Geographic and Racial/Ethnic Differences in Access to Methamphetamine Detoxification Services, United States, 2021.

INTRODUCTION: Methamphetamine detoxification before entering formal and longer term treatment may have a positive impact on treatment retention and success. Understanding geographic distribution of methamphetamine specialty detox services and differential access by race/ethnicity is critical for establishing policies that ensure equitable access across populations. METHODS: We used the Mental health and Addiction Treatment Tracking Repository to identify treatment facilities that offered any substance use detoxification in 2021 (N=2346) as well as the census block group in which they were located. We sourced data from the US Census Bureau to identify the percentage of a census block group that was White, Black, and Hispanic. We used logistic regression to model the availability of methamphetamine-specific detox, predicted by the percentage of a block group that was Black and Hispanic. We adjusted for relevant covariates and defined state as a random effect. We calculated model-based predicted probabilities. RESULTS: Over half (60%) of detox facilities offered additional detox services specifically for methamphetamine. Sixteen states had <10 methamphetamine-specific detox facilities. The predicted probability of methamphetamine-specific detox availability was 60% in census block groups with 0%-9% Black residents versus only 46% in census block groups with 90%-100% Black residents, and was 61% in census block groups with 0%-9% Hispanic residents versus 30% in census block groups with 90%-100% Hispanic residents. CONCLUSIONS: During an unprecedented national methamphetamine crisis, access to a critical health care service was disproportionately lower in communities that were predominately Black and Hispanic. We orient our findings around a discussion of health disparities, residential segregation, and the upstream causes of the systematic exclusion of minoritized communities from health care.

Evaluating the substance use disorder treatment gap in the United States, 2016-2019: A population health observational study.

BACKGROUND AND OBJECTIVES: Only 10% of Americans with substance use disorders (SUDs) receive treatment with insufficient treatment access and screening practices proposed and potential contributing factors. METHODS: This retrospective cross-sectional study used National Survey on Drug Use and Health (NSDUH) data to assess individuals with SUDs receiving treatment between 2016 and 2019 (survey n = 12,111; weighted n = 12,394,214). Demographic, access, and screening characteristics were investigated as predictors of treatment receipt using time-series logistic regression analyses to test trends and assessed treatment receipt odds, controlling for demographic and treatment characteristics. RESULTS: For those with past-year SUDs, 13.0% reported receiving past-year SUD treatment (survey n = 1605; weighted n = 1,612,154). The SUD treatment receipt rate remained statistically stable from 2016 to 2019, with a nonsignificant treatment receipt trend declining from 14% to 12%. Treatment changes were notable among Native Americans (+53.80%), Pacific Islanders (+94.10%), multiracial (-59.96%), ages 65+ (-70.18%), and ages 12-17 (-50.70%). In the regression model, race, sex, age, insurance status, and receiving mental health treatment were associated with SUD treatment receipt. DISCUSSION AND CONCLUSIONS: The treatment gap remains substantial and stable. Annually, about 87% of Americans with SUDs are not receiving the treatment they need. Asian Americans were less likely and those attending general mental health services were more likely to receive treatment. SCIENTIFIC SIGNIFICANCE: We present an updated SUD treatment gap evaluation, and identify access and screening characteristics associated with SUD treatment receipt. Policymakers, clinicians, and researchers must continue improving access and identification of those in need of care.

National and state-level trends in the availability of mental health treatment services tailored to individuals ordered to treatment by a court: United States, 2016, 2018, and 2020.

