Receipt of Medications for Alcohol Use Disorder in the Veterans Health Administration: Comparison of rates at the intersections of racialized and ethnic identity with both sex and transgender status

Abstract

Objectives.

Medications for Alcohol Use Disorder (MAUD) are recommended for patients with AUD yet are under-prescribed. Consistent with Minority Stress and Intersectionality theories, persons with multiple socio-demographically marginalized identities (e.g., Black women) often experience greater barriers to care and have poorer health outcomes. We use data from the Veterans Health Administration (VHA) to assess disparities in FDA-approved MAUD and all effective MAUD between the following groups: racialized and ethnic identity, sex, transgender status, and their intersections.

Methods.

Among all VHA outpatients between 8/1/15-7/31/17 with documented alcohol screenings and an ICD diagnosis for AUD in the 0-365 days prior (N=308,238), we estimated the prevalence and 95% confidence intervals (CI) of receiving FDA-approved MAUD and any MAUD in the following year and compared them using Chi-Square or Fisher’s exact tests. Analyses are unadjusted to present true prevalence and group differences.

Results.

The overall prevalence for MAUD was low (FDA-MAUD=8.7%, any MAUD=20.0%). Within sex, Black males had the lowest rate of FDA-MAUD (7.3%, [7.1-7.5]) whereas American Indian/Alaskan Native females had the highest (18.4%, [13.8-23.0]). Among those identified as transgender, Asian and Black transgender persons had the lowest rates of FDA-MAUD (0%; 4.3%, [1.8-8.5], respectively) whereas American Indian/Alaskan Native transgender patients had the highest (33.3%, [2.5-64.1]). Similar patterns were observed for any MAUD, with higher rates overall.

Conclusions.

Substantial variation exists in MAUD prescribing, with marginalized veterans disproportionately receiving MAUD at lower and higher rates than average. Implementation and quality improvement efforts are needed to improve MAUD prescribing practices and reduce disparities.

Introduction

Although estimates vary, approximately 5-14% of U.S. adults meet criteria for alcohol use disorder (AUD) in any given year.¹ AUD leads to severe deficits in functioning (e.g., occupational and social impairment), and can result in devastating health-related consequences, including morbidity and mortality.² Evidence-based behavioral and pharmacologic treatments for AUD are available³, however pharmacological treatments can be more readily delivered across clinical settings, making them a more accessible treatment option for AUD. The potential for medication prescribing to reach patients beyond those seen in specialty AUD care makes medication treatment particularly appealing, as greater access to treatment may help de-stigmatize and reduce barriers to AUD care (which is commonly siloed to specialty substance use settings).⁴ Several medications for AUD (MAUD) have been shown to be effective in helping individuals either abstain from or reduce their drinking and include both Federal Drug Administration (FDA) approved (naltrexone, acamprosate, and disulfiram) and non-FDA approved (topiramate, gabapentin, and baclofen) pharmacotherapy.^3,5–7 Although some medications are recommended more strongly (e.g., naltrexone, topiramate) than others (e.g., disulfiram, baclofen), incorporating MAUD into AUD treatment plans is recommended by VA/DoD SUD Clinical Guidelines (barring contraindications; e.g., pregnancy).³ However, despite clinical guidelines and theoretical ease of access, pharmacotherapy treatment rates for AUD are low, with national data from SAMHSA indicating that only around 2.2% of adults with an AUD are prescribed an evidence-based medication.⁸

