Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Mar 21.
Published before final editing as: J Subst Use Addict Treat. 2026 Feb 24:209926. doi: 10.1016/j.josat.2026.209926

A Longitudinal Analysis of Overall and Telehealth Addiction Treatment Utilization Three Years after Pandemic-Related Policy Changes

Subarna Bhattacharya a, Vanessa A Palzes a, Monique B Does a, Ninad R Athale b, Cynthia I Campbell a,c,d
PMCID: PMC13003766  NIHMSID: NIHMS2152115  PMID: 41748073

Abstract

Purpose

Studies have identified differences by patient characteristics in addiction treatment utilization in the early COVID-19 pandemic period, yet an understanding of longitudinal changes in utilization patterns remains unclear. We examined treatment utilization trends over three years post-pandemic, with a particular focus on differences by age and race and ethnicity.

Methods

Using electronic health record data, this retrospective cohort study examined overall and telehealth addiction treatment initiation and engagement 3 years pre- and 3 years post-pandemic (3/16/2020) following identification of 170,618 episodes involving problematic substance use among 124,413 adults in a large, integrated Northern California health system. Interrupted time series models were fit to examine annual utilization rates during pre- and post-pandemic periods (3/1/2017–1/15/2020 and 5/17/2020–2/28/2023, respectively), and level- and trend-changes in utilization from pre- to post-pandemic, overall and by age group and race and ethnicity.

Results

Overall treatment initiation decreased from 27.0% to 24.2% during the pre-pandemic period by approximately 2% annually (RR [95% CI]=0.98 [0.96, 0.99]), increased by 6% after the onset of the pandemic (1.06 [1.03, 1.10]), and then decreased by 2% annually in the post-pandemic period to 23.0% (0.98 [0.79, 0.99]). Telehealth initiation increased from 1.9% to 2.6% during the pre-pandemic period by 12% annually (RR [95% CI]=1.12 [1.06, 1.19]), increased five-fold immediately after pandemic onset (RR [95% CI]=5.14 [4.62, 5.72]), and then decreased by 10% annually (RR [95% CI]=0.90 [0.88, 0.92]). Overall and telehealth engagement followed similar patterns. Pre- to post-pandemic trends in utilization varied by age group and slightly by race and ethnicity, which may have been primarily driven by initial increases in utilization at the onset of the pandemic.

Conclusions

Following immediate increases in treatment initiation and engagement during the pandemic, utilization via telehealth decreased slightly over time. Availability of telehealth was not associated with increased or sustained utilization over time. Despite some variation in trends over time by age group and race and ethnicity, we did not find strong evidence of differences across groups.

1. Introduction

In 2023, 59% of Americans aged 12 years and older used substances and 17.1% were diagnosed with a substance use disorder (SUD); the highest prevalence was in young adults, with 27.1% of 18–25 year olds diagnosed with an SUD (Substance Abuse and Mental Health Services Administration, 2024).

Effective treatments for addiction exist, including psychotherapy and medications for opioid use disorder (OUD) and alcohol use disorder (AUD). Prior to the pandemic, specialty addiction treatment was primarily delivered in person (Cleveland Clinic, 2024; Uscher-Pines et al., 2020). However, the pandemic disrupted the typical model of care at a time when increased social isolation and added stressors (such as financial and employment-related barriers) led to an increase in substance use throughout the United States; 13% of Americans reported increasing substance use to cope with pandemic-related stressors (Abramson, 2021).

During the SARS-CoV-2 pandemic, local and state policies were implemented to mitigate the spread of infection (e.g., lockdowns, limits on in-person gatherings), which led in-person group therapy to shift to virtual formats (Oesterle et al., 2020). At the same time, federal and state laws expanded the coverage of telehealth services and relaxed restrictions on buprenorphine prescribing for OUD (Centers for Medicare and Medicaid Services, 2020), encouraging healthcare systems and providers to pivot to providing care primarily via telehealth modalities (Cantor et al., 2022). The availability for addiction treatment programs to provide telehealth options increased by about 143% from 2020 to 2021 (Cantor et al., 2022). However, these telehealth options may not have fully reached those who had limited or no access to technology or internet; telehealth was generally more accessible for younger patients, while those in older age groups less accustomed to navigating virtual modalities faced barriers in accessing telehealth (Lin et al., 2022).

Many studies have examined the early impacts of pandemic-related policy changes, such as restrictions on in-person care and a transition to telehealth visits, on addiction treatment utilization in different settings, with mixed findings (Choi et al., 2024; C. Kruse et al., 2020; C. S. Kruse et al., 2017; Palzes et al., 2023). A previous study by this research team in an insured population found no variation in overall and telehealth addiction treatment initiation by race and ethnicity or socioeconomic status 10 months after the onset of the pandemic, although telehealth treatment initiation increased for younger adults compared to their older counterparts (Palzes et al., 2023). However, another study of Medicaid patients in New York found that SUD medication initiation was disrupted for racial/ethnic minority groups during the early pandemic, particularly for the non-Hispanic Black community (Perry et al., 2023). Similarly, a systematic review focusing on the Asian American, Native Hawaiian, and Pacific Islander population found that this group remained underrepresented in addiction treatment throughout the pandemic (Choi et al., 2024). A study of Massachusetts patients with SUD found an increase in addiction treatment visits during the immediate pandemic surge (March 2020) only among non-Hispanic White patients, with a decrease in visits during the pandemic reopening period (June 2020) among Hispanic and non-Hispanic Black patients (Yang et al., 2020). Additionally, age differences remain prevalent particularly in terms of telehealth treatment utilization, as older patients tend to have lower rates of telehealth use due to limited experience and levels of comfort with technology (C. Kruse et al., 2020). Understanding these differences further can help inform ways to broaden the reach of telehealth across age groups, such as by addressing technological distrust and cost (C. S. Kruse et al., 2017).

