Trends in Birth Rates After Elimination of Cost Sharing for Contraception by the Patient Protection and Affordable Care Act

Key Points

Question  Was the elimination of cost sharing for contraception under the Patient Protection and Affordable Care Act associated with a change in birth rates among women in different income groups?

Findings  In this cross-sectional study, the estimated probability of birth decreased most precipitously among women in the lowest income group from 8.0% in 2014 to 6.2% in 2018. The probability of a birth also decreased significantly among women in higher income groups, but this decrease was smaller in magnitude.

Meaning  These findings suggest that contraception insurance coverage without consumer cost sharing may be associated with decreased income-related disparities in unintended birth rates.

Importance  Reducing out-of-pocket costs is associated with improved patterns of contraception use. It is unknown whether reducing out-of-pocket costs is associated with fewer births.

Objective  To evaluate changes in birth rates by income level among commercially insured women before (2008-2013) and after (2014-2018) the elimination of cost sharing for contraception under the Patient Protection and Affordable Care Act (ACA).

Design, Setting, and Participants  This cross-sectional study used data from Clinformatics Data Mart database from January 1, 2008, to December 31, 2018, for women aged 15 to 45 years who were enrolled in an employer-based health plan and had pregnancy benefits for at least 1 year. Women without household income information and women with evidence of having undergone a hysterectomy were excluded.

Exposure  Section 2713 of the ACA.

Main Outcomes and Measures  The primary outcome was the proportion of reproductive-aged women with a live birth by year (measured yearly from 2008 to 2018 [11 time points]) within 3 income categories. The secondary outcome was the distribution of contraceptive method fills in 3 categories by year: (1) most effective methods (long-acting reversible contraception or sterilization), (2) moderately effective methods (pill, patch, ring, and injectable), and (3) no prescription or surgical method.

Results  The analytic sample included 4 590 989 women (mean [SD] age; 30.8 [9.1] years in 2013; 3 069 053 White [66.9%]) enrolled in 47 721 health plans. A total of 500 898 participants (40.8%) resided in households with incomes less than 400% of the federal poverty level in 2013. In all 3 years (2008, 2013, and 2018), women in the lowest income category were younger than women in the other income groups (median range, 21-22 years vs 30-34 years) and in households with a higher median number of dependents (9-10 vs 2-4). There was an associated decrease in births in all income groups in the period after the elimination of out-of-pocket costs. The estimated probability of birth decreased most precipitously among women in the lowest income group from 8.0% (95% CI, 7.4%-8.5%) in 2014 to 6.2% (95% CI, 5.7%-6.7%) in 2018, representing a 22.2% decrease (P < .001). The estimated probability decreased in the middle income group by 9.4%, from 6.4% (95% CI, 6.3%-6.4%) to 5.8% (95% CI, 5.7%-5.8%) (P < .001), and in the highest income group by 1.8%, from 5.6% (95% CI, 5.6%-5.7%) to 5.5% (95% CI, 5.4%-5.5%) (P < .001) in the period after the elimination of cost sharing.

Conclusions and Relevance  In this cross-sectional study, the elimination of cost sharing for contraception under the ACA was associated with improvements in contraceptive method prescription fills and a decrease in births among commercially insured women. Women with low income had more precipitous decreases than women with higher income, suggesting that enhanced access to contraception may address well-documented income-related disparities in unintended birth rates.

Nearly half of pregnancies in the US are unplanned, and marked income-related disparities in unintended pregnancy rates are well described.¹ Unintended pregnancies are associated with delayed prenatal care, reduced likelihood of breastfeeding, maternal depression, physical violence, and higher maternal and infant mortality rates.^2,3 Undesired pregnancies have societal costs, including an estimated $5 billion per year in direct and indirect costs for the US health care system.⁴ On the basis of evidence that contraception is effective at preventing pregnancy, cost-effective, and consistent with Healthy People 2020 goals, the Institute of Medicine identified contraception as an essential preventive health service for women.⁵ As a result, contraception was included on the list of preventive care services required to be covered without consumer cost sharing by most insurers under §2713 of the Patient Protection and Affordable Care Act (ACA).⁶

Studies have consistently shown that removal of out-of-pocket costs (OOPCs) for contraception is associated with increased consistency of use, method continuation, and selection of the most effective methods.^7-11 The ACA effectively eliminated cost sharing for most commercially insured women by 2014, and its implementation was associated with increased use of prescription contraception, particularly the use of long-acting reversible contraception (LARC) (eg, intrauterine device or implant).^8-10,12 It remains unknown, however, whether these changes in contraceptive use were associated with fewer births. Furthermore, it is also unknown whether this policy might specifically benefit women with low income, who may be more sensitive to prices and also experience higher rates of unintended births compared with women with higher income.¹ Accordingly, our objective was to examine changes in birth rates by income level among commercially insured women before (2008-2013) and after (2014-2018) the elimination of cost sharing for contraception under the ACA.

