Telehealth Adoption by Mental Health and Substance Use Disorder Treatment Facilities in the COVID-19 Pandemic

The Substance Abuse and Mental Health Services Administration (SAMHSA) collects data from nearly all mental health and substance use disorder treatment facilities in the United States by using the National Mental Health Services Survey (N-MHSS) and National Survey of Substance Abuse Treatment Services (N-SSATS). SAMHSA’s Behavioral Health Treatment Services Locator reports responses from the facilities that have agreed to be listed. The locator is updated continuously with information such as facility name, address, telephone number, and services offered (26). N-SSATS facilities were asked, “Which of the following clinical/therapeutic approaches listed below are used frequently at this facility?” One possible answer was “Telemedicine/telehealth therapy (including Internet, Web, mobile, and desktop programs).” For N-MHSS, the same response was possible; however, the question was worded slightly differently: “Which of these mental health treatment approaches are offered at this facility, at this location?” Facilities that marked a checkmark for this option were defined as offering telehealth.

The data set utilized for this study, the Mental health and Addiction Treatment Tracking Repository, draws daily updates from SAMHSA (27). The data were collected in 2020 and 2021. We excluded facilities that are part of the U.S. Department of Veterans Affairs system, given that they have a unique telehealth system that differs from those of civilian providers and that has existed since before the COVID-19 pandemic. We also restricted our analysis to outpatient treatment facilities, given that in-person visits to these facilities have been reduced in response to the physical distancing precautions required to combat the spread of COVID-19 (28).

We identified outpatient facilities in both the N-MHSS and N-SSATS on the basis of their response to two different questions. In the N-MHSS, the question is as follows: “Mental health treatment is provided in which of the following service settings at this facility, at this location?” A possible response is “outpatient.” In the N-SSATS the question is as follows: “Does this facility offer OUTPATIENT substance abuse services at this location, that is, the location listed on the front cover?” Facilities responding “yes” were classified as an outpatient facility. Our sample included 15,691 outpatient behavioral health treatment facilities—6,601 outpatient mental health and 9,090 outpatient substance use disorder treatment facilities.

We calculated the number and percentage of outpatient mental health and substance use disorder treatment facilities that offered telehealth on January 20, 2020, and January 20, 2021.

We included measures of facility- and county-level characteristics. The facility-level characteristics were from responses to the N-MHSS and N-SSATS within the Behavioral Health Treatment Services Locator, including the service setting of the treatment facility (outpatient only versus inpatient and outpatient), whether the treatment facility accepts Medicaid as a form of payment, whether the treatment facility accepts private insurance as a form of payment, whether the facility provides care for adults only, and the ownership of the facility (public, private not for profit, and private for profit).

County-level characteristics included several features by which we expected that health care service availability may differ on the basis of evidence from the existing literature (29–31). Specifically, they included number of COVID-19 cases per 10,000 population from USAFacts (32), urbanicity (33), the share of the county population that speaks only English in the household from the American Community Survey (ACS) (34), the share of the county population without broadband Internet from the Federal Communications Commission (35), the percentage of the population below the federal poverty level from the ACS, the percentage of the population that is black non-Hispanic from the ACS, the percentage of the population that is white non-Hispanic from the ACS, and finally the percentage of the population that is Hispanic from the ACS.

We calculated changes in the number and percentage of the two types of treatment facilities that offered telehealth between January 20, 2020, and January 20, 2021. As a subsequent step, we stratified these estimates according to facility- and county-level characteristics, as described above.

To identify risk factors for facilities not offering telehealth care on January 20, 2021, we executed multivariable mixed-effects linear probability regression analysis. In addition to the facility- and county-level variables as described above, the models also included state fixed effects. Standard errors were clustered at the state level.

Finally, to aid interpretation of these findings, we generated marginal means from the regression models (36). This provides predicted probabilities of not offering telehealth services by January 2021, which can help identify risk factors for not offering telehealth in 2021. All analyses were executed with Stata, version 16.0 (37). The study was deemed exempt by the RAND Corporation’s Human Subjects Protection Committee. All analyses were done in 2021.

As of January 2020, only 33% (N=5,113) of the 15,691 outpatient behavioral health treatment facilities offered telehealth: 42% (N=2,791) of the 6,601 mental health treatment facilities and 26% (N=2,322) of the 9,090 substance use disorder treatment facilities. By January 2021, the percentage of outpatient behavioral health treatment facilities offering telehealth rose to 61% (N=9,639) overall, with 68% (N=4,460) among mental health facilities and 57% (N=5,179) among substance use disorder treatment facilities.

