Background Women seeking abortion services need to access services in a timely fashion. Quick and appropriate referrals to abortion providers are critical to this process.
Methods The objective of this study was to determine the quality and quantity of referrals for abortion services from reproductive health care facilities that do not provide abortion services. USA states were ranked by restrictiveness of abortion, and a simulated patient made calls to the five most and six least restrictive states. Referrals were considered direct if the name or telephone number of a facility that provided abortion services was given; indirect when Planned Parenthood was suggested without additional details; and inappropriate if the referral did not provide abortion services.
Results Of 142 calls, 77 (52.4%) were made to least restrictive states and 62 (45.8%) were made to most restrictive states. Among all calls, even after prompting staff members for a referral, 45.8% resulted in a direct referral, 19.0% resulted in an indirect referral, 8.5% resulted in an inappropriate referral and 26.8% resulted in no referral. Facilities in least restrictive states were significantly more likely to provide unprompted direct referrals (p=0.006) and significantly less likely to provide no referral (p<0.001) than facilities in most restrictive states, though these differences disappeared after prompting the staff member to provide a referral.
Conclusions A simulated patient received a direct referral for abortion services less than half the time, even after prompting a staff member to provide one. All facilities providing women's health care should have appropriate referrals readily available for patients seeking abortion services.
Statistics from Altmetric.com
Half of all pregnancies in the USA are unplanned, and of these nearly half end in induced abortion.1 Although one in three women will have an induced abortion at some point in her lifetime, access to abortion services remains low.2 In what has been deemed a neglected health disparity, minority and low-income women have disproportionately higher rates of abortion than non-minority women and women of higher socioeconomic status, yet have more limited access to abortion services.3 Overall, 87% of counties in the USA lack an abortion provider, and these counties are home to 35% of American women.4 Some states have as few as two providers, with up to 96% of women residing in a county without an abortion provider.4 While the incidence of abortion increased 1% between 2005 and 2008, from 19.4 to 19.6 abortions per 1000 women aged 15–44 years, the total number of abortion providers has not increased.4 In fact, in the USA between 1982 and 2000, the number of abortion providers decreased from a high of 2900 to 1800.5 ,6 This 39% decline in providers may be linked to fewer residency programmes offering meaningful training in elective abortion, despite a requirement made in 1995 by the Accreditation Council for Graduate Medical Education (ACGME) that routine training opportunities in abortion be offered to residents by all obstetrics and gynaecology (OB/GYN) residency programmes seeking accreditation.7 Although more recent data show abortion training to be positively correlated with providing abortion in future practice, regardless of one's intention to provide abortion prior to residency,8 the number of providers in 2008 is virtually unchanged from the number in 2000.
Several professional organisations recognise that abortion is a key component of women's health care and acknowledge the importance of including abortion in student curricula. In addition to the ACGME requirements, the Association of Professors of Gynecology and Obstetrics, which sets national standards for medical student education, lists abortion as a priority 1 core educational objective, meaning that students must know the objectives.9 Also, the Residency Review Committee in Family Medicine requires that all residents receive training to competently provide pregnancy options counselling for unintended pregnancies.10
The Institute of Medicine defines access to health services as “the timely use of personal health services to achieve the best possible health outcomes”.11 Healthy People 2020 identifies four components of access to care, all of which are important for abortion care: coverage, timeliness, services and workforce.12 Service means that people have a source of care. This is important for abortion because even if a woman has a regular source of primary care, the low number of abortion providers nationwide make it unlikely that her primary care provider (PCP) will provide abortion services, thus forcing her to find another source of care. Ideally, if her PCP does not provide abortion services, she will be given a referral to a facility that does provide abortion services. Timeliness is the system's ability to provide care quickly after a need is recognised. Abortion is a time-sensitive procedure, as both the costs and risks increase with gestational age.1 ,13 Appropriate referrals for women who are seeking abortion services are necessary to ensure that patients' needs are met quickly.
