy.6 To determine whether this is a barrier to provision, we asked participants who reported not providing medication abortion the following: “Currently, if you want to provide medical abortion, you must stock the medications in your office. Would you offer medical abortion to your patients if you could write a prescription for mifepristone and misoprostol, and your patient could obtain both medications at a pharmacy?” Participants could select yes, no, or not sure.
To ensure data entry accuracy, 5% of all mailed questionnaires were randomly selected for review. We found only one consistent data entry error (a coding error on the shortened paper surveys), prompting all shortened surveys to be reviewed.
We conducted data analyses with Stata 15.1. Respondents who answered the primary outcome questions on whether they provided medication or surgical abortion in the past 12 months were included in the analysis. Responses that were incorrectly given (eg, anyone who did not follow the skip logic correctly on the paper survey) were removed, and any respondents who stated they were retired in the open responses were removed and recategorized as ineligible.
We categorized the state where respondents practiced into U.S. regions based on U.S. Census Bureau definitions (Northeast, Midwest, South, and West), and Puerto Rico was categorized as “South.” We coded participants' practice setting in the following hierarchy if more than one response was selected: university faculty practice, partnership or group, health maintenance organization or staff model, solo or private practice, other. We coded the community type where participants' practice was located in the following hierarchy if more than one response was selected: urban inner city; urban non–inner-city; suburban; midsize town, rural, or military.
If more than one response was provided for the location where induced abortions were performed, we coded abortion location in the following manner: abortion provided only in an ambulatory surgical center or hospital setting, any abortion provided in an outpatient office but not at a Planned Parenthood or specialized clinic, and any abortion in a Planned Parenthood or specialized clinic. We calculated the median number of abortions for each site where abortion was performed among those who provided any abortion.
Statistical tests assumed significance at P value less than .05. Descriptive statistics were calculated separately for respondents who did and did not report providing abortion in the prior year and among participants overall. Comparisons of categorical variables were analyzed using χ2 tests. Univariable and multivariable logistic regression analyses were performed to estimate odds of providing abortion in the prior 12 months (1 means provided medication or surgical abortion, 0 means did not provide induced abortion) by background characteristics. All variables were used as binary or categorical predictors, with reference groups based on sample size or meaningful comparison. We excluded missing data. We included all independent variables from Table 1 in an initial multivariable regression model. Sequentially, we removed from the model extraneous variables with a P value greater than .2. Age and gender were forced into the model a priori because they were found to be significant predictors of abortion provision in the representative survey of U.S. ob-gyns performed in 2008–2009.2 The Allendale Investigational Review Board approved this study.
Of the original sample of 1,000 members of the Collaborative Ambulatory Research Network, 18 were found to be no longer an ACOG member or not in practice, and two were unreachable by mail, resulting in a sample of 980 members. Among this sample, 655 respondents provided data (response rate 67%). Nonresponders were similar to respondents in terms of age and geographic region of practice, but slightly more women responded to the survey (61% of respondents vs 55% of nonresponders, P=.06). Of the 655 respondents, 99% reported seeing patients of reproductive age, and of these, 72% reported having a patient in the prior year who needed or wanted an induced abortion. Participants who reported seeing patients of reproductive age who answered the questions on whether they provided medication or surgical abortion (N=597) were included in this analysis. Overall, 515 (86%) took the long version of the survey, and this proportion was similar across geographic regions.
Table 1 shows the demographic characteristics of participants, separated by those who did and did not provide abortion in the prior year. Overall, 23.8% (95% CI 20.5%–27.4%) reported providing any type of induced abortion in the prior year. In univariable analyses, male physicians and physicians residing in the Midwest and South had lower odds of provision compared with female physicians and physicians residing in the Northeast, respectively. Physicians in a university faculty practice had higher odds of provision compared with those in a partnership or group practice, as did physicians practicing in urban or suburban communities compared with those practicing in midsize towns, rural areas, or military settings.
In multivariable regression analysis, controlling for age and gender, those practicing in the Midwest (adjusted odds ratio [AOR] 0.31, 95% CI 0.16–0.60) and the South (AOR 0.22, 95% CI 0.11–0.42) compared with participants living in the Northeast had lower odds of providing induced abortion. Those practicing in an urban inner city (AOR 2.71, 95% CI 1.31–5.60) or urban non–inner-city area (AOR 2.89, 95% CI 1.48–5.64) had higher odds of providing abortion compared with those in a midsize town, rural area, or who practiced exclusively in the military.
