Engagement with and acceptability of telephone-based telemedicine in addiction care in Vancouver, Canada, 2020–2023

Abstract

Objectives:

Telemedicine-based addiction medicine care expanded during the COVID-19 pandemic. We characterize the engagement with and acceptability of these services in Vancouver, Canada.

Methods:

Using longitudinal self-report data from participants who were prescribed medications to address cravings and withdrawal of any substances in 2020–2023 in three cohorts of community-recruited people who use drugs in Vancouver, we performed a multivariable generalized estimating equation analysis to identify factors associated with telemedicine use for substance use medications in the past six months (vs. in-person care). We also explored participant perceptions of their telemedicine engagement.

Results:

Among 871 eligible participants, 428 (49.1%) were prescribed substance use medications through telemedicine at the first interview during the study period. Compared with those who did not use telemedicine, telemedicine use in a six-month period was positively associated with continuing to be prescribed substance use medications from the previous six-month period (adjusted odds ratio [AOR] = 1.53; 95% confidence interval [CI]: 1.11–2.11) and negatively associated with homelessness (AOR = 0.71, 95% CI 0.54–0.92) and daily opioid use (AOR = 0.78, 95% CI 0.64–0.94). With each subsequent assessment, participants were less likely to have used telemedicine (AOR = 0.86, 95% CI 0.81–0.91). Of the 255 participants who preferred telemedicine (59.6%), 214 (83.9%) cited convenience as the reason.

Discussion:

Telemedicine use was preferred to in-person care among most of our participants but less likely among participants showing markers of structural vulnerability and higher intensity substance use. Future research should investigate the efficacy of telemedicine vs. in-person addiction medicine care.

INTRODUCTION

The concurrent public health emergencies of the toxic drug crisis and COVID-19 led to rapid changes in substance use treatment services across North America, including the widespread adoption of telemedicine.¹ New strategies to address the toxic drug supply were developed rapidly and upscaled in many jurisdictions with an aim to improving the accessibility of substance use disorder treatment and treatment retention.¹

In British Columbia (BC), Canada, opioid toxicity has become the second-leading cause of loss of years of life.² The early COVID-19 pandemic and public health guidance to socially isolate created a challenge to traditional models of substance use disorder care, typically conducted in-person, and presented the threat of further opioid toxicity deaths.³ Telemedicine-based substance use disorder care, understood here as appointments with a prescriber by telephone or video conferencing, presented an opportunity to provide care more indiscriminately concerning location, and allowed for ongoing care compliant with COVID-19 pandemic public health recommendations.³

Previous publications have described the potential of telemedicine to expand access to substance use care^4,5 and noted efficacy in reduction in opioid toxicity death,⁶ while acknowledging that evidence is scarce and continues to evolve.^7,8 Concern has been raised about challenges with health equity, recognizing that disparities in access to specialized substance use care may be exacerbated by factors such as the digital divide (inequality in access to, and capability with, technology) and care provider preference in terms of which patients are offered telemedicine-based interventions.^5,9

In BC, three long-running harmonized prospective cohorts have followed people who use drugs (PWUD), primarily in Vancouver’s Downtown Eastside (DTES) neighbourhood, a community characterized by the socioeconomic marginalization of its residents, its relatively open illicit drug market and its long history of drug user advocacy, substance use research, and novel service implementation.¹⁰ Using these cohorts, in this study we aimed to examine the participant-reported engagement with, and acceptability of, telemedicine-based addiction medicine care for substance use disorders, and the changes in the frequency of use of these services between 2020 and 2023.

METHODS

Study design

This study was conducted using survey data from the Vancouver Injection Drug Users Study (VIDUS), AIDS Care Cohort to Evaluate Exposure to Survival Services (ACCESS), and At-Risk Youth Study (ARYS) linked cohorts. Eligibility criteria differ for each cohort, and are described in previous publications.^11,12 Briefly, VIDUS includes HIV-seronegative adults (≥18 years of age) reporting injection use of illicit drugs in the month leading up to enrolment; ACCESS includes adults living with HIV reporting any illicit drug use (other than cannabis) in the month leading up to enrolment; and ARYS includes youth (age 14–26 at enrollment) reporting any illicit drug use (other than cannabis) in the month leading up to enrolment. Individuals provide written informed consent upon study recruitment through street outreach and word-of-mouth in Vancouver. The three cohorts are linked or harmonized in that no participant is enrolled in more than one cohort at any given time and that the cohorts administer the same questionnaire at the same frequency (every six months), which allows for pooled analyses. At baseline and semi-annually thereafter, participants complete an interviewer-administered questionnaire collecting data on patient sociodemographic characteristics, substance use, service utilization, health status, and social-structural exposures. Since December 2020, the questionnaire has also assessed telemedicine engagement and experiences. Participants receive an honorarium of $50 for completing a study visit.