This cross-sectional study used deidentified patient data from Clinformatics Data Mart (Optum) to examine the association of the elimination of OOPCs with contraceptive provision patterns and births among commercially insured women. The Clinformatics Data Mart database includes patient demographic information, including household income and data on patient cost sharing. Race/ethnicity is included in this sample and is populated using a combination of sources including self-report, public records, and demographic derivations. This study was deemed exempt by the University of Michigan institutional review board. Informed consent was waived because this was a retrospective review of existing deidentified data. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

The analytic sample was drawn from a population of 7 761 568 women aged 15 to 45 years enrolled in employer-based health plans between January 1, 2008, and December 31, 2018. We restricted the sample to women with continuous enrollment in a single employer-based health plan with pregnancy benefits for at least 1 year. Women without household income information and women with evidence of having undergone a hysterectomy were excluded. The sample selection is described in detail elsewhere⁹ (eFigure in the Supplement). All outcomes and patient characteristics were collected at the woman-year level.

Our hypothesis was that the removal of financial barriers (ie, OOPCs) to contraception would be associated with a greater proportion of women using either moderately effective or the most effective contraceptive methods, all of which require a prescription or a procedure and can be costly to initiate. We believed that improved contraception use patterns would be associated with a decrease in births because a large proportion of births follow an unintended conception. Previous work reported that OOPCs decreased rapidly after mandated first dollar coverage was implemented in similar health plans.^8,10,13,14 The preperiod (2008-2013) and postperiod (2014-2018) that we chose were based on observed decreases in OOPCs for LARC methods because they were known to have the highest level of OOPCs at baseline.^9,13

Our primary outcome was the proportion of commercially insured women with a birth by income level between 2008 and 2018. Information on household income and number of covered dependents was included in our data source. Household income for the primary beneficiary was demographically derived at the household level. Yearly guidelines for poverty-level computations from the US Department of Health and Human Service’s Office of the Assistant Secretary for Planning and Evaluation¹⁵ were used with the household income and number of covered dependents data to categorize women into 3 income groups relative to the federal poverty level (FPL): less than 100% FPL; 100% to 399% FPL; and greater than or equal to 400% FPL.

Our secondary outcome, contraceptive fill patterns, was derived to provide additional evidence that either supported or refuted our hypothesis. We assessed whether there were parallel trends in contraception fill patterns consistent with observed changes in births. We did not consider contraception fill patterns as a primary outcome because changes in contraception use after the implementation of the ACA have been reported previously.^9,10,13,14 Although incidence of use (eg, LARC placements, sterilization procedures) are reliably identified in claims, estimating prevalence of use is less accurate for long-acting methods, especially sterilization. We categorized contraceptive method fill patterns by effectiveness: (1) most effective methods (LARC or sterilization), (2) moderately effective methods (pill, patch, ring, or injectable), and (3) no prescription or surgical method. Method fills were identified using device, procedure, diagnosis, and pharmacy codes (eTable 1 in the Supplement). Long-acting or sterilization methods were identified using International Classification of Diseases, Ninth Revision, International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, and Current Procedural Terminology, Fourth Edition (Healthcare Common Procedure Coding System) codes. Both procedure and device codes were required to indicate LARC use. Pharmacy fills in a given calendar year were used to identify moderately effective method users. Women with evidence of more than 1 method category fill in a year (eg, pills, sterilization) were categorized in the most effective method category for that calendar year.

We evaluated changes in births before (2008-2013) and after (2014-2018) the elimination of OOPCs for contraception under the ACA by household income category (<100% FPL, 100%-399% FPL, and ≥400% FPL) with a comparative interrupted time-series model conducted at the woman-year level.¹⁶ This design approach compares outcomes before and after an intervention while accounting for trends over time. Our comparative interrupted time-series approach expanded on the standard interrupted time-series design with the inclusion of comparison groups based on household income. The interactions among time, household income category, and the elimination of OOPCs were of primary interest within the comparative interrupted time-series framework, and the following interactions were included in our model: time by elimination of OOPC, time by household income, elimination of OOPC and household income, and time by household income by elimination of OOPC. The comparative interrupted time-series model was fit with a generalized estimating equation to account for repeated observations per woman over the study period and controlled for age, race/ethnicity, number of dependents, geographic region, and health plan type. Model-based adjusted predictions were obtained. Birth trends and levels before and after the elimination of OOPCs for contraception were compared within and between household income groups. Similar models and methods assessed contraceptive fill patterns. All statistical tests were 2-sided, and an α level of .05 was used to determine statistical significance. Claims data management was performed in SAS, version 9.2 (SAS Institute), and statistical analyses were conducted in Stata, version 14.1 (StataCorp).

