State-level abortion regulations have increased since 2011,1 which make it difficult for women to access care in early pregnancy. In July 2013, Texas enacted an omnibus abortion law, House Bill 2. The law had four provisions: physicians providing abortion had to have admitting privileges at a hospital within 30 miles of the facility, the provision of medication abortion had to follow the U.S. Food and Drug Administration–approved label for mifepristone, most abortions after 20 weeks postfertilization were prohibited, and all abortion facilities had to meet ambulatory surgical center standards. The first three provisions went into effect by November 1, 2013. The ambulatory surgical center provision was enforced briefly in October 2014 before the U.S. Supreme Court temporarily blocked enforcement.
The number of Texas facilities providing abortion declined from 41 to 19 after the law's passage and implementation.2 The total number of abortions decreased by 17% from 2012 to 2014, and the decrease was greater in counties where women needed to travel farther to an open facility.3 State vital statistics from 2014 documented an increase in second-trimester abortion compared with prior years.4 This is concerning because later abortion, although safe, is associated with a higher risk of complications; it is also more expensive, and fewer facilities offer the service.5–7
We calculated the change in second-trimester abortion after implementation of House Bill 2. Facility closures and health care provider shortages may have contributed to increases in second-trimester abortion; therefore, we also examined whether indicators of limited access to services accounted for changes observed.
This study draws on three data sources: annual vital statistics data, mystery client calls to abortion facilities, and county-level population estimates from the American Community Survey (https://www.census.gov/programs-surveys/acs/). We obtained individual-level vital statistics data on abortions performed in Texas each year between 2011 and 2014 from the Texas Department of State Health Services. These annual data included information on a woman's age, race or ethnicity, number of previous live births, county of residence, date of abortion, gestational age at the time of abortion, and type of abortion. Information on the location where a woman obtained her abortion was not provided.
Since 2012, we have tracked the number and location of Texas facilities providing abortions and dates clinics closed by collecting and confirming information through interviews with clinic staff, requests to the Department of State Health Services on licensed facilities, and reports in the press. Beginning in January 2014, we also started conducting monthly mystery client calls to all facilities that were open in July 2013, when the Texas legislature passed House Bill 2, or that opened subsequently. The purpose of the calls was to determine whether the facility was providing abortions and the number of days until the next available appointment for the consultation and ultrasound visit, which was required at least 24 hours before an abortion after implementation of a 2011 Texas law. Female research assistants posing as women seeking information about abortion services called facilities during regular business hours in the middle of each month. They made up to four calls the same day to reach someone at the facility; if no one answered, the research assistant called twice the next day. We considered the facility closed if clinic staff did not answer the phone on two consecutive days, the call was rerouted to another facility, the phone number was no longer in service, or if there was a voicemail message that confirmed the clinic was closed. The research assistants entered onto a standardized form information about whether the clinic was open or closed and, if open, days until the next available consultation appointment. The institutional review boards at the authors' universities approved all components of the study, and the Texas Department of State Health Services approved the use of vital statistics data.
Our primary outcome was the occurrence of abortions in the second trimester of pregnancy, which we defined as any procedure occurring at or after 12 weeks from last menstrual period (LMP), given that cervical preparation is often used at this gestational age.8 Before 2014, facilities reported gestational age by weeks from LMP. In 2014, Texas required facilities to begin reporting abortions based on weeks postfertilization, the measurement used for dating pregnancies in House Bill 2. To account for this change in reporting, we estimated gestational age at the time of abortion by adding 2 weeks to postfertilization age for all abortions taking place in 2014.
We assessed the service environment by focusing on three indicators of access: accessibility of services (measured as distance to facilities), availability of services (size of the facility network in a region), and ability of open facilities to accommodate women seeking care (number of days to appointments, or timeliness of care).9
To measure the one-way distance women may have traveled for care, we used the population-weighted centroid of women's county of residence as the starting location.10 We did not know the location where women obtained care because this information was not included in the vital statistics data we received. Based on a 2014 survey of abortion patients,2 we found that 73% of women seeking care at Texas abortion facilities traveled to their nearest open facility or obtained care within 10 miles of that facility. Therefore, in these analyses, we assumed that women traveled to the nearest open Texas facility. We estimated the distance to women's nearest facility by computing the driving distance between their starting location and the location of all Texas facilities that were open on the date of their abortion, using Open Source Routing Machine 4.911 and our records of the date of facility closures, and identifying the smallest value. Based on a visual inspection of a LOWESS plot, we categorized one-way distance as less than 10 miles, 10 to 49 miles, 50 to 99 miles, and 100 or more miles. We differentiated women living 100 or more miles from a facility (vs less than 100 miles), because the mandatory waiting period between the consultation and procedure for these women is reduced from 24 to 2 hours.
