Impact of Medicaid Expansion and State-Level Racial Diversity on Breast Cancer Endocrine Therapy Prescriptions: A quasi-experimental, comparative interrupted time series study

Abstract

Aims:

To determine whether Medicaid expansion impacted racially more diverse states similarly as racially less diverse states in endocrine therapy (ET) prescriptions.

Methods:

A quasi-experimental, comparative interrupted time series study of Medicaid-financed ET prescriptions from 2011–2018 Medicaid State Drug Utilization Database. The exposures were state’s Medicaid expansion and racial diversity status. The outcome was state’s quarterly number ET prescriptions per 100,000 non-elderly adult females (NAFs).

Results:

During the year of expansion, ET prescriptions increased sharply in expansion states but remained flat in nonexpansion states (slope: 11.96 vs. 0.43 prescriptions per 100,000 NAFs per quarter, p<0.001). After that, the slopes were similar between expansion and nonexpansion states (1.75 vs. 0.24, p=0.057) but the level of prescriptions in expansion states maintained at a higher level. When stratified by state’s racial diversity status, the slope of increase in the first year was sharper for racially more diverse expansion states (16.49, p=0.008) than racially less diverse expansion states (8.46, p<0.001), resulting in significant differences in ET prescriptions between racially more diverse expansion and nonexpansion states but largely nonsignificant differences between racially less diverse expansion and nonexpansion states.

Conclusions:

Although Medicaid expansion significantly increased ET prescriptions in expansion vs. nonexpansion states, this difference was only observed among racially more diverse states. Racially more diverse nonexpansion states had the lowest rates of ET prescriptions and the gaps from racially more diverse expansion states significantly widened after expansion.

Policy Summary:

Our study shows that, before expansion, racially more diverse nonexpansion states had the lowest rates of ET prescriptions. After expansion, the gaps between these states and racially more diverse expansion states significantly widened. These results highlighted the importance of continuing to examine the health impacts of states not expanding Medicaid, including the health equity impacts for low income racial/ethnic minority populations with cancer and other life-threatening diseases.

Background

Breast cancer is the most common noncutaneous cancer among women in the United States and the leading cause of cancer death in females second only to lung cancer.[1] Although the overall incidence rate is highest in non-Hispanic White (NHW) women in the US, among women aged < 45 years, breast cancer incidence was highest among non-Hispanic Black (NHB) women.[2,3] NHB women also have the highest mortality rate from breast cancer among all US racial/ethnic groups.[1] Despite advances in prevention and treatment, disparity in mortality between NHW and NHB women persists.[4] While there are biological differences across racial groups, they are not likely to fully explain this variation in breast cancer mortality.[5,6]

Because most breast cancers are hormone (estrogen and/or progestogen) receptor positive (HR+),[7] adjuvant endocrine therapy (ET) is used to prevent recurrence.[8] The most commonly used ETs are tamoxifen and aromatase inhibitors (AI: anastrozole, exemestane, letrozole). Using them for 5–10 years post-surgery is recommended and can reduce recurrence and death from breast cancer.[8] Despite the known efficacy, previous studies have found underuse of adjuvant ET, including noninitiation (i.e. never starting), nonadherence (i.e. not taking as prescribed), and nonpersistence (i.e. early discontinuation), with only 50% of patients completing ET as prescribed over year 5, which has been associated with shorter time to recurrence, lower quality of life, and increased costs.[9] Underuse of ET was found to be disproportionately higher among racial/ethnic minority populations, which may contribute to disparities in breast cancer mortality.[10]

Medicaid is the primary health insurer for low-income nonelderly people in the U.S. Prior to the Affordable Care Act (ACA), Medicaid eligibility for nonelderly adults in most states was limited to people with disabilities, pregnant women, and low-income parents. Women with cancer can also gain Medicaid coverage if diagnosed under the Centers for Disease Control and Prevention’s National Breast and Cervical Cancer Early Detection Program (NBCCEDP). Medicaid is an important source of coverage for people with cancer with 1 in 7 nonelderly women with breast cancer covered by Medicaid from 2011–2014.[11]