Our analytic sample included 4 590 989 unique women (mean [SD] age, 30.8 [9.1] years in 2013) enrolled in 47 721 health plans between 2008 and 2018. Women were predominately White (3 069 053 [66.9%]), and 500 898 of the participants in 2013 (40.8%) resided in households with incomes below 400% FPL (Table). Compared with women in the higher income groups, women in the lowest income category were younger (median range, 21-22 years vs 30-34 years), resided in households with a larger number of dependents (median range, 9-10 vs 2-4), and were more racially diverse (Black proportion range, 16.9%-25.3% vs 6.7%-19.2%). In this sample, the percentage of LARC placements that incurred OOPCs decreased rapidly from 66.3% (11 304 of 17 039) in 2008 to 6.7% (1845 of 27 426) in 2014 and 3.2% (973 of 29 954) in 2018. During the study period, we identified 766 087 births among 20 455 355 observations (woman-years).

There was a significantly associated decrease in births in all income groups in the period after the elimination of OOPCs (Figure 1). The estimated probability of birth decreased most precipitously among women in the lowest income group from 8.0% (95% CI, 7.4%-8.5%) in 2014 to 6.2% (95% CI, 5.7%-6.7%) in 2018, representing a 22.2% decrease. The estimated probability also decreased in the middle income group by 9.4%, from 6.4% (95% CI, 6.3%-6.4%) to 5.8% (95% CI, 5.7%-5.8%), and in the highest income group by 1.8%, from 5.6% (95% CI, 5.6%-5.7%) to 5.5% (95% CI, 5.4%-5.5%) after the elimination of OOPCs (eTable 2 in the Supplement). Within each income group, annual birth rates before and after the elimination of OOPCs differed significantly. For example, the change from 2013 to 2014 in the less than 100% FPL group was –0.0087 (95% CI, –0.017 to –0.0003), in the 100% to 399% FPL group was –0.002 (95% CI, –0.0028 to –0.001), and in the 400% or greater FPL group was –0.0008 (95% CI, –0.001 to –0.0003). Annual birth rates after the elimination of OOPCs also differed significantly among the household income groups (<100% FPL vs 100%-399% FPL, 0.0032 [95% CI, 0.0009-0.005]; <100% FPL vs ≥400% FPL, 0.0054 [95% CI, 0.003-0.008]; 100%-399% FPL vs ≥400% FPL, 0.0022 [95% CI, 0.0019-0.0025]). Birth rates decreased more rapidly in low and middle income categories than in the high income group (Figure 1). Differences in birth rates between women in the highest and lowest income groups decreased by 62.2% between 2008 and 2018.

Changes in contraception fill patterns between 2008 and 2018 were consistent with the observed changes in birth rates. Within each income group, trends in annual rates of not filling a prescription method differed significantly before and after the elimination of OOPCs in 2014 (<100% FPL, –0.007 [95% CI, –0.012 to –0.002]; 100%-399% FPL, –0.009 [95% CI, –0.0097 to –0.0088]; ≥400% FPL, –0.008 [95% CI, –0.0084 to –0.0075]).

Trends in annual rates of not filling a prescription method of contraception after 2014 also differed significantly across income groups, with the 2 lower household income groups demonstrating a more rapid decrease than the higher income group (Figure 2 and eTable 3 in the Supplement). Conversely, shifts in adjusted prescription contraception method fills in the higher income group were largely between different categories of prescription methods (eg, pills to LARC) (Figure 3).

Results from our study of this large sample of commercially insured women suggest that the elimination of cost sharing was associated with an increase in the birth rate reduction within all income groups but most notably in the 2 lower income groups. These observed trends were associated with a smaller difference in birth rates between women in the highest and lowest income categories. Observed changes in contraceptive method fills were consistent with changes in births. Trends in prescription contraception fills, especially for the most effective methods, were positively associated with the elimination of cost sharing. Furthermore, the observed differences in method fills across income groups were consistent with our hypothesis that women with low income may be more sensitive to the price of contraception. These findings suggest that the elimination of cost sharing for contraception may be associated not only with improved contraception use patterns but also with fewer births. Furthermore, the observed differences in birth rates across income groups suggest that expanded coverage of prescription contraception may be associated with a reduction in income-related disparities in unintended pregnancy rates.

Reducing unintended pregnancies has been a public health goal in the US for decades because there are serious consequences for women, families, and society. In 2011, 42% of unintended pregnancies (excluding miscarriages) ended in abortion¹ and two-thirds of unplanned births were funded by public insurance programs such as Medicaid.¹⁷ There are also serious long-term consequences of mistimed births for women,^18-23 for the life opportunities of their children,^21-24 and for society. Removing cost-related barriers for birth control (via the ACA mandate or otherwise) alone may not eliminate unintended pregnancy, but it may be an important component of the multifaceted approach needed to address this important public health concern.

