Association of Travel Distance to Nearest Abortion Facility With Rates of Abortion | Health Disparities | JAMA Network Open | JAMA Network
[Skip to Navigation]
Sign In
Figure 1.  Reported Abortion Rate per 1000 Women of Reproductive Age, by County of Residence (2015)
Reported Abortion Rate per 1000 Women of Reproductive Age, by County of Residence (2015)

Includes female residents aged 15 to 44 years.

Figure 2.  Estimated Abortion Rate per 1000 Women of Reproductive Age, by County of Residence (2015)
Estimated Abortion Rate per 1000 Women of Reproductive Age, by County of Residence (2015)

Includes female residents aged 15 to 44 years.

Table 1.  Selected Sociodemographic Characteristics of Study Population by Whether Counties Report Residents’ Abortions (2015)
Selected Sociodemographic Characteristics of Study Population by Whether Counties Report Residents’ Abortions (2015)
Table 2.  Decline in County-Level Abortion Rate in a Spatial Poisson Model, by Travel Distance to the Nearest Abortion Care Facility (n = 1948)
Decline in County-Level Abortion Rate in a Spatial Poisson Model, by Travel Distance to the Nearest Abortion Care Facility (n = 1948)
Table 3.  Estimated Increases in Abortions and Abortion Rate Under Travel Distance Scenarios, by State (2015)a
Estimated Increases in Abortions and Abortion Rate Under Travel Distance Scenarios, by State (2015)a
1.
Onitilo  AA, Liang  H, Stankowski  RV,  et al.  Geographical and seasonal barriers to mammography services and breast cancer stage at diagnosis.   Rural Remote Health. 2014;14(3):2738.PubMedGoogle Scholar
2.
Engelman  KK, Hawley  DB, Gazaway  R, Mosier  MC, Ahluwalia  JS, Ellerbeck  EF.  Impact of geographic barriers on the utilization of mammograms by older rural women.   J Am Geriatr Soc. 2002;50(1):62-68. doi:10.1046/j.1532-5415.2002.50009.x PubMedGoogle ScholarCrossref
3.
Huang  B, Dignan  M, Han  D, Johnson  O.  Does distance matter? distance to mammography facilities and stage at diagnosis of breast cancer in Kentucky.   J Rural Health. 2009;25(4):366-371. doi:10.1111/j.1748-0361.2009.00245.x PubMedGoogle ScholarCrossref
4.
Gumpertz  ML, Pickle  LW, Miller  BA, Bell  BS.  Geographic patterns of advanced breast cancer in Los Angeles: associations with biological and sociodemographic factors (United States).   Cancer Causes Control. 2006;17(3):325-339. doi:10.1007/s10552-005-0513-1 PubMedGoogle ScholarCrossref
5.
Anderson  AE, Henry  KA, Samadder  NJ, Merrill  RM, Kinney  AY.  Rural vs urban residence affects risk-appropriate colorectal cancer screening.   Clin Gastroenterol Hepatol. 2013;11(5):526-533. doi:10.1016/j.cgh.2012.11.025 PubMedGoogle ScholarCrossref
6.
Shoff  C, Chen  VY, Yang  TC.  When homogeneity meets heterogeneity: the geographically weighted regression with spatial lag approach to prenatal care utilization.   Geospat Health. 2014;8(2):557-568. doi:10.4081/gh.2014.45 PubMedGoogle ScholarCrossref
7.
Kozhimannil  KB, Hung  P, Henning-Smith  C, Casey  MM, Prasad  S.  Association between loss of hospital-based obstetric services and birth outcomes in rural counties in the United States.   JAMA. 2018;319(12):1239-1247. doi:10.1001/jama.2018.1830 PubMedGoogle ScholarCrossref
8.
Jones  RK, Jerman  J.  Population group abortion rates and lifetime incidence of abortion: United States, 2008-2014.   Am J Public Health. 2017;107(12):1904-1909. doi:10.2105/AJPH.2017.304042 PubMedGoogle ScholarCrossref
9.
Policy Surveillance Program. State abortion laws. Updated March 1, 2021. Accessed March 23, 2021. https://lawatlas.org/datasets/abortion-laws
10.
Gerdts  C, Fuentes  L, Grossman  D,  et al.  Impact of clinic closures on women obtaining abortion services after implementation of a restrictive law in Texas.   Am J Public Health. 2016;106(5):857-864. doi:10.2105/AJPH.2016.303134 PubMedGoogle ScholarCrossref
11.
Venator  J, Fletcher  J.  Undue Burden Beyond Texas: An Analysis of Abortion Clinic Closures, Births and Abortions in Wisconsin. National Bureau of Economic Research; 2019. doi:10.3386/w26362
12.
Barr-Walker  J, Jayaweera  RT, Ramirez  AM, Gerdts  C.  Experiences of women who travel for abortion: a mixed methods systematic review.   PLoS One. 2019;14(4):e0209991. doi:10.1371/journal.pone.0209991 PubMedGoogle Scholar
13.
Upadhyay  UD, Johns  NE, Meckstroth  KR, Kerns  JL.  Distance traveled for an abortion and source of care after abortion.   Obstet Gynecol. 2017;130(3):616-624. doi:10.1097/AOG.0000000000002188 PubMedGoogle ScholarCrossref
14.
Khan  AA.  An integrated approach to measuring potential spatial access to health care services.   Socioecon Plann Sci. 1992;26(4):275-287. doi:10.1016/0038-0121(92)90004-O PubMedGoogle ScholarCrossref
15.
Johns  NE, Foster  DG, Upadhyay  UD.  Distance traveled for Medicaid-covered abortion care in California.   BMC Health Serv Res. 2017;17(1):287. doi:10.1186/s12913-017-2241-0 PubMedGoogle ScholarCrossref
16.