Though underdiagnosed across systems, rates of documented AUD in the Veterans Health Administration (VHA) are similar to rates seen in the general adult population. Approximately 6-7% of veterans (n = ~419,832) seeking care in the VHA are diagnosed with an AUD in any given year, with rates having increased slightly over the past decade.⁹ However, rates of documented AUD differ among certain subgroups of the VHA patient population. Recent VHA research estimates approximately 4-5% of women patients (n = ~17,463),⁹ ~8% of transgender patients (n = ~737),¹⁰ ~9% of Black patients (n = ~79,468), ~7% of Hispanic patients (n = ~21,465), and ~5% of White patients (n = ~202,530)¹¹ have a documented AUD diagnosis. Rates of pharmacotherapy for AUD have increased in the VHA over the years and range from ~3-12%, depending on the data source.^12–15 Higher rates in VHA may be due to academic detailing efforts in 2013 to increase medications for mental health disorders across VHA, including for AUD.¹² Research has also revealed that MAUD rates vary across VHA by individual- and facility-level factors. For example, veterans who attend treatment in a specialty substance use clinic, have a co-occurring psychiatric disorder, have more severe AUD, and/or receive treatment from certain VHA medical centers are more likely to receive MAUD.^16,17 Additionally, some research has found that MAUD receipt varies by demographic factors, including gender (male/female, transgender status), race, and ethnicity. For instance, studies have found that female veterans are more likely to receive MAUD than male veterans ^16,17 and Black veterans with AUD are less likely than White veterans to receive MAUD.¹⁴ Other more recent VHA research has found that veterans identified as transgender in the electronic health record (EHR) are more likely than cisgender veterans to receive pharmacotherapy for AUD.¹⁸

Low treatment rates and disparities in receipt of care for AUD have been a longstanding problem in the field of alcohol-related care.¹⁹ Moreover, despite the effectiveness and other advantages of MAUD (e.g., more efficient use of clinician resources), research has consistently documented barriers to MAUD prescribing, including provider comfort with and knowledge of MAUD, stigma surrounding treating patients with AUD, and patient preferences and misinformation about MAUD.^20,21 In addition, individuals with intersecting marginalized identities (e.g., Black women, Hispanic transgender persons) who also have an AUD may be more vulnerable to both barriers to MAUD and the harmful effects of AUD.^22–24

Several theories, including Minority Stress Theory²⁵ and intersectionality theory,^26,27 put forth hypotheses as to why persons with multiple socio-demographically marginalized identities are often at the greatest risk of experiencing the poorest health/health-related outcomes and experiencing the greatest barriers to care. For example, intersectionality theory, developed from the perspective of Black feminist theory, asserts that race and gender are mutually constitutive categories wherein relative privilege or disadvantage on one axis (e.g., race) fundamentally changes a person’s lived experience of other identities (e.g., gender).²⁶ Examining disparities as they pertain to health and healthcare without considering intersectionality or accounting for a more nuanced understanding of one’s identity (e.g., Black women vs. Black individuals) will limit researchers’ ability to draw specific conclusions and recommendations to healthcare systems attempting to reduce these disparities. Therefore, the present study uses electronic health record (EHR) data from the nation’s largest integrated health system, the VHA, to investigate disparities in both FDA-approved MAUD and all effective MAUD, between the following groups: racialized and ethnic identity, sex, transgender status, and the intersection of race and ethnicity with sex or transgender status. This study is ‘first generation’ within the traditional three-generation disparity research pipeline:²⁸ 1) detect disparity, 2) explore drivers of the disparity, and 3) develop and test interventions to address the disparity. No studies, to our knowledge, have assessed disparities in MAUD through an intersectional lens.

Methods

Data source and study sample

Data for the present study included EHR data extracted from the VHA’s Corporate Data Warehouse (CDW). The CDW is a data repository that contains administrative and clinical data for all VHA patients, including demographics, diagnoses, clinical encounters, and prescription information. Data were extracted for all patients with an outpatient visit between 8/1/15 -7/31/17 with a documented alcohol screen (i.e., an AUDIT-C score). As patients could have more than one visit during the study period, each patient’s most recent visit was considered the index visit. Patients were included in the sample for this analysis if they had a documented diagnosis for AUD—International Classification of Disease (ICD)-9 and 10 codes—in the 0-365 days prior to the index visit. This current study is a secondary analysis of data extracted for a previous study that sought to investigate alcohol use and related care among transgender veterans, but did not specifically examine intersectionality in patients with AUD.¹⁰ Study procedures were approved by the Institutional Review Board at the VA Puget Sound Health Care System and the University of Washington.