To estimate the size of the facility network in a woman's region of residence, we divided the number of open facilities in each of Texas' eight health service regions by the region's population of women aged 15–44 years, using 5-year estimates from the 2011 to 2015 American Community Survey. We then multiplied this value by 250,000 women to create a measure that would be comparable across regions and would account for differences in the effect of the number of facilities that closed and the population size of reproductive-aged women in potential need of services. We recalculated the relative size of the network in a region at each time point that a facility closed (or opened). By linking a woman's county of residence to the health service region, we determined the size of the facility network in the area on the date of her abortion. Based on a visual inspection of a LOWESS plot and distribution of observations in the prepolicy and postpolicy periods, we categorized the regional network size as less than one facility per 250,000 reproductive-aged women, 1–1.4 facilities, and 1.5 or more facilities.
We estimated the timeliness with which women were likely to receive care using data from the mystery client calls on the number of days until the next available consultation visit (ie, wait time). First, we computed the average wait time each month in a metropolitan area by summing the number of days until the next available consultation at all facilities in the area and dividing that value by the total number of open facilities in the area that month. Because we did not begin the mystery calls until January 2014, we used open facilities' January data to approximate wait times in November and December 2013. Mystery client calls were not conducted in May 2014; therefore, we estimated the May wait time using the average number of days to schedule a consultation visit in April and June 2014. To minimize sampling variation, we created a moving average for each metropolitan area using 3 months of data that included 1 month on either side of the index month. For example, the average wait time in a metropolitan area in February 2014 was estimated using the number of days until the next available consultation visit reported in January, February, and March. Finally, we used quartiles to categorize the expected number of days until the next consultation visit in a woman's nearest metropolitan area in the month of her abortion: less than 3 days, 3–4 days, 5–9 days, 10 days or more.
We compared changes that occurred in the year after House Bill 2 was implemented (November 1, 2013–October 31, 2014) with a matched 12-month period before the law was introduced (November 1, 2011–October 31, 2012). We chose this timeframe for the comparison period because any changes that occurred in the service environment were unlikely due to the anticipated effects of abortion regulations proposed in the 2013 legislative session. We compared the mean gestational age at abortion in each period, using t-tests, and computed the distribution of abortions according to procedure type and gestational age (medication at less than 12 weeks of gestation, surgical at less than 12 weeks, any procedure at 12–15 weeks, any procedure at 16 weeks or more). We also computed the total number of abortions and the percentage of abortions performed at 12 weeks of gestation or more, overall and by women's age (younger than 18 years, 18–24, 25–29, 30 or older), race or ethnicity (Hispanic, white, black, Asian, other or unknown), number of previous live births (0, 1–2, 3 or more), one-way distance to the nearest Texas facility, and facility network size in a woman's region of residence. In each period, we assessed demographic differences in second-trimester abortion using χ2 tests.
To assess the association between receiving abortion care after implementation of House Bill 2 and obtaining an abortion at 12 weeks of gestation or more, we used mixed-effects logistic regression, which enabled us to account for the clustering of observations within counties. We estimated three nested models to examine the change in the association between the policy periods and second-trimester abortion. Model 1 adjusted for women's demographic characteristics (ie, age, race or ethnicity, number of live births) that have been associated with obtaining abortion care later in pregnancy in other studies.12,13 Model 2 included additional adjustments for one-way distance to the nearest open Texas facility at the time of a woman's abortion. Model 3 adjusted for the variables in Model 2 and the facility network size in a woman's region of residence at the time of her abortion. We conducted mediation analyses to determine the degree to which demographic characteristics, distance, and facility network size explained the change in rates of second-trimester abortion associated with the postpolicy compared with prepolicy period.14,15 Specifically, we calculated the percent change in the beta coefficients between nested models [eg, (βModel1 – βUnadjusted Model)/βUnadjusted Model]. We estimated nonparametric 95% CIs for the percent change between estimates using a 1,000-iteration bootstrap, sampling with replacement at the county level.
