Consideration of and Reasons for Not Obtaining Abortion Among Women Entering Prenatal Care in Southern Louisiana and Baltimore, Maryland

We recruited English and Spanish-speaking pregnant women 18 years and older at their first prenatal care visit at three prenatal care facilities in Southern Louisiana and one prenatal care facility in Baltimore, Maryland, between June 2015 and June 2017. Participants completed a self-administered iPad survey, followed by an in-clinic structured interview with a University of California, San Francisco research coordinator. Ethical approval was obtained from the University of California, San Francisco Institutional Review Board and the Louisiana State University Health Sciences Campus Institutional Review Board. The University of Maryland Institutional Review Board relied on the approval of the University of California, San Francisco Institutional Review Board.

We initially selected Southern Louisiana in order to be able to study experiences of women who might be unable to obtain an abortion if Louisiana’s Hospital Admitting Privileges Law (Guttmacher Institute, 2014) were to have gone into effect. The Admitting Privileges Law was signed into law, but did not go into effect during the study time period due to a series of injunctions (Glenza, 2014; McConnaughey, 2017). We also selected Louisiana because it had been classified as one of the most restrictive states with respect to abortion when the study began (Burns, Dennis, & Douglas-Durham, 2014). Louisiana’s multiple laws restricting abortion include mandatory waiting periods and ultrasounds, a parental involvement law, a 20-week ban, and lack of state-funding to pay for abortion for low-income women (Guttmacher Institute, 2018a). At the time we began the study in 2015, Louisiana had five abortion clinics (Guttmacher Institute, 2018a), three in the southern part of the state; by the time we finished recruitment, Louisiana had three abortion clinics, with two in the southern part of the state.

As the likelihood of Louisiana’s Hospital Admitting Privileges Law going into effect during our study period became less certain, we added a comparison state. The goal of adding a comparison state was to explore if experiences of women who consider, but do not have abortions, vary across states with different abortion policy and abortion service availability environments. We selected Baltimore, Maryland—an urban area with similar demographics in terms of race/ethnicity, poverty, and birth rate to our sites in Southern Louisiana—but with very different abortion policy and access. Maryland has one law restricting abortion, a parental involvement law (Guttmacher Institute, 2018b), and importantly, Maryland uses state Medicaid funds to cover abortion (Guttmacher Institute, 2015). Prior to the start of data collection, there were 25 abortion clinics in the state of Maryland (Guttmacher Institute, 2018b) and one of these clinics closed during the study time period (Jacob, 2017). During the entire study time period, there were six abortion clinics in or directly adjacent to Baltimore, MD (ANSIRH, 2017).

The prenatal care facilities where we recruited participants were affiliated with local universities and served primarily low-income pregnant women, many of whom were eligible for Medicaid insurance for their prenatal care.

In each recruitment facility, an onsite research coordinator approached all women over 18 who spoke English and presented for their first prenatal care appointment at one of the participating sites during the study time period. Midway through the first year of recruitment, we began recruiting women who spoke Spanish. Women who were younger than 18, were not pregnant, were there for a subsequent prenatal care visit, did not speak and read English or Spanish, or were incarcerated were ineligible. The onsite research coordinator recruited all potentially eligible participants, screened them for eligibility, obtained written informed consent, and enrolled them in the study. The research coordinator then instructed participants on how to complete the self-administered iPad survey and left them to complete it independently. The research coordinator then conducted a 5 to 15-min in-clinic structured interview with participants. Participants received a $30 gift card.

Based upon our initial study design, we calculated a sample size to be able to compare the proportion of pregnancies that were unintended from before to after Louisiana’s Hospital Admitting Privileges Law, if that law had gone into effect, hypothesizing that the law would impact that proportion. Following the recognition that the law was unlikely to go into effect and recognizing our need for a comparison site, we estimated our needed sample based on preliminary estimates of the proportion of Louisiana participants who reported that they had called an abortion clinic (~ 16%). We hypothesized that the proportion would be lower in MD, and estimated that we would need 269 participants in LA and 296 participants in MD to detect an 8% difference between the two states (Fig. 1).