Grossman  D, White  K, Hopkins  K, Potter  JE.  Change in distance to nearest facility and abortion in Texas, 2012 to 2014.   JAMA. 2017;317(4):437-439. doi:10.1001/jama.2016.17026 PubMedGoogle ScholarCrossref
17.
Joyce  TJ, Tan  R, Zhang  Y.  Back to the Future? Abortion Before and After Roe. National Bureau of Economic Research; 2012. doi:10.3386/w18338
18.
Cunningham  S, Lindo  JM, Myers  C, Schlosser  A.  How Far Is too Far? New Evidence on Abortion Clinic Closures, Access and Abortions. National Bureau of Economic Research; 2017.
19.
Bearak  JM, Burke  KL, Jones  RK.  Disparities and change over time in distance women would need to travel to have an abortion in the USA: a spatial analysis.   Lancet Public Health. 2017;2(11):e493-e500. doi:10.1016/S2468-2667(17)30158-5 PubMedGoogle ScholarCrossref
20.
Cartwright  AF, Karunaratne  M, Barr-Walker  J, Johns  NE, Upadhyay  UD.  Identifying national availability of abortion care and distance from major US cities: systematic online search.   J Med Internet Res. 2018;20(5):e186. doi:10.2196/jmir.9717 PubMedGoogle Scholar
21.
Myers  C, Jones  R, Upadhyay  U.  Predicted changes in abortion access and incidence in a post-Roe world.   Contraception. 2019;100(5):367-373. doi:10.1016/j.contraception.2019.07.139 PubMedGoogle ScholarCrossref
22.
Brown  BP, Hebert  LE, Gilliam  M, Kaestner  R.  Distance to an abortion provider and its association with the abortion rate: a multistate longitudinal analysis.   Perspect Sex Reprod Health. 2020;52(4):227-234. doi:10.1363/psrh.12164 PubMedGoogle ScholarCrossref
23.
National Conference of State Legislatures. Insurance carriers and access to healthcare providers: network adequacy. Updated February 1, 2018. Accessed June 14, 2019. https://www.ncsl.org/research/health/insurance-carriers-and-access-to-healthcare-providers-network-adequacy.aspx
24.
Jatlaoui  TC, Boutot  ME, Mandel  MG,  et al.  Abortion surveillance—United States, 2015.   MMWR Surveill Summ. 2018;67(13):1-45. doi:10.15585/mmwr.ss6713a1 PubMedGoogle ScholarCrossref
25.
Huber  S, Rust  C.  osrmtime: Calculate travel time and distance with OpenStreetMap data using the Open Source Routing Machine (OSRM).   Stata J. 2016;16(2):416-423. doi:10.1177/1536867X1601600209 Google ScholarCrossref
26.
US Census Bureau. Centers of population. Updated October 11, 2019. Accessed May 7, 2018. https://www.census.gov/geographies/reference-files/time-series/geo/centers-population.html
27.
National Abortion Rights Action League. Who decides? the status of women’s reproductive rights in the United States. January 2016. Accessed August 31, 2018. https://www.prochoiceamerica.org/wp-content/uploads/2017/01/WHODecides-2016-Final.pdf
28.
Upadhyay  UD, McCook  AA, Bennett  AH, Cartwright  AF, Roberts  SCM.  State abortion policies and Medicaid coverage of abortion are associated with pregnancy outcomes among individuals seeking abortion recruited using Google Ads: a national cohort study.   Soc Sci Med. 2021;274:113747. doi:10.1016/j.socscimed.2021.113747 PubMedGoogle Scholar
29.
Wood  SN.  Generalized Additive Models: An Introduction With R. CRC Press; 2017. doi:10.1201/9781315370279
30.
Brown  BP, Hebert  LE, Gilliam  M, Kaestner  R.  Association of highly restrictive state abortion policies with abortion rates, 2000-2014.   JAMA Netw Open. 2020;3(11):e2024610. doi:10.1001/jamanetworkopen.2020.24610 PubMedGoogle Scholar
31.
Cohen  DS, Joffe  C.  Obstacle Course: The Everyday Struggle to Get an Abortion in America. University of California Press; 2020.
32.
Krieger  N, Chen  JT, Waterman  PD, Soobader  MJ, Subramanian  SV, Carson  R.  Geocoding and monitoring of US socioeconomic inequalities in mortality and cancer incidence: does the choice of area-based measure and geographic level matter? the Public Health Disparities Geocoding Project.   Am J Epidemiol. 2002;156(5):471-482. doi:10.1093/aje/kwf068 PubMedGoogle ScholarCrossref
33.
Fuentes  L, Jerman  J.  Distance traveled to obtain clinical abortion care in the United States and reasons for clinic choice.   J Womens Health (Larchmt). 2019;28(12):1623-1631. doi:10.1089/jwh.2018.7496 PubMedGoogle ScholarCrossref
34.
Upadhyay  UD.  Innovative models are needed for equitable abortion access in the USA.   Lancet Public Health. 2017;2(11):e484-e485. doi:10.1016/S2468-2667(17)30181-0 PubMedGoogle ScholarCrossref
35.
Centers for Disease Control and Prevention. 10 Essential public health services. Updated March 18, 2021. Accessed August 31, 2019. https://www.cdc.gov/publichealthgateway/publichealthservices/essentialhealthservices.html
36.
Roberts  SCM, Fuentes  L, Berglas  NF, Dennis  AJA.  A 21st-century public health approach to abortion.   Am J Public Health. 2017;107(12):1878-1882. doi:10.2105/AJPH.2017.304068 PubMedGoogle ScholarCrossref
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    Views 4,156
    Citations 0
    Original Investigation
    Obstetrics and Gynecology
    July 6, 2021