No studies to our knowledge have examined the long-term impacts of pandemic related policy changes on addiction treatment utilization, particularly by age and race and ethnicity. This study examines longitudinal changes in overall and telehealth addiction treatment utilization, comparing trends from three years prior to the pandemic to three years post-pandemic, overall and by age group and race and ethnicity.

2. Methods

2.1. Setting

This study (NIDA Clinical Trials Network-0117) used electronic health records (EHR) data from Kaiser Permanente Northern California (KPNC), an integrated health care delivery organization with a large, diverse, community-based population of over 4 million patients, who are representative of the local and statewide insured population (Davis et al., 2023). KPNC has 21 Addiction Medicine and Recovery Services (AMRS) specialty departments in 15 different service areas, which provide a group-based model of care, medication for AUD and OUD, and individual services as needed. Patients do not need prior physician approval to seek AMRS treatment. KPNC offered telehealth psychotherapeutic visits relatively quickly after the pandemic onset – the capacity and workflows for doing so improved over time (Ridout et al., 2021). In-person options for medical detox and prescribing visits were always available.

This study was approved by the KPNC Institutional Review Board and waived informed consent. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines for cohort studies (von Elm et al., 2008).

2.2. Study Design

In this retrospective cohort study, we compared trends in addiction treatment utilization between pre-pandemic (3/1/2017–1/15/2020) and post-pandemic (5/17/2020–2/28/2023) periods using interrupted time series analysis. We defined the pandemic as beginning on March 16, 2020, the day lockdowns went into effect in the region (San Francisco Chronicle, 2021). We followed the Healthcare Effectiveness Data and Information Set 2023 (HEDIS) Specifications for Initiation and Engagement of Substance Use Treatment to identify the study sample, specify the inclusion/exclusion criteria, set follow-up duration, and measure addiction treatment utilization outcomes (National Committee for Quality Assurance, 2022). HEDIS is a performance improvement tool created by the National Committee for Quality Assurance that standardizes the measurement of important care quality metrics across health care systems.

2.3. Sample

We identified index episodes of problematic substance use among KPNC patients 18 years of age and older between March 1, 2017 to February 28, 2023, Based on Healthcare Effectiveness Data and Information Set 2023 (HEDIS) specifications (National Committee for Quality Assurance, 2022). Problematic substance use was defined as receiving either a drug use disorder (International Classification of Diseases, 10th Revision [ICD-10] diagnosis codes F11–16 and F18–19, excluding remission codes), or an “unhealthy substance use behavior” diagnosis at any outpatient, partial hospitalization or intensive outpatient, acute or non-acute inpatient, emergency department, or telehealth (telephone or video) encounter, or in a claim. Drug use disorder codes included those for opioids, cannabis, sedatives, cocaine, stimulants, inhalants, and other psychoactive substance use. Unhealthy substance use behavior diagnoses were identified using healthcare system-specific diagnosis codes that standardize to an ICD-10 code of Z72.89. The unit of analysis was an episode of problematic substance use; allowing patients to have multiple episodes. We utilized a negative diagnosis and medication history of 194 days to identify new episodes of problematic substance use (National Committee for Quality Assurance, 2022). Episode index dates were defined as the new diagnosis date (or discharge date for inpatient stays). All episodes were followed for 47 days after the index date (National Committee for Quality Assurance, 2022). We excluded episodes for patients who were not continuously enrolled in the 194 days before the index date, as well as those who died or disenrolled during the 47-day follow up period. We also implemented a washout period by excluding episodes that occurred in the two months before and after the start of the pandemic (January 16 – May 16, 2020) to prevent pre-pandemic outcomes from crossing over into the post- pandemic period, and to account for the time before telehealth treatment options became available in the healthcare system. The resulting analytical sample included 170,618 episodes of problematic substance use.

2.4. Exposure

Our primary exposure of interest was time, which was segmented into pre-pandemic and post-pandemic onset periods to capture multiple local, state, and healthcare system policy changes that could affect access and utilization of addiction treatment (e.g. regional lockdowns, expanded coverage of telehealth services, relaxed buprenorphine prescribing restrictions, increased healthcare system capacity to deliver telehealth services).

2.5. Moderators

We assessed potential differences by age group (18–34, 35–49, 50–64, and 65+ years old at index) and race and ethnicity as documented in pre-determined categories in the EHR (American Indian/Alaska Native, Asian/Native Hawaiian/Pacific Islander, Black, Latino/Hispanic, White, and Other/Unknown).

2.6. Outcomes

Outcomes were based on HEDIS measures for initiation and engagement of addiction treatment (“HEDIS Measures and Technical Resources,” 2023). Treatment for any substance use problem was examined, regardless of the index episode diagnosis.

Overall addiction treatment initiation was defined as initiating treatment through an eligible encounter (e.g. inpatient admission, outpatient visit, intensive outpatient encounter, partial hospitalization, or telehealth visit) or receiving pharmacotherapy for an AUD or OUD within 14 days of the episode index date (National Committee for Quality Assurance, 2022). Outpatient visits (via in-person or telehealth) included individual and group therapy. Pharmacotherapy events included dispensation or administration of the following medications: acamprosate, disulfiram, naltrexone (oral and injectable), buprenorphine (sublingual or injectable), buprenorphine/naloxone, and methadone (included through claims data from opioid treatment programs). If the episode was an inpatient stay, we automatically considered it as initiation of treatment, per HEDIS specifications.

Telehealth addiction treatment initiation was defined as initiating treatment through a telehealth visit (any telephone or video visit, e-visit, or virtual check-in) within 14 days of the episode index date (National Committee for Quality Assurance, 2022).