Open-ended responses to in-clinic structured interview questions about why the participant did not obtain an abortion were coded by two coders (SR & JH) separately, and discrepancies were resolved through consensus. The responses were classified into a-priori categories informed by the social ecological model (Sallis & Owen, 2002) of personal reason, interpersonal reason, healthcare/other organization interaction, and policy-related reason. Responses could fall into more than one category.

Analyses were primarily descriptive. To assess for potential confounding by different participant characteristics across states, we first conducted comparisons of participant characteristics across states using t tests for continuous and chi-square tests for dichotomous and categorical variables. We then compared outcomes across states, first using chi-square tests, then logistic regression with state as the main predictor, adjusting for participant characteristics that varied across states at a p < 0.10 level, as well as those that we had conceptual reasons to include in the model (i.e., age and relationship with the man involved in the pregnancy), and using clustered standard errors [using Stata’s vce(cluster) command] to account for non-independent observations within recruitment facility. We did not include education in the multivariable models as it was too closely related to other measures of socioeconomic status included in the model (employment, housing insecurity, and public assistance receipt). We did not use multivariable modeling for exploratory subgroup analyses that considered outcomes of preferring abortion now and self-induction among those who had considered abortion, due to small sample sizes.

We approached 386 women in Southern Louisiana (LA) and 367 in Baltimore, Maryland (MD), which represented 97% and 100% of all potentially eligible women who presented for their first prenatal care appointment during study recruitment. Of those approached, 14% in LA and 4% in MD were ineligible due to age, not speaking English or Spanish, or having a nonviable pregnancy. Of those who were eligible, 86% in each state (285/331 in LA and 304/352 in MD) consented to participate. Of those who consented, 99% (n = 586) initiated the iPad survey, 98% (n = 579) completed the iPad survey, and 95% (n = 559) completed both the iPad survey and the in-clinic interview. Reporting having considered abortion in the iPad survey was not associated with completing either the iPad survey or in-clinic interview.

Across both states, most study participants reported their race as Black, had a high school education or attended some college, and received public assistance and/or Medicaid insurance (see Table 1). Most had been pregnant previously and had had one or more previous live births; a substantial minority reported a previous abortion and about 10% had been involved with Child Protective Services in the past. A substantial minority were food and/or housing insecure. The man involved in the pregnancy was most commonly a boyfriend or partner. Women were mostly ambivalent about their pregnancies, but were, on average, sure about their decision to continue the pregnancy. Substance use in the past year was common, with more than a third reporting binge drinking, more than a fourth reporting tobacco use, and about a fifth reporting illicit drug use. About one in seven reported anxiety or depression.

Samples were mostly similar across states, with a few exceptions (see Table 1). There were more Black women, more women completing high school or having some college, more previous births, and more previous abortions in Maryland. Women in Maryland were also more likely to report a history of depression, employment, public assistance receipt, and Medicaid insurance.

Overall, 28% (n = 74) of participants completing both the iPad survey and the in-clinic interview in Louisiana and 34% (n = 100) in Maryland reported having considered abortion in response to one or more of the questions in the iPad survey and in-clinic interview (see Fig. 2). There did not appear to be a mode difference in reporting, with 94% consistency in Louisiana and 93% in Maryland.

Regarding concrete actions taken toward obtaining an abortion, overall, 17% (n = 46) of Louisiana participants and 21% (n = 62) of Maryland participants completing both the iPad survey and the in-clinic interview said they considered abortion, but did not call an abortion clinic. Ten percent (n = 28) of Louisiana participants and 13% (n = 38) of Maryland participants called an abortion clinic. Two percent (n = 6) of Louisiana participants and 3 % (n = 10) of Maryland participants went to an abortion clinic appointment. (For the Louisiana participants, it was for the first of the two required visits, often called the abortion information visit; no participants in Louisiana went to the second visit, i.e., the abortion appointment.) While the overall distribution differed, there was no statistically significant difference in either the unadjusted or adjusted models across states in the proportion who called a clinic or in the proportion who went to an abortion clinic (see Table 2).