    Association of Travel Distance to Nearest Abortion Facility With Rates of Abortion

    Author Affiliations
    • 1Bixby Center for Global Reproductive Health, Department of Obstetrics, Gynecology and Reproductive Sciences, University of California, San Francisco
    • 2Department of Epidemiology and Biostatistics, University of California, San Francisco
    • 3Advancing New Standards in Reproductive Health, Bixby Center for Global Reproductive Health, Department of Obstetrics, Gynecology and Reproductive Sciences, University of California, San Francisco
    JAMA Netw Open. 2021;4(7):e2115530. doi:10.1001/jamanetworkopen.2021.15530
    Key Points

    Question  Is there an association between median travel distance to an abortion facility and abortion rate?

    Findings  In this cross-sectional geographic analysis of US counties, increases in median travel distance to the nearest abortion care facility were associated with significant reductions in median abortion rate (21.1 per 1000 female residents of reproductive age for <5 miles; 3.9 per 1000 female residents of reproductive age for ≥120 miles). Reductions in travel distances were associated with significant increases in the median abortion rate (telemedicine simulation, 10.2 per 1000 female residents of reproductive age).

    Meaning  In this study, the abortion rates declined as travel distance to an abortion care facility increased, and modeling suggests the need for abortion care can be only partially met through service delivery innovations.

    Abstract

    Importance  Travel distance to abortion services varies widely in the US. Some evidence shows travel distance affects use of abortion care, but there is no national analysis of how abortion rate changes with travel distance.

    Objective  To examine the association between travel distance to the nearest abortion care facility and the abortion rate and to model the effect of reduced travel distance.

    Design, Setting, and Participants  This cross-sectional geographic analysis used 2015 data on abortions by county of residence from 1948 counties in 27 states. Abortion rates were modeled using a spatial Poisson model adjusted for age, race/ethnicity, marital status, educational attainment, household poverty, nativity, and state abortion policies. Abortion rates for 3107 counties in the 48 contiguous states that were home to 62.5 million female residents of reproductive age (15-44 years) and changes under travel distance scenarios, including integration into primary care (<30 miles) and availability of telemedicine care (<5 miles), were estimated. Data were collected from April 2018 to October 2019 and analyzed from December 2019 to July 2020.

    Exposures  Median travel distance by car to the nearest abortion facility.

    Main Outcomes and Measures  US county abortion rate per 1000 female residents of reproductive age.

    Results  Among the 1948 counties included in the analysis, greater travel distances were associated with lower abortion rates in a dose-response manner. Compared with a median travel distance of less than 5 miles (median rate, 21.1 [range, 1.2-63.6] per 1000 female residents of reproductive age), distances of 5 to 15 miles (median rate, 12.2 [range, 0.5-23.4] per 1000 female residents of reproductive age; adjusted coefficient, −0.05 [95% CI, −0.07 to −0.03]) and 120 miles or more (median rate, 3.9 [range, 0-12.9] per 1000 female residents of reproductive age; coefficient, −0.73 [95% CI, −0.80 to −0.65]) were associated with lower rates. In a model of 3107 counties with 62.5 million female residents of reproductive age, 696 760 abortions were estimated (mean rate, 11.1 [range, 1.0-45.5] per 1000 female residents of reproductive age). If abortion were integrated into primary care, an additional 18 190 abortions (mean rate, 11.4 [range, 1.1-45.5] per 1000 female residents of reproductive age) were estimated. If telemedicine were widely available, an additional 70 920 abortions were estimated (mean rate, 12.3 [range, 1.4-45.5] per 1000 female residents of reproductive age).

    Conclusions and Relevance  These findings suggest that greater travel distances to abortion services are associated with lower abortion rates. The results indicate which geographic areas have insufficient access to abortion care. Modeling suggests that integrating abortion into primary care or making medication abortion care available by telemedicine may decrease unmet need.

    Introduction

    In the US, increasing travel distance or travel time to a health care clinician is associated with less use of preventive care and poorer health outcomes for women, including reduced use of mammography,1,2 later stage at diagnosis of breast cancer,3,4 and reduced use of risk-appropriate colonoscopy.5 County-level analyses of pregnancy-related outcomes have shown spatial relationships in rates of prenatal care use,6 and the closure of rural hospitals not adjacent to urban areas was associated with increased preterm births in the following year.7

    Abortion is a common reproductive health care service, with 1 in 4 US women obtaining this care during their lifetime.8 However, many states have implemented policies restricting abortion care clinicians and facilities (hereinafter referred to as abortion providers).9 Studies of these policies have documented clinic closures and women unable to obtain abortion care, with disproportionate effects on low-income women and non-White women.10,11 Increased travel for an abortion is associated with delays in care, increased costs, and stress.10,12 Even when women are able to obtain abortion care, greater travel distance has been associated with decreased odds of returning to the abortion facility for follow-up care and increased odds of visiting an emergency department.13

    Research in a variety of settings has indicated that the farther a woman lives from an abortion care facility, the less likely she is to obtain that care. These studies used distance or travel time to an abortion provider as a measure of potential rather than realized access.14 Regional research has focused on California, Texas, New York, and Wisconsin11,15-18; national analyses have focused on disparities in access.19-21 One longitudinal, econometric study in 18 states22 found an association between travel distance and abortion rate but did not generate interpretable abortion rates.