Overall addiction treatment engagement was defined as meeting criteria for overall addiction treatment initiation and having any of the following on the day after the initiation encounter (or inpatient discharge) through 34 days after the initiation event: ≥1 weekly or monthly opioid treatment service with medication administration, ≥1 long-acting medication administration events, or ≥2 eligible treatment encounters or medication treatment events (National Committee for Quality Assurance, 2022).

Telehealth addiction treatment engagement was defined as meeting criteria for overall addiction treatment initiation and having at least one eligible treatment encounter post-initiation via telehealth.

2.7. Covariates

The following covariates were included: age group (18–34, 35–49, 50–64, 65+ years), sex (male, female), race and ethnicity (White, Asian/Native Hawaiian/Pacific Islander, Black, Latino/Hispanic, American Indian/Alaska Native, Other/Unknown), co-occurring mental health diagnoses identified using ICD-10 diagnosis codes (attention-deficit/hyperactivity disorder, anxiety disorder, obsessive-compulsive disorder, post-traumatic stress disorder, any eating disorder, depression, any other mood disorder, or any psychotic disorder), Charlson comorbidity index (0, 1–2, 3+) (Deyo et al., 1992), and index diagnosis (SUD diagnosis or unhealthy substance use behavior diagnosis). An indicator for month of the year (1–12) was included to control for seasonality. Insurance type (Medicaid, Medicare, commercial, other/unknown) and an indicator for whether the patient enrolled in KPNC via the California Healthcare Exchange were included to account for differences in how people enrolled in the health plan over time. The neighborhood deprivation index (NDI) was included as a measure of neighborhood socioeconomic status, calculated from census tract-level geocoded data of income/poverty, education, employment, housing, and occupation (Messer et al., 2006). NDI was categorized based on the distribution of KPNC members, where higher quartiles indicate lower socioeconomic status; missing values were imputed to the mean and categorized into the 3rd quartile.

2.8. Statistical Analysis

For each addiction treatment utilization outcome, we conducted interrupted time series analysis with segmented regression models. To account for within-patient correlations due to some patients having multiple episodes, we utilized generalized estimating equations. To allow estimation of rate ratios (RRs), we utilized a Poisson distribution with robust standard errors. Each of the models included a variable for day (incrementing for each day since the start of the pre-pandemic period) and a time indicator for days after the start of the pandemic period, allowing estimation of the change in the daily rate of treatment utilization during the pre- pandemic period (i.e., the pre-pandemic trend), immediate level change in the daily rate from pre- to post- pandemic onset (i.e., level change), and change in the daily rate comparing the two periods (i.e., trend change) (Wagner et al., 2002). Models were adjusted for all covariates listed previously. From the models, we estimated rate ratios and 95% confidence intervals (CIs) for pre- and post-pandemic trends (slopes) and scaled them by 365 days to enable interpretation of annual trends. We also compared pre- and post-pandemic annual trends by estimating the ratio of rate ratios (RRR); their p-values represent the magnitude of the trend change in utilization. Line plots for outcomes were generated using predicted probabilities from each model, averaged by week. Subgroup differences were examined by including interaction terms between time variables (day and days post-pandemic) and the covariate (either age group or race and ethnicity). Tests of interaction terms with p-values <0.05 represented significant differences between each subgroup and a reference subgroup. All statistical analyses were conducted using SAS Version 9.4, and all data visualizations were created using RStudio Version 4.3.1.

3. Results

3.1. Sample Characteristics

The analytical samples included 85,701 episodes of problematic substance use among 71,665 patients in the pre-pandemic period (03/01/2017–01/15/2020), and 84,917 episodes among 71,821 patients in the post-pandemic period (05/17/2020–02/28/2023); 12,831 patients had episodes in both time periods (Table 1; Figure S1). Across both periods, 170,618 episodes of problematic substance use among 124,413 patients were included.

Table 1.

Characteristics of episodes with problematic substance use during pre-pandemic and post-pandemic periods, Kaiser Permanente Northern California

Characteristic Pre-Pandemic, N = 85,701 Post-Pandemic, N = 84,917 p-valuea
Unique Patients (n) 71,665 71,821
Age (years), mean (sd) 40 (17) 40 (17) 0.084
Age Group (years)
    18–34 40,432 (47.2%) 40,630 (47.8%)
    35–49 18,214 (21.3%) 19,514 (23.0%)
    50–64 17,229 (20.1%) 15,222 (17.9%)
    65+ 9,826 (11.5%) 9,551 (11.2%)
Sex, n (%) <0.001
    Male 49,632 (57.9%) 47,510 (55.9%)
    Female 36,069 (42.1%) 37,407 (44.1%)
Race and ethnicity, n (%) <0.001
    White 45,535 (53.1%) 42,063 (49.5%)
    Asian/Native Hawaiian/Pacific Islander 6,424 (7.5%) 6,774 (8.0%)
    Black 13,021 (15.2%) 13,285 (15.6%)
    Latino/Hispanic 17,583 (20.5%) 19,073 (22.5%)
    American Indian/Alaska Native 1,440 (1.7%) 1,285 (1.5%)
    Other/Unknown 1,698 (2.0%) 2,437 (2.9%)
Insurance Type, n (%) <0.001
    Medicaid 12,541 (14.6%) 15,168 (17.9%)
    Medicare 14,695 (17.1%) 11,229 (13.2%)
    Commercial 58,160 (67.9%) 58,281 (68.6%)
    Other/Unknown 305 (0.4%) 239 (0.3%)
California Healthcare Exchange, n (%) 5,322 (6.2%) 7,340 (8.6%) <0.001
Neighborhood Deprivation Index Quartile, n (%) <0.001
    1st (lowest) 13,176 (15.4%) 14,560 (17.1%)
    2nd 23,684 (27.6%) 22,323 (26.3%)
    3rd 25,458 (29.7%) 25,177 (29.6%)
    4th (highest) 23,383 (27.3%) 22,857 (26.9%)
Alcohol Use Disorder Diagnosis, n (%) 7,785 (9.1%) 7,759 (9.1%) 0.7
Mental Health Diagnosis (any), n (%) 46,788 (54.6%) 48,592 (57.2%) <0.001
Any Substance Use Disorder Diagnosis, n (%) 59,249 (69.1%) 55,267 (65.1%) <0.001
Cannabis 32,627 (38.1%) 32,043 (37.7%) 0.2
Cocaine 5,004 (5.8%) 4,060 (4.8%) <0.001
Hallucinogen 530 (0.6%) 604 (0.7%) 0.018
Inhalant 119 (0.1%) 133 (0.2%) 0.3
Opioid 12,121 (14.1%) 9,418 (11.1%) <0.001
Sedative/Hypnotic/Anxiolytic 2,430 (2.8%) 1,851 (2.2%) <0.001
Stimulant 9,408 (11.0%) 9,062 (10.7%) 0.042
Other DUD 8,099 (9.5%) 7,353 (8.7%) <0.001
Unhealthy Substance Use Behavior Diagnosis, n (%) 26,824 (31.3%) 29,930 (35.2%) <0.001
Charlson Comorbidity Index, n (%) <0.001
    0 57,891 (67.5%) 58,386 (68.8%)
    1–2 18,182 (21.2%) 17,734 (20.9%)
    3+ 9,628 (11.2%) 8,797 (10.4%)
Open in a new tab
a

