The impact of state Medicaid coverage of abortion on people accessing care in three states

Abstract

Context

Medicaid is a major funder of reproductive health services, including family planning and pregnancy‐related care, especially for people with limited income and people of color. Federal Medicaid funds cannot be used for abortion however 16 states allow state Medicaid funds to pay for abortion. In recent years, Illinois and Maine implemented, and West Virginia discontinued, state Medicaid coverage of abortion.

Methodology

With retrospective procedure‐ and patient‐level data obtained from clinics in these three states, we used an interrupted time series design, multivariable regression models, and descriptive statistics to assess changes in procedure volume and patients' share of total procedure price (patient price).

Results

In Maine and Illinois, implementing state Medicaid coverage of abortion contributed to an immediate overall increase in abortion access (as seen by a rise in monthly procedure volume at the time of the policy's implementation), a decrease in patient price (by 36% in Maine and 44% in Illinois) after policy implementation as compared to pre‐implementation, and overall improved access among people of color. Conversely, when West Virginia discontinued coverage, access to care decreased, patient price increased by 130%, and the share of abortion procedures among people of color decreased.

Conclusions

In the fragmented abortion access landscape of the post‐Roe era, our study provides new evidence that financial assistance offered through state Medicaid policies that cover abortion may be most helpful to those facing traditional structural inequities to access, while discontinuation of Medicaid coverage of abortion further burdens those already economically marginalized.

INTRODUCTION

Medicaid is the major public funder of family planning^* services in the United States for people ^† with limited income who may become pregnant, ¹ and covers a wide range of other reproductive health care, including prenatal services, childbirth, and postpartum care—all without cost‐sharing. In the context of systematic barriers and discrimination, a disproportionately higher number of people of color who may become pregnant are enrolled in Medicaid as their primary form of health coverage, ² and a higher percentage of abortions are obtained by people of color than by white individuals. ³ Yet the Hyde Amendment restricts the use of federal funds for abortion, including federal matching funds for state Medicaid programs, except in cases of rape, incest, and life endangerment. ^{‡ , 4} This further constrains access to abortion services for people with limited income, potentially furthering the racial health equity gap. States have been left to fill this gap. Existing literature shows that state Medicaid restrictions delay and reduce access to abortion services among Medicaid‐eligible pregnant people in the United States^{5, 6}—especially among people of color^{7, 8}—and that including abortion services under state Medicaid coverage can result in earlier access to care. ⁹ In recent years, states have moved in different directions regarding this issue, with some, such as Illinois and Maine, committing state Medicaid funds, whereas others, such as West Virginia, have discontinued state Medicaid coverage of abortion care. This has resulted in a complicated, fragmented policy and funding landscape in the U.S., especially in the wake of the 2022 Dobbs versus Jackson Women's Health Organization ruling that overturned the 1973 Roe versus Wade decision that abortion was a constitutionally protected right.

The objective of this study is to better understand the impact of the state Medicaid funding policy changes in Illinois, Maine, and West Virginia on (a) the volume of abortion procedures, which serves as a proxy for access to abortion care; and (b) the share of total procedure price paid by the patient (patient price). In Illinois in 2017, the passage of House Bill 40 (HB40) lifted funding bans on abortion coverage for people enrolled in Medicaid and further aimed to improve access to services by permitting advanced practice clinicians to perform abortions. Full implementation of the Medicaid policy took 30 months, during which several national organizations that support abortion care provided transitional funding to bridge the gap until the law was fully implemented. Maine implemented similar Medicaid policies. In 2019, Legislative Document (LD) 820 required Maine's Medicaid program and private health insurers to cover abortion services. Concurrently, Maine enacted LD 1261, allowing advanced practice clinicians to perform abortions. ^§ Conversely, in 2018, West Virginia retracted Medicaid coverage of abortion through the “No Constitutional Right to Abortion” Amendment to the state constitution.^** To address our aim, we analyzed clinic administrative data from 39 clinics across all three states.

Previous researchers have studied the implementation of HB40 in Illinois by analyzing electronic medical records at one major medical center ¹⁰ or by using rich qualitative data.^{11, 12} Analyzing 2018 data from one major medical center in Chicago, Commito and colleagues ¹⁰ found an increase in patients' gestational age at the time of procedure and a reduction in patients paying out of pocket after the Illinois policy shift. Their article concludes by inviting future studies to examine trends over several years after the policy change implementation and across multiple providers in the state. ¹⁰ In this spirit, to our knowledge, our study is the first to analyze the policy shift in Illinois with quantitative data collected from clinics across the state. Thus, our study complements research carried out by prior authors^{10, 12, 13} by (a) including clinic administrative data from multiple clinical sites, and (b) analyzing contemporaneous Medicaid policy shifts related to abortion coverage in two additional states. This study draws on data from a larger project about the impact of Medicaid policy changes on patients, clinics, and local abortion funds in Illinois, Maine, and West Virginia.