    We conducted a national analysis to test the hypothesis that greater travel distance to the nearest abortion facility is associated with lower abortion rates and to provide estimated abortion rates under actual conditions and alternate assumptions of abortion access. We extend the literature by estimating changes in abortion rate under 2 travel distance scenarios: less than 30 miles (48 km), a common definition of network adequacy for primary care,23 and less than 5 miles (8 km), a simulation of medication abortion by telemedicine.

    Methods
    Study Design

    This geographic analysis used publicly available data from 27 states, the American Community Survey, and the US Census to calculate county-level abortion rates per 1000 female residents of reproductive age (15-44 years) in 2015, the most recent year of county-level data available when the study began. We compare abortion rates across counties with varying median travel distances to the nearest abortion facility. We estimated the abortion rate for 48 states and estimated the effect of different travel distance scenarios on the abortion rate in a multivariable model. Data were collected from April 2018 to October 2019. The University of California, San Francisco, institutional review board approved this study and waived the need for informed consent for the use of publicly available data. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

    Study Population

    A county or an equivalent administrative unit is the smallest geographic unit for which abortions are reported and was the unit of analysis for this study. We limited our analyses to the 48 contiguous US states (n = 3107 counties) because modes of travel other than by road are common in Hawaii and Alaska. To develop the model for estimations, we used data from 27 states that publicly report abortion by county of residence (n = 1948 counties). When states’ reports disagreed with the Centers for Disease Control and Prevention abortion estimates by state of residence for 2015 by greater than 1%, we adjusted the number of abortions.24 Adjustments increased the abortion count by 4.8% overall, with the largest changes in states reporting fewer than 10 000 abortions. Seven states suppressed abortion counts for counties reporting fewer than a specific number of abortions (eg, <10). In these 267 counties, we replaced the suppressed count with the midpoint between 0 and the specified limit (eg, 5) when it did not create a county abortion rate exceeding the state mean; otherwise, the county abortion count was missing. To calculate county-level abortion rate per 1000 female residents aged 15 to 44 years, we used 2015 American Community Survey 5-year population estimates.

    Exposures and Outcomes

    The study outcomes were numbers of abortions and abortion rate per 1000 female residents of reproductive age by county of residence. The main exposure was the travel distance to the nearest abortion facility. We used the Advancing New Standards in Reproductive Health national abortion facility database and Open Source Routing Machine to determine travel distance and time by car.20,25 We included publicly advertised abortion facilities operating in 2015, as verified by telephone calls, local news reports, or a website active throughout the year. We identified latitude/longitude coordinates for each abortion facility (n = 789), calculated travel distance from the population-weighted center of each census tract (n = 72 539), selected the facility with the shortest travel distance, and calculated the median travel distance per county.26 We excluded census tracts with no residents and those with only institutionalized male populations. We used a categorical variable for travel distance because previous research has shown a nonlinear relationship between travel distance and abortion rates.11,15-18 To determine the cut points for the categorical variable, we used state legislatures’ common definitions of health care network adequacy.23

    For scenario testing, the main exposures were travel distance of less than 30 miles, a common definition of network adequacy for primary care, or less than 5 miles, simulating medication abortion by telemedicine. We considered the shortest travel distance in our model to simulate medication abortion by telemedicine because patients still spend time seeking and obtaining these services.

    Control Variables

    Analyses included covariables identified as associated with the abortion rate: age, race/ethnicity, marital status, educational attainment, household income, and nativity.1 Race/ethnicity data were based on self-report in the US Census. Other covariables were from the American Community Survey 5-year estimates for 2015. We used a 5-level state policy grade (A, B, C, D, or F) developed by the National Abortion Rights Action League for 2015 that captures “the cumulative burden each state imposes on access to reproductive health care.”27(p78) The grade has been used in previous studies on abortion policies28 and takes into account enacted laws and regulatory activity, such as whether advanced practice clinicians may provide early abortion care, whether state Medicaid covers abortion care, whether waiting periods or in-person counseling are mandated, and whether laws target abortion care facilities or ban some types of abortion.

    Statistical Analysis

    Data were analyzed from December 2019 to July 2020. To model abortion rates, we used a spatial Poisson generalized additive model. We modeled rates using numbers of abortions as the outcome and population of female residents of reproductive age as an offset term. We included a priori covariables as linear terms (county proportions of female residents aged 25-29 years; non-Hispanic Black, Asian, Native American, Alaska Native, Native Hawaiian, Pacific Islander, or multiracial female residents aged 15-44 years; married women older than 18 years; women aged 18-44 years with a high school degree or equivalent; households with female residents aged 15-44 years below the federal poverty level; and foreign-born female residents), except travel distance and state policy grade, which were categorical. To account for residual spatial autocorrelation, we fit a spatial smooth across counties using a Markov random field (MRF).29 To tune the parameters of the MRF, including the snap parameter defining the neighborhood matrix and the smoothing parameter, we used cross-validation. We created 27 cross-validation folds using counties grouped by state, each with a validation set of counties belonging to the same state. We took this approach because missing county data were mostly the result of entire states not reporting. We compared models using cross-validated mean squared error and residual spatial autocorrelation (Moran I statistic) and by examining estimated rates to identify models that fell outside a plausible maximum (>100 per 1000 female residents of reproductive age).