Table 2 shows details of participants' abortion practice in the prior year. Overall, 9.4% provided only surgically induced abortion, 4.0% provided only medication abortion, and 10.4% provided both surgical and medication abortion. Approximately 20% provided surgical abortion in the prior year, with about half of these performing electrical vacuum aspiration, one quarter using manual vacuum aspiration, and one quarter using sharp curettage. Fourteen percent provided medication abortion in the prior year; among these providers, 58.1% used the combined mifepristone–misoprostol regimen, 41.9% used misoprostol alone, and 10.5% used methotrexate and misoprostol. Participants were not specifically asked the gestational age range of patients provided medication abortion.
Participants who completed the long version of the survey and who provided abortion in the prior year gave information about where they provided abortion (n=120). A total of 32.5% reported providing abortion only in an ambulatory surgical center or hospital setting, 47.5% reported providing in an outpatient office but not at a specialized clinic or Planned Parenthood, and 10.8% reported providing at least some abortions at a specialized clinic or Planned Parenthood. However, the number of abortions performed by participants varied by site. The median number of abortions performed in the past year by those who only provided at an ambulatory surgical center or hospital was 5.5, and the median was 8.0 for those who provided in their office but not at a specialized clinic. Among those who provided any abortion at a specialized clinic, the median number of procedures was 112.0 in the prior year (Table 3).
Respondents who reported having patients seeking abortion but did not provide medication abortion in the prior year were asked about reasons why they did not provide (n=368). About one third (34%) cited personal, religious, or moral beliefs against abortion, 19% pointed to practice setting restrictions against abortion provision, and 16% mentioned office staff attitudes. Ten percent said there was no perceived need, and 8% said their patients had access to another provider or they referred out (Table 4). Eleven percent cited a lack of training as the reason for not providing medication abortion, and a similar proportion cited the requirement to stock the medications in their clinic. Nine percent cited the requirement to sign the provider agreement with the manufacturer of mifepristone. Some of the responses in the “other” category, each of which was cited by less than 4% of participants, included community attitudes, not having an ultrasound scanner in the office, lack of surgical backup, and laws and regulations.
When those who did not provide medication abortion but reported having patients seeking abortion were asked whether they would offer the service if they could write prescriptions for the medications, 28% said they would offer medication abortion, 47% said they would not, and 22% said they were not sure (Table 4). The number of respondents who said they would offer medication abortion by prescription (n=102) is more than the number in the survey who reported providing medication abortion in the prior year (n=86). This suggests that the proportion of ob-gyns offering induced medication abortion might increase from 14% currently to as much as 31% if it were not required to stock the medication in one's office.
This study aimed to describe induced abortion provision among practicing ob-gyns in the United States and found that 24% of them had performed abortion in the prior year. Factors associated with provision included practicing in the Northeast or West rather than the South or Midwest and practicing in an urban setting. This inequitable geographic distribution is similar to findings in previous research, which may be related to more restrictions on abortion provision in Southern and Midwestern states.2,7,8
The proportion of ob-gyns who reported providing induced abortion in our survey was higher than the 14% reported in a national survey conducted in 2008–2009.2 Our results were similar to a survey of ob-gyns who became board certified between 1998 and 2001, which found that 22% performed an induced abortion in the prior year.9 The Accreditation Council for Graduate Medical Education has required training in induced abortion in obstetrics and gynecology residency programs since 1996, which may have contributed to an increase in provision.
Among those providing induced abortion in the prior year, approximately one third reported providing the service only in an ambulatory surgical center or hospital setting. Given the safety of providing abortion in an outpatient setting,10 more research is needed to understand the reasons why ob-gyns choose to provide the service in a hospital. These physicians provided a small number of procedures in the prior year (median of 5.5), which is consistent with national data.3
Desai et al recently reported on a survey of ob-gyns in private practice, which found that 7% reported performing an induced abortion in 2013 or 2014.7 That survey focused on provision of abortions at the physician's office and excluded specialized clinics providing abortion care. About half of the ob-gyns in our survey who provided abortion reported doing so in an outpatient setting, which makes our findings comparable with those of Desai.