Our study also contributes to the knowledge base on abortion care access during COVID‐19.^{13, 14, 15} A significant portion of our study period overlapped with the emergence of the COVID‐19 pandemic and its resulting effects on health, health care access, employment and income, access to health insurance, and Medicaid eligibility. In the wake of the pandemic, many states classified abortion procedures as a nonessential use of personal protective equipment and issued orders that delayed abortion access for 6–8 weeks during the first wave of pandemic shutdowns. ¹⁶ During the shutdown period, 33 states suspended all elective medical procedures. This affected abortion access in West Virginia, for example, where all medication and instrumentation abortions not deemed medically necessary were prohibited. Additionally, 15 states attempted to prevent timely abortions through legislative action or executive order, including West Virginia and three states bordering Illinois. This context made it challenging to evaluate the impact of Medicaid policy shifts on access and cost outcomes, given the pandemic's potentially confounding effects and the states' unique pandemic responses that contextualized care delivery.

In their analysis of the implementation of state Medicaid coverage of abortion care in Illinois, Zuniga and colleagues ¹¹ noted that their data collection ended just before the pandemic arrived. Unable to ask abortion providers about the wide‐ranging effects of the pandemic on health care access, they cited a need for future research to explore the intersection of cost and insurance barriers with other factors affecting access to abortion care. ¹¹ In this vein, as depicted in Figure 1, we identified, a priori, various co‐occurring state‐ and national‐level policy changes related to abortion care that may also have affected abortion care access during our study period. At the national level, these policies included:

• Approval of generic mifepristone. A generic version of mifepristone was approved in 2019, ¹⁷ reducing the price of mifepristone to providers.

• Removal of in‐person requirement for dispensing mifepristone. In 2021, as part of the pandemic public health emergency, the FDA temporarily implemented enforcement discretion of the requirement for in‐person dispensing of mifepristone for medication abortion. The FDA risk evaluation and mitigation strategy (REMS) for mifepristone for reproductive health indications was modified in 2023 to retain this change. ¹⁸

FIGURE 1.

Timeline of abortion care policies enacted at the state and national levels during the study period (January 2016 through June 2021). * The Hyde Amendment, enacted in 1977, restricts the federal portion of federal‐state matches for Medicaid funding to cover abortion only in cases of rape, incest, or life endangerment. ²⁸ In Maine, LD 1261 and LD 820 were passed in June 2019 but enacted concurrently in September 2019. In Illinois, HB 40 was passed in 2017, but not enacted until January 1, 2018; and not fully funded until June 2020 (see the bar for “Medicaid fully implemented”). Between January 2018 and June 2020, several national organizations provided transitional funding (see the “Transitional Funding” bar) for Medicaid‐eligible patients to Illinois clinics that opted to accept these funds in lieu of billing Medicaid while the State worked toward fully funded implementation. Since transitional funding was a proxy for Medicaid funds in this timeframe, our study begins the post‐policy period in Illinois with the enactment of HB 40 on January 1, 2018. The pre‐policy analysis period in Illinois begins in 2017 because 2016 data were not available for all Illinois clinics. In November 2018, West Virginia immediately retracted Medicaid coverage of abortion after a vote to amend the state constitution (see the “No Constitutional Right to Abortion” bar).

State‐level policies included:

• Presumptive eligibility. In Maine, state Medicaid (MaineCare) moved to a presumptive eligibility model in 2021, which streamlined the Medicaid eligibility and scheduling processes (i.e., coverage could be assumed at time of scheduling the appointment). In Illinois, in 2019, the Reproductive Health Act granted pregnant people, mothers, and babies Medicaid coverage in a presumptive eligibility status that expedited their Medicaid application.

• Pandemic‐related response. There were state‐specific health care policy responses to the COVID‐19 public health emergency, including variation in the use of telehealth for medication abortions.

• Scope of practice laws. There were state‐specific expansions to the types of medical professionals who could perform abortions. As previously mentioned, the passage of HB40 in Illinois ¹⁹ and LD 1261 in Maine allowed advanced practice clinicians such as nurse practitioners and physician assistants to perform abortions.

These policies serve as a contextual framework for interpreting patient price and access outcomes in relation to state Medicaid coverage of abortion.

METHODOLOGY

Research questions

To understand the impact of Medicaid policy changes in Illinois, Maine, and West Virginia, we developed research questions focused on two outcomes of interest: procedure volume and patient price:

1. How has the number of abortion procedures (procedure volume) changed statewide after the policy shift, and how does this differ by patient demographic characteristics (race, financial need)?

2. How has the patient's share of the total procedure price (patient price) changed statewide after the policy shift?

Data collection and variables

We collected retrospective procedure‐ and patient‐level data (clinic administrative data) from January 2016 through June 2021 from 39 clinics that provided abortion services in Illinois, Maine, and West Virginia. Participating clinics included both stand‐alone abortion clinics and health centers with a focus to provide abortion and other reproductive health services. For each state, the date the Medicaid policy shift took effect ^†† demarcated the pre‐ and post‐policy analysis timeframe (see Figure 1). Because each state implemented its policy change (of implementing or discontinuing Medicaid coverage for abortion care) at a different time and the period for which data was available differed by state, the length of the pre‐ and post‐policy analysis periods was also unique to each state. Therefore, we present our findings in terms of monthly counts or average monthly percentages for a fairer comparison of pre‐ and post‐policy outcomes within a given state. ^‡‡

To minimize burden on clinics, we requested administrative data that they routinely collect as part of their operational processes. ^§§ We conducted quality checks of the clinic data to identify missing values and possible data errors and followed up with each clinic to address any issues with their data through an iterative process.