The ACA expanded Medicaid eligibility to nonelderly adults with income <138% of the federal poverty level. However, Medicaid expansion is optional to states due to a 2012 Supreme Court decision.[12] As of January 2020, 37 states (including DC) had expanded their Medicaid programs under the ACA. Previous studies have found that nonexpansion states are more likely to be racially more diverse and more Black.[13] According to the Kaiser Family Foundation, of the uninsured adults who could gain coverage through expansion in the remaining non-expansion states, 60% are people of color (Hispanics, 29%; NHBs, 23%) and 96% of them resides in the South region.[14] Women with breast cancer may be at elevated risk of not receiving high-quality care in more racially diverse states that have not expanded Medicaid. Figure 1 illustrates the racial diversity and breast cancer mortality rate of each state by color, with Medicaid expansion status indicated by overlaid lines or dots. A group of eight states in the southern half of the U.S., in dark blue on the map, are categorized as having both high breast cancer mortality (above national average) and being more racially diverse.[15] Of these states, only New Mexico and Louisiana have expanded Medicaid by 2020 (Table 1).

Figure 1: State’s racial diversity, breast cancer mortality rate, and Medicaid expansion status (states with no stripes or dots did not expand Medicaid prior to 2018)*.

* State’s racial diversity was estimated using the average racial diversity index (RDI) of each state across the 8 years of the study (2011–2018) and used the median of the average RDIs of all states to classify the states as racially more (>median) or less (<=median) diverse states. Breast cancer mortality was from the Centers for Disease Control and Prevention, National Center for Health Statistics. Underlying Cause of Death 1999–2019 on [CDC WONDER Online Database](http://wonder.cdc.gov/), released 2021. Data are from the Multiple Cause of Death Files, 1999–2019, as compiled from data provided by the 57 vital statistics jurisdictions through the Vital Statistics Cooperative Program. Accessed at http://wonder.cdc.gov/ucd-icd10.html on February 26, 2021.

Table 1.

States’ Racial Diversity Status by Expansion Status

Expansion			Nonelderly Adult Females	Non-expansion			Nonelderly Adult Females
State	Expanded by 2018	Expansion Date	Racial More Diverse	State	Expanded by 2018	Expansion Date	Racially More Diverse
AK	1	9/1/2015	1	AL	0	none	1
AR	1	1/1/2014	0	FL	0	none	1
AZ	1	1/1/2014	1	GA	0	none	1
CA	1	1/1/2014	1	ID	0	1/1/2020	0
CO	1	1/1/2014	0	KS	0	none	0
CT	1	1/1/2014	1	ME	0	1/10/2019	0
HI	1	1/1/2014	1	MO	0	none	0
IA	1	1/1/2014	0	MS	0	none	1
IL	1	1/1/2014	1	NC	0	none	1
IN	1	2/1/2015	0	NE	0	10/1/2020	0
KY	1	1/1/2014	0	OK	0	none	1
LA	1	7/1/2016	1	SC	0	none	1
MD	1	1/1/2014	1	SD	0	none	0
MI	1	4/1/2014	0	TN	0	none	0
MN	1	1/1/2014	0	TX	0	none	1
MT	1	1/1/2016	0	UT	0	1/1/2020	0
ND	1	1/1/2014	0	VA	0	1/1/2019	1
NH	1	8/15/2014	0	WI	0	none	0
NJ	1	1/1/2014	1	WY	0	none	0
NM	1	1/1/2014	1
NV	1	1/1/2014	1
OH	1	1/1/2014	0
OR	1	1/1/2014	0
PA	1	1/1/2015	0
RI	1	1/1/2014	0
WA	1	1/1/2014	1
WV	1	1/1/2014	0
Total	27		12	Total	19		9
% of All expansion states			44%	% of All nonexpansion states			47%