    To make estimations, including the travel distance scenarios, we first established which counties lay outside the spatial limits of the observed data. For those counties outside the convex hull of observed data, we made estimations based on the covariables alone. This process was based on experiments that showed that extrapolating the spatial effect can lead to implausible estimated rates. For counties within the convex hull of the observed data, the estimations use the full spatial model. As a sensitivity analysis, we used a categorical variable for travel time as the main exposure.

    We used R, version 3.5.3, and the mgcv package for modeling and estimation, version 1.8-27 (R Foundation for Statistical Computing). Statistical tests were 2 sided and used P < .05 to indicate statistical significance.

    Results

    Counties that reported (n = 1948) vs those that did not report (n = 1160) residents’ abortions had similar mean numbers of female residents aged 25 to 29 years (15.6% vs 15.6%), high school–educated women aged 18 to 44 years (27.2% vs 27.8%), foreign-born female residents (4.5% vs 4.4%), and female residents of reproductive age living in households below the federal poverty level (22.1% vs 22.1%) (Table 1). Meaningful differences between states that report vs those that do not report county-level abortions were the proportion of Black residents of reproductive age (10.4% vs 6.1%) or those of other race/ethnicity (3.4% vs 5.5%), the proportion of married female residents older than 15 years (50.7% vs 51.7%), and the distribution of reproductive health policy grades (eg, grade F, 41.7% vs 15.2%).

    In 27 states and 1948 counties with 37.3 million female residents of reproductive age, there were 428 720 reported abortions (eTable 1 in the Supplement and Figure 1). The mean abortion rate was 11.5 per 1000 female residents of reproductive age; the median rate was 9.9 per 1000 female residents of reproductive age (range, 0-63.6 per 1000 female residents of reproductive age). The Moran I statistic showed a strong spatial correlation of reported abortions, at 0.49. In 48 states, the population-weighted mean travel distance to the nearest facility offering abortion care was 25.6 miles (41 km); median travel distance, 8.2 miles (13 km) (range, 1-383 miles [1.6-613 km]) (eFigure 1 in the Supplement).

    In a multivariable spatial Poisson model, greater travel distance was associated with declining abortion rates in a dose-response association (Table 2). Compared with a median county travel distance of less than 5 miles, the abortion rate declined by 0.05 (95% CI, −0.07 to −0.03) per 1000 female residents of reproductive age at 5 to less than 15 miles (8 to <24 km), 0.22 (95% CI, −0.24 to −0.19) per 1000 female residents of reproductive age at 15 to less than 30 miles (24 to <48 km), 0.34 (95% CI, −0.37 to −0.31) per 1000 female residents of reproductive age at 30 to less than 60 miles (48 to <96 km), 0.43 (95% CI −0.47 to −0.39) per 1000 female residents of reproductive age at 60 to less than 120 miles (96 to <193 km), and 0.73 (95% CI −0.80 to −0.65) per 1000 female residents of reproductive age at 120 miles or more (≥193 km). The cross-validated mean squared error for this model was 0.01, and the Moran I statistic was −0.01.

    Based on this model, we estimated the abortion rate in the 48 contiguous states using 2015 travel distance conditions. The national estimate is for 3107 counties with 62.5 million female residents of reproductive age. We estimated 696 760 abortions with a mean abortion rate of 11.1 and median abortion rate of 8.7 per 1000 female residents of reproductive age (range, 1.0-45.5) (Table 3 and Figure 2).

    In estimates under 2 alternate travel scenarios, lesser travel distance was associated with changes in the abortion rate (Table 3 and eTable 2 in the Supplement). With all travel distances less than 30 miles, a common definition of network adequacy for primary care,23 we estimated 714 660 abortions, a 2.6% increase (18 190 additional abortions) (eFigure 2 in the Supplement). In this scenario, the mean abortion rate was 11.4 per 1000 female residents of reproductive age; the median rate was 9.3 per 1000 female residents of reproductive age (range, 1.1-45.5). In the scenario with all travel distances of less than 5 miles, a simulation of medication abortion by telemedicine, we estimated 767 390 abortions, a 10.2% increase (70 920 additional abortions) (eFigure 3 in the Supplement). In this scenario, the mean abortion rate was 12.3 per 1000 female residents of reproductive age and the median rate was 10.2 per 1000 female residents of reproductive age (range, 1.4-45.5). Under both travel distance scenarios, the largest increases in abortion rate were in South Carolina, Mississippi, New Mexico, Louisiana, Wyoming, Texas, North Carolina, Georgia, South Dakota, and Arizona. States with the smallest estimated increases in abortion rate were California, New York, Colorado, Iowa, Connecticut, Maryland, Nevada, and Washington. In sensitivity analyses using travel time as the main exposure, the results were similar (eTable 3 in the Supplement).