Procedure volume is a continuous variable that uses the monthly totals of completed abortion procedures (medication and instrumentation) in each state. Patient price consists of the share (i.e., percentage) of the total procedure price that is paid by the patient out‐of‐pocket after insurance payments and philanthropic contributions, if any. We calculated it using each procedure's price and payment information. It includes all services received by (a) full self‐pay patients and (b) insured patients with deductibles, copays, or coinsurance costs paid out‐of‐pocket. For the subgroup analyses, we constructed race and financial need as dichotomous variables based on patient demographic information and procedure payment data provided by the clinics. Regarding race, procedures were divided between “white” and “people of color”***; and financial need was defined as being eligible for or enrolled in Medicaid.

Data analysis

We first estimated changes in patient demographics, procedure volume and patient price, without controlling for other factors, using descriptive statistics and t‐tests, in order to identify statistically significant differences between pre‐ and post‐Medicaid funding periods. We then used scatter plots of monthly averages of both outcomes with associated linear time trends across pre‐ and post‐Medicaid funding periods to identify seasonal patterns, outliers, and anomalies in the data. This data visualization approach informed the interrupted time series (ITS) specifications we used in the regression analysis to evaluate the impact of the policy change controlling for other factors. We adjusted our regression models for individual‐ and clinic‐level variation, as well as variability in COVID‐19 prevalence (as measured by county‐specific monthly COVID‐19 cases and deaths).

The ITS method compared the trend of each outcome variable after the policy shift—either implementing or discontinuing state Medicaid funding—with the outcome trend that would have occurred if the trend observed in the pre‐policy analysis period had continued after the policy shift. The unit of analysis was patient‐month. We conducted the ITS analysis using the entire sample (all procedures) in each state. In Illinois and Maine, we also did so for procedure volume by race and financial need to identify heterogeneous effects of the policy shift on procedure volume across these two subgroups. Due to data outliers and incomplete race and financial need data for some months, we did not conduct heterogeneity analyses for subgroups in West Virginia.

We used the following regression model for the ITS analysis:

$$Y_{\mathit{\text{ipcm}}}=\begin{array}{l}{\beta }_0+{\beta }_1{\mathit{\text{time}}}_m+{\beta }_2{\mathit{\text{shift}}}_{\mathit{pm}}+{\beta }_3\mathit{\text{time}}\_\mathit{\text{after}}\_{\mathit{\text{shift}}}_{\mathit{pm}}+X_i\\ +Z_p+C_c+{\mathit{\text{COVID}}}_c+{ϵ}_{\mathit{\text{ipcm}}}.\end{array}$$

Brief descriptions of the variables used in the model follow:

• $Y_{\mathit{\text{ipcm}}}$ is the outcome (procedure volume, patient price) for individual $i$, who received care at clinic $p$, in county $c$, during month $m$.

• ${\mathit{\text{time}}}_m$ indicates the number of months since the beginning of the study period.

• ${\mathit{\text{shift}}}_{\mathit{pm}}$ is a binary variable that turns on with the shift from/to state Medicaid funding for clinic $p$.

• $\mathit{\text{time}}\_\mathit{\text{after}}\_{\mathit{\text{shift}}}_{\mathit{pm}}$ takes the value of 0 before the shift from/to state Medicaid funding, and after the shift it indicates the number of months since the policy change for clinic $p$. For example, the variable takes the value of 1 during the first month after the shift, 2 during the second month after the shift, and so on.

• $X_i$ is a binary variable that takes the value of 1 if the procedure took place in‐state.

• $Z_p$, and $C_c$ are clinic and county fixed effects.

• ${\mathit{\text{COVID}}}_c$ is a vector of county‐level COVID‐19 indicators, including cases and deaths per 100,000. ^†††

• ${ϵ}_{\mathit{\text{ipcm}}}$ is the idiosyncratic error term.

The interpretation of the model's coefficients follows:

• ${\beta }_0$ represents the baseline level of the outcome at $t$ = 0.

• ${\beta }_1$ represents the underlying trend during the pre‐policy analysis period.

• ${\beta }_2$ represents the change in the level of outcome following the shift from/to Medicaid funding.

• ${\beta }_3$ represents the change in the trend during the post‐policy analysis period.

RESULTS

This section presents results for procedure volume and patient price for each state. Table 1 shows patient demographic characteristics, averages, and t‐test significances for before and after the policy shift in Illinois and Maine (states that implemented the use of state Medicaid funds to cover abortions) and West Virginia (which discontinued the use of state Medicaid funds to cover abortions). In Illinois, the increase in procedure volume was proportionally greater for people of color than for white patients. Also, patients obtaining an abortion after the policy shift in Illinois had a significantly lower average income than those obtaining an abortion before the policy shift. In Maine, though representing a small number of procedures overall, people of color obtained significantly more procedures after the policy shift than before it. Conversely, as also seen in Table 1, the opposite pattern emerged in West Virginia: after the removal of abortion coverage under state Medicaid, abortion procedure volume among white patients remained unaffected—it even increased—whereas it decreased significantly among people of color.