A recent systematic review of studies examining the impact of Medicaid expansion across the cancer continuum found that Medicaid expansion has increased insurance coverage among cancer patients and survivors, improved access to screening and preventative care, and shifted diagnoses towards earlier stage.[16] However, few studies have examined Medicaid expansion’s impact on cancer treatment except cancer-directed surgeries and even fewer studies have assessed its impact on racial disparities in cancer treatment.[16] To the best of our knowledge, only one study has examined Medicaid expansion’s impact on ET use; this study reported an increase in Medicaid-financed ET prescriptions after expansion.[11] Building on this study, we examined the differences in Medicaid-financed ET prescriptions, comparing states that expanded Medicaid to those that have not by the end of 2018, stratified by the racial diversity status of each state. Since more previously-uninsured minority racial/ethnic populations gained coverage under the ACA than NHWs and the gains were larger in Medicaid expansion states,[17] we hypothesized that expansion states that are racially more diverse may see bigger impacts on increasing ET prescriptions following expansion than do states that are less racially diverse.

Methods

Data Source

ET prescription data were from Medicaid State Drug Utilization Database (SDUD), which includes all filled outpatient prescriptions purchased online or through retail pharmacies and covered under the Medicaid Drug Rebate Program financed by Medicaid.[18] The SDUD has been used in previous studies of ACA-Medicaid expansion effects on prescription fills.[11,19–24] For this study we used data from quarter 1 (Q1) of 2011 to Q4 of 2018. We excluded years before 2011 because Medicaid managed care plans were added in March 2010 and many states transitioned in 2010 to including managed care Medicaid during that year. Four states (DE, MA, NY and VT) and DC were excluded from the analysis because they expanded Medicaid prior to 2011.[11] Expansion (n=27) and nonexpansion (n=19) states were classified by whether a state expanded Medicaid by the end of 2018 (Table 1). To account for differences in population size, we estimated the number of nonelderly adult females (NAFs) in each state for each year using the Annual Social and Economic Supplement (ASES) of the Current Population Survey. The ASES was also used to estimate other state-level characteristics of NAFs as well as NAF Medicaid enrollees (percentages (i.e. 0–100) of populations in different race/ethnicity groups [Hispanic, non-Hispanic white (NHW), non-Hispanic black (NHB), non-Hispanic Asian (NHA) and other races including individuals of multiple races], age under 40, full-time unemployed, full-time students, living in metropolitan areas, and of foreign birth). Due to a redesign of the processing system for the ASES in 2019, we did not include 2019’s data in our analysis.[25]

Outcome Measures

The outcome measure was state’s quarterly number of Medicaid-financed ET prescriptions per 100,000 NAFs. ETs included tamoxifen, anastrozole, exemestane, letrozole, toremifene, and fulvestrant.

State’s Racial Diversity Status

We developed a racial diversity index score (RDI) of each state’s NAF population. We did not use racial and ethnic composition of the NAF Medicaid enrollees due to concerns of its endogeneity to expansion status because Medicaid enrollment changed after expansion. The RDI is based on the Simpson’s Index of Diversity that has been used to measure racial diversity in past literature.[26,27] The index is calculated as following:

$$RDI=1-\%Hispani{c}^2-\%NH{W}^2-\%NH{B}^2-\%NH{A}^2-\%OtherRace{s}^2$$

It ranges from 0 to 1, with higher values denoting greater racial diversity. Since the state-level racial diversity of NAFs did not change significantly during the study period, to reduce estimation variation, we calculated the average RDI of each state across the 8 years of the study and used the median of the average RDIs of all states to define the more (>median) or less (<=median) racially diverse states.