The pre-pandemic period was March 1, 2017 to January 15, 2020, and the post-pandemic period was May 17, 2020 to February 28, 2023.

b

P-values from Wilcoxon rank sum and Pearson’s Chi-squared tests.

In the pre-pandemic sample of SUD episodes, the mean age was 40 years (SD = 17); 47.2% were aged 18–34 years (Table 1). The pre-pandemic sample was 57.9% male, 53.1% White, 7.5% Asian/Native Hawaiian/Pacific Islander, 15.2% Black, 20.5% Latino/Hispanic, 1.7% American Indian/Alaska Native, and 2.0% had another or unknown race and ethnicity (Table 1). In the post-pandemic sample, the average age was 40 years (SD = 17); 47.8% were aged 18–34 years (Table 1). The post-pandemic sample was 55.9% male, 49.5% White, 8.0% Asian/Native Hawaiian/Pacific Islander, 15.6% Black, 22.5% Latino/Hispanic, 1.5% American Indian/Alaska Native, and 2.9% had another or unknown race and ethnicity (Table 1). About half of the samples had a mental health diagnosis (54.6% and 57.2% in the pre- and post-pandemic periods, respectively), and most had a SUD diagnosis (69.1% and 65.1% in the pre- and post- periods, respectively) (Table 1).

3.2. Treatment Initiation

3.2.1. Overall

Overall addiction treatment initiation decreased from 27.0% to 24.2% of episodes during the pre- pandemic period, by about 2% annually (RR [95% CI]=0.98 [0.96, 0.99]) (Figure 1). Initiation rates increased by 6% after the onset of the pandemic (level change RR [95% CI]=1.06 [1.03, 1.10]). Initiation rates decreased from 26.8% to 23.0% in the post-pandemic period, by 2% annually (RR [95% CI]=0.98 [0.79, 0.99]); pre- and post-pandemic annual trends did not significantly differ (trend change p=0.727) (Table S1).

Figure 1. Average weekly initiation and engagement of overall and telehealth addiction treatment.

Figure 1.

Open in a new tab

Data points represent the weekly unadjusted proportion of initiation or engagement. Lines represent the weekly adjusted proportion of initiation or engagement estimated from segmented Poisson regression models. The pre-pandemic period spans from March 1, 2017 to January 15, 2020. The post-pandemic period spans from May 17, 2020 to February 28, 2023. Episodes occurring during the washout period (January 16 – May 16, 2020, yellow dashed lines) were excluded. Estimated pre- and post-pandemic trend rate ratios (RRs), and pandemic level change RR, with 95% confidence intervals (CI), are also shown in the table.

Initiation trends varied by age group (Figure 2, Table S2). Patients aged 18–34 years had 22% higher initiation after the onset of the pandemic (level change RR [95% CI]=1.22 [1.13, 1.31]), which was greater than level changes observed in the groups aged 50–64 and 65+ years (p-values <0.001). There were no trend differences between patients aged 18–34 and 35–49 years. Initiation rates decreased similarly in both the pre- and post-pandemic periods for patients aged 18–34 years (RR [95% CI]=0.95 [0.93, 0.98], and 0.93 [0.90, 0.96], respectively; trend change p=0.209), while initiation rates for other age groups (35–49, 50–64, and 65+ years) did not significantly change over time in either time period. Initiation trends did not vary by race and ethnicity (Table S3, Figure S2).

Figure 2. Average weekly initiation and engagement of overall and telehealth addiction treatment by age group.

Figure 2.

Open in a new tab

Lines represent the weekly adjusted proportion of initiation or engagement estimated from segmented Poisson regression models. The pre-pandemic period spans from March 1, 2017 to January 15, 2020. The post-pandemic period spans from May 17, 2020 to February 28, 2023. Episodes occurring during the washout period (January 16 – May 16, 2020, yellow dashed lines) were excluded.

3.2.2. Telehealth

Telehealth addiction treatment initiation increased from 1.9% to 2.6% during the pre-pandemic period, by 12% annually (RR [95% CI]=1.12 [1.06, 1.19]) (Figure 1). There was a five-fold increase in telehealth initiation rates after the onset of the pandemic (level change RR [95% CI]=5.14 [4.62, 5.72]). Telehealth initiation rates decreased by 10% annually during the post- pandemic period (RR [95% CI]=0.90 [0.88, 0.92]); pre- and post-pandemic trends differed significantly (trend change p<0.001) (Table S1).