Thirteen percent (n = 38) of those who initiated the iPad survey reported preferring abortion for this pregnancy at the time of pregnancy discovery in Louisiana and 17% (n = 52) in Maryland (not statistically significant) (see Table 2). In the full sample, 2% (n = 5) of Louisiana participants and 1% (n = 3) of Maryland participants reported preferring abortion now, i.e., upon prenatal care entry (not statistically significant). Restricting the sample to those who reported preferring abortion upon pregnancy discovery, 11% (n = 4/38) of Louisiana and 2% (n = 1/52) of Maryland participants still preferred abortion now (p = 0.078 in unadjusted model) (see Table 2); one Louisiana participant and two Maryland participants had not preferred abortion initially, but preferred it now.

Three percent (n = 9) of Louisiana participants who initiated the iPad survey and 1% (n = 3) of Maryland participants reported having attempted to self-induce an abortion (p < 0.001 in adjusted model) (see Table 2). When limited to those who reported considering abortion on the iPad survey, this was 11% (n = 7/66) in Louisiana and 3% (n = 3/96) in Maryland (p = 0.052 in unadjusted model). Two participants in Louisiana reported attempting self-induction and did not report having considered abortion. Two participants in each state reported attempting self-induction and that they considered abortion, but did not report preferring abortion upon discovering pregnancy. In Louisiana, of the nine participants who reported attempting self-induction, 0 reported using misoprostol, three other herbs, three hit self in abdomen, and three other/not stated. In Maryland, of the three participants who reported attempting self-induction, 0 reported misoprostol, two other herbs, 0 hit self in abdomen, and one other/not stated.

Participants gave reasons for not obtaining an abortion that fell into all four of the a-priori categories of the social ecology model: personal, interpersonal, healthcare/other organization interaction, and policy-related. Personal reasons included those related to the participant’s preferences, including deciding that she wanted to have the baby and/or parent as well as deciding that they did not want to have an abortion (Table 3).

“Because I want a baby. I may not be ready, but you know whatever happens happens and it will all fall into place” – White, 22 yrs, 0 previous births, Louisiana

“Cause in my damned mind I decided that I wanted to keep them” – Black, 31 yrs, 2 previous births, Maryland

“I don’t believe in [abortion]” – Black, 26 yrs, 0 previous births, Maryland

“I wouldn’t be able to live with myself” – Other/multiracial, 30 yrs, 2 previous births, Maryland

“I just don’t see myself doing that again. It’s not a good experience” – Black, 25 yrs, 2 previous births, Maryland

Interpersonal reasons included instances of others swaying them as well as instances of women making joint decisions with another person, most typically the man involved in the pregnancy. They were typically based on actual interactions, but sometimes also included a perception of what others would think. Like personal reasons, they also included reasons related to affirmatively deciding to have the baby and/or parent as well as those related to deciding to not have an abortion.

“My kids, they want another sibling. They were excited to hear the news, even if I wasn’t as excited.” – Other/multiracial, 37 yrs, 2 previous births, Louisiana

“My aunt, I asked her about it and she begged me not to do it. She told me to send my baby to her. I felt bad. She begged me and if she begged me, then I felt like this child was important. You shouldn’t abort your children. I said, well, I want my baby.” – Black, 22 yrs, 1 previous birth, Louisiana

“Me and my husband talked about it and decided even though it was unplanned God had a plan” – Black, 28 yrs, 1 previous birth, Maryland

“Cause my husband snatched the phone [when I was making the appointment for the abortion]. It wasn’t what he wanted” – Black, 25 yrs, 2 previous births, Maryland

Healthcare/other organizational reasons included interactions with a health care provider or other organization that swayed the participant away from having an abortion.