Table 1 summarizes these increases for outpatient mental health and substance use disorder treatment facilities, according to facility- and county-level characteristics. Although the level of telehealth availability between 2020 and 2021 increased substantially for each census region, the increase was not even across regions. The largest increase in the percentage of mental health and substance use disorder treatment facilities offering telehealth occurred in the Northeast (205% and 246% increase, respectively). The magnitude of increase—for almost all characteristics—was higher for substance use disorder treatment facilities than for mental health treatment facilities. For outpatient mental health treatment facilities, percentage changes in nonmetropolitan counties (26%) were smaller than for substance use disorder treatment facilities (79%). For outpatient substance use disorder facilities, higher percentage changes occurred in counties with lower percentages of English-only speaking households (lowest quartile, 147%). For both types of facility, we also found that as the percentage of individuals in poverty in a county increased, so did the availability of telehealth.

Uptake of telehealth services increased in many areas of the country. (Two U.S. maps showing the distribution of telehealth coverage among outpatient mental health treatment facilities at the county level on January 20, 2020, and on January 20, 2021, are included in an online supplement to this article.) The shifting color gradient indicates counties that saw increases of facilities offering telehealth. A total of 246 counties, with a combined population of 23,117,451, that did not have a mental health treatment facility that offered telehealth in 2020 gained at least one that did so in 2021. Nevertheless, as of January 2021, a large percentage of counties (43% of all counties), with a total population of 35,800,015 individuals (11% of the total population), still lacked any mental health treatment facilities offering telehealth.

Adoption of telehealth services by substance use disorder treatment facilities increased across the country, especially in the Midwest and South (see second set of maps in online supplement). A total of 475 counties, with a combined population of 37,992,885, that did not have a substance use disorder treatment facility that offered telehealth in 2020 gained at least one that did so in 2021. Nevertheless, as of January 2021, a large percentage of counties (49% of all counties), with a total population of 36,068,677 individuals (11% of the total population), still lacked any substance use disorder treatment facilities offering telehealth.

Table 2 presents the predicted probabilities of not offering telehealth in 2021 by facility- and county-level characteristics. We report some of the results with statistically significant differences below. The likelihood of not offering telehealth in 2021 differed by region. Compared with mental health facilities in the Northeast (predicted probability of not offering telehealth=0.460, 95% confidence interval [CI]=0.419–0.500), facilities in the three other census regions were all more likely to offer telehealth in 2021: West, predicted probability=0.257, 95% CI=0.218–0.296; Midwest, predicted probability=0.308, 95% CI=0.254–0.363; and South, predicted probability=0.277, 95% CI=0.236–0.318. The difference between the Northeast and the other regions was significant (p<0.001). In contrast, substance use disorder treatment facilities in the Midwest were less likely to offer telehealth in 2021 (predicted probability=0.460, 95% CI=0.412–0.508), compared with such facilities in the South (predicted probability=0.400, 95% CI=0.376–0.425). The difference was significant (p=0.028).

Mental health treatment facilities that did not accept Medicaid as a form of payment were more likely to not offer telehealth in 2021 (predicted probability of not offering telehealth=0.432, 95% CI=0.368–0.500), compared with facilities that accepted Medicaid (predicted probability=0.312, 95% CI=0.291–0.335). The difference was statistically significant (p<0.001). We also found that mental health treatment facilities that did not accept private insurance as payment had a higher probability of not offering telehealth in 2021 (predicted probability=0.409, 95% CI=0.346–0.472), compared with those that accepted private insurance (predicted probability=0.306, 95% CI=0.283–0.329). The difference was statistically significant (p=0.002). We found similar results for substance use disorder treatment facilities that differed only slightly in their magnitude but that were also significant (p<0.001).

Mental health treatment facilities in metropolitan areas had a higher probability of not offering telehealth in 2021 (predicted probability=0.342, 95% CI=0.315–0.369), compared with facilities in nonmetropolitan areas (predicted probability=0.269, 95% CI=0.233–0.304). The difference was statistically significant (p=0.001). For substance use disorder treatment facilities, we did not find a statistically significant difference in the likelihood of not offering telehealth in 2021 by urbanicity.

No significant differences were noted for either mental health or substance use disorder treatment facilities on the basis of a facility’s ownership or the county-level measures of percentage of households speaking only English and the percentage of the population below the federal poverty level.