PURPOSE: We sought to identify trends and characteristics associated with the availability of tailored mental health services for individuals involved in the criminal justice system and ordered to treatment by a court, nationally in the US and by state. METHODS: We used National Mental Health Services Survey to identify outpatient mental health treatment facilities in the US (2016 n = 4744; 2018 n = 4626; 2020 n = 4869). We used clustered multiple logistic regression to identify changes over time as well as facility- and state-level factors associated with the availability of specialty court-ordered services. RESULTS: Slightly more than half of the outpatient mental health treatment facilities offered specialized services for individuals ordered to treatment by a court, with wide variation between states. Nationally, there was a significant increase in the odds of offering court-ordered treatment in 2020 compared to 2016 (aOR = 1.16, 95% CI = 1.06-1.27, p < 0.01). Notable associations included offering integrated substance use treatment (versus none, aOR = 2.95, 95% CI = 2.70-3.22, p < 0.0001) and offering trauma therapy (versus none, aOR = 2.05, 95% CI = 1.85-2.27, p < 0.0001). CONCLUSION: The availability of mental health services for individuals ordered to treatment by a court is growing nationally but several states are lagging behind. Court ordered treatment is a promising strategy to improve health and reduce reliance on the carceral system as a healthcare provider. At the same time, we express caution around disparities within behavioral health courts and advocate for equity in access to incarceration alternatives.

Comprehensive and Integrated Services in Specialty Mental Health Treatment Facilities in the US: Differences by the Racial/Ethnic Composition of the Facility's Clientele, 2020.

The integration of multiple ancillary services into mental health treatment settings may improve outcomes, but there are no national studies addressing whether comprehensive services are distributed equitably. We investigated whether the availability of a wide range of service types differs based on the facility's racial/ethnic composition. We used the 2020 National Mental Health Services Survey to identify twelve services offered in outpatient mental health treatment facilities (N = 1,074 facilities). We used logistic regression to model each of the twelve services, predicted by the percentage of a facility's clientele that was White, Black, and Hispanic, adjusted for covariates. Facilities with the highest proportions of Black and Hispanic clientele demonstrated the lowest predicted probabilities of offering comprehensive and integrated services. Our findings offer context around upstream factors that may, in part, drive treatment disparities. We orient our findings around frameworks of structural racism and inequities in mental healthcare.

Temporal Trends in Telehealth Availability in Mental Health Treatment Settings: Differences in Growth by State Rurality, 2015-2020.

We sought to investigate temporal trends in telehealth availability among outpatient mental health treatment facilities and differences in the pace of telehealth growth by state urbanicity and rurality. We used the National Mental Health Services Survey (2015-2020) to identify outpatient mental health treatment facilities in the US (N = 28,989 facilities; 2015 n = 5,018; 2020 n = 4,889). We used logistic regression to model telehealth, predicted by time, state rurality (1 to 10% rural, 10 to < 20%, 20 to < 30%, or [Formula: see text] 30%), and their interaction, and adjusted for relevant covariates. We estimated the predicted probability of telehealth based on our model. We estimated effects with and without data from 2020 to assess whether the rapid and widespread adoption of telehealth during the COVID-19 pandemic changed the rural/urban trajectories of telehealth availability. We found that telehealth grew fastest in more urban states (year*rurality interaction p < 0.0001). Between 2015 and 2020, the predicted probability of telehealth in more urban states increased by 51 percentage points (from 9 to 61%), whereas telehealth in more rural states increased by 38 percentage points (from 23 to 61%). Predicted telehealth also varied widely by state, ranging from more than 75% of facilities (RI, OR) to below 20% (VT, KY). Health systems and new technological innovations must consider the unique challenges faced by urban populations and how best practices may be adapted to meet the growing urban demand. We framed our findings around the need for policies that minimize barriers to telehealth.

A State Profile of Disparities in Telehealth Utilization Among Medicare Beneficiaries: An Intersection Between Race/Ethnicity, Rurality, and Chronic Conditions-Arkansas, 2019.