Measures

Predictors

Groups of interest were based on racialized and ethnic identity, sex, transgender status, and the intersection of race and ethnicity with sex or transgender status. Race, ethnic identity, and sex were extracted from the EHR and are based on self-report information obtained from veterans when they enroll in VHA care. Veterans were categorized into the following racialized and ethnic groups for the purposes of the current study: Black non-Hispanic only, White non-Hispanic only, Hispanic (any race), Asian Pacific Islander (API) non-Hispanic only, American Indian/Alaskan Native (AI/AN) non-Hispanic only, Multiple race non-Hispanic, and Unknown. Sex was defined as male/female based on EHR documentation of “sex at birth”. Cisgender/Transgender designation was identified using relevant ICD-9 and 10, Clinical Modification (ICD-9-CM and ICD-10-CM, respectively) codes, using methods developed and validated with VHA EHR data.^10,18,29 These methods have been found to be highly accurate in identifying transgender patients.³⁰ Patients were identified as transgender if their EHR contained ≥1 transgender-related diagnostic code from the start of the CDW (January 1, 1999) to the end of the study period (July 31, 2017; see Supplemental Table A from Williams et al., 2021 for codes used).¹⁰

Outcomes

MAUD were categorized into two groups: (1) FDA-approved MAUDs, which included a filled prescription for acamprosate, disulfiram, and/or naltrexone (oral or injectable) during the 0-365 days following the index visit, and (2) Any MAUD, which included a filled prescription for acamprosate, disulfiram, naltrexone (oral or injectable), topiramate, gabapentin, and/or baclofen during the 0-365 days following the index visit.

Descriptives

Sociodemographic information was extracted to help describe the sample. This included age, marital status, and veteran co-pay status.

Data Analytic Plan

Sociodemographic information was calculated (e.g., frequencies) for the full sample and for those receiving MAUD. Prevalence estimates and their 95% confidence intervals (CI) were obtained for both FDA-approved MAUD and any effective MAUD for patient groups based on sex, racialized or ethnic identity, transgender status, and intersections of racialized or ethnic identity with both sex and transgender status. We then compared rates across intersectional groups using Chi-Square or Fisher’s exact tests. Analyses were solely descriptive and unadjusted to present true prevalence and group differences as they exist in VHA.

Results

Data extraction resulted in 308,238 veterans with a documented AUD during the study timeframe (see Table 1 for sociodemographic information). The overall prevalence for FDA-approved MAUD was 8.7% (n=26,941) and for any MAUD was 20.0% (n=61,502). The prevalence of MAUD, stratified by FDA-approved vs. any, and delineated by sex, racialized or ethnic identity, and transgender status is presented in Table 2. Prevalence for MAUD was lowest among men (FDA-MAUD=8.5%, 95% CI=8.4%-8.6%; any MAUD=19.5%, 95% CI=19.3-19.6%), Black patients (FDA-MAUD=7.5%, 95% CI=7.3-7.7%; any MAUD=19.2%, 95% CI=18.9-19.4%), and cisgender persons (FDA-MAUD=8.7%, 95% CI=8.6-8.8%; any MAUD=19.9%, 95% CI=19.8-20.1%). Prevalence was highest among females (FDA-MAUD=13.2%, 95% CI=12.7-13.7%; any MAUD=28.2%, 95% CI=27.5-28.9%), American Indian/Alaskan Natives (FDA-MAUD=13.3%, 95% CI=12.1-14.5%; any MAUD=25.0%, 95% CI=23.5-26.5%), and transgender persons (FDA-MAUD=12.8%, 95% CI=9.9-15.6%; any MAUD=24.6%, 95% CI=20.9-28.2%). All chi-square tests were significant at p < .001.

Table 1.