Trend Analysis

We used comparative interrupted time series (CITS) analysis to determine the trend change in ET prescriptions. Similar to standard ITS, CITS compares outcomes before vs. after policy change (i.e. Medicaid expansion) accounting for time trends; unlike ITS, it further evaluates whether the policy impact in the treatment group (i.e. expansion states) deviated from the comparison group (i.e. nonexpansion states).[28–30] To enable comparison of differences in the levels and trends pre- and post-expansion, all nonexpansion states were assigned 2014 Q1 as the expansion quarter since most states expanded Medicaid in that quarter. A continuous time variable in the units of quarters were defined centered at the expansion. We excluded quarters beyond 12 quarters before and 20 quarters (including the expansion quarter) after expansion because states with data in those quarters were fewer and estimates from these quarters may not be reliable. Observation of the raw data indicated a different slope within the first year of implementation. Therefore, we estimated a different trend in the first year post expansion. The model adjusted for baseline state-level characteristics averaged over the three years prior to the “expansion quarter” and indicators of Q2-Q4 (Q1 as the reference) to account for seasonal variations (see Supplement-Methods for details).

We conducted this CITS analysis using all states to estimate the overall expansion effect. We then carried out separate analysis by state-level racial diversity status to determine the differential impact of ACA’s Medicaid expansion among states that were more diverse and those that were less diverse.

Adjusted mean number of ET prescriptions

CITS assumes linear trends before and after expansion. To relax the linear trend assumption, we compared ET prescriptions between expansion and nonexpansion states at each time point by racial diversity status adjusting for baseline state-level population characteristics prior to expansion as following:

$$\begin{gathered} +\sum _{t=1}^T{\tau }_t*Tim{e}_t+\sum _{t=1}^T{\alpha }_q*\left(Tim{e}_t*Expansio{n}_i\right)+\sum _{t=1}^T{\beta }_q*(Tim{e}_t*Rac{e}_{Diversit{y}_i}) \\ +\sum _{t=1}^T{\gamma }_q*(Tim{e}_t*Expansio{n}_i*Rac{e}_{Diversit{y}_i})+\sum _{k=1}^K{\varnothing }_k*X_k+{\epsilon }_{it} \end{gathered}$$

Detailed description is included in Supplement-Methods. From this model, we predicted the adjusted mean ET prescriptions by expansion status for each quarter and plotted these differences (and their 95% CIs) for all states as well as stratifying by state’s racial diversity status.

Sensitivity Analysis

We restricted analyses to states that expanded Medicaid in 2014 Q1, which were the majority of expansion states (n=21/27) and compared them to the non-expansion states. This created a completely balanced panel with all states having equal number of observations.

All analysis used robust standard errors adjusting for clustering of observations from the same state. Statistical analyses were performed using SAS, version 9.4 (SAS Institute Inc, Cary, NC) and STATA version 16.1 (Stata Corp, College Station, TX). Since both SDUD and ASES data are de-identified and publicly available, no IRB approval was needed.

Results

Population Characteristics

Of the 46 states included in our analysis, 21 states had a racially more diverse NAF population and 25 states had a racially less diverse NAF population, compared to the median RDI of all states across the 8 years (Table 1). Expansion and nonexpansion states were generally comparable in baseline state-level characteristics of NAFs, although expansion states had lower percentage of NHBs but higher percentage of Hispanics and NHAs (differences non-significant) (Table 2). Stratified analysis by racial diversity group showed few statistically significant differences by Medicaid expansion status within each group.

Table 2.

Baseline Socio-Demographic Characteristics of States’ Nonelderly Adult Females in 3 Years Prior to Expansion