Introduction: The COVID-19 pandemic brought about renewed interest and investment in telehealth, while also highlighting persistent health disparities in the Southern states. Little is known about the characteristics of those utilizing telehealth services in Arkansas, a rural Southern state. We sought to compare the characteristics of telehealth utilizers and nonutilizers among Medicare beneficiaries in Arkansas before the COVID-19 public health emergency to provide a baseline for future research investigating disparities in telehealth utilization. Methods: We used Arkansas Medicare beneficiary data (2018-2019) to model telehealth use. We included interactions to assess how the association between the number of chronic conditions and telehealth was moderated by race/ethnicity and rurality, adjusted for covariates. Results: Overall telehealth utilization in 2019 was low (n = 4,463; 1.1%). The adjusted odds of utilizing telehealth was higher for non-Hispanic Black/African Americans (vs. white, adjusted odds ratio [aOR] = 1.34, 95% confidence interval [CI] = 1.17-1.52), rural beneficiaries (aOR = 1.99, 95% CI = 1.79-2.21), and those with more chronic conditions (aOR = 1.23, 95% CI = 1.21-1.25). Race/ethnicity and rurality were significant moderators, such that the association between the number of chronic conditions and telehealth was strongest among white and among rural beneficiaries. Discussion: Among the 2019 Arkansas Medicare beneficiaries, having more chronic conditions was most strongly associated with telehealth use among white and rural individuals, while the effect was not as pronounced for Black/African American and urban individuals. Our findings suggest that advances in telehealth are not benefiting all Americans equally, with aging minoritized communities continuing to engage with more strained and underresourced health systems. Future research should investigate how upstream factors such as structural racism perpetuate poor health outcomes.

Individual and Health Policy Factors Associated With Positive Heroin and Opioid Treatment Response: United States, 2018.

Objectives. To identify client- and state-level factors associated with positive treatment response among heroin and opioid treatment episodes in the United States. Methods. We used national data from 46 states using the Treatment Episode Dataset‒Discharges (2018) to identify heroin and opioid treatment episodes (n = 162 846). We defined positive treatment response as a decrease in use between admission and discharge. We used multivariable regression, stratified by race/ethnicity, to identify demographic, pain-related, and state-level factors associated with positive treatment response. Results. Lower community distress was the strongest predictor of better treatment outcomes across all racial/ethnic groups, particularly among White and American Indian/Alaska Native episodes. A primary opioid of heroin was associated with worse outcomes among White and Hispanic episodes. Legislation limiting opioid dispensing was associated with better outcomes among Hispanic episodes. Buprenorphine availability was strongly associated with better outcomes among Black episodes. Conclusions. State-level variables, particularly community distress, had greater associations with positive treatment outcomes than client-level variables. Public Health Implications. Changes in state-level policies and increased resources directed toward areas of high community distress have the potential to improve opioid use disorder treatment and reduce racial/ethnic disparities in treatment outcomes. (Am J Public Health. 2022;112(S1):S66-S76. https://doi.org/10.2105/AJPH.2021.306503).

Deployment of Health Equity Strike Teams to Address COVID-19 Vaccine Disparities in Arkansas, 2021.

Minority populations have been disproportionately affected by the COVID-19 pandemic, and disparities have been noted in vaccine uptake. In the state of Arkansas, health equity strike teams (HESTs) were deployed to address vaccine disparities. A total of 13 470 vaccinations were administered by HESTs to 10 047 eligible people at 45 events. Among these individuals, 5645 (56.2%) were African American, 2547 (25.3%) were White, and 1068 (10.6%) were Hispanic. Vaccination efforts must specifically target populations that have been disproportionately affected by the pandemic. (Am J Public Health. 2022;112(1):29-33. https://doi.org/10.2105/AJPH.2021.306564).

Homeless Opioid Treatment Clients Transitioning to Dependent and Independent Housing: Differential Outcomes by Race/Ethnicity.