TABLE 1.

Outcome and procedure characteristics summary before and after the policy shift, by state.

Demographics
	Before policy shift	After policy shift	Difference (t‐test) significance
Maine
Time period	Jan. 2016–Sept. 2019	Oct. 2019–June 2021
Total number of procedures (N)	6980 over 45 months	3602 over 21 months
Average total procedure price	$756	$687	−68**
Procedures funded by Medicaid (%)	6.7	35.8	n/a
Procedures paid fully out‐of‐pocket (%)	6.8	4.4	−2.4**
Procedures paid by private insurance, partially out‐of‐pocket, and/or by other funds (not Medicaid) (%)	86.5	59.8	n/a
Age (mean)	26.9	27.4	0.5**
White %	88.6	86.2	−2.4**
People of color %	11.4	13.8	2.4**
Hispanic %	2.6	4.5	1.9**
Out of state a %	5.5	6.6	1.1*
Annual income (mean)	$22,901	$24,529	1628*
Illinois
Time period	Jan. 2017–Dec. 2017	Jan. 2018–June 2021
Total number of procedures (N)	16,557 over 12 months	73,760 over 42 months
Average total procedure price	$1651	$1683	32**
Procedures funded by Medicaid (%)	0.2	37.1	n/a
Procedures paid fully out‐of‐pocket (%)	0.0	0.0	0
Procedures paid by private insurance, partially out‐of‐pocket, and/or by other funds (not Medicaid) (%)	99.8	62.9	n/a
Age (mean)	26.8	26.8	0
White %	51.9	43.7	−8.2**
People of color %	48.1	56.3	8.2**
Hispanic %	21.8	21.1	0.7
Out of state a %	18.1	16.9	−1.2**
Annual income (mean)	$23,498	$10,115	−13,383**
West Virginia
Time period	Jan. 2016–Oct. 2018	Nov. 2018–June 2021
Total number of procedures (N)	3979 over 34 months	2848 over 32 months
Average total procedure price	$394.5	$396.9	2.4
Procedures funded by Medicaid N (%)	59.3	0.0	n/a
Procedures paid fully out‐of‐pocket (%)	37.6	71.6	34.0**
Procedures paid by private insurance, partially out‐of‐pocket, and/or by other funds (not Medicaid) (%)	3.2	28.4	n/a
Age (mean)	26.8	27.2	0.4**
White %	79.9	83.4	3.6**
People of color %	20.3	16.6	−3.6**
Hispanic %	0.9	1	0.1
Out of state a %	13.2	16.2	3.0**
Annual income (mean)	NA	NA	NA

Outcomes
	Before policy shift	After policy shift	Difference
Average monthly procedure volume
Maine	156.7	167.7	11.0
Illinois	1379.8	1756.2	376.4**
West Virginia	116.5	87.9	(−)29.1**
Share of total price paid by patient out‐of‐pocket (patient price)
Maine	52.3%	33.3%	(−)19.0**
Illinois	28.9%	16.2%	(−)12.7**
West Virginia	38.6%	88.8%	50.2**

Note: Denominators for the descriptive statistics in this table are the total number of procedures occurring in each state either before (“before policy shift”) or after (“after policy shift”) the policy shift. For example, in the first column, “Procedures funded by Medicaid (%)” is the share of procedures occurring before the policy shift where clinic administrative data indicated at least partial payment of the procedure by Medicaid, divided by the total number of procedures in that state that occurred before the policy shift. Similarly, “White %,” in the first column, is the share of procedures occurring before the policy shift where clinic administrative data indicated that the patient was white, divided by the total number of procedures. It was not applicable to run statistical tests of difference for “Procedures funded by Medicaid (%)” because the nature of the policy shift dictated whether state Medicaid was available to pay for procedures before the shift or after. It was not applicable to run statistical tests of difference for “Procedures paid by private insurance, partially out‐of‐pocket, and/or by other funds (not Medicaid) (%)” given the variability and combination of payer sources that funded this category of procedures, the lack of standard reporting of these payments across clinics and states, and resultant lack of interpretability of the change metric.

Abbreviations: NA, not available; n/a, not applicable.

^* p < 0.05;

^** p < 0.01.

^a Percentage of total procedures that were obtained by patients who live out‐of‐state (outside Maine, Illinois and West Virginia, respectively).