	All States			Racially Less Diverse States			Racially More Diverse States
	Nonexpansion	Expansion		Nonexpansion	Expansion		Nonexpansion	Expansion
	3-year before			3-year before			3-year before
	(N=228)	(N=324)		(N=120)	(N=180)		(N=108)	(N=144)
	Mean %	Mean %	P*	Mean %	Mean %	P*	Mean %	Mean %	P*
Age
Under 40	46.5	45.7	0.276	47.1	45.2	0.167	45.9	46.2	0.733
Race/Ethnicity			0.063			0.549			0.321
Hispanic	8.2	10.9		6.1	5.2		10.4	18.0
Non-Hispanic White	72.6	70.6		83.8	83.2		60.1	54.9
Non-Hispanic Black	13.6	8.4		5.0	6.5		23.2	10.8
Non-Hispanic Asian All Other/Multiple	2.4	5.6		2.0	2.5		2.8	9.5
Races	3.2	4.5		3.1	2.7		3.4	6.8
Education			0.757			0.776			0.238
No High School Degree/Diploma	9.8	9.3		8.4	8.5		11.4	10.3
High School	27.5	27.4		27.0	28.2		28.1	26.3
College or Above	62.6	63.3		64.6	63.3		60.5	63.4
Other
Metro	69.8	77.9	0.104	64.1	70.2	0.308	76.2	87.5	0.054
Full time unemployed	34.3	33.6	0.660	31.3	32.5	0.600	37.7	35.0	0.102
Full time Students	7.4	7.6	0.458	7.5	7.6	0.903	7.3	7.7	0.360
Born in US	90.4	86.2	0.091	92.3	91.9	0.771	88.2	79.1	0.046

^*P-values of likelihood tests from logistic regression models regressing expansion status each variable; standard errors were adjusted for clustering within states

Changes in Level and Rate (Slope) of ET Prescriptions following Medicaid Expansion

Table 3 reports the CITS analysis. Prior to expansion, there were no difference in the level/slope of ET prescriptions between expansion and nonexpansion states. During the year of expansion, expansion states had a positive slope in ET prescriptions while the slope of nonexpansion states remained flat, resulting in a significant difference (11.96 vs. 0.43 prescriptions per 100,000 NAFs per quarter, p<0.001). However, after the first year, the slope of ET prescriptions in expansion states largely returned to its pre-expansion level such that its difference from the nonexpansion states was no longer statistically significant (1.75 vs. 0.24, p=0.057).

Table 3.

Trend Analysis Using Comparative Interrupted Times Series Analysis ^*

Quarterly ET prescriptions per 100,000 NAFs	All States				Racially Less Diverse States				Racially More Diverse States
	Coef.	95% CI		P	Coef.	95% CI		P	Coef.	95% CI		P
t	0.86	−0.41	2.12	0.182	−0.14	−1.77	1.49	0.865	1.97	0.11	3.83	0.039
z	−10.81	−37.82	16.19	0.424	−52.11	−88.77	−15.45	0.007	35.74	4.88	66.60	0.025
z_t	0.92	−0.82	2.67	0.292	2.29	−0.32	4.89	0.082	−0.63	−2.84	1.57	0.556
x0	2.15	−9.47	13.76	0.711	10.87	−9.70	31.44	0.286	−7.76	−13.66	−1.86	0.013
x0_t	−0.42	−1.91	1.06	0.568	−0.51	−2.12	1.09	0.516	−0.23	−2.98	2.51	0.861
x0_z	−3.59	−18.34	11.16	0.626	−9.02	−31.61	13.57	0.418	1.86	−17.09	20.81	0.840
x0_z_t	10.60	4.61	16.60	0.001	6.83	2.51	11.15	0.003	15.39	2.84	27.93	0.019
x1	−6.43	−12.92	0.07	0.052	−3.62	−9.22	1.98	0.195	−10.24	−22.72	2.24	0.103
x1_t	−0.19	−1.73	1.34	0.802	0.69	−1.56	2.93	0.533	−1.18	−3.27	0.91	0.252
x1_z	−5.87	−24.66	12.93	0.533	−4.77	−20.57	11.03	0.539	−7.39	−46.74	31.95	0.699
x1_z_t	−10.01	−15.90	−4.13	0.001	−6.98	−10.77	−3.19	0.001	−13.72	−26.25	−1.19	0.033
Slope during implementation year
		95%				95%				95%
	Coef.	CI		P	Coef.	CI		P	Coef.	CI		P
Expansion	11.96	6.52	2.01	0.000	8.46	4.70	12.22	0.000	16.49	4.91	28.06	0.008
Nonexpansion	0.43	−1.15	17.37	0.585	−0.65	−2.76	1.46	0.531	1.73	−0.52	3.99	0.124
Diff	11.53	5.68	17.37	0.000	9.11	5.09	13.14	0.000	14.75	2.36	27.14	0.022
Slope after first year of implementation
		95%				95%				95%
	Coef.	CI		P	Coef.	CI		P	Coef.	CI		P
Expansion	1.75	0.27	0.82	0.022	2.17	−0.42	4.77	0.097	1.58	0.83	2.34	0.000
Nonexpansion	0.24	−0.34	3.08	0.408	0.04	−0.67	0.74	0.911	0.55	−0.42	1.52	0.249
Diff	1.51	-0.05	3.08	0.057	2.13	-0.53	4.80	0.112	1.03	-0.19	2.26	0.094