Purpose: Homeless opioid treatment clients who transition into housing generally demonstrate better outcomes, but housing environments vary widely and may not benefit racial/ethnic minority populations equally. We sought to identify how race/ethnicity moderates the association between positive opioid treatment response and moving into dependent or independent living environments. Methods: We used the Treatment Episode Dataset-Discharges (2018-2019) to identify outpatient treatment clients who were homeless at admission and indicated heroin or other opioids as their primary drug of choice (n = 20,021). We defined positive treatment response as a reduction in opioid use between admission and discharge. We used multivariable logistic regression to model treatment response. We included an interaction between housing at discharge (remained homeless [reference], dependent living, or independent living) and race/ethnicity, and adjusted for relevant confounders. Results: Transitioning from homeless to dependent living was positively associated with treatment response among White (aOR = 3.57, 95% CI = 3.15-4.06), Hispanic (aOR = 2.11, 95% CI = 1.55-2.86), and Black clients (aOR = 1.79, 95% CI = 1.41-2.27), but no association was observed for homeless American Indian/Alaska Native clients. Transitioning from homeless to independent living was strongly associated with treatment response among all groups with the strongest association observed among White clients (aOR = 4.70, 95% CI = 4.26-5.19). Conclusions: Interventions aimed at improving OUD treatment outcomes among homeless clients should identify individual and structural factors that support moving into fully housed and independent living environments. Dependent living offers needed support during crises, but should be temporary and priority should be placed on independent, permanent, and autonomous living environments.

Identifying community needs of the Hispanic faith community to develop a research agenda.

OBJECTIVE: To conduct a community health needs assessment among the Hispanic faith community population to develop a community-partnered research agenda. DESIGN: A cross-sectional design was used to conduct a community needs assessment using a community-based participatory research approach SAMPLE: Hispanic faith community members in Central Arkansas. MEASUREMENT: Data collection was led by Hispanic faith leaders using an audience response system at places of worship. An 88-item Community Health Needs Assessment survey was used that included demographic questions and questions related to five domains including community concerns, community resources, healthcare access, health concerns, and hunger and nutrition. RESULTS: There were a total of 100 participants in the community needs assessment. Hunger and nutrition was the highest ranking community concern followed by healthcare access. CONCLUSION: Based upon the results of the study, the university researchers and Hispanic faith community members have begun the initial steps to developing a research agenda to address the major concerns of the community.

The Role of State Medicaid Expansions in Integrating Comprehensive Mental Health Services into Opioid Treatment Programs: Differences Across the Rural/Urban Continuum.

Co-occurring mental health (MH) problems are common among those with opioid use disorders (OUDs). However, most opioid treatment programs (OTPs) do not provide MH services. We measured the association between state level characteristics (Medicaid expansion status and rurality) and MH/OUD services integration. We used a generalized linear model to estimate how the association between integration and Medicaid expansions varied across levels of rurality (National Survey on Substance Abuse Treatment Services; 2018; n = 1507 OTPs). The predicted probability of OTPs offering MH services decreased as rurality increased, and the strength of the negative association was greater in non-expansion states ([Formula: see text]=-0.038, SE = 0.005, p < 0.0001) than in expansion states ([Formula: see text]=-0.020, SE = 0.003, p < 0.0001). Access to integrated MH services was lowest in rural non-Medicaid expansion states, despite the high risk of opioid misuse and a high need for MAT and MH services in this population.

US trends in mask wearing during the COVID-19 pandemic depend on rurality.