Table 2 displays the coefficients for the ITS analyses for procedure volume in each state overall as well as procedure volume ITS coefficients by subgroup (race and financial need) for Illinois and Maine. ^‡‡‡ Table 3 contains ITS coefficients for patient price in each state overall. The ITS analyses assess (a) the significance and directionality of the trend in the outcome before the policy shift, (b) the magnitude and significance of any immediate change ^§§§ in the level of the outcome following the policy shift, and (c) the significance and directionality of any change in trend after the policy shift relative to the post‐shift trend predicted by the pre‐shift trend in the absence of the policy shift. It is important to note that most of the post‐policy analysis time period in Maine (see Figure 1 and Table 1) occurred during the peak months of the COVID‐19 pandemic, as its policy shift took effect a few months before the first COVID‐19 cases were identified in the U.S. Further, any decreases observed throughout 2020 and 2021 in monthly procedure volume across the three states (panel a, Figures 2, 4, and 5) should be interpreted in the context of the COVID‐19 pandemic and the restrictions placed on health care access, transportation, social distancing, and the financial effect on individuals and households.

TABLE 2.

ITS coefficients for procedure volume (all states).

	ITS coefficients: monthly procedure volume
	Trend before policy shift	Immediate change at policy start	Change in trend after policy shift
Maine	−0.2	42.0**	−3.0
Illinois	−14.8*	413.9**	19.6**
West Virginia	0.7*	−28.5*	−1.8**

	White			People of color
	Trend before policy shift	Immediate change at policy start	Change in trend after policy shift	Trend before policy shift	Immediate change at policy start	Change in trend after policy shift
Maine	−0.2	40.2**	−2.7	−0.2	42.1**	−2.2
Illinois	−4.3	62.8*	5.6	−10.1**	327.7**	16.1**

	No financial need			Financial need
	Trend before policy shift	Immediate change at policy start	Change in trend after policy shift	Trend before policy shift	Immediate change at policy start	Change in trend after policy shift
Maine	−0.2	42.3**	−2.2	−0.2	42.6**	−3.3
Illinois	−7.1	189.2**	4.7	−4.5	115.8*	14.2**

Note: Heterogeneity analyses for West Virginia were not conducted due to incomplete demographic data for this state. The Immediate change at policy start refers to the first full calendar month nearest to the implementation date of the policy. The implementation date was specific to the implementing state. In Illinois, where the policy change took effect on January 1, 2018, the first full month post‐policy was defined as January 2018. In Maine, where the policy change took effect closer to the end of a month (i.e., September 19, 2019), the first full month post‐policy was defined as October 2019. In West Virginia, where the policy change was both passed and put into immediate effect toward the beginning of a month (i.e., November 6, 2018), the first full month post‐policy was defined as November 2018.

^* p < 0.05;

^** p < 0.01.

TABLE 3.

ITS coefficients for patient price (all states).

	ITS coefficients: share of total price paid by the patients out‐of‐pocket
	Trend before policy shift	Immediate change at policy start	Change in trend after policy shift
Maine	0.00	−13.14**	−0.55**
Illinois	−0.34**	−9.23**	0.31**
West Virginia	0.07	48.6**	0.00

^* p < 0.05;

^** p < 0.01.

FIGURE 2.

Trend in Illinois procedure volume. This figure includes three panels: Panel (a) shows the overall monthly trend in procedure volume in Illinois; panel (b) shows the monthly trend in procedure volume in Illinois by race; panel (c) shows the monthly trend in procedure volume in Illinois by financial need. The vertical red line in every graph indicates the time of the policy shift.

FIGURE 4.

Trend in Maine procedure volume. This figure includes three panels: panel (a) shows the overall monthly trend in procedure volume in Maine; panel (b) shows the monthly trend in procedure volume in Maine by race; panel (c) shows the monthly trend in procedure volume in Maine by financial need. The vertical red line in every graph indicates the time of the policy shift.

FIGURE 5.

Trend in West Virginia procedure volume. This figure shows the overall monthly trend in procedure volume in West Virginia. The vertical red line indicates the time of the policy shift. While Figures 2 and 4 include additional panels displaying the monthly trend in procedure volume by race and financial need in Illinois and Maine, respectively, we did not conduct heterogeneity analysis for West Virginia due to incomplete demographic data for this state.

Illinois

Monthly procedure volume

In Illinois, the average monthly procedure volume increased from approximately 1380 (1379.8) before the policy shift to 1756 (1756.2) after the policy shift (Table 1). This increase in monthly average of approximately 376 procedures is statistically significant at a 99% confidence level (p < 0.01). In the period before the policy shift, the trend in monthly procedure volume was decreasing by around 15 procedures per month (Table 2 and panel a, Figure 2). This downward trend in monthly procedure volume reversed after the policy shift, increasing significantly, at a rate of approximately 20 procedures per month, following an initial increase in average volume of approximately 414 procedures at the time of the policy shift.

Even though procedure volume in Illinois increased among both white individuals and people of color, panel b in Figure 2 shows that the increase in overall procedure volume was driven mainly by the increase among people of color. Among people of color, the number of monthly procedures was decreasing before the policy shift by around 10 procedures per month. This downward trend reversed in the post‐policy period by about 16 procedures per month, following an initial increase of approximately 328 monthly procedures at the time of the policy shift.

Similarly, the overall increase in monthly procedure volume in Illinois seems to have been driven by the volume increase among people with financial need (panel c, Figure 2). In the post‐policy period, monthly procedure volume among people with financial need was increasing by 14 procedures per month compared to the pre‐policy period, while no such change in the trend existed among people who were not in financial need. Both groups, nonetheless, experienced a statistically significant increase in procedure volume around the time of the policy shift.