Among ob-gyns who reported seeing patients who needed or requested an abortion, the most commonly cited reasons for not providing medication abortion were personal reasons or practice restrictions. These results are similar to a survey of ob-gyns in New Mexico in 2008.11 Qualitative research has also highlighted how practice restrictions prevent trained physicians from providing the service.12
Beyond these personal and practice explanations, some of the reasons for not providing medication abortion could be addressed through training and policy changes. Eleven percent cited a lack of training, suggesting medication abortion teaching in residencies and continuing education might increase uptake of the method. A similar proportion reported that the requirement to stock mifepristone in their clinics was a reason they did not provide the method, and our findings suggest that the number of ob-gyns providing medication abortion might at least double if they could write a prescription for mifepristone. Pharmacy dispensing of Mifeprex by prescription is currently prohibited by the drug's Risk Evaluation and Mitigation Strategy imposed by the U.S. Food and Drug Administration.6 A recent analysis found that the mifepristone Risk Evaluation and Mitigation Strategy is not justified given the positive safety record of the drug, and the authors argued for its withdrawal.13 Our survey suggests that the Risk Evaluation and Mitigation Strategy is a barrier to provision of medication abortion, which should add new urgency to the push to remove this medically unnecessary restriction.
This study has several limitations. Although the response rate for this survey was 67% and nonresponders were similar to respondents, there is a risk of nonparticipation bias. Of note, participants did not know the survey focused on induced abortion practice when they were invited to participate. Still, it is likely that nonparticipation bias results in an overestimation of abortion provision in this survey. Our survey asked about abortion provision in the prior year, whereas the 2008–2009 survey asked “do you provide abortion services.” Although our wording is more precise, it may overestimate provision for physicians who recently stopped offering abortion. The hypothetical question about prescribing mifepristone may overestimate the effect of allowing pharmacy dispensing of the drug, because other barriers, such as practice restrictions and pharmacist refusals, may still limit expansion of medication abortion. In addition, Collaborative Ambulatory Research Network members may represent a subset of more engaged ACOG Fellows who might be more likely to change their practice if the policy regarding mifepristone dispensing were changed.
Quality of induced abortion care depends on all women seeking the service being able to do so in a timely fashion in a safe and effective way. State-based abortion regulations, federal laws restricting provision of medication abortion, individual providers' personal, religious, or moral beliefs, and practice and community factors affect the availability of abortion services and pose barriers to abortion quality. Expanding opportunities for professional development and reversing restrictive state and federal policies may help to improve the quality of abortion care.
1. National Academies of Sciences, Engineering, and Medicine. The safety and quality of abortion care in the United States. Washington, DC: The National Academies Press; 2018.
2. Stulberg DB, Dude AM, Dahlquist I, Curlin FA. Abortion provision among practicing obstetrician–gynecologists. Obstet Gynecol 2011;118:609–14.
3. Jones RK, Jerman J. Abortion incidence and service availability in the United States, 2014. Perspect Sex Reprod Health 2017;49:17–27.
4. Finer LB, Wei J. Effect of mifepristone on abortion access in the United States. Obstet Gynecol 2009;114:623–30.
5. Leddy MA, Lawrence H, Schulkin J. Obstetrician-gynecologists and women's mental health: findings of the Collaborative Ambulatory Research Network 2005–2009. Obstet Gynecol Surv 2011;66:316–23.
6. U.S. Food and Drug Administration. Risk evaluation and mitigation strategy for Mifeprex (mifepristone) tablets, 200 mg. Available at: https://www.accessdata.fda.gov/drugsatfda_docs/rems/Mifeprex_2016-03-29_REMS_full.pdf
. Retrieved July 4, 2018.
7. Guttmacher Institute. An overview of abortion laws. Available at: https://www.guttmacher.org/state-policy/explore/overview-abortion-laws
. Retrieved July 6, 2018.
8. Desai S, Jones RK, Castle K. Estimating abortion provision and abortion referrals among United States obstetrician-gynecologists in private practice. Contraception 2018;97:297–302.
9. Steinauer J, Landy U, Filippone H, Laube D, Darney PD, Jackson RA. Predictors of abortion provision among practicing obstetrician-gynecologists: a national survey. Am J Obstet Gynecol 2008;198:39.e1–6.
10. White K, Carroll E, Grossman D. Complications from first-trimester aspiration abortion: a systematic review of the literature. Contraception 2015;92:422–38.
11. Espey E, Leeman L, Ogburn T, Skipper B, Eyman C, North M. Has mifepristone medical abortion expanded abortion access in New Mexico? A survey of OB-GYN and family medicine physicians. Contraception 2011;84:178–83.
12. Freedman L, Landy U, Darney P, Steinauer J. Obstacles to the integration of abortion into obstetrics and gynecology practice. Perspect Sex Reprod Health 2010;42:146–51.
13. Mifeprex REMS Study Group, Raymond EG, Blanchard K, Blumenthal PD, Cleland K, Foster AM, et al. Sixteen years of overregulation: time to unburden Mifeprex. N Engl J Med 2017;376:790–4.