^*Models adjusted for quarters and state-level fixed baseline characteristics in the 3 years prior to expansion quarter (% of NAFs in different race ethnicity groups [Hispanic, NHW, NHB, NHA and Other races including individuals of multiple races], age under 40, full-time unemployed, full-time students, living in metropolitan areas, and of foreign birth)

When stratified by state’s racial diversity status, prior to expansion, racially less diverse expansion states had a significantly lower level of ET prescriptions than racially less diverse nonexpansion states (−52.11, p=0.007). In contrast, racially more diverse expansion states had a significantly higher level of ET prescriptions than racially more diverse nonexpansion states (35.74, p=0.025). Following expansion, the increase in the first year was sharper for racially more diverse expansion states (Slope: 16.49, p=0.008) than racially less diverse expansion states (Slope: 8.46, p<0.001).

Changes in the Number of ET Prescriptions following Medicaid Expansion

Overall, a statistically significant difference in ET prescriptions was observed starting in the third quarter of expansion year between expansion and nonexpansion states (Figure 1.a). Stratifying by state-level racial diversity status showed a different impact of Medicaid expansion on the number of ET prescriptions (Figure 1.b). Among racially less diverse states, the difference in the mean number of ET prescriptions between expansion and non-expansion states was not statistically significant until near the end of the available data (quarter 31,32). In contrast, among racially more diverse states, the difference in ET prescriptions between expansion and non-expansion states was statistically significant starting in the second quarter following expansion.

In sensitivity analysis, excluding states that expanded later than 2014 Q1 did not substantially affect the estimates from CITS analysis (Supplemental Table 1).

Discussion

Using the Medicaid outpatient prescription drug database, we found that Medicaid expansion was associated with significant increases in ET prescriptions in expansion states. Most of the rate of change in ET prescriptions (i.e., the difference in slopes) was observed in the first year post-expansion. After the first year, changes in ET prescription rates leveled off but the absolute number of ET prescriptions in expansion states remained at a higher level. On the other hand, few changes were observed in ET prescriptions among nonexpansion states over the corresponding time period. Stratifying by racial diversity status of the states’ NAF populations, more racially diverse states showed similar patterns, with significant increases in ET prescriptions among expansion states relative to nonexpansion states. In contrast, among racially less diverse states, expansion states had a smaller and mostly nonsignificant increase in ET prescriptions compared to the nonexpansion states in the same group during the post-expansion period.