For women who obtained abortions after House Bill 2, we also assessed the relationship between the expected timeliness of care and second-trimester abortion. We first plotted the monthly average number of days until the next available consultation visit for each of the metropolitan areas with at least one facility that remained open for 6 or more months after House Bill 2 to facilitate a visual comparison of variation in wait times. Then we estimated a mixed-effects logistic regression model for second-trimester abortion that included the number of days until the next available consultation visit, one-way distance to the nearest open Texas facility, and facility network size, as well as women's demographic characteristics. All analyses were conducted using Stata 15.1.
Of the 123,429 abortions included in the two 12-month periods before and after implementation of House Bill 2, we excluded records for non-Texas residents and that were missing information on women's county of residence (n=5,048), gestational age (n=31), or other demographic characteristics (n=274). This resulted in a final sample of 118,076 abortions. The sample included 64,902 Texas-resident abortions before House Bill 2 was introduced (November 2011–October 2012) and 53,174 abortions after implementation of the law (November 2013–October 2014).
The mean gestational age at abortion increased from 7.3 weeks to 8.3 weeks between the two periods (P<.001). Overall, 6,813 (10.5%) abortions occurred at 12 weeks of gestation or more before House Bill 2 and 7,720 (14.5%) abortions occurred at 12 weeks of gestation or more after implementation of the law (P<.001; Fig. 1). Additionally, the number of first-trimester medication abortions decreased from 17,739 (27.3% of all procedures) to 4,593 (8.6%). The number of procedures at 22 weeks of gestation or more also decreased from 220 (0.3%) before House Bill 2 was introduced to 39 (0.1%) after the law's implementation (not shown).
In both policy periods, women who were younger than 18 years of age, black, lived further from an open Texas facility at the time of their abortion, and lived in a region with a smaller facility network were more likely to have a second-trimester abortion (all P<.01; Table 1). Additionally, more women who obtained abortions lived less than 10 miles from the nearest facility before implementation of House Bill 2 than after (78.2% vs 61.6%), and the percentage living 100 or more miles from the nearest facility more than doubled between periods (from 2.9% to 7.1%). The percentage of women who obtained abortions living in a region with 1.5 or more facilities per 250,000 reproductive-aged women also decreased from 66.9% before House Bill 2 to 12.2% after the law, and the percentage of women who lived in a region with less than one facility per 250,000 reproductive-aged women increased (2% vs 40%).
The 21 facilities that closed for 6 months or more after passage of House Bill 2 were distributed throughout the state (Fig. 2). Regions with less than one facility per 250,000 reproductive-aged women after House Bill 2 included less populated areas, such as the Panhandle (Lubbock), South (McAllen and Harlingen), and East Texas, as well as North Central Texas, where there were numerous facility closures in the Dallas–Fort Worth metropolitan area.
In the unadjusted mixed-effects logistic regression model, the odds ratio (OR) for obtaining an abortion at 12 weeks of gestation or more after implementation of House Bill 2 compared with the period before passage of the law was 1.45 (95% CI 1.40–1.50). After adjusting for women's demographic characteristics only (Table 2, Model 1), the OR for second-trimester abortion after implementation of House Bill 2 was 1.44 (95% CI 1.39–1.50). In the model that also included one-way distance to the nearest open Texas facility at the time of a woman's abortion, the OR for period after House Bill 2 was 1.40 (95% CI 1.34–1.45); the OR was 1.17 (95% CI 1.10–1.25) after further adjustment for facility network size. In the fully adjusted model (Table 2, Model 3), women who lived 50–99 miles from the nearest open Texas facility had higher odds of second-trimester abortion than those whose nearest facility was less than 10 miles away; living 100 or more miles away was not significantly associated with second-trimester abortion. Compared with women living in a region with 1.5 or more facilities per 250,000 reproductive-aged women, those in regions with 1–1.4 open facilities and with less than one open facility had higher odds of second-trimester abortion.
Demographics did not account for the higher odds of obtaining an abortion at 12 weeks of gestation or more after compared with before implementation of House Bill 2 (percent attenuation: −1.9%; 95% CI −4.5% to 0.1%). Distance and facility network size accounted for 9.4% (95% CI −28.8% to −2.9%) and 57.7% (95% CI −88.8% to −22.8%) of the increased odds of obtaining an abortion at 12 weeks of gestation or more after House Bill 2, respectively.