“Because I had so many C sections, [having an abortion] was a health risk” –Black, 29 yrs, 2 previous births, Maryland

“They told me that I would have to go to Shreveport for the pill and even after that ultrasound that they may decide that I can’t do the pill if they don’t like what they see…They also don’t…offer any kind of anesthesia or pain pills. The doctor also plays some pretty raunchy rap music while she does the procedures.” – White, 27 yrs, 0 previous births, Louisiana

“Because I wanted to be put to sleep and they couldn’t do that.” – Black, 24 yrs, 2 previous births, Maryland

Policy-related reasons women reported for not having an abortion included citing factors related to the policy and abortion service availability environment such as lack of money to pay for the abortion or insurance not covering abortion (due to legal restrictions on what Medicaid and private insurance plans can cover), running up against gestational limits (specified by law and law-related clinic policy), having to make—and pay for—two separate visits to the clinic (due to a two-visit legal requirement), and lack of local clinics (due to the complexity of opening and keeping open an abortion clinic in a complicated regulatory environment). One woman in Maryland reported that she had been unable to obtain an abortion earlier in her pregnancy because she was incarcerated and the jail/prison did not facilitate her abortion care. Policy-related reasons of difficulty paying for the abortion and exceeding gestational limits predominated in LA.

“…It was going to cost $2600.” – Black, 23 yrs, 0 previous births, Louisiana

“My insurance wouldn’t cover it” –Black, 26 yrs, 2 previous births, Maryland

“Because the lady said it was $125 [for the ultrasound to check my gestational age] and if I was over 18 weeks [and therefore couldn’t get the abortion], that it would be nonrefundable...Probably the only reason is because I was over the amount of days or weeks or whatever” – Black, 21 yrs, 1 previous birth, Louisiana

“I went to the counseling session and left because I didn’t want to stay 6 hours and [pay] $50.” – Black, 35 yrs, 2 previous births, Louisiana

Among all women entering prenatal care and completing both the iPad survey and the in-clinic interview, personal reasons were the most common reason for not having an abortion across both settings [19% (n = 52) Louisiana and 28% (n = 81) Maryland, p < 0.001 in adjusted models]. There was no difference in invoking of interpersonal reasons [6% (n = 15) Louisiana, 5% (n = 14) Maryland] or healthcare/other organization reasons [3% (n = 7) Louisiana, 3% (n = 9) Maryland] across sites. Policy reasons were more common in Louisiana than in Maryland [6% (n = 15) Louisiana, 1% (n = 2) Maryland, p < 0.001 in adjusted models]. Restricting the denominator to those who reported considering abortion, the overall pattern of findings is similar, although the difference for personal reasons between states is no longer statistically significant. The difference in rates of reporting a policy-related reason for not having an abortion, however, is stark: almost one-fourth (22%, n = 15) of women who reported considering abortion in the in-clinic interview in Louisiana reported a policy reason for not having an abortion, as compared to 2% (n = 2) in Maryland (p < 0.001 in adjusted model).

Although the open-ended format of the question allowed women to offer multiple reasons for not obtaining an abortion, of women considering abortion, most (75%, n = 118) reported reason(s) that fell into only one category. About one-quarter (24%, n = 37) offered reasons that fell into two categories and 1% (n = 2) offered reasons that fell into three categories. Offering reasons that fell into just one category was less common in Louisiana than Maryland [66%, 31%, and 3% (n = 43, 20, 2) in Louisiana vs. 81%, 19%, and 0% (n = 72, 17, 0) in Maryland]. Forty-nine percent (n = 32) of Louisiana participants and 73% (n = 65) of Maryland participants who reported considering abortion in the in-clinic interview reported only a personal reason(s), 9% (n = 6) of Louisiana and 1% (n = 1) of Maryland only a policy reason(s), and 8% (n = 5) of Louisiana and 0% (n = 0) of Maryland personal and policy reasons.

Conceptualizing the concrete actions women could take toward abortion as a sequential process (i.e., calling a clinic, going to a clinic appointment, having the abortion), personal reasons for discontinuing action to obtain an abortion were the most common across all concrete actions in both Louisiana and Maryland. The exception to this pattern is among women who went to the clinic in Louisiana, for whom the frequency of policy reasons and of personal reasons was similar. While women reported policy reasons for discontinuing the process of seeking an abortion at all stages of the process in Louisiana, policy reasons were most common among women who got as far as presenting for care at an abortion clinic. No woman who presented for care at a clinic in Maryland provided a policy reason for not having an abortion, and policy reasons, already rare in the Maryland sample, were typically only reported among Maryland women who called a clinic and made an appointment [results not shown].