Sociodemographic information for the full sample (N=308,238) of patients with a documented alcohol use disorder

	N	%
Age
18-29	18,582	6.03
30-44	60,282	19.56
45-64	142,377	46.19
65+	86,997	28.22
Sex
Male	291,958	94.72
Female	16,280	5.28
Race and ethnicity
Black	75,334	24.44
White	192,531	62.46
Hispanic (any race)	22,192	7.20
Asian/Pacific Islander	3,909	1.27
American Indian/Alaskan Native	3,076	1.00
Multiple race	2,598	0.84
Unknown	8,598	2.79
Gender Identity
Cisgender	307,705	99.83
Transgender	533	0.17
Marital Status
Divorced/separated	123,950	40.21
Married	104,442	33.88
Never married/single	67,304	21.84
Widowed	10,877	3.53
Unknown/missing	1,665	0.54
VA Copay Status
Copay required due to means	39,768	12.9
No copay required due to disability	82,387	26.73
No copay required due to means/other	119,690	38.83
Unassigned	66,393	21.54
Medication for AUD (MAUD)
Any MAUD	61,502	19.95
FDA-Approved MAUD	26,941	8.74

Note. Veterans identified as Hispanic only categorized in the “Hispanic” subgroup

Table 2.

Prevalence rates and 95% C.I.s of MAUD by sex, race and ethnicity, and transgender status independently, stratified by any effective MAUD versus FDA-approved MAUD

	Any MAUD N (%)	95% C.I.	FDA-approved MAUD N (%)	95% C.I.
Sex
Male	56,906 (19.5%)	19.3-19.6	24,794 (8.5%)	8.4-8.6
Female	4,596 (28.2%)	27.5-28.9	2,147 (13.2%)	12.7-13.7
Race and ethnicity
Black	14,431 (19.2%)	18.9-19.4	5,655 (7.5%)	7.3-7.7
White	38,450 (20.0%)	19.8-20.1	17,346 (9.0%)	8.9-9.1
Hispanic (any race)	4,755 (21.4%)	20.9-22.0	2,205 (9.9%)	9.5-10.3
Asian/Pacific Islander	758 (19.4%)	18.2-20.6	338 (8.7%)	7.8-9.5
American Indian/Alaskan Native	769 (25.0%)	23.5-26.5	409 (13.3%)	12.1-14.5
Multiple race	560 (21.6%)	20.0-23.1	240 (9.2%)	8.1-10.4
Unknown	1,779 (20.7%)	19.8-21.5	748 (8.7%)	8.1-9.3
Transgender status
Cisgender	61,371 (19.9%)	19.8-20.0	26,873 (8.7%)	8.6-8.8
Transgender	131 (24.6%)	20.9-28.2	68 (12.8%)	9.9-15.6

Note. MAUD = Medication for Alcohol Use Disorder. All chi-square omnibus tests significant at p < .001

Figure 1 and Supplemental Table 1 present prevalence rates and their 95% C.I.s from an intersectional perspective for FDA-approved MAUD. Within sex, findings revealed that Black males had the lowest rate of FDA-approved MAUD (7.3%, 95% CI=7.1-7.5%) whereas American Indian/Alaskan Native females had the highest rate (18.4%, 95% CI=13.8-23.0%). Within transgender status, Asian/Pacific Islander and Black transgender persons had the lowest rates of FDA-MAUD (0%; 4.3%, 95% CI=1.8-8.5, respectively) compared with other patients identified as transgender or cisgender. American Indian/Alaskan Native transgender patients had the highest FDA-approved MAUD rates (33.3%, 95% CI=2.5-64.1%). All chi-square tests (p<.05) and Fisher’s exact test (p=.026) were significant at p < .05.

Figure 1.

Prevalence rates by race, ethnicity and sex or transgender status for FDA-approved MAUD.

Note. MAUD = Medications for Alcohol Use Disorder; API = Asian/Pacific Islander, AI/AN = American Indian/Alaskan Native. All chi-square omnibus tests significant at p < .05; Fisher’s exact test significant at p=.026 for comparisons among individuals identifying as transgender. Ns for each group (i.e., numerator) as follows: Black males (n=5,126), Black females (n=529), Black cisgender individuals (n=5,651), Black transgender individuals (n=4), White males (n=16,061), White females (n=1,285), White cisgender individuals (n=17,290), White transgender individuals (n=56), Hispanic males (n=2,035), Hispanic females (n=170), Hispanic cisgender individuals (n=2,202), Hispanic transgender individuals (n=3), API males (n=309), API females (n=29), API cisgender individuals (n=338), API transgender individuals (n=0), AI/AN males (n=358), AI/AN females (n=51), AI/AN cisgender individuals (n=406), AI/AN transgender individuals (n=3), Multiple race males (n=215), Multiple race females (n=25), Multiple race cisgender individuals (n=239), Multiple race transgender individuals (n=1), Unknown race/ethnicity males (n=690), Unknown race/ethnicity females (n=58), Unknown race/ethnicity cisgender individuals (n=747), Unknown race/ethnicity transgender individuals (n=1).