    Discussion

    This study provides a national model of abortions by county of residence and confirms the association between greater travel distance and lower abortion rates previously found in state-level and regional analyses.11,15-22 The declines in the median abortion rate by travel distance category are clinically meaningful and have meaningful consequences for people’s lives. This study generated interpretable state- and county-level abortion rates that have utility for policy makers. It also estimates the unmet need for abortion services associated with large travel distances from facilities. As states continue to enact laws restricting abortion access, it is likely that the distance to care will grow for more US women. Implementing policies to reduce spatial inequalities will become even more important in the event that the US Supreme Court returns to states the right to ban abortion. Even with Roe v Wade intact, we estimate there would be more than 70 000 additional abortions annually if spatial inequalities were eliminated.

    Our findings regarding the association between travel distance and abortion rate are consistent with previous regional studies, suggesting this phenomenon is robust to differing analytic approaches.11,15-18 Most regional analyses used differences-in-differences to estimate the decrease in abortion rate after state policy changes and abortion facility closures. Unlike Brown et al,22 we did not find evidence of a threshold association between distance to abortion provider and the abortion rate, possibly because our distance measure was more precise, based on road travel to the address of the nearest facility rather than the straight-line distance between population-weighted centers of 2 counties. This analysis expands what is known about the association between travel distance and abortion rates by estimating the effect of public health interventions that could reduce travel for abortion care. We estimate the number of individuals who would want abortion care but are unable to access it due to long travel distances.

    One critique of this analysis is that reverse causality may be at play—that is, facilities providing abortion care may locate where there is demand, making the observed association between abortion rate and travel distance a function of declining demand. In a study designed to address reverse causality, Brown et al22 showed that increased distance to abortion providers accounted for their observed declines in the abortion rate. This is plausible because state laws and regulations of abortion providers directly affect where they can locate, constraining a demand-driven response.30 To address this critique, we account for state abortion policies by including a covariable in our model. When spatial inequalities are eliminated in the telemedicine scenario, all but 1 of the states with the greatest increases in abortion rate have highly restrictive abortion policies (grade D or F): New Mexico (grade B) had a large increase in abortion rate and supportive policies, likely due to large travel distances to an abortion provider (73% of counties have median travel distance >60 miles). Conversely, states with the smallest increases in abortion rate all have policies supportive of access to abortion care and low numbers of women of reproductive age per facility.

    A periodic census of US abortion providers suggests a larger number of abortions than our model estimates, although the census does not provide data for 2015.1 Our model’s underestimation of abortions in 2 populous states at the edges of the map, Florida and California, accounts for most of the difference between these 2 sources of data. Our model underestimates abortion in these states because neither California nor Florida report abortions by county of residence and because the MRF smooth term trends toward zero along adjacency map edges. We were able to partially address this by using the MRF smooth term only in counties that fall within the convex hull of reported data. This underestimation means that our estimates of the number of abortions in all scenarios are conservative.

    Limitations

    Some limitations of this study are shared by all abortion-related research. Abortions are underreported by states for a variety of reasons, so our model likely underestimates the abortion rate. Our model includes more states with more complete data than previous studies and had fewer differences in female sociodemographic characteristics by counties’ abortion reporting status. Travel distance to an abortion facility is not the only barrier faced by people seeking abortion care; financial limitations, restrictive laws, religious beliefs, and stigma also create barriers.31 The model partly addresses these barriers by including household income and reproductive health state policy grades as covariables.27

    Some limitations of this study are specific to spatial analyses. Outcomes are sometimes sensitive to the level of geography used in an analysis.32 We had no option but to use county as the geographic unit of analysis owing to the way abortion data are reported. Our results are likely to be conservative because the model used travel distance rather than travel time; travel time may be high when a trip is taken on public transit, despite a short distance. The model assumes that people would travel to the nearest facility, but considerations such as preexisting health conditions, financial limitations, or gestational age may affect the choice of a facility. However, a national survey of patients undergoing abortion33 showed that 80% went to the nearest facility or to one that was within 15 miles of the nearest facility.

    Conclusions

    In this cross-sectional study, greater travel distance to the nearest abortion care facility was associated with a lower abortion rate, suggesting that reducing travel distances to abortion facilities would increase access, even in states without restrictive laws. These results identify geographic areas with insufficient access to abortion care and could inform decisions about the location of new facilities. However, given the low population density of many counties with poor access, innovative strategies to meet the need for care are warranted.34 The Centers for Disease Control and Prevention included in its 10 essential public health services “assur[ing] an effective system that enables equitable access to the individual services and care needed to be healthy.”35 A public health approach to abortion would use strategies that have increased access to other forms of medical care, including expanding the types of clinicians who can offer care, mobile clinics, telemedicine, dispensing via lockboxes, and mail order delivery of medications.36 Such public health approaches will become even more important if the US Supreme Court gives states more leeway to regulate or ban abortion. New models of abortion care could rely less on facilities and increase access even for remote counties, leading to more equitable reproductive health care nationwide.

    Back to top
    Article Information

    Accepted for Publication: April 29, 2021.

    Published: July 6, 2021. doi:10.1001/jamanetworkopen.2021.15530

    Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2021 Thompson KMJ et al. JAMA Network Open.

    Corresponding Author: Kirsten M. J. Thompson, MPH, Bixby Center for Global Reproductive Health, Department of Obstetrics, Gynecology and Reproductive Sciences, University of California, San Francisco, 1001 Potrero Ave, Ward 6D-12, San Francisco, CA 94110 (kirsten.thompson@ucsf.edu).

    Author Contributions: All authors contributed equally to the study design, literature search, data interpretation, and writing of this report. Ms Thompson had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

    Concept and design: Thompson, Foster, Upadhyay.