INTRODUCTION: Face masks are widely recommended as a COVID-19 prevention strategy. State mask mandates have generally reduced the spread of the disease, but decisions to wear a mask depend on many factors. Recent increases in case rates in rural areas following initial outbreaks in more densely populated areas highlight the need to focus on prevention and education. Messaging about disease risk has faced challenges in rural areas in the past. While surges in cases within some communities are likely an impetus for behavior change, rising case rates likely explain only part of mask-wearing decisions. The current study examined the relationship between county-level indicators of rurality and mask wearing in the USA. METHODS: National data from the New York Times' COVID-19 cross-sectional mask survey was used to identify the percentage of a county's residents who reported always/frequently wearing a mask (2-14 July 2020). The New York Times' COVID-19 data repository was used to calculate county-level daily case rates for the 2 weeks preceding the mask survey (15 June - 1 July 2020), and defined county rurality using the Index of Relative Rurality (n=3103 counties). Multivariate linear regression was used to predict mask wearing across levels of rurality. The model was adjusted for daily case rates and other relevant county-level confounders, including county-level indicators of age, race/ethnicity, gender, political partisanship, income inequality, and whether each county was subject to a statewide mask mandate. RESULTS: Large clusters of counties with high rurality and low mask wearing were observed in the Midwest, upper Midwest, and mountainous West. Holding daily case rates and other county characteristics constant, the predicted probability of wearing a mask decreased significantly as counties became more rural (β=-0.560; p<0.0001). CONCLUSION: Upticks in COVID-19 cases and deaths in rural areas are expected to continue, and localized outbreaks will likely occur indefinitely. The present findings highlight the need to better understand the mechanisms underlying perceptions of COVID-19 risk in rural areas. Dissemination of scientifically correct and consistent information is critical during national emergencies.

Racial/Ethnic Differences in Drug- and Alcohol-Related Arrest Outcomes in a Southwest County From 2009 to 2018.

Objectives. To estimate the association between race/ethnicity and drug- and alcohol-related arrest outcomes.Methods. We used multinomial logistic regression and general estimating equations to estimate the association between race/ethnicity and arrest outcomes in 36 073 drug- and alcohol-related arrests obtained from administrative records in a Southwest US county from 2009 to 2018. Results were stratified by charge type.Results. Among misdemeanor drug- and alcohol-related arrests, American Indian/Alaska Native (AI/AN; adjusted odds ratio [AOR] = 3.60; 95% confidence interval [CI] = 3.32, 3.90), Latino (AOR = 1.53; 95% CI = 1.35, 1.73), and Black persons (AOR = 1.28; 95% CI = 1.05, 1.55) were more likely than White persons to be booked into jail as opposed to cited and released. AI/AN (AOR = 10.77; 95% CI = 9.40, 12.35), Latino (AOR = 2.63; 95% CI = 2.12, 3.28), and Black persons (AOR = 1.84; 95% CI = 1.19, 2.84) also were more likely than White persons to be convicted and serve time for their misdemeanor charges. Results were similar for felony drug- and alcohol-related arrests aggregated and stratified.Conclusions. Our results suggest that race/ethnicity is associated with outcomes in drug-related arrests and that overrepresentation of racial/ethnic minorities in the criminal justice system cannot be attributed to greater use of drugs and alcohol in general.

Gender Differences in Use of Alcohol Treatment Services and Reasons for Nonuse in a National Sample.

BACKGROUND: The majority of adults with alcohol use disorders do not obtain help, and women are less likely to utilize alcohol services than men. We sought to quantify gender differences in alcohol services utilization, overall and by type, using national longitudinal data and to explore potential gender differences in perceived need for help and reasons for not seeking help. METHODS: We analyzed data from the National Epidemiologic Survey on Alcohol and Related Conditions from White, African American, and Hispanic adults (n = 2,592) who met DSM-IV criteria for alcohol abuse or dependence at Wave 1 (2000 to 2001). We tested gender differences in Wave 2 (2004 to 2005) services utilization, perceived need for help, and treatment barriers using Rao-Scott chi-square tests and assessed predictors of outcomes in multivariable logistic regression, adjusting for problem severity, co-occurring disorders, and demographics. RESULTS: Women had much lower odds than men of utilizing any alcohol service (adjusted odds ratio [aOR] 0.53; 95% confidence interval [95% CI]: 0.33, 0.86), specialty services (aOR 0.41; 95% CI 0.19, 0.87), and 12-step groups (aOR 0.39; 95% CI 0.21, 0.71). Perceived need for help among those who had not used any services was very low (5%), with no gender difference. Further, men and women reported equivalent numbers of treatment barriers and the same rank order for the most frequently endorsed barriers. However, women were twice as likely as men to think a problem would get better by itself-the most frequent reason for not seeking help (47% vs. 24%, p < 0.001), and men were more likely than women to report unsuccessful past help-seeking and not thinking anyone could help (19% vs. 3%, p < 0.001 and 17% vs. 5%, p = 0.001, respectively). CONCLUSIONS: Consistent with previous studies, women were less likely to utilize alcohol services than men. Future interventions should address low problem recognition, and tailoring to gender-specific barriers may help close the disparity in services utilization.