Contraception is a clinically efficacious and cost-effective strategy for reducing unintended pregnancy. A policy that eliminates OOPCs for contraception is consistent with our clinical goals of ensuring that all individuals can decide whether and when to have children. This policy may also be an effective strategy to address well-documented income-related disparities in unintended pregnancy rates in the US. Recent court decisions, including the 2020 Supreme Court decision upholding the rules that expand exemptions from the contraceptive mandate,²⁵ could roll back improvements in access for some women. We believe that continued monitoring is warranted.

This study has strengths. These data offer advantages not available from other sources, including the size of the sample and information on household income.

This study also has limitations. First, this study could not establish a causal relationship between the elimination of cost sharing or changes in contraception use and decreases in births. However, it adds to a growing body of evidence that consistently concludes that the elimination of cost sharing is associated with improved contraceptive use patterns.^9,10,13,14 Changes in the population of women covered by employer health plans might also explain the observed decrease in births. Furthermore, in administrative data, we were not able to observe services or procedures without some level of insurance coverage or services obtained outside the member’s health plan. Because we could not directly measure pregnancy intention, we assumed that changes in contraception use patterns reflected pregnancy desires and that decreases in births represented unwanted pregnancies averted. In addition, we cannot generalize our findings to all commercially insured women or to women in general. This limitation might be especially true of the lowest income group in the sample because they appeared to have unique characteristics.

In this cross-sectional study, the elimination of cost sharing for contraception under the ACA was associated with improvements in contraceptive method prescription fills and a decrease in births among commercially insured women. Women with low income had more precipitous decreases than women with higher income, suggesting that enhanced access to contraception may address well-documented income-related disparities in unintended birth rates.

Accepted for Publication: September 8, 2020.

Published: November 6, 2020. doi:10.1001/jamanetworkopen.2020.24398

Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2020 Dalton VK et al. JAMA Network Open.

Corresponding Author: Vanessa K. Dalton, MD, Department of Obstetrics and Gynecology, University of Michigan, L4100 University Hospital South, 1500 E Medical Center Dr, Ann Arbor, MI 48109 (daltonvk@med.umich.edu).

Author Contributions: Dr Dalton and Ms Kolenic had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Dalton, Moniz, Bailey, Admon, Kolenic, Fendrick.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Dalton, Bailey, Admon, Kolenic, Tilea.

Critical revision of the manuscript for important intellectual content: Dalton, Moniz, Bailey, Admon, Fendrick.

Statistical analysis: Bailey, Kolenic, Tilea.

Obtained funding: Dalton.

Administrative, technical, or material support: Dalton, Bailey.

Supervision: Dalton, Admon, Fendrick.

Conflict of Interest Disclosures: Dr Dalton reported receiving research support from the Agency for Healthcare Research & Quality (AHRQ), the National Institutes of Health, the American Association of Obstetricians and Gynecologists Foundation, the Laura and John Arnold Foundation, the National Institute for Reproductive Health, and the Blue Cross Blue Shield Foundation; serving as a paid contributing editor for the Medical Letter and as an author for UpToDate; serving as a consultant for Bind; participating on study sections for the National Institutes of Health and the AHRQ; receiving personal fees from the Society of Family Planning and the Society of Maternal Fetal Medicine; and serving as a medical expert in vaccine litigation for Merck. outside the submitted work. Dr Moniz reported receiving grants from the AHRQ and receiving personal fees from Society of Family Planning outside the submitted work. Dr Fendrick reported receiving personal fees or equity interest from AbbVie, Amgen, Centivo, the Community Oncology Association, Covered California, EmblemHealth, Exact Sciences Corp, Freedman Health System, GRAIL, Harvard University, Health and Wellness Innovations, Health at Scale Technologies, MedZed, Penguin Pay, Risalto Health, Sempre Health, the State of Minnesota, the US Department of Defense, the Virginia Center for Health Innovation, Wellth, and Zanzors; receiving research support from AHRQ, Boehringer Ingelheim, the Gary and Mary West Health Policy Center, the National Pharmaceutical Council, the Patient-Centered Outcomes Research Institute, the Pharmaceutical Research and Manufacturers of America, the Robert Wood Johnson Foundation, and the State of Michigan/Centers for Medicare & Medicaid Services; serving as coeditor of the American Journal of Managed Care; and being a member of the Medicare Evidence Development and Coverage Advisory Committee and V-BID Health outside the submitted work. No other disclosures were reported.

Funding/Support: This study was supported by grant K08 HS025465 (Dr Moniz) and grant R01 HS023784 (Dr Dalton) from the AHRQ.

Role of the Funder/Sponsor: The AHRQ 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.