Patient price

The share of total price paid by the patient (patient price) in Illinois declined significantly (p < 0.01), by 13 percentage points (see Table 1) after the implementation of state Medicaid coverage. Before the policy shift, patients paid out‐of‐pocket about 29% of the total procedure price, on average. After the policy shift, they paid about 16%. In Illinois, the median price, in dollars, paid by the patient out‐of‐pocket prior to the policy shift was $470; after the policy shift, the median price paid by the patient out‐of‐pocket decreased to $60—a change of $410, which is mainly due to the increase in the number of procedures funded by Medicaid (see Table 1). These gains were offset by a small but significant increase of $32 (see Table 1) in average total procedure price.

As shown in the ITS results (Table 3 and panel a, Figure 3), patient price in Illinois was declining before the policy shift and further declined significantly by 9 percentage points at the time of the policy shift compared to what it most likely would have been had the state not enacted the policy. Although the average patient price continued to decrease after the policy shift (visible in the negative slope of the post‐policy trendline in panel a, Figure 3), the decline was not as steep as the one observed in the pre‐policy shift trendline; as indicated by the positive ITS estimate for the trend post‐policy (0.3 percentage point), the downward trend before the policy shift in Illinois did not continue at the same pace (see Table 3).

FIGURE 3.

Trend in patient price (all states). This figure includes three panels: panel (a) shows the trend in average patient price (the share of the total price paid by the patient out‐of‐pocket) in Illinois; panel (b) shows the trend in average patient price in Maine; panel (c) shows the trend in average patient price in West Virginia. The vertical red line in every graph indicates the time of the policy shift. In Illinois and Maine (panels a and b) the policy shift was the state's coverage of abortion under state Medicaid; in West Virginia (panel c) the policy shift was the discontinuation of allowing state Medicaid funds to cover abortion.

Maine

Monthly procedure volume

The average number of abortion procedures per month in Maine increased from approximately 157 (156.7) to 168 (167.7) after the policy shift (Table 1), however this increase of an average of 11 procedures per month is not statistically significant at a 95% confidence level (p = 0.057). Before the policy shift, the trend in monthly procedure volume in Maine was flat, as shown in the linear trendline in panel a, Figure 4 (and the non‐significant point estimate of −0.2 in Table 2). At the time of the policy shift, there was an immediate and statistically significant increase in monthly procedure volume of 42 procedures (Table 2 and panel a, Figure 4). However, monthly procedure volume did not continue at this increased level during the post‐policy analysis period. Instead, there was a slight and statistically insignificant downward trend in the number of monthly procedures, resulting in a mean decrease of 3 procedures per month compared to the trend observed before the policy went into effect.

The heterogeneity visible in the monthly procedure volume results for Illinois by race and financial need is not present in the Maine data (Tables 2 and 3 and panels b and c, Figure 4). As shown in Table 2, there is a similar statistically significant and immediate increase of about 40–42 procedures per month at the time of policy implementation across all Maine subgroups (i.e., white, people of color, people with financial need, people without financial need). The change in trend of monthly procedure volume after the policy shift, compared to before the policy shift, was not significant at the p < 0.05 significance level for any of the Maine subgroups.

Patient price

As in Illinois, the patient price, or share of the total procedure price paid by the patient, in Maine declined significantly (p < 0.01) by 19 percentage points after the policy shift (Table 1). Whereas before the policy shift, patients were paying roughly half the price of their abortions out‐of‐pocket, after it, patients paid about a third of the total price. Unlike Illinois, in Maine, total procedure price decreased together with patient price, producing a benefit for patients by both measures. Specifically, in Maine, the median price paid by the patient out‐of‐pocket, in dollars, prior to the policy shift was $400; after the policy shift, this amount decreased to $200—a change of $200. The increase in the number of procedures funded by Medicaid (see Table 1) was mainly responsible for the decrease in the median price paid by the patient out‐of‐pocket. This benefit to the patient was additive on top of a significant decrease of $68 in average total procedure price following the policy shift ^**** (see Table 1).

As noted in the ITS findings (Table 3), at the time of the policy shift, patient price immediately declined significantly (by 13 percentage points) and the trend continued to decline (panel b, Figure 3). Linear trends in average patient price before and after the policy shift show this statistically significant decrease in the share of total price paid out‐of‐pocket by patients, followed by a declining rate of approximately 0.5 percentage point per month through the end of the post‐policy analysis period, a trend that differed significantly from the expected trend had the policy shift not occurred.

West Virginia

Monthly procedure volume

In contrast to Maine and Illinois, West Virginia, which discontinued state Medicaid funding for abortion care, experienced a statistically significant (p < 0.01) decrease in monthly procedure volume from an average of approximately 117 procedures per month before the policy shift to a monthly average of approximately 88 after the policy shift (see Table 1). As noted in the ITS findings (Table 2) and seen in Figure 5, West Virginia experienced a significant, increasing trend in monthly procedure volume, which was halted at the time of the policy shift by an immediate and significant drop in average monthly procedure volume of approximately 29 procedures. The trend resumed its upward trajectory in the 3‐year period after the policy shift, but the pace of increase—approximately 2 fewer procedures per month (−1.8)—declined compared to the expected value in the absence of the policy shift (see Table 2).