Figure 2 and Supplemental Table 2 present prevalence rates and their 95% C.I.s from an intersectional perspective for any effective MAUD. Patterns of prescribing for any MAUD were similar to patterns for FDA-approved MAUD, with higher prevalence rates overall, except for transgender patients identifying as American Indian/Alaskan Native or Asian/Pacific Islander. Prescribing remained low for Asian/Pacific Islander transgender persons (0%) and high for American Indian/Alaskan Native transgender persons (33.3% 95% CI=2.5-64.1%), as compared with other patients identified as transgender or cisgender. All chi-square tests were significant at p < .001, however, the Fisher’s exact test examining MAUD among those identifying as transgender was not significant (p= 0.794).

Figure 2.

Prevalence rates by race, ethnicity and sex or transgender status for any effective MAUD.

Note. MAUD = Medications for Alcohol Use Disorder; API = Asian/Pacific Islander, AI/AN = American Indian/Alaskan Native. All chi-square omnibus tests significant at p < .001 except for the Fisher’s exact test examining difference in MAUD in individuals identifying as transgender (p= 0.794). Ns for each group (i.e., numerator) as follows: Black males (n=13,115), Black females (n=1,316), Black cisgender individuals (n=14,409), Black transgender individuals (n=22), White males (n=35,873), White females (n=2,577), White cisgender individuals (n=38,357), White transgender individuals (n=93), Hispanic males (n=4,411), Hispanic females (n=344), Hispanic cisgender individuals (n=4,749), Hispanic transgender individuals (n=6), API males (n=680), API females (n=78), API cisgender individuals (n=758), API transgender individuals (n=0), AI/AN males (n=678), AI/AN females (n=91), AI/AN cisgender individuals (n=766), AI/AN transgender individuals (n=3), Multiple race males (n=506), Multiple race females (n=54), Multiple race cisgender individuals (n=557), Multiple race transgender individuals (n=3), Unknown race/ethnicity males (n=1,643), Unknown race/ethnicity females (n=136), Unknown race/ethnicity cisgender individuals (n=1,775), Unknown race/ethnicity transgender individuals (n=4).

Discussion

The current study examined disparities in rates of pharmacotherapy for AUD through an intersectional lens via race, ethnicity, and sex or transgender status. We found substantial variation in receipt of medications for AUD among VHA patients with a documented AUD diagnosis. Some of our findings confirm past work examining the association between demographic variables and rates of MAUD in the VHA,^16,17 but extend this work by investigating rates for groups with overlapping marginalized identities who may be most vulnerable to the negative effects of AUD. Overall, rates of prescriptions, for both FDA-approved MAUD and any MAUD, were lower among men, which is in line with past work discussed above^16,17 showing that women are more likely to receive MAUD than men. Additionally, in line with previous research, patients in the present study categorized as Black and patients categorized as cisgender had lower rates of MAUD than other racial and ethnic groups as well as transgender patients, respectively. American Indian/Alaskan Native veterans appear to have the highest rates of MAUD overall and Black patients appear to have the lowest rates, except when compared with Asian/Pacific Islander transgender veterans, who have the lowest rates of any MAUD. Findings suggest that minoritized veterans are both disproportionately more likely and less likely to receive effective medications for AUD, depending on the minoritized group.