Telehealth initiation trends varied by age group (Figure 2, Table S2). Among patients aged 18–34 years, telehealth initiation rates increased by 10% annually in the pre-pandemic period (RR [95% CI]=1.10 [1.01, 1.19]), increased by 725% after the onset of the pandemic (level change RR [95% CI]=7.25 [6.14, 8.55]), then decreased by 14% annually in the post-pandemic period (RR [95% CI]=0.86 [0.83, 0.89]); annual trends significantly differed (p<.001). Patients aged 35–49 and 50–64 years had similar trajectories as patients aged 18–34 with significant annual trend changes (trend change p-values: <0.001 and 0.002, respectively), but less of an immediate pandemic effect (p-values <.001). Telehealth initiation trends for patients aged 65+ remained relatively stable between the two time periods. Telehealth initiation trends did not vary by race and ethnicity (Table S3, Figure S2).

3.3. Treatment Engagement

3.3.1. Overall

Overall addiction treatment engagement decreased from 8.1% to 7.5% of episodes during the pre-pandemic period, with no significant annual trend change (Figure 1). Engagement rates increased by 19% after the onset of the pandemic (level change RR [95% CI] = 1.19 [1.11, 1.28]). Engagement rates decreased from 9.3% to 8.1% in the post-pandemic period. Pre- and post-pandemic annual trends did not significantly differ (trend change p = 0.112) (Table S1).

All age groups had stable engagement trends across the pre- and post-pandemic periods (Figure 2, Table S2). Patients aged 18–34 years had a significant increase in engagement rates after the onset of the pandemic (level change RR [95% CI]=1.33 [1.19, 1.49]), which was greater than the level change in the group aged 50–64 years (comparison p-value = 0.032).

Engagement trend changes differed between Asian/Native Hawaiian/Pacific Islander and White patients (p-value=0.049) (Table S3). Among Asian/Native Hawaiian/Pacific Islander patients, engagement rates did not significantly change during the pre-pandemic period, increased by 44% immediately after pandemic onset (level change RR [95% CI]: 1.44 [1.10, 1.90]), then decreased by 12% annually in the post-pandemic period (RR [95% CI]: 0.88 [0.80, 0.98], trend change p = 0.013). In contrast, White patients had relatively stable engagement trends across pre- and post- periods, with an increase of 19% at the onset of the pandemic (level change RR [95% CI]=1.19 [1.07, 1.31]. The other racial/ethnic groups observed stable engagement trends over time, with no significant differences between groups.

3.3.2. Telehealth

Telehealth addiction treatment engagement increased from 1.2% to 2.5% of episodes during the pre-pandemic period, by 36% annually (RR [95% CI]=1.36 [1.28, 1.45]) (Figure 1). Telehealth engagement rates increased three-fold after the onset of the pandemic (level change RR [95% CI]=3.05 [2.71, 3.43]). Telehealth engagement rates then began to decrease by 5% during the post- period (RR [95% CI]=0.95 [0.92, 0.98]); pre- and post- trends significantly differed (trend change p < 0.001) (Table S1).

Telehealth engagement trends varied slightly by age group. For patients aged 18–34, 35–49, and 50–64, telehealth engagement rates were increasing during the pre-pandemic period, which increased more sharply at the onset of the pandemic and then decreased during the post- period (Figure 2, Table S2). Patients aged 35–49 years had a greater increase in telehealth engagement than those aged 18–34 years at the onset of the pandemic (p-value=0.009). In contrast, trends in telehealth engagement for patients aged 65+ years remained relatively stable across the two time periods, despite an increase at the onset of the pandemic.

Asian/Native Hawaiian/Pacific Islander, Black, and Latino/Hispanic had similar trend changes for telehealth treatment engagement as White patients, with increases in the pre-pandemic period followed by decreases post-pandemic (trend change p-values < 0.001) (Table S3). In contrast, telehealth engagement trends for American Indian/Alaska Native patients and patients in the Other/Unknown group remained stable across the two time periods. American Indian/Alaska Native patients differed from White patients in their pre-pandemic annual trends (comparison p-value = 0.006), but had similar post-pandemic trends; trend changes over time significantly differed between groups (comparison p-value = 0.038).

4. Discussion

In this cohort of adults with problematic substance use, we observed no major differences in overall addiction treatment initiation and engagement trends comparing the three years before and after the pandemic. However, there was a sharp increase in telehealth initiation and engagement immediately after the pandemic onset, followed by a gradual decrease in the following three years to almost pre-pandemic levels. Findings indicate that patients continued to utilize addiction treatment at low levels, including via telehealth, at the height of the pandemic when lockdown orders were implemented (Mehtani et al., 2021) as well as in the years after.

The gradual decrease in utilization rates over time to pre-pandemic levels could be related to decreased demand for addiction treatment; however national epidemiologic studies suggest otherwise. Specifically the prevalence of substance use increased from 14.5% in 2020 to 17.1% in 2023 (SAMHSA, 2021) (SAMHSA, 2024). Decreases in utilization rates over time could also be driven by patient preference to the treatment modalities that were available and accessible. For example, patients who preferred in-person to telehealth treatment may not have initiated care when in-person options were more limited. The health system eventually established a hybrid model of care in all its addiction treatment programs; the timing and availability of options varied by clinic. While some clinics offered in-person psychotherapeutic treatment as early as July 2021, others did not move to a hybrid model until early 2023. Finally, findings suggest that despite the increases in telehealth utilization observed, it did not solve the wider issue of low overall addiction treatment utilization rates.