Patient price

While the total procedure price in West Virginia remained around $400 both before and after the policy shift (see Table 1), patient price increased significantly after West Virginia discontinued Medicaid coverage for abortion. On average, in the pre‐policy period, patients contributed out‐of‐pocket about 39% of the total price of their abortion (Table 1 and panel c, Figure 3). Immediately after the policy shift, the mean share of their out‐of‐pocket costs increased over 50 percentage points, resulting in patients paying, on average, nearly 89% of the total price of an abortion after the policy shift. Panel c, Figure 3 displays the linear trends in patient price before and after the policy shift. In dollars, prior to the policy shift, the median price paid by the patient out‐of‐pocket was $0; after the state discontinued Medicaid coverage of abortion, the median price paid by the patient out‐of‐pocket escalated to $350. This change reflects the absence of any procedures funded by Medicaid after the policy shift (see Table 1).

As shown in the ITS coefficients (Table 3), the trend in patient price was flat before the policy shift (0.07) and remained flat after the policy shift (0.00), but as shown in panel c of Figure 3, at a much higher level. The large and immediate percentage point surge at the time of the policy shift significantly increased (p < 0.01) the financial burden for patients as state Medicaid coverage ended, with patient price remaining at that increased level for the remainder of the study.

DISCUSSION

Our study indicates that implementing state Medicaid coverage of abortion contributed to increased access to care in both Maine and Illinois, as measured by changes in procedure volume. There is also initial evidence from changes observed in the composition of demographic characteristics of pregnant people who obtained procedures before and after the policy shifts in these states that access improved even more than the statewide average among people with financial need in Illinois and people of color in both states. In contrast, in West Virginia, the share of abortion procedures obtained by people of color significantly decreased following the state's discontinuation of abortion as a state Medicaid‐covered service. Our findings are consistent with prior literature that has noted structural inequities affecting abortion access ²⁰ among people of color and people who are economically marginalized, who face especially burdensome financial barriers when unable to use Medicaid to cover an abortion procedure.^{12, 21} Our findings are also relevant considering the growing body of evidence that suggests an association between abortion‐restrictive policies (e.g., prohibiting state Medicaid from covering abortion, having licensed physician requirements or mandatory waiting periods) and negative maternal and infant health outcomes.^{22, 23, 24}

The regression analyses found noteworthy trends in abortion procedures overall, and among different subgroups in Illinois, controlling for other personal and state factors. Although procedure volume increased significantly at the time of the policy shift for white people and people of color in Illinois and Maine, post‐shift trendlines diverged for these demographic groups in Illinois (panel b, Figure 2), suggesting that Medicaid funding of abortion care may have helped to address a previously unmet need for abortion care among people of color in Illinois. In Maine, the close temporal proximity of the policy shift to the emergence of COVID‐19 in the U.S., combined with the relatively short post‐policy analysis period, did not permit many contiguous months in which to observe whether procedure volumes in Maine regulated as COVID‐19 restrictions lifted, or whether time would reveal evidence of an access gap like that found in Illinois.

In both states, there was some evidence of converging trendlines among people with different levels of financial need. In Illinois, this took the form of a steadily increasing trend in procedure volume after the policy shift for people with financial need until early 2020, when, around the time of COVID‐19's arrival and related pandemic shut‐downs, the trendline began to merge with that of people who did not need financial assistance (panel c, Figure 2). In Maine, the variability observed from month to month among people with financial need and those who did not need financial assistance, as well as the difference in average monthly procedure volumes between these two groups, narrowed after the policy shift, with the trendlines for both groups merging fairly consistently for the remainder of the post‐policy analysis period (panel c, Figure 4). This convergence may be evidence of the policy shift in Maine having achieved the alignment of abortion care access across people with different financial needs. For Illinois, the convergence may suggest that the contextual effects of COVID‐19 reduced access disparities among people with financial need and those who did not need financial assistance, that is, abortion access challenges historically exacerbated for people with financial need became a more universal experience, regardless of financial need, during COVID‐19.

Our findings show a significant reduction in patient price after state Medicaid coverage of abortion was implemented. Specifically, following the policy shift, the share of total price paid by the patient out‐of‐pocket (i.e., patient price) decreased, on average, by 36% in Maine and 44% in Illinois. After the shift, patients paid out‐of‐pocket, on average, only one‐sixth (Illinois) to one‐third (Maine) of the price of an abortion procedure, compared to the expected percentages without the policy shifts. In contrast, in West Virginia, following the discontinuation of state Medicaid coverage for abortion care, access decreased as procedure volume fell, and patients faced significantly higher out‐of‐pocket costs (as a proportion of the total price). Average patient price increased by 130% in West Virginia, leaving patients responsible, on average, for nearly 89% of the total price after abortion coverage was discontinued.