Previous research has demonstrated the effects of the ACA on racial/ethnic disparities in health insurance coverage. For example, while rates of being uninsured decreased for all racial/ethnic groups between 2013–2017, health insurance coverage disparities between NHB or Hispanic individual and NHW individuals decreased substantially during this period with the greatest reduction occurred in Medicaid expansion states.[17,31] Medicaid expansion can thus be an important tool to increase health equity by providing access to health insurance coverage for low-income individuals that are disproportionally from racial/ethnic minority populations. As results from the present study indicate, racially more diverse states that expand Medicaid can have had even greater impacts on access to care. To help explain whether changes in ET prescriptions post-expansion (if any) could be attributed to increases in Medicaid enrollments or to more ET prescriptions by the existing enrollees (if no significant increases in enrollment was found), we examined the changes in mean number of NAF Medicaid enrollees 3 years before and 3 years after expansion estimated using data from the ASES, stratified by state’s expansion and racial diversity status. We found that increases in NAF Medicaid enrollment were larger in racially more diverse states than in racially less diverse states (Supplemental Table 2). The combined change in ET prescription results and change in NAF enrollment results suggest that Medicaid expansion in racially more diverse states may have been particularly important for previously-uninsured women with breast cancer who may have gained access to this potentially life-saving treatment following Medicaid expansion.

Limitations

In this study, we used the universe of Medicaid outpatient prescriptions databases, and thereby, were able to accurately assess the changes in ET prescriptions by Medicaid enrollees before and after Medicaid expansion across the states. We adjusted for baseline state-level characteristics of NAFs between states. However, a major limitation is the lack of individual-level data, which can provide more direct comparison of ET utilization between individuals in the expansion vs. nonexpansion states by race/ethnicity. Lack of information on individual patients or providers characteristics limits our ability to understand how individual-level factors affect changes in prescription rates. In this study, ET prescriptions were weighted by states’ NAF populations to adjust for differences in state population sizes. NAFs may not accurately reflect the base populations where use of ET is indicated. Although ETs are used primarily for breast cancer treatment, they may be used for other conditions such as prevention of breast cancer among women at high risk or treatment of gynecomastia in men.[33,34] However, it is likely that the vast majority of ET use is for breast cancer treatment in females. We relied on prescriptions fills as a proxy for medication use and may have overestimated the actual use of ETs. Although we used models that relaxed the linear trends of CITS, our results are still subject to other assumptions of our models and are valid only if these assumptions are satisfied.

Our study contributes to the literature of Medicaid expansion’s impact on racial disparity in cancer survivors. Apart from assessing Medicaid expansion’s impact on racial disparity in uninsurance rates among individuals with cancer, few studies have examined the effect of expansion on racial disparity in cancer treatment.[16] Among surgical cancer patients, Crocker et al. found no statistically significant effect of expansion on racial disparities.[32] Our study of Medicaid-financed ET prescriptions suggests that racial disparity may have been reduced in expansion states but could have been widened in the nonexpansion states in racially more diverse states. Future individual-level analysis using nationally representative data should be conducted to confirm this.

Conclusion

We found that Medicaid expansion was associated with significant increases in ET prescriptions in expansion states vs. nonexpansion states after expansion. However, in stratified analyses, this difference was only observed among racially more diverse states. Racially more diverse nonexpansion states had the lowest rates of ET prescriptions and the gaps from racially more diverse expansion states significantly widened after expansion. Future studies using patient-level data are warranted to assess the effect of this disparity directly and its association with health outcomes and costs among Medicaid beneficiaries with breast cancer.

Policy Summary

Our study shows that, prior to expansion, racially more diverse nonexpansion states had the lowest rates of ET prescriptions. After expansion, the gap in ET prescriptions between these states and racially more diverse expansion states significantly widened. Results from the present study highlight the importance of continuing to examine the health impacts of states not expanding Medicaid, including the health equity impacts for low income racial/ethnic minority populations with cancer and other life-threatening diseases.

Figure 2. Differences in adjusted mean quarterly endocrine therapy prescriptions between expansion and nonexpnasion states (quarter 12 Indicates the quarter before “expansion” quarter): (A) all states (B) stratified by state’s racial diversity status.

Highlights:

• Medicaid expansion was associated with significant increases in ET prescriptions in expansion states vs. nonexpansion states

• However, when stratified by state’s racial diversity status, differences in ET prescriptions between expansion and nonexpansion states was only observed among racially more diverse states.

• Racially more diverse nonexpansion states had the lowest rates of ET prescriptions and the gaps from racially more diverse expansion states significantly widened after expansion.