The American Congress of Obstetricians and Gynecologists (ACOG) states in a 2007 Committee Opinion entitled The Limits of Conscientious Refusal in Reproductive Medicine that “physicians and other health care providers have the duty to refer patients in a timely manner to other providers if they do not feel that they can in conscience provide the standard reproductive services that patients request”.14 A recent study by Rasinski et al.15 examined the opinions of 1800 OB/GYN physicians regarding these recommendations and found physician support for providing referrals. Using a vignette where an OB/GYN physician refused a requested induced abortion, the authors found that while 70% of respondents rated the vignette doctor as acting appropriately when a referral was made, only 12% believed the doctor acted appropriately when the doctor disclosed personal objections to abortion and refused to provide a referral.15
While the ACOG recommendations reinforce to physicians their responsibility to provide a referral to any patient who requests one, there is no explicit recommendation for frontline staff members. This is problematic because women who call a practice seeking abortion services will likely speak with a frontline staff member before seeing a clinician, and if the facility does not provide abortion services this staff member may be her only contact with the facility. The findings of Rasinski et al.15 show support for referrals among OB/GYN physicians, but even if a physician is willing to provide a referral to a woman who presents seeking abortion services, it is unknown whether this woman would receive a referral from the frontline staff member over the telephone at a facility that does not provide abortion services.
The primary aim of this study was to assess quality and quantity of referrals from a frontline staff member for a simulated patient seeking abortion services over the telephone at facilities that provide reproductive health care but not abortion services.
Selection of states
States were categorised as most restrictive or least restrictive based on the number of abortion laws and regulations present in those states as of February 2010.16 For convenience, we considered laws and regulations listed by the Guttmacher Institute in their monthly publication State Policies in Brief: An Overview of Abortion Laws. States were classified for the purpose of investigating regional differences in referrals for abortion, and the number of laws and regulations was chosen as a proxy for political climate, with the theory that states that passed more laws and regulations regarding abortion may be more opposed to abortion and thus less willing to provide referrals for abortion services. One point was given for each law or regulation and the states were ranked by points. These laws and regulations included: (1) abortions must be performed by a licensed physician; (2) abortions must be performed in a hospital in certain cases; (3) a second physician must participate in certain cases; (4) abortion is prohibited at some point with the exception of life or health; (5) individuals may refuse to participate in abortion services; (6) institutions may refuse to participate in abortion services; (7) mandatory waiting periods exist; (8) abortion for a minor requires parental consent; and (9) abortion for a minor requires parental notification. The six least restrictive states and the five most restrictive states were chosen in an attempt to balance the number of abortion providers in the two categories. All states classified as most restrictive had eight or nine points, while all states classified as least restrictive had zero or two points. All states that were not classified as most restrictive or least restrictive had between three and seven points. The phone calls were made from May 2010 through January 2011 and were conducted by a single investigator acting as a simulated patient. All of the study states remained in the least restrictive or most restrictive categories throughout the study period.
Selection of non-providers
The address of each abortion provider was obtained from the National Abortion Federation (NAF) website.17 Membership in NAF is voluntary, therefore not all abortion clinics are NAF members. NAF members consist of a mix of private practices, Title X clinics and hospitals. The practice types of the study facilities were not recorded. Google Maps was used to locate the five reproductive health care facilities that did not provide abortion services (non-providers) and were geographically closest to each abortion provider. This was done to ensure that each non-provider was located near a NAF-affiliated abortion facility to which a patient could potentially be referred. Non-provider facilities were excluded if they were on the NAF list of abortion providers, if they clearly did not provide reproductive health care (listings for law firms, medical organisations, etc.) or if they appeared to be private residences. One of the five closest clinics was then randomly selected by the investigator acting as the simulated patient. After a non-provider was chosen it was excluded from the lists for all subsequent matches. If new matches were required due to non-working telephone numbers, an additional non-provider was added to the list and a new match was randomly chosen. Like the providers, non-providing facilities consisted of private practices, Title X clinics and hospitals.
Assessment of referral
Each non-provider was contacted on three separate occasions at least 1 week apart. The simulated patient spoke with whoever answered the phone and began the call by asking the staff member if his or her office provided abortions. When the staff member said “No”, the simulated patient paused to allow the staff member time to offer a referral. If no referral were offered the simulated patient prompted the staff member by asking: “Do you know somewhere I could call?” The referral was considered to be direct if it resulted in the name or telephone number of a facility where the patient could obtain abortion services. If the referral was to a facility that was not on the list of NAF providers an investigator called the facility to confirm whether they provided abortion services. Referrals were considered to be indirect if the referral was to Planned Parenthood without any specific details about the clinic, such as telephone number or location, or to a national hotline. Inappropriate referrals were referrals to facilities that did not provide abortion services, which was confirmed by telephone by an investigator. Referrals to the Internet or the telephone book were considered to be no referral. The simulated patient also recorded a subjective measure of the respondent's attitude towards her on a scale of 1 to 10, with 1 being extremely hostile and 10 being extremely helpful. Calls that resulted in the simulated patient being hung up on were not rated.