Current literature shows that telehealth generally contributed to increased addiction treatment retention; for example, one study of Medicare patients in Kentucky and Ohio found that telehealth initiation for OUD was associated with higher odds of retention (Hammerslag et al., 2023). A similar study of Medicare beneficiaries found that OUD-related telehealth services during the pandemic were associated with increased odds of retention (Jones et al., 2022). While our treatment engagement measure and cohort with problematic substance use (not just OUD) differ from prior studies focused on OUD treatment retention, we found little evidence that the increases in engagement observed early in the pandemic – potentially related to expanded telehealth availability - were sustained over time.

All age groups showed a sharp increase in telehealth treatment initiation soon after the pandemic onset. The largest increase was seen in patients 18–34 years, with the magnitude decreasing with increasing age. This trend is consistent with previous reports of the role of age in telehealth addiction treatment utilization (Palzes et al., 2023). Older adults may be less inclined to utilize telehealth treatment due to lower levels of comfort and access to technology, reinforcing the idea of a ‘digital divide’ for this population (Kalicki et al., 2021). This contrasts with a previous study quantifying OUD treatment during the pandemic, which found that younger male adults were less likely to receive telehealth services, although this study examined patients with OUD in the Veterans Health Administration (Frost et al., 2022). Our study was conducted in a commercially-insured population of patients with all types of substance use disorders, not just OUD, thus the findings cannot be directly compared. Over time, the rates of telehealth initiation decreased for patients of all ages during the post- period, although patients aged 35–49 years continued to have the highest rates compared to other age groups. Patients aged 18–34 years had the greatest decline in telehealth initiation during the post-pandemic period of any age group; this may be attributed to having a greater initial increase at the onset of the pandemic, leaving more room for a decline over time.

Trends in addiction treatment utilization did not vary markedly by race and ethnicity. While overall initiation trends remained stable from pre- to post-pandemic for all racial and ethnic groups, telehealth initiation generally increased pre-pandemic and decreased post-pandemic. Overall engagement trends followed a similar pattern across racial and ethnic groups, although Asian/Native Hawaiian/Pacific Islander patients had a more pronounced change in engagement trends than White patients, suggesting a greater decline in engagement post-pandemic. Literature on addiction treatment engagement for Asian Americans has been conflicting, with some evidence of lower utilization (Von Gunten & Wu, 2021). Studies often combine Asian American groups into one category, masking the heterogeneity in these populations, which could explain disparate findings (Choi et al., 2024). Barriers to treatment may include cultural factors or language barriers, which could be the target for future efforts to improve treatment initiation and engagement (Choi et al., 2024). While the overall lack of differences by race and ethnicity is reassuring, the relatively lower engagement rates across racial and ethnic groups reinforce the challenge of engaging patients with problematic substance use in treatment. A similar study assessing telehealth addiction treatment utilization in 4 New York City programs during the pandemic reported fewer Black patients were initiating treatment in 2020 compared to previous years; the authors highlight various sociopolitical factors that exacerbated ongoing barriers this community faces when attempting to access care (Tomlinson et al., 2023). This study also found that White and Asian patients were more likely to receive telehealth and hybrid care during the pandemic (Tomlinson et al., 2023). This contrast to our findings may be due to differences in patient population (e.g. outpatient SUD clinics within a New York health system vs primarily commercially insured). In addition, this study period examined treatment utilization from 2019–2021 while we employed a longer time frame (2017–2023).

Our findings suggest that having expanded telehealth availability was not associated with sustained increases in addiction treatment utilization over time and additional or more interventions may be needed. Further studies assessing treatment utilization rates for specific substance use disorders may also provide insight into how to best target such interventions.

4.1. Limitations

This study had limitations, including that our data are restricted to the Northern California insured population in one healthcare system. Therefore, our findings may not be generalizable to other settings or patient populations. However, due to the size and demographic diversity of KPNC’s member population, this sample is generally similar to commercially insured populations (Davis et al., 2023).

We identified SUDs based on HEDIS standardized measures that utilize ICD diagnosis codes; while these methods are detailed, all patients with SUDs may not be captured. We note that our health system universally screens for alcohol problems in primary care but not drug use. The underdiagnosis of SUDs in healthcare systems is a known limitation (Campbell et al., 2024), potentially due to stigma about substance use and fear of discrimination (NIDA, 2022). We addressed this limitation by including diagnosis codes for “unhealthy substance use behavior”, which are healthcare system-specific and typically used in non-specialty care settings where providers may not have the time or expertise to make a formal SUD diagnosis. While the inclusion of the “unhealthy substance use behavior” codes allowed us to identify a larger sample of patients with problematic substance use, we acknowledge that these codes are not typically used in standardized measures of healthcare system performance, such as HEDIS. Therefore, the utilization rates reported in our study may be lower than metrics previously published and are not official health system HEDIS rates. However, we do not expect this reporting difference to impact patient groups differentially.

We defined treatment initiation as beginning treatment within 14 days of the index diagnosis, consistent with standardized HEDIS specifications to measure quality of care (National Committee for Quality Assurance, 2022) and other prior studies (Bensley et al., 2017; Chavez et al., 2022; Lind et al., 2019). Therefore, patients who started treatment after this 14-day window were not considered to have initiated treatment in our study, suggesting our estimates may be more conservative than some studies in the literature. Various factors could have impacted levels of treatment initiation and engagement, including patient and provider preference. Our EHR data does not include this information, although our primary goal was to assess trends over time rather than differences in preference regarding treatment.

Data for this project came from electronic health records, which includes self-reported and non-self-reported sources for demographic data. Therefore, racial/ethnic data was incomplete; 2.4% of this cohort was missing racial/ethnic information. Additionally, these pre-determined categories did not allow for distinguishing between Latino and Hispanic individuals.

5. Conclusion

This cohort study of patients with problematic substance use from a large, integrated health care system found that while telehealth addiction treatment utilization increased immediately after pandemic-related policy changes were implemented, utilization rates decreased to nearly pre-pandemic levels over time. Findings suggest that these policy changes for the delivery of addiction treatment, including telehealth-delivered care, did not have major long-term impacts on differences in treatment utilization by age and race and ethnicity.

