Understanding Dobbs’ Impacts on Abortions

The 2022 Dobbs v. Jackson Women’s Health Organization decision sparked profound change in reproductive health care, and there has been perpetual interest since in the effects of the ruling. One potential way of doing so is through analysis of health care claims data, which are ubiquitous.

For instance, claims data were used to conclude that abortion is equally as safe when provided in office-based or ambulatory settings, that complications rates including emergency department visits are rare, and that restricting federal funds for abortion is associated with adverse outcomes. These claims-based studies were effective because they used a rigorous process to identify abortion in claims, and appropriately assessed selection bias.

However despite enthusiasm for using claims data to study sexual and reproductive health services, particularly abortion, use of the data poses challenges. Researchers should be aware of limitations to claims data when using it to identify trends in abortion care—and the policy changes that could help overcome these limitations.

Limitation #1: Federal limitations on abortion coverage. First, since 1977, the Hyde Amendment has prohibited the use of federal funds to finance abortion care except in cases of rape, incest, or immediate threat to the pregnant person’s life. This means federal insurance programs like Medicaid (the largest payor of pregnancy care and the primary source of coverage for most people seeking abortion), and other programs like Medicare, Marketplace plans, Federal Employees Health Benefits, TRICARE, and the Indian Health Service cannot be used to pay for abortion care outside of these limited circumstances.

Limitation #2: High numbers of abortions are paid out-of-pocket. Approximately 65% to 70% of people seeking abortion care pay out-of-pocket. Reasons included confidentiality concerns, lack of insurance coverage, and clinics performing the procedure not accepting insurance. Of the 30% to 35% that use insurance, Medicaid is twice as likely to be the payor compared to private insurance, but even in circumstances in which Medicaid will cover abortion care, coverage is rarely applied. Furthermore, when patients have insurance that covers abortion and want to use insurance to pay for the procedure, clinics providing it do not universally accept insurance. The likelihood of clinics accepting insurance varies by state policy on abortion. As a result, the number of abortions that are associated with an insurance claim is dramatically lower than the number of actual abortions, introducing significant selection bias into any cohort created from insurance claims.

Limitation #3: Travel for abortion is becoming more common. Increasingly restrictive state abortion policies mean more people now travel out-of-state for abortion, visit telemedicine medication abortion clinics in states with shield laws, and self-manage abortion with pills obtained outside the formal health setting through community networks or online vendors. In all these circumstances, patients are less likely to use insurance, further contributing to underestimation of abortion in claims data.

Policy Changes That Would Make Claims Data More Useful

Two straightforward policy changes would make a meaningful difference for capturing more accurate data on abortions.

A first step would be to standardize reporting in the national, comprehensive Transformed Medicaid Statistical Information System (T-MSIS) Analytic Files (TAF). For example, in states that use state funds to cover abortion for people on Medicaid, submission of claims to TAF should include all fee-for-service abortion claims, managed care abortion claims, and any abortions that fall under the Hyde exceptions. This standardized approach would increase representation and reduce sampling bias at the state and individual levels. Because service data appear more complete in individual state Medicaid claims data, ensuring reporting to TAF for wider analysis is critical for across-state comparisons.

A second policy change would be to include abortion, as well as other reproductive care like contraception, in ongoing efforts to harmonize the all-payer claims databases at the Agency for Healthcare Research and Quality (AHRQ). Inclusion of such data in the data harmonization effort will be central to understanding the impact of differing policies on sexual and reproductive health. Until abortion care is covered universally across health plans and states, comparisons across time and geography will remain challenging. As such, researchers should also estimate the proportion of abortions that are visible in claims relative to incidence reported to the CDC and the Abortion Provider Census.

A version of this piece originally appeared as “Measuring Abortion in Claims Data: What is the State of the Science?” in the November 14, 2024 issue of Contraception.