It is known that some practices only provide abortion services to established patients and thus may provide better referrals than non-providers. Non-providers were considered to be patient-only facilities if the staff member asked the simulated patient during any of the calls if she were an established patient. In order to examine whether these practices influenced the overall quality of referrals, a subanalysis was performed excluding the patient-only facilities. For the purposes of the primary analysis patient-only facilities were classified as non-providers.
After all calls were completed, a debriefing letter was sent to all non-providers who appeared on the lists of the five geographically closest facilities. The letter informed the facility that they may have been contacted as part of the study and described the objectives and methods of the study. The letter did not include any study results and informed the facilities that all of their identifying information had been destroyed after the letters were addressed and mailed.
Random selection of non-providers and data analysis were conducted with SAS 9.2 (SAS Institute, Cary, NC, USA). Data are presented as proportions or medians with interquartile range (IQR). Comparisons between least restrictive and most restrictive states were made using Chi square (χ2), Fisher's exact and Mann-Whitney U tests. All tests were two sided, and p values <0.05 were considered to be statistically significant. The geodetic distance was calculated using the zipcitydistance function in SAS.
A total of 142 telephone calls were made to 48 non-provider facilities; 46 facilities were contacted three times and two were contacted twice. Of these, 77 (54.2%) calls were made to non-providers in least restrictive states and 65 (45.8%) calls were made to non-providers in most restrictive states. The non-provider facilities were located a median of 2.9 geodetic miles (IQR 0.0–4.7) from the abortion provider to which they were matched. All but one (97.9%) non-provider were located within 19.5 geodetic miles of a provider; one (2.1%) non-provider was located 44.5 geodetic miles from a provider. The median distance between the matched facilities was similar in most restrictive (3.1 miles) and least restrictive (2.6 miles) states (p=0.46).
Without a prompt from the simulated patient, only 28.2% of staff members provided the simulated patient with a direct referral, and more than half (54.9%) gave no referral (Table 1). The proportion of direct referrals increased to 45.8% and the proportion of no referrals decreased to 26.8% after the simulated patient prompted the staff member to provide a referral. Without a prompt, referral patterns in least restrictive states were significantly different from those in most restrictive states (p=0.002). Non-providers in least restrictive states were significantly more likely than non-providers in most restrictive states to provide direct referrals (p=0.006) compared to all other referral types, and significantly less likely to provide no referral (p<0.001) compared to all other referral types. This difference disappeared when the simulated patient prompted the staff member to provide a referral. However, even with a prompt, more than a quarter (26.8%) of calls resulted in no referral being given. During five (3.5%) calls, the staff member hung up on the simulated patient before she could ask for a referral. All five of these calls were to facilities in least restrictive states. Including prompted and unprompted referrals, 32.6% of non-providers gave a direct referral at all calls and 13.0% of non-providers gave no referral at all calls.
Twelve (8.5%) calls resulted in the simulated patient being given an inappropriate referral. In all 12 calls referrals were made to reproductive health care facilities that did not offer abortion services; none were to crisis pregnancy centres.
Six non-providers, accounting for 18 (12.7%) phone calls, asked the simulated patient if she were a patient at that facility. Four of the non-providers were in least restrictive states and two were in most restrictive states. Excluding these patient-only facilities did not substantially change the results.
The median subjective attitude of staff members was 7.0 (IQR 6.0–7.0) and did not differ significantly between most restrictive and least restrictive states (p=0.18). Scores of 5 and below were given to staff members who acted negatively towards the simulated patient; only three (2.1%) staff members received a score of 5.