Better understanding the effects of Medicaid coverage on patient price and procedure access helps inform decisions made by policymakers and other stakeholders in states considering similar policy actions. Existing literature has shown that financial assistance for abortion through state Medicaid programs enables patients to reallocate scarce resources to other necessities (such as food or rent) or cover additional abortion‐related costs, like travel and lodging. ¹¹ Similarly, initiating state Medicaid coverage of abortion can spur local abortion funds to support more patients in new ways by re‐directing funds that previously were needed to pay for procedures that are now covered by Medicaid. With a change in median patient out‐of‐pocket costs of approximately $400 in Illinois and $200 in Maine, our findings suggest that many patients in Illinois and Maine may have found greater freedom and fewer economic constraints to finance additional costs when seeking care after the policy shift. Conversely, the discontinuation of state Medicaid coverage in West Virginia suggests that patients who would previously have been able to afford an abortion, could not obtain one without new financial support from another source. The need to raise additional funds from other sources may result in delayed access to care and abortions taking place at a later gestational age, which increases the cost, the risk of clinical complications, or may even make patients forgo a needed procedure. Pregnant people whose abortions are delayed or denied have been shown to experience increased economic insecurity and household poverty ²⁵ ; a higher likelihood of staying in contact with a violent partner ²⁶ ; and more life‐threatening complications like eclampsia and postpartum hemorrhage. ²⁷

Analysis choices and considerations

Transitional funding

As noted in the Introduction and Figure 1, low state Medicaid reimbursement rates and administrative challenges in Illinois resulted in a period of transitional funding between January 2018 and June 2020. Transitional funding was a direct response to the policy shift and served as a proxy for state Medicaid funding while clinics and the state navigated the new payment environment. In our analysis, we treated the start of transitional funding as the beginning of the policy shift that allowed state Medicaid coverage of abortion. This approach gave us a longer period to explore how more financial assistance for abortion care affected abortion access and cost. As a robustness check, we conducted additional analyses in which we treated transitional funding and the later availability of state Medicaid funds as separate interventions. The results of these robustness checks did not provide further insight into the effect of the policy shift.

COVID‐19

While our regression analyses controlled for monthly average COVID‐19 cases and deaths, these were likely insufficient controls, given the pervasiveness of the pandemic. These indicators, however, were an attempt to acknowledge COVID‐19's potential range of impacts on abortion care, not just on availability of health care services, but also on personal finances due to lost wages, lost jobs, lost private insurance coverage; as well as impacts on available personal time (e.g., due to caring for children in the home due to schools shifting to remote learning). In short, we selected COVID‐19 controls that featured prevalence as a proxy of the risk of experiencing any of these impacts.

As noted above, a larger issue for our study was that the Maine policy (LD 860) had not been in effect for long before the arrival of COVID‐19 in the U.S.; thus, its effect on outcomes are confounded by COVID‐19. Decreases in procedure volume in 2020 are evident in each state (Table 3; panel a, Figures 2 and 4; and Figure 5) and should be considered in light of the many general restrictions on health care access during this period.

LIMITATIONS

Clinics invited to participate in the study were either standalone clinics known to provide abortion services during the study period or health centers with a focus on providing abortion and other reproductive health services. Although the census of known clinics of these types that provided abortion services during the study period in Maine and West Virginia participated in the study, seven Illinois clinics that were invited to participate either declined (two clinics), did not respond (four clinics), or did not follow through after initial engagement (one clinic) and were dropped from the study. We estimate that the omitted clinics accounted for 11% of abortions conducted in Illinois in 2020. These clinics serve geographic areas and patients similar to the clinics included in the study. However, there may be nuances related to lower volume and perhaps smaller clinics' experience of the policy shift that we could not capture. For example, one of the omitted facilities closed since the policy shift, and another experienced an administrative change around the time of the policy shift that rendered the facility temporarily unable to perform abortions. ^†††† These circumstances could have affected the usefulness of clinic data had we been able to obtain them from these two clinics.

In Maine, the passage of LD 1261, concurrently enacted with LD 820 in September 2019, allowed certain APCs to perform medication and instrumentation abortions. It is difficult to isolate the effect of this policy, confounding the impact of state Medicaid coverage of abortion care on outcomes.

Finally, while we were able to examine aggregate changes in the demographic characteristics of patients who obtained abortions both before and after the policy shift in Maine and Illinois, we did not have access to income data in West Virginia and race was often missing. Furthermore, analysis of race and income data in Maine and Illinois did not include the full set of clinics that contributed clinic administrative data, because one clinic in Maine could provide neither race nor income data, and one clinic in Illinois could not provide income data. Further, we aimed to limit burden on participating clinics. The lack of standard collection of financial indicators across clinics led us to rely heavily on payment data to determine Medicaid status and financial need, thus we could not follow a set of Medicaid beneficiaries both before and after the respective policy shifts in each state.

Heil SKR, Caglayan K, Castillo G, et al. The impact of state Medicaid coverage of abortion on people accessing care in three states. Perspect Sex Reprod Health. 2024;56(3):255‐268. doi: 10.1111/psrh.12275