In addition, Black females and Asian/Pacific Islander females evidenced lower rates of MAUD as compared to females from other racial and ethnic groups. This is in line with findings from the 2000, 2005, and 2010 National Alcohol Surveys, which looked at intersectional (race, ethnicity and sex) differences in alcohol service utilization and found that Black and Hispanic females with AUD were less likely than White females with AUD to report seeking alcohol-related services.¹⁹ Current findings also align with some of our more recent work examining intersectional disparities in receipt of brief alcohol interventions for veterans endorsing unhealthy alcohol use during VHA care.²² Our findings also revealed that American Indian/Alaskan Native transgender veterans showed much higher rates of FDA-approved MAUD as compared to other transgender patients of differing racial and ethnic groups and compared to cisgender patients. However, given the wide confidence intervals around the estimate of MAUD for American Indian/Alaskan Native transgender patients, inferential results should be interpreted with caution. Nonetheless, these findings are in stark contrast to MAUD prescribing rates among Black and Asian/Pacific Islander transgender veterans, who had much lower rates of FDA-approved MAUD than other transgender veterans of differing racial and ethnic groups. Overall, MAUD rates, particularly FDA-approved MAUD rates, appear lowest among all Black veterans with AUD, regardless of sex or transgender status. As a first-generation study, it is important to detect these existing disparities as the first step to addressing them.

A crucial next step involves understanding drivers (or reasons) for these disparities. It is possible that patient-level factors are contributing to disparities. For example, certain patient subgroups may decline offerings of MAUD at higher rates than other patient subgroups, contributing to disparities. Medication refusals may arise from differential patient preferences (e.g., worries about adverse effects from medications), perceived low warmth/empathy from providers, or a belief that pharmacotherapy is not effective. For example, some qualitative research suggests Black women with a history of mental health and/or substance use treatment may believe that medications for these conditions are not explained well by providers which can lead to feelings of frustration, beliefs that medications are ineffective or cause harm, and ultimately higher rates of refusal and lower prescribing.²⁰ This work suggests the importance of building a therapeutic alliance with patients in the context of medication prescribing and that therapist common factors such as alliance, therapist empathy, and/or positive regard, may help increase medication initiation and/or adherence (just as it helps engage patients in behavioral therapy). Future qualitative research focused on MAUD among a range of marginalized subgroups is needed to understand if these thoughts and behaviors may also be true for racially and ethnically marginalized patients across sex and transgender status with AUD (e.g., Black men, Asian/Pacific Islander transgendered patients).

Other factors at the provider-, healthcare system-, and societal-level could additionally be leading to and/or compounding disparities. Provider-level factors may include interpersonal racism/transphobia. For example, research has shown that transgender people of color report experiencing negativity from healthcare providers that is either directed at their race and/or ethnicity or gender identity,³¹ which often contributes to mistrust in the healthcare system and reduced receipt of effective care. Healthcare system-level factors might include inequitable availability of care/resources (especially resources devoted for marginalized groups), while system-level factors may include structural racism/transphobia in patients’ communities.³² These factors may be particularly true for Black patients given the decades of research showing disparities in healthcare for Black individuals (both AUD-related and non). Given these possibilities, a deeper understanding is needed (perhaps through qualitative methods) to inform efforts at overcoming barriers at multiple levels in order to improve care and prevent dire outcomes for the most vulnerable patients (e.g., Black transfeminine individuals have increased mortality risk compared to Black cisgender and other Black transgender individuals).³³

Clinical trial and implementation research have tested whether various strategies (e.g., VHA academic detailing, clinical champions, patient education) can increase the uptake of prescribing for MAUD in VHA.³⁴ For example, VHA academic detailing deployed in 2013 was found to increase MAUD in intervention sites (from 4% to 8%) while other MAUD-focused intervention research (e.g., in primary care clinics) have had mixed success.^13,34,35 This limited success has led to research aimed at understanding why various implementation efforts have evidenced mixed results. For example, Hagedorn et al.³⁶ used qualitative interviews with primary care providers pre- and post-implementation intervention to better understand why efforts to increase MAUD prescribing in three VHA primary care clinics were of limited success (see Harris et al.¹³ for trial findings). Interviews revealed that implementation efforts addressed barriers related to knowledge/education of prescribing but were not able to satisfactorily address: (1) barriers related to the structure of the medical appointment (i.e., visits are time-limited and providers still felt that deviating from clinical reminders and patient concerns made it difficult to address AUD) and (2) available clinical resources (i.e., difficulty triaging/collaborating with behavioral health).