Microaggressions and marijuana use among college students.

This study examines the association between exposure to microaggressions and marijuana use, using original survey data from a sample of racial/ethnic minority college students (n = 332) from a large Division I university in the United States. Nearly all of our sample (96%) reported at least one experience with microaggressions in the past 6 months, while 33% reported using marijuana regularly. We modeled regular use of marijuana using multiple logistic regression, with consideration of sex, age, race/ethnicity, and microaggression scale scores as covariates. Age, sex, the microinvalidations subscale score, and the full microaggression scale score were significantly associated with marijuana use in our full models (p < .01; p = .01; p = .02; p = .03, respectively). With each additional experience of microaggression, the odds of regular marijuana use increase. Academic communities may consider the primary prevention of discriminatory behavior when addressing student substance use.

A brief report on Hispanic youth marijuana use: Trends in substance abuse treatment admissions in the United States.

Increases in Hispanic youth admissions to substance abuse treatment programs for marijuana use are a growing public health concern. In this study, we investigated trends in Hispanic youth from 1995 to 2012 utilizing the Treatment Episode Data Set-Admissions of the Substance Abuse Mental Health Services Administration. Hispanic youth marijuana admissions are associated with youth 15-17 years old, in high school, and living in a dependent situation. Notably, female admissions increased at greater rates than males. Results also point to decreasing tolerance of minor marijuana use by schools and community agencies. Findings highlight the need for targeted, culturally specific, and cost-effective treatment and prevention efforts.

We are complex beings: comparison of statistical methods to capture and account for intersectionality.

OBJECTIVES: Intersectionality conceptualises how different parts of our identity compound, creating unique and multifaceted experiences of oppression. Our objective was to explore and compare several quantitative analytical approaches to measure interactions among four sociodemographic variables and interpret the relative impact of axes of marginalisation on self-reported health, to visualise the potential elevated impact of intersectionality on health outcomes. DESIGN: Secondary analysis of National Epidemiologic Survey on Alcohol and Related Conditions-III, a nationally representative cross-sectional study of 36 309 non-institutionalised US citizens aged 18 years or older. PRIMARY OUTCOME MEASURES: We assessed the effect of interactions among race/ethnicity, disability status, sexual orientation and income level on a self-reported health outcome with three approaches: non-intersectional multivariate regression, intersectional multivariate regression with a single multicategorical predictor variable and intersectional multivariate regression with two-way interactions. RESULTS: Multivariate regression with a single multicategorical predictor variable allows for more flexibility in a logistic regression problem. In the fully fitted model, compared with individuals who were white, above the poverty level, had no disability and were heterosexual (referent), only those who were white, above the poverty level, had no disability and were gay/lesbian/bisexual/not sure (LGBQ+) demonstrated no significant difference in the odds of reporting excellent/very good health (aOR=0.90, 95% CI=0.71 to 1.13, p=0.36). Multivariate regression with two-way interactions modelled the extent that the relationship between each predictor and outcome depended on the value of a third predictor variable, allowing social position variation at several intersections. For example, compared with heterosexual individuals, LGBQ+ individuals had lower odds of reporting better health among whites (aOR=0.94, 95% CI=0.93 to 0.95) but higher odds of reporting better health among Black Indigenous People of Color (BIPOC) individuals (aOR=1.13, 95% CI=1.11 to 1.15). CONCLUSION: These quantitative approaches help us to understand compounding intersectional experiences within healthcare, to plan interventions and policies that address multiple needs simultaneously.