In this longitudinal analysis of 15,691 outpatient mental health and substance use disorder treatment facilities throughout the United States, we found that telehealth service availability among mental health treatment facilities increased by approximately 77% from January 2020 to January 2021. The increase over the same period was 143% among substance use disorder treatment facilities. We also found differential increases in telehealth adoption across regions. Although facilities in the Northeast experienced the largest increase in telehealth adoption, the region still lagged behind others in terms of telehealth availability per population, especially among outpatient mental health treatment facilities. We also found that outpatient mental health treatment facilities in metropolitan areas were less likely to offer telehealth in 2021, compared with facilities in nonmetropolitan areas. We suspect that a potential cause of the differences in telehealth adoption geographically could be attributable to disproportionate workforce disruptions and existing workforce availability issues. A survey conducted by the National Council for Behavioral Health in April 2020 found that almost half of responding behavioral health organizations throughout the country reported furloughing or discharging employees as a result of COVID-19 (38).

To a large extent, the rapid adoption of telehealth at outpatient mental health and substance use disorder treatment facilities demonstrates the responsiveness of the U.S. health systems to accommodate the constraints introduced by the COVID-19 pandemic. Several case studies have delineated health systems’ efforts to expand telehealth services across the nation during the pandemic (18, 39). For facilities that successfully achieved the transition to telehealth, questions remain about whether and to what extent federal and state governments will maintain policies that support telehealth for behavioral health services (13, 40, 41). Recently, the Centers for Medicare and Medicaid Services announced that Medicare will continue reimbursement codes for telehealth even after the pandemic has been controlled (42). It will be interesting for future studies to examine whether such policies will help maintain or further expand the current telehealth coverage, given our finding that 32% of mental health treatment facilities and 43% of substance use disorder treatment facilities in our sample still did not offer telehealth services in January 2021, nearly 1 year into the pandemic.

We found several significant risk factors that predicted a lack of uptake in telehealth services in 2021, including forms of payment accepted. For both mental health and substance use disorder treatment facilities, not accepting Medicaid was a predictor of not offering telehealth in 2021. One potential explanation for this is that Medicaid predominantly serves low-income, high-needs individuals who may contend with transportation and other barriers to care (43, 44). As such, facilities accepting Medicaid payment may have had a stronger motivation to increase telehealth services. Historically, changes in Medicaid reimbursement have led to changes in telehealth adoption (45). Among substance use disorder treatment facilities, not accepting private insurance payment was a significant predictor of not offering telehealth in 2021. Again, changes in reimbursement could be responsible for offering telehealth by facilities that accept private insurance as payment. For example, 17 states required reimbursement parity for telehealth and in-person services for non–COVID-19 patients (46). Further research is needed to track changes in federal- and state-level policies regarding private insurance coverage of telehealth and to assess how the changes affect telehealth adoption by behavioral health facilities.

This study had several limitations. First, not all outpatient mental health and substance use disorder treatment facilities are in SAMHSA’s Behavioral Health Treatment Services Locator database. Although the vast majority of facilities choose to be listed in the database, it remains unclear what percentage of all facilities decline to be included in it. Second, the locator data do not contain information on treatment capacity or quality, and thus we were not able to determine how many patients could be treated virtually by the facilities that offered telehealth or to assess the quality of care. Third, the Behavioral Health Treatment Services Locator data do not differentiate between telehealth provided by telephone or by video.

Looking forward, outpatient mental health and substance use disorder treatment facilities may consider the long-term implications of maintaining telehealth operations, even if only as part of a broader portfolio of in-person and community-based services. Several studies have indicated reductions in no-show rates for telehealth visits, compared with in-person visits (47, 48); an anticipated potential reduction in overhead for physical infrastructure (49); and new costs necessary to convert providers’ personal spaces (50). From an analytic perspective, our results also underscore the relevance of SAMHSA’s Behavioral Health Treatment Services Locator for regularly tracking service availability at the facility level across the nation. Prior studies have utilized this information cross-sectionally (29–31, 51–56). Future research may use it to monitor changes in telehealth availability among facilities as the pandemic evolves.

During the COVID-19 pandemic, the percentage of outpatient mental health and substance use disorder treatment facilities offering telehealth has grown dramatically. However, our analyses also indicated that considerable proportions of mental health and substance use disorder treatment facilities still did not offer telehealth as of January 2021, and we identified significant factors associated with this finding. If policy makers wish to increase the number and percentage of behavioral health facilities that provide telehealth, they should consider collecting additional information about why facilities did not institute telehealth during the COVID-19 pandemic. Separately, more work needs to be undertaken to ascertain how increased telehealth availability affects behavioral health care utilization and quality.

The authors thank Matthew Cefalu, Ph.D., for providing insight on the analytic approach.