Given these challenges, research on improving prescribing for opioid use disorder (OUD) in the VHA may help shed light on what could be done to improve MAUD. For example, a larger VHA trial focused on improving prescribing for OUD in eight VHA facilities used the lessons learned from the above-mentioned MAUD implementation trial and was able to significantly increase medication treatment for OUD by working at the facility- as opposed to the clinic-level.³⁷ Although all facilities had the same overarching goal (increase MOUD), implementation strategies were tailored to each facility based on specific goals/action steps put forth by local teams of providers. Similar facility-level, tailored implementation efforts could be used to improve MAUD with an eye toward an understanding that increased and equitable MAUD prescribing is the ultimate goal. Additionally, top-down, national VHA efforts to improve equitable MAUD are most likely needed.

Limitations

There are several limitations to the present study that should be acknowledged. Data extraction ended in 2017 and it is conceivable that patterns in care have since changed. For instance, it is possible that patterns of MAUD care changed over the course of the COVID-19 pandemic; however, recent work examining AUD care in VHA suggests that pharmacotherapy rates for AUD stayed the same during the early years of this pandemic and even slightly increased (while behavioral treatments for AUD during the same time period decreased, underscoring the potential for MAUD to be a more accessible form of treatment).³⁸ Moreover, given that there have been no national implementation efforts to increase MAUD in the VHA since 2013, there is little reason to believe that patterns have changed substantially since this time period. A subsequent limitation is that the analyses presented here are descriptive and unadjusted. While this approach is recommended for disparities research to show true patterns without adjusting out pathways (e.g., adjustment for facility-level differences in care resultant from variation in resources stemming from structural racism would possibly alter findings), some may argue that adjustment for age (which varies by sex and racialized group in VA) is an important next step. Future research should examine whether other patient demographics and facility-/clinic-level factors (e.g., provider training) are related to disparities in MAUD receipt. In addition, some of the intersectional transgender patient subgroups were quite small (e.g., 3 American Indian/Alaskan Native transgender individuals received FDA-approved MAUD). More work replicating our findings will be necessary before definitive conclusions can be made, particularly related to transgender subgroups. Moreover, AUD is often underdiagnosed in healthcare settings, including in the VHA,³⁹ and severity of AUD cannot be easily captured. Thus, our sample almost certainly does not capture all VHA patients with an AUD nor does it tell us which patients may benefit most from MAUD given severity. However, despite this limitation, EHR studies are one of the most effective ways to conduct population-level research on health services when it is not possible to administer structured clinical assessments to every individual. Furthermore, future work may want to compare findings to intersectional differences in rates of engaging in behavioral treatment for AUD, as MAUD is often paired with behavioral therapy. Related to our outcomes, we include analyses examining medications that are not FDA approved for the treatment of AUD. Although these medications (gabapentin, topiramate, baclofen) are recommended for treating AUD in VA/DoD SUD clinical practice guidelines to varying degrees (baclofen less so),³ we do not know the number of patients with AUD who are receiving these medications solely for AUD treatment. Thus, results must be viewed with this in mind. It would be beneficial for future research to investigate prescribing rates of these medications solely for AUD to understand how often they are being used for this purpose. Last, the current data do not capture MAUD that may have been prescribed outside of the VHA, which limits the generalizability to non-VHA settings.

Conclusions and Clinical Relevance

Findings from this first-generation disparities study indicate that there is substantial variation in MAUD prescribing in the VHA. Marginalized and minoritized veterans are disproportionately receiving and not-receiving MAUD, and Black patients with AUD, in particular, appear to have the lowest rates of this care. Given recent calls for equity related to prescribing (e.g., “pharmacoequity”),⁴⁰ exploring intersectional disparities in MAUD prescribing in second-generation studies will be important. After better understanding drivers, thoughtful, tailored, evidence-based implementation strategies that engage the VHA at the facility-level while focusing on health equity may help improve MAUD prescribing overall and reduce these disparities in care.