Women seeking induced abortions in the USA face various barriers to access, one of which is lack of information about where to obtain an abortion. While an ACOG opinion states that all physicians have a duty to refer patients to an abortion provider if they are unable to provide the service, there is no similar recommendation for the frontline staff members who are often the sole contact for women seeking induced abortions. Additionally, all health care personnel in the USA, including frontline staff members, are protected by conscience protection statutes, which allow individuals to refuse to participate in abortions.18 Because these laws protect frontline staff members who refuse to provide referrals, they further restrict information for women seeking abortion services.
Overall, direct referrals for a simulated patient seeking abortion services was low, even after prompting a staff member to provide a referral. A previous study surveyed OB/GYN physicians about their opinions on abortion referral and found that 70% supported referrals,15 which is much higher than the proportion of direct referrals obtained by the simulated patient in the present study. This may reflect a difference in the willingness of physicians and frontline staff to provide referrals or simply a lack of knowledge of appropriate referral facilities among frontline staff.
Reproductive health facilities that do not provide abortions in most restrictive states are less likely than their counterparts in least restrictive states to offer immediate direct or indirect referrals. However, the combination of many laws and regulations with a low level of direct referrals creates a double burden for vulnerable patients seeking abortion services. It is possible that facilities that do not provide abortion services are unaware of facilities that do provide services, especially in rural areas. We attempted to take this into account by assuring that 97.9% of non-providers called were within 19.5 geodetic miles of an abortion provider listed on the NAF website. However, geodetic distances were measured from the centroids of zip codes, the geographic areas of which vary widely, and thus are only approximations of distance. Geodetic distances also underestimate the travel distance between two places. It is also possible that the overall attitudes towards induced abortion are more negative in most restrictive states, therefore staff members may be less willing to provide a referral unless directly prompted.
Staff members in both least restrictive and most restrictive states were generally polite and professional; only three (2.1%) calls were rated as ‘negative’ by the investigator acting as the simulated patient. However, five calls resulted in the staff member hanging up on the simulated patient before she could ask for a referral. These five calls were not given a subjective rating. Together, these calls would be most harmful for women who have difficulty advocating for themselves, particularly low-income and minority women who are already disproportionately at risk of needing an abortion.
A strength of this study is that we minimised the variation between calls by using a single caller posing as the simulated patient with a standardised script. A limitation is the possibility that a facility we classified as a non-provider might in fact provide abortion services, perhaps to current patients. We attempted to address this potential for bias by conducting a subanalysis in which we excluded patient-only facilities; however, this did not substantially change the results. No attempts were made to speak with different staff members at each contact; their names, if given, were not recorded. Although each call was treated independently, it is possible that the same staff member answered the phone each time the facility was contacted. We were concerned with simulating an actual patient experience and thus did not attempt to speak with different staff members.
The ethics of referral for abortion services remain professionally controversial. Chervenak and McCullough argued that in the case of a healthy woman with an unwanted pregnancy, simply providing the woman with referral information but not ensuring that the referral was made is all that is ethically required.19 Kaunitz et al. disagreed with this opinion, citing the authors' acknowledgement that this type of referral does not ensure that the patient's clinical needs will be met in a timely fashion.20 In the present study, direct and indirect referrals resulted in 65% of all calls, meaning that even with this lower standard of referral defined by Chervenak and McCullough, roughly one-third of the time the simulated patient did not receive care that is considered to be ethical.
While this study demonstrated low quality and quantity of referrals from frontline staff members, it is unclear to what extent this prevents women seeking abortion services from obtaining timely care. It is also unclear whether these findings reflect the referral practices of clinicians within these facilities. Future directions include the development of materials to improve referrals in all facilities that provide reproductive health care and evaluation of the effectiveness of these materials.
A simulated patient seeking abortion services from a facility that did not provide them received a direct referral less than half the time, even after prompting the staff member to provide one. As induced abortion is a common procedure that will be used by approximately one-third of American women at some point in their lifetime, all facilities that provide women's reproductive health care should have contact information for appropriate referrals readily available for women seeking abortion services.
The authors are grateful to Dr Milton Kotelchuck for his valuable assistance in the development of this project.
Funding This project was conducted with grant support from the Harvard University William F Milton Endowment.
Competing interests None.
Ethics approval Ethics approval was granted by the Committee on Clinical Investigations at Beth Israel Deaconess Medical Center, Boston, MA, USA.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.