Supplementary Material

Supplementary materials

Funding/Support:

This work was supported by the National Drug Abuse Treatment Clinical Trials Network of the National Institute on Drug Abuse (UG1DA040314).

Conflicts of Interest:

Dr. Campbell and Ms. Does have received support managed through their institution from the Industry PMR Consortium, a consortium of companies working together to conduct post marketing studies required by the Food and Drug Administration that assesses risks related to opioid analgesic use.

References

  1. Abramson A (2021, March 1). Substance use during the pandemic. Https://Www.Apa.Org. https://www.apa.org/monitor/2021/03/substance-use-pandemic [Google Scholar]
  2. Bensley KM, Harris AHS, Gupta S, Rubinsky AD, Jones-Webb R, Glass JE, & Williams EC (2017). Racial/ethnic differences in initiation of and engagement with addictions treatment among patients with alcohol use disorders in the veterans health administration. Journal of Substance Abuse Treatment, 73, 27–34. 10.1016/j.jsat.2016.11.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Campbell AG, Naz S, Gharbi S, Chambers J, Denne S, Litzelman DK, & Wiehe SE (2024). The Concordance of Electronic Health Record Diagnoses and Substance use Self-Reports Among Reproductive Aged Women Enrolled in a Community-Based Addiction Reduction Program. Inquiry: A Journal of Medical Care Organization, Provision and Financing, 61, 00469580241237051. 10.1177/00469580241237051 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Cantor J, McBain RK, Kofner A, Hanson R, Stein BD, & Yu H (2022). Telehealth Adoption by Mental Health and Substance Use Disorder Treatment Facilities in the COVID-19 Pandemic. Psychiatric Services (Washington, D.C.), 73(4), 411–417. 10.1176/appi.ps.202100191 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Centers for Medicare and Medicaid Services. (2020). Centers for Medicare & Medicaid Services (CMS) and Substance Abuse and Mental Health Services Administration (SAMHSA): Leveraging Existing Health and Disease Management Programs to Provide Mental Health and Substance Use Disorder Resources During the COVID-19 Public Health Emergency (PHE)1. [Google Scholar]
  6. Chavez LJ, Steelesmith DL, Bridge JA, & Fontanella CA (2022). Predictors of substance use disorder treatment initiation and engagement among adolescents enrolled in Medicaid. Substance Abuse, 43(1), 1260–1267. 10.1080/08897077.2022.2074603 [DOI] [PubMed] [Google Scholar]
  7. Choi S, Hong S, Gatanaga OS, Yum AJ, Lim S, Neighbors CJ, & Yi SS (2024). Substance use and treatment disparities among Asian Americans, Native Hawaiians, and Pacific Islanders: A systematic review. Drug and Alcohol Dependence, 256, 111088. 10.1016/j.drugalcdep.2024.111088 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Cleveland Clinic. (2024). What Is Substance Use Disorder? Cleveland Clinic. https://my.clevelandclinic.org/health/diseases/16652-drug-addiction-substance-use-disorder-sud [Google Scholar]
  9. Davis AC, Voelkel JL, Remmers CL, Adams JL, & McGlynn EA (2023). Comparing Kaiser Permanente Members to the General Population: Implications for Generalizability of Research. The Permanente Journal, 27(2), 87–98. 10.7812/TPP/22.172 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Deyo RA, Cherkin DC, & Ciol MA (1992). Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. Journal of Clinical Epidemiology, 45(6), 613–619. 10.1016/0895-4356(92)90133-8 [DOI] [PubMed] [Google Scholar]
  11. Frost MC, Zhang L, Kim HM, & Lin L (Allison). (2022). Use of and Retention on Video, Telephone, and In-Person Buprenorphine Treatment for Opioid Use Disorder During the COVID-19 Pandemic. JAMA Network Open, 5(10), e2236298. 10.1001/jamanetworkopen.2022.36298 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Hammerslag LR, Mack A, Chandler RK, Fanucchi LC, Feaster DJ, LaRochelle MR, Lofwall MR, Nau M, Villani J, Walsh SL, Westgate PM, Slavova S, & Talbert JC (2023). Telemedicine Buprenorphine Initiation and Retention in Opioid Use Disorder Treatment for Medicaid Enrollees. JAMA Network Open, 6(10), e2336914. 10.1001/jamanetworkopen.2023.36914 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Jones CM, Shoff C, Hodges K, Blanco C, Losby JL, Ling SM, & Compton WM (2022). Receipt of Telehealth Services, Receipt and Retention of Medications for Opioid Use Disorder, and Medically Treated Overdose Among Medicare Beneficiaries Before and During the COVID-19 Pandemic. JAMA Psychiatry, 79(10), 981–992. 10.1001/jamapsychiatry.2022.2284 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Kalicki AV, Moody KA, Franzosa E, Gliatto PM, & Ornstein KA (2021). Barriers to telehealth access among homebound older adults. Journal of the American Geriatrics Society, 69(9), 2404–2411. 10.1111/jgs.17163 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Kruse C, Fohn J, Wilson N, Patlan EN, Zipp S, & Mileski M (2020). Utilization Barriers and Medical Outcomes Commensurate With the Use of Telehealth Among Older Adults: Systematic Review. JMIR Medical Informatics, 8(8), e20359. 10.2196/20359 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Kruse CS, Mileski M, & Moreno J (2017). Mobile health solutions for the aging population: A systematic narrative analysis. Journal of Telemedicine and Telecare, 23(4), 439–451. 10.1177/1357633X16649790 [DOI] [PubMed] [Google Scholar]