    Acquisition, analysis, or interpretation of data: Thompson, Sturrock, Upadhyay.

    Drafting of the manuscript: Thompson, Sturrock.

    Critical revision of the manuscript for important intellectual content: All authors.

    Statistical analysis: Thompson, Sturrock.

    Administrative, technical, or material support: Upadhyay.

    Supervision: Foster, Upadhyay.

    Conflict of Interest Disclosures: Ms Thompson reported receiving personal fees from GenBioPro outside the submitted work. No other disclosures were reported.

    Funding/Support: This study was supported by grant 2019-7861 from the William and Flora Hewlett Foundation and by an anonymous foundation for general operating support (Ms Thompson).

    Role of the Funder/Sponsor: The sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

    References
    1.
    Onitilo  AA, Liang  H, Stankowski  RV,  et al.  Geographical and seasonal barriers to mammography services and breast cancer stage at diagnosis.   Rural Remote Health. 2014;14(3):2738.PubMedGoogle Scholar
    2.
    Engelman  KK, Hawley  DB, Gazaway  R, Mosier  MC, Ahluwalia  JS, Ellerbeck  EF.  Impact of geographic barriers on the utilization of mammograms by older rural women.   J Am Geriatr Soc. 2002;50(1):62-68. doi:10.1046/j.1532-5415.2002.50009.x PubMedGoogle ScholarCrossref
    3.
    Huang  B, Dignan  M, Han  D, Johnson  O.  Does distance matter? distance to mammography facilities and stage at diagnosis of breast cancer in Kentucky.   J Rural Health. 2009;25(4):366-371. doi:10.1111/j.1748-0361.2009.00245.x PubMedGoogle ScholarCrossref
    4.
    Gumpertz  ML, Pickle  LW, Miller  BA, Bell  BS.  Geographic patterns of advanced breast cancer in Los Angeles: associations with biological and sociodemographic factors (United States).   Cancer Causes Control. 2006;17(3):325-339. doi:10.1007/s10552-005-0513-1 PubMedGoogle ScholarCrossref
    5.
    Anderson  AE, Henry  KA, Samadder  NJ, Merrill  RM, Kinney  AY.  Rural vs urban residence affects risk-appropriate colorectal cancer screening.   Clin Gastroenterol Hepatol. 2013;11(5):526-533. doi:10.1016/j.cgh.2012.11.025 PubMedGoogle ScholarCrossref
    6.
    Shoff  C, Chen  VY, Yang  TC.  When homogeneity meets heterogeneity: the geographically weighted regression with spatial lag approach to prenatal care utilization.   Geospat Health. 2014;8(2):557-568. doi:10.4081/gh.2014.45 PubMedGoogle ScholarCrossref
    7.
    Kozhimannil  KB, Hung  P, Henning-Smith  C, Casey  MM, Prasad  S.  Association between loss of hospital-based obstetric services and birth outcomes in rural counties in the United States.   JAMA. 2018;319(12):1239-1247. doi:10.1001/jama.2018.1830 PubMedGoogle ScholarCrossref
    8.
    Jones  RK, Jerman  J.  Population group abortion rates and lifetime incidence of abortion: United States, 2008-2014.   Am J Public Health. 2017;107(12):1904-1909. doi:10.2105/AJPH.2017.304042 PubMedGoogle ScholarCrossref
    9.
    Policy Surveillance Program. State abortion laws. Updated March 1, 2021. Accessed March 23, 2021. https://lawatlas.org/datasets/abortion-laws
    10.
    Gerdts  C, Fuentes  L, Grossman  D,  et al.  Impact of clinic closures on women obtaining abortion services after implementation of a restrictive law in Texas.   Am J Public Health. 2016;106(5):857-864. doi:10.2105/AJPH.2016.303134 PubMedGoogle ScholarCrossref
    11.
    Venator  J, Fletcher  J.  Undue Burden Beyond Texas: An Analysis of Abortion Clinic Closures, Births and Abortions in Wisconsin. National Bureau of Economic Research; 2019. doi:10.3386/w26362
    12.
    Barr-Walker  J, Jayaweera  RT, Ramirez  AM, Gerdts  C.  Experiences of women who travel for abortion: a mixed methods systematic review.   PLoS One. 2019;14(4):e0209991. doi:10.1371/journal.pone.0209991 PubMedGoogle Scholar
    13.
    Upadhyay  UD, Johns  NE, Meckstroth  KR, Kerns  JL.  Distance traveled for an abortion and source of care after abortion.   Obstet Gynecol. 2017;130(3):616-624. doi:10.1097/AOG.0000000000002188 PubMedGoogle ScholarCrossref
    14.
    Khan  AA.  An integrated approach to measuring potential spatial access to health care services.   Socioecon Plann Sci. 1992;26(4):275-287. doi:10.1016/0038-0121(92)90004-O PubMedGoogle ScholarCrossref
    15.
    Johns  NE, Foster  DG, Upadhyay  UD.  Distance traveled for Medicaid-covered abortion care in California.   BMC Health Serv Res. 2017;17(1):287. doi:10.1186/s12913-017-2241-0 PubMedGoogle ScholarCrossref
    16.
    Grossman  D, White  K, Hopkins  K, Potter  JE.  Change in distance to nearest facility and abortion in Texas, 2012 to 2014.   JAMA. 2017;317(4):437-439. doi:10.1001/jama.2016.17026 PubMedGoogle ScholarCrossref
    17.