As documented in the mystery client calls, the next consultation visit was available the same day at some open facilities but could be as long as 26 days from the date of the call during the 12-month period after implementation of House Bill 2 (not shown). The average number of days until the next consultation visit varied over time and across the five metropolitan areas with at least one facility that remained open for 6 or more months (Fig. 3). Facilities in Dallas–Fort Worth had the longest average wait times; they were unable to schedule women for consultations for at least 13 days between February and April 2014. During these months, the average wait times in Austin and San Antonio ranged from 9 to 11 days and then decreased beginning in May 2014. Although wait times fluctuated over the 12-month period at facilities in Houston and El Paso, women could consistently schedule their consultation visits in 4 days or less, on average.
In the multivariable-adjusted regression model among Texas residents who obtained abortion care after implementation of House Bill 2, the number of days until the next available consultation visit was significantly associated with having an abortion at 12 weeks of gestation or more (Table 3). Compared with women whose nearest city had an average wait time less than 3 days during the month of their abortion, those whose nearest metropolitan area had wait times between 5 and 9 days and 10 days or more had higher odds of having a second-trimester abortion. The associations between one-way distance to the nearest Texas facility, facility network size in a woman's region of residence, and having a second-trimester abortion were similar to those from the analyses that included women obtaining care before and after implementation of House Bill 2. However, the OR for women whose nearest facility was 50–99 miles was not statistically significant.
The number of abortions occurring at 12 weeks of gestation or more increased by 13% in the year after implementation of Texas House Bill 2 compared with the 12-month period before the law's introduction and passage, despite an overall decrease in the total number of abortions. Although the decrease in abortions at 22 weeks of gestation or more was likely due to the law's provision prohibiting abortions after this point in pregnancy, our analyses indicate that changes in facility network size after clinic closures explained a substantial portion of the overall increase in second-trimester abortion. Difficulties finding a health care provider have been associated with obtaining abortion care later in pregnancy.16 Indeed, as reported in other studies from Texas,17–19 women struggled to find up-to-date information about services online or did not receive an accurate referral from health care providers in the initial months after House Bill 2; some had their consultation or procedure appointments canceled, which contributed to delays reaching another open facility.
We also observed that women living farther from an open Texas facility were more likely to obtain abortions at 12 weeks of gestation or more. As documented in several studies,18–20 women traveling longer distances for care face difficulties arranging transportation, take time off work, possibly have to stay overnight and find childcare, and cover the increased cost of these expenses, which may further delay their care. However, unlike prior research from Texas demonstrating a linear trend between increasing distance to a facility and a decreasing number of abortions after House Bill 2,3 there was not a similar increase in second-trimester abortion as distance to a facility increased. Specifically, women living 100 or more miles from a facility were not more likely to have a second-trimester abortion than those living near a facility. This result may be attributable to these women making fewer visits to a facility because the mandatory waiting period between the consultation and abortion was reduced from 24 to 2 hours. Furthermore, although distance attenuated the association between policy period and second-trimester abortion, it had a smaller effect than other variables, suggesting that other indicators of abortion access may play a more important role in women's ability to obtain timely care.
Another novel aspect of our study is that we were able to document the timeliness with which women were likely able to make their initial consultation visit appointments and model the association between wait time and obtaining a second-trimester abortion. We observed considerable variation in wait times across metropolitan areas after implementation of House Bill 2—as well as within some cities, which was probably due to women's demand for services exceeding health care provider availability. This expands on previous research21 demonstrating an association between second-trimester abortion and appointment wait times by illustrating the role that facility-level delays have on women obtaining care later in pregnancy.
Although there is an increased need for second-trimester abortion services, Texas passed legislation in 2017 prohibiting physicians from performing dilation and evacuation, the most commonly used procedure in the second trimester. The law is enjoined as of this writing, but if it is allowed to go into effect, health care providers will have few evidence-based options for later abortion care. Given the limited number of U.S. health care providers offering care at 16 weeks of pregnancy or more, women, particularly those with fewer resources, may be forced to continue unwanted pregnancies.22,23
This study has several limitations. Our approach to account for the change in reporting gestational age may bias our estimates of second-trimester abortion if health care providers still reported weeks from LMP rather than postfertilization age. However, given the large decrease in the number of abortions at 22 weeks of gestation or more, it is unlikely that our findings are primarily due to a bias in our estimating approach. We also may have underestimated the distance women traveled and the number of days until the consultation visit, because other studies have reported that, after House Bill 2, women did not obtain care at the nearest facility or traveled out of state.2,17,19 Because we do not have information on women's preference for medication abortion or gestational age when they presented for care, we do not know whether the nearest facility could provide the services they needed. Finally, we were unable to account for some of the increase in second-trimester abortion, which may be related to changes in the cost of care or further delays between the consultation and procedure visit.