  17. Lin C, Clingan SE, Cousins SJ, Valdez J, Mooney LJ, & Hser Y-I (2022). The impact of COVID-19 on substance use disorder treatment in California: Service providers’ perspectives. Journal of Substance Abuse Treatment, 133, 108544. 10.1016/j.jsat.2021.108544 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Lind BK, McCarty D, Gu Y, Baker R, & McConnell KJ (2019). Predictors of Substance Use Treatment Initiation and Engagement Among Adult and Adolescent Medicaid Recipients. Substance Abuse, 40(3), 285–291. 10.1080/08897077.2018.1550467 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Mehtani NJ, Ristau JT, Snyder H, Surlyn C, Eveland J, Smith-Bernardin S, & Knight KR (2021). COVID-19: A Catalyst for Change in Telehealth Service Delivery for Opioid use Disorder Management. Substance Abuse, 42(2), 205–212. 10.1080/08897077.2021.1890676 [DOI] [PubMed] [Google Scholar]
  20. Messer LC, Laraia BA, Kaufman JS, Eyster J, Holzman C, Culhane J, Elo I, Burke JG, & O’Campo P (2006). The development of a standardized neighborhood deprivation index. Journal of Urban Health: Bulletin of the New York Academy of Medicine, 83(6), 1041–1062. 10.1007/s11524-006-9094-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. National Committee for Quality Assurance. (2022). HEDIS Measures and Technical Resources. NCQA. https://www.ncqa.org/hedis/measures/ [Google Scholar]
  22. NIDA. (2022, June 1). Stigma and Discrimination | National Institute on Drug Abuse (NIDA). https://nida.nih.gov/research-topics/stigma-discrimination [Google Scholar]
  23. Oesterle TS, Kolla B, Risma CJ, Breitinger SA, Rakocevic DB, Loukianova LL, Hall-Flavin DK, Gentry MT, Rummans TA, Chauhan M, & Gold MS (2020). Substance Use Disorders and Telehealth in the COVID-19 Pandemic Era. Mayo Clinic Proceedings, 95(12), 2709–2718. 10.1016/j.mayocp.2020.10.011 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Palzes VA, Chi FW, Metz VE, Sterling S, Asyyed A, Ridout KK, & Campbell CI (2023). Overall and Telehealth Addiction Treatment Utilization by Age, Race, Ethnicity, and Socioeconomic Status in California After COVID-19 Policy Changes. JAMA Health Forum, 4(5), e231018. 10.1001/jamahealthforum.2023.1018 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Perry A, Wheeler-Martin K, Hasin DS, Terlizzi K, Mannes ZL, Jent V, Townsend TN, Pamplin JR, Crystal S, Martins SS, Cerdá M, & Krawczyk N (2023). Utilization and disparities in medication treatment for opioid use disorder among patients with comorbid opioid use disorder and chronic pain during the COVID-19 pandemic. Drug and Alcohol Dependence, 253, 111023. 10.1016/j.drugalcdep.2023.111023 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Ridout KK, Alavi M, Ridout SJ, Koshy MT, Harris B, Dhillon I, Awsare S, Weisner CM, Campbell CI, & Iturralde E (2021). Changes in Diagnostic and Demographic Characteristics of Patients Seeking Mental Health Care During the Early COVID-19 Pandemic in a Large, Community-Based Health Care System. The Journal of Clinical Psychiatry, 82(2), 20m13685. 10.4088/JCP.20m13685 [DOI] [PubMed] [Google Scholar]
  27. San Francisco Chronicle. (2021). The Bay Area’s yearlong battle against the coronavirus. The San Francisco Chronicle. https://projects.sfchronicle.com/2021/coronavirus-timeline/ [Google Scholar]
  28. Substance Abuse and Mental Health Services Administration. (2021). Key Substance Use and Mental Health Indicators in the United States: Results from the 2020 National Survey on Drug Use and Health. [Google Scholar]
  29. Substance Abuse and Mental Health Services Administration. (2024). Key Substance Use and Mental Health Indicators in the United States: Results from the 2023 National Survey on Drug Use and Health. [Google Scholar]
  30. Tomlinson MF, McCann-Pineo M, Thomas MP, & Polydorou S (2023). Demographic differences in services utilization across in-person (2019), telehealth (2020), and hybrid (2021) outpatient substance use services in New York. Journal of Substance Use and Addiction Treatment, 150, 209047. 10.1016/j.josat.2023.209047 [DOI] [PubMed] [Google Scholar]
  31. Uscher-Pines L, Cantor J, Huskamp HA, Mehrotra A, Busch A, & Barnett M (2020). Adoption of telemedicine services by substance abuse treatment facilities in the U.S. Journal of Substance Abuse Treatment, 117, 108060. 10.1016/j.jsat.2020.108060 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP, & STROBE Initiative. (2008). The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: Guidelines for reporting observational studies. Journal of Clinical Epidemiology, 61(4), 344–349. 10.1016/j.jclinepi.2007.11.008 [DOI] [PubMed] [Google Scholar]
  33. Von Gunten CD, & Wu L-T (2021). Comorbid Substance Use Disorder Profiles and Receipt of Substance Use Disorder Treatment Services: A National Study. Journal of Studies on Alcohol and Drugs, 82(2), 246–256. 10.15288/jsad.2021.82.246 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Wagner et al. (2002). Segmented regression analysis of interrupted time series studies in medication use research—PubMed. https://pubmed.ncbi.nlm.nih.gov/12174032/ [DOI] [PubMed] [Google Scholar]
  35. Yang J, Landrum MB, Zhou L, & Busch AB (2020). Disparities in outpatient visits for mental health and/or substance use disorders during the COVID surge and partial reopening in Massachusetts. General Hospital Psychiatry, 67, 100–106. 10.1016/j.genhosppsych.2020.09.004 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary materials
NIHMS2152115-supplement-Supplementary_materials.pdf (404.1KB, pdf)

RESOURCES