    Joyce  TJ, Tan  R, Zhang  Y.  Back to the Future? Abortion Before and After Roe. National Bureau of Economic Research; 2012. doi:10.3386/w18338
    18.
    Cunningham  S, Lindo  JM, Myers  C, Schlosser  A.  How Far Is too Far? New Evidence on Abortion Clinic Closures, Access and Abortions. National Bureau of Economic Research; 2017.
    19.
    Bearak  JM, Burke  KL, Jones  RK.  Disparities and change over time in distance women would need to travel to have an abortion in the USA: a spatial analysis.   Lancet Public Health. 2017;2(11):e493-e500. doi:10.1016/S2468-2667(17)30158-5 PubMedGoogle ScholarCrossref
    20.
    Cartwright  AF, Karunaratne  M, Barr-Walker  J, Johns  NE, Upadhyay  UD.  Identifying national availability of abortion care and distance from major US cities: systematic online search.   J Med Internet Res. 2018;20(5):e186. doi:10.2196/jmir.9717 PubMedGoogle Scholar
    21.
    Myers  C, Jones  R, Upadhyay  U.  Predicted changes in abortion access and incidence in a post-Roe world.   Contraception. 2019;100(5):367-373. doi:10.1016/j.contraception.2019.07.139 PubMedGoogle ScholarCrossref
    22.
    Brown  BP, Hebert  LE, Gilliam  M, Kaestner  R.  Distance to an abortion provider and its association with the abortion rate: a multistate longitudinal analysis.   Perspect Sex Reprod Health. 2020;52(4):227-234. doi:10.1363/psrh.12164 PubMedGoogle ScholarCrossref
    23.
    National Conference of State Legislatures. Insurance carriers and access to healthcare providers: network adequacy. Updated February 1, 2018. Accessed June 14, 2019. https://www.ncsl.org/research/health/insurance-carriers-and-access-to-healthcare-providers-network-adequacy.aspx
    24.
    Jatlaoui  TC, Boutot  ME, Mandel  MG,  et al.  Abortion surveillance—United States, 2015.   MMWR Surveill Summ. 2018;67(13):1-45. doi:10.15585/mmwr.ss6713a1 PubMedGoogle ScholarCrossref
    25.
    Huber  S, Rust  C.  osrmtime: Calculate travel time and distance with OpenStreetMap data using the Open Source Routing Machine (OSRM).   Stata J. 2016;16(2):416-423. doi:10.1177/1536867X1601600209 Google ScholarCrossref
    26.
    US Census Bureau. Centers of population. Updated October 11, 2019. Accessed May 7, 2018. https://www.census.gov/geographies/reference-files/time-series/geo/centers-population.html
    27.
    National Abortion Rights Action League. Who decides? the status of women’s reproductive rights in the United States. January 2016. Accessed August 31, 2018. https://www.prochoiceamerica.org/wp-content/uploads/2017/01/WHODecides-2016-Final.pdf
    28.
    Upadhyay  UD, McCook  AA, Bennett  AH, Cartwright  AF, Roberts  SCM.  State abortion policies and Medicaid coverage of abortion are associated with pregnancy outcomes among individuals seeking abortion recruited using Google Ads: a national cohort study.   Soc Sci Med. 2021;274:113747. doi:10.1016/j.socscimed.2021.113747 PubMedGoogle Scholar
    29.
    Wood  SN.  Generalized Additive Models: An Introduction With R. CRC Press; 2017. doi:10.1201/9781315370279
    30.
    Brown  BP, Hebert  LE, Gilliam  M, Kaestner  R.  Association of highly restrictive state abortion policies with abortion rates, 2000-2014.   JAMA Netw Open. 2020;3(11):e2024610. doi:10.1001/jamanetworkopen.2020.24610 PubMedGoogle Scholar
    31.
    Cohen  DS, Joffe  C.  Obstacle Course: The Everyday Struggle to Get an Abortion in America. University of California Press; 2020.
    32.
    Krieger  N, Chen  JT, Waterman  PD, Soobader  MJ, Subramanian  SV, Carson  R.  Geocoding and monitoring of US socioeconomic inequalities in mortality and cancer incidence: does the choice of area-based measure and geographic level matter? the Public Health Disparities Geocoding Project.   Am J Epidemiol. 2002;156(5):471-482. doi:10.1093/aje/kwf068 PubMedGoogle ScholarCrossref
    33.
    Fuentes  L, Jerman  J.  Distance traveled to obtain clinical abortion care in the United States and reasons for clinic choice.   J Womens Health (Larchmt). 2019;28(12):1623-1631. doi:10.1089/jwh.2018.7496 PubMedGoogle ScholarCrossref
    34.
    Upadhyay  UD.  Innovative models are needed for equitable abortion access in the USA.   Lancet Public Health. 2017;2(11):e484-e485. doi:10.1016/S2468-2667(17)30181-0 PubMedGoogle ScholarCrossref
    35.
    Centers for Disease Control and Prevention. 10 Essential public health services. Updated March 18, 2021. Accessed August 31, 2019. https://www.cdc.gov/publichealthgateway/publichealthservices/essentialhealthservices.html
    36.
    Roberts  SCM, Fuentes  L, Berglas  NF, Dennis  AJA.  A 21st-century public health approach to abortion.   Am J Public Health. 2017;107(12):1878-1882. doi:10.2105/AJPH.2017.304068 PubMedGoogle ScholarCrossref
    ×