Implementation of House Bill 2 adversely affected multiple dimensions of women's access to abortion care, which contributed to the observed increases in second-trimester abortion. States considering similar laws restricting access to abortion can expect clinically significant changes in women's health outcomes.
1. Nash E, Gold RB, Mohammed L, Ansari-Thomas Z, Cappello O. Policy trends in the states, 2017. Available at: https://www.guttmacher.org/article/2018/01/policy-trends-states-2017
. Retrieved June 9, 2018.
2. Gerdts C, Fuentes L, Grossman D, White K, Keefe-Oates B, Baum SE, et al. The impact of clinic closures on women obtaining abortion services after implementation of a restrictive law in Texas. Am J Public Health 2016;106:857–64.
3. Grossman D, White K, Hopkins K, Potter JE. Change in distance to nearest facility and abortion in Texas, 2012 to 2014. JAMA 2017;317:437–9.
4. Grossman D. The use of public health evidence in whole woman's health v Hellerstedt. JAMA Intern Med 2017;177:155–6.
5. Upadhyay UD, Desai S, Zlidar V, Weitz TA, Grossman D, Anderson P, et al. Incidence of emergency department visits and complications after abortion. Obstet Gynecol 2015;125:175–83.
6. Zane S, Creanga AA, Berg CJ, Pazol K, Suchdev DB, Jamieson DJ, et al. Abortion-related mortality in the United States 1998–2010. Obstet Gynecol 2015;126:258–65.
7. Jones RK, Ingerick M, Jerman J. Differences in abortion service delivery in hostile, middle-ground and supportive states in 2014. Womens Health Issues 2018;28:212–18.
8. Allen RH, Goldberg AB. Cervical dilation before first-trimester surgical abortion (<14 weeks' gestation). Contraception 2016;93:277-91.
9. Penchansky R, Thomas JW. The concept of access: definition and relationship to consumer satisfaction. Med Care 1981;19:127–40.
10. United States Census Bureau. Centers of population. Available at: https://www.census.gov/geo/reference/centersofpop.html
. Retrieved June 9, 2018.
11. Huber S, Rust C. Calculate travel time and distance with OpenStreetMap data using the Open Source Routing Machine (OSRM). Stata J;16:416–23.
12. Finer LB, Frohwirth LF, Dauphinee LA, Singh S, Moore AM. Timing of steps and reasons for delays in obtaining abortions in the United States. Contraception 2006;74:334–44.
13. Colman S, Joyce T. Regulating abortion: impact on patients and providers in Texas. J Pol Anal Manage 2011;30:775–97.
14. Hayes A. Beyond Baron and Kenny: statistical mediation analysis in the new millennium. Commun Monogr 2009;76:408–20.
15. Preacher K, Hayes A. Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behav Res Methods 2008;40:879–91.
16. Drey EA, Foster DG, Jackson RA, Lee SJ, Cardenas LH, Darney PD. Risk factors associated with presenting for abortion in the second trimester. Obstet Gynecol 2006;107:128–35.
17. Fuentes L, Lebenkoff S, White K, Gerdts C, Hopkins K, Potter JE, et al. Women's experiences seeking abortion care shortly after the closure of clinics due to a restrictive law in Texas. Contraception 2016;93:292–7.
18. Baum SE, White K, Hopkins K, Potter JE, Grossman D. Women's experience obtaining abortion care in Texas after implementation of restrictive abortion laws: a qualitative study. PLoS One 2016;11:e0165048.
19. Jerman J, Frohwirth LF, Kavanaugh ML, Blades N. Barriers to abortion care and their consequences for patients traveling for services: qualitative findings from two states. Perspect Sex Reprod Health 2017;49:95–102.
20. White K, deMartelly V, Grossman D, Turan JM. Experiences accessing abortion care in Alabama among women traveling for services. Womens Health Issues 2016;26:298–304.
21. Jones RK, Jerman J. Time to appointment and delays in accessing care among US abortion patients. Available at: https://www.guttmacher.org/report/delays-in-accessing-care-among-us-abortion-patients
. Retrieved June 9, 2018.
22. Jones RK, Jerman J. Abortion incidence and service availability in the United States, 2014. Perspect Sex Reprod Health 2017;49:17–27.
23. Upadhyay UD, Weitz TA, Jones RK, Barar R, Foster DG. Denial of abortion because of provider gestational age limits in the United States. Am J Public Health 2014;104:1687–94.