A multilevel analysis of changing telehealth availability in opioid use disorder treatment settings: Conditional effects of rurality, the number and types of medication for opioid use disorder available, and time, US, 2016-2023.

PURPOSE: The opioid overdose crisis requires strengthening treatment systems with innovative technologies. How people use telehealth for opioid use disorder (OUD) is evolving and differs in rural versus urban areas, as telehealth is emerging as a local resource and complementary option to in-person treatment. We assessed changing trends in telehealth and medication for OUD (MOUD) and pinpoint locations of low telehealth and MOUD access. METHODS: We used national data from the Mental health and Addiction Treatment Tracking Repository (2016-2023) to identify specialty outpatient SUD treatment facilities in the United States (N = 83,988). We modeled the availability of telehealth using multilevel multivariable logistic regression, adjusting for covariates. We included a 3-way interaction to test for conditional effects of rurality, the number of MOUD medication types dispensed, and year. We included two random effects to account for clustering within counties and states. FINDINGS: We identified 495 facilities that offered both telehealth and all three MOUD medication types (methadone, buprenorphine, naltrexone) in 2023, clustered in the eastern United States. We identified a statistically significant 3-way interaction (p < 0.0001), indicating that telehealth in facilities that did not offer MOUD shifted from more telehealth in rural facilities in earlier years to more telehealth in urban facilities in later years. CONCLUSIONS: Treatment facilities that offer both telehealth and all three MOUD medication types may improve access for hard-to-reach populations. We stress the importance of continued health system strengthening and technological resources in vulnerable rural communities, while acknowledging a changing landscape of increased OUD incidence and MOUD demand in urban communities.

The association between rurality, dual Medicare/Medicaid eligibility and chronic conditions with telehealth utilization: An analysis of 2019-2020 national Medicare claims.

INTRODUCTION: Telehealth services have the potential to increase healthcare access among underserved populations, such as rural residents and racial/ethnic minority groups. The COVID-19 public health emergency led to unprecedented growth in telehealth utilization, but evidence suggests the growth has not been equitable across all patient populations. This study aimed to explore whether telehealth utilization and expansion changed equitably from 2019 to 2020 among sub-groups of Medicare beneficiaries. METHODS: We conducted an analysis of telehealth utilization among a 20% random sample of 2019-2020 Medicare beneficiaries on a national level. We fit multivariable logistic regression models and calculated average marginal effects (AME) to assess the association between demographic and clinical characteristics on telehealth utilization. RESULTS: We found telehealth utilization was less likely among non-Hispanic Black/African-American (2019: adjusted odds ratio [aOR] = 0.77, AME = -0.15; 2020: aOR = 0.85, AME = -3.50) and Hispanic (2019: aOR = 0.79, AME = -0.13; 2020: aOR = 0.87, AME = -2.89) beneficiaries, relative to non-Hispanic White beneficiaries in both 2019 and 2020, with larger disparities in 2020. Rural beneficiaries were more likely to utilize telehealth than urban beneficiaries in 2019 (aOR = 2.62, AME = 0.84), but less likely in 2020 (aOR = 0.57, AME = -14.47). In both years, dually eligible Medicare/Medicaid beneficiaries were more likely than non-dually eligible beneficiaries to utilize telehealth (2019: aOR = 4.75, AME = 0.84; 2020: aOR = 1.34, AME = 2.25). However, the effects of dual eligibility and rurality changed in both models as the number of chronic conditions increased. DISCUSSION: We found evidence of increasing disparities in telehealth utilization among several Medicare beneficiary sub-groups in 2020 relative to 2019, including individuals of minority race/ethnicity, rural residents, and dually eligible beneficiaries, with disparities increasing among individuals with more chronic conditions. Although telehealth has the potential to address health inequities, our findings suggest that many of the patients in greatest need of healthcare are least likely to utilize telehealth services.