Between December 2020 and March 2022, study interviews were conducted remotely via telephone or videoconferencing (in order to be compliant with public health measures). Participants who did not have access to a telephone were provided with the option to use a study-owned pre-paid cellphone for the interview. If a participant did not have access to a private space to complete the interview, a space was provided for them to use. Since March 2022, we resumed in-person interviews while keeping the option of remote interviews for those who may have had difficulty visiting the study offices. The cohort receives annual ethics approval by the Providence Health Care/University of British Columbia Research Ethics Board (ACCESS) and the Simon Fraser University Research Ethics Board (VIDUS and ARYS).

Participants were eligible for inclusion in the present study if they completed at least one interview between December 2020 and May 2023, and reported that they had received medications for substance use care (prescriptions for withdrawal management, opioid agonist therapy, medications for relapse prevention, “risk mitigation,” etc.) in the prior six-months, whether by telemedicine, or by an in-person clinic visit. While practice and terminology are rapidly evolving, “risk mitigation” (including “prescribed safer supply/prescribed pharmaceutical alternatives to the toxic drug supply”) in this context is different from conventional substance use treatment and refers to a provincial policy of prescribing daily-dispensed, unwitnessed tablet-form opioids (and, less frequently, stimulants) intending to reduce the loss of life associated with the concurrent public health impact of COVID-19 and the toxic drug crisis since 2020.^13,14 However, in February 2025, the provincial government changed the unwitnessed program to a witnessed one.¹⁵

Study measures

To characterize engagement with telemedicine addiction care, the main outcome variable of interest was use of telemedicine, defined as having received medications for substance use care from a physician using telemedicine in the past six months.

To further describe participant experiences, two secondary outcome variables were assessed. The first was the adequacy of telemedicine-based substance use care, expressed as how often participants were able to obtain the advice/medication/support that they were seeking through telemedicine. This variable was dichotomized as “always” or “usually” vs. “sometimes,” “occasionally,” or “never.” The second was participant-expressed preference for telemedicine over exclusive in person, clinic-based care (yes vs. no).

Explanatory variables of interest were selected based on clinical experience of authors (DR and PB), the existing literature, and the available data. This included age (in years), gender (cisgender or transgender man vs. other), ethnicity/ancestry (white vs. Black, Indigenous, or other persons of colour), homelessness (self-defined by participants but typically including both sheltering outdoors/on the street and being temporary sheltered; yes vs. no), employment (yes vs. no), non-medical use of opioids, crystal methamphetamine, powder/crack cocaine, and alcohol, respectively (daily vs. not daily), non-fatal overdose (yes vs. no), and injection drug use (any use vs. no use). All variables except for ethnicity/ancestry referred to the six-month period prior to interviews. As a measure of previous treatment exposure, we also included a lagged variable assessing whether participants were receiving medications for substance use care in the immediate previous period of assessment, each assessment being six months in duration. Anticipating changes in the availability of telemedicine-based care over time, the six-month period of assessment itself was also included as a variable, where the first period started in December 2020 and ended in May 2021. The same set of explanatory variables was used to identify factors associated with the primary and secondary outcomes in each of the three parallel analyses.

Statistical analyses

Initially we examined the described variables of interest at baseline (i.e., the first study visit during the study period), stratified by engagement with telemedicine care, using the Mann-Whitney U test for numerical variables and the Chi-square test for categorical variables. Further, among participants using telemedicine addiction care, descriptive statistics were gathered at baseline on the medications and care types provided, platforms used to access care, and participant preferences and perceptions on the adequacy and acceptability of telemedicine care.

Three parallel analyses were performed, for first the primary outcome, Model 1 (Engagement with Telemedicine Care), then subsequently for Model 2 (Expressed Adequacy of Telemedicine Care), and Model 3 (Expressed Preference for Telemedicine Care). To estimate the bivariate associations between each explanatory variable and each outcome, a bivariate generalized estimating equation (GEE) with the logit link function and an exchangeable correlation structure was used. All variables, except for ethnicity/ancestry, were treated as time-varying, and as the analyses involved serial measure for each participant, we used a GEE that provided standard errors adjusted by multiple observations per person using an exchangeable correlation structure. Next, for each of Models 1–3, we fit a full multivariable model using covariates significantly associated with the outcome at p < 0.1 in the bivariate analyses. Reduced models were then sequentially constructed by removing covariates with the highest p-value one by one until there were no further variables left to remove, and the model with the lowest Quasi Information Criterion (QIC) was selected as the final model. Variance inflation factors were calculated to assess for multicollinearity. Using a cut-off score of 5.0,¹⁶ we did not identify any multicollinearity in the final or interim models. All p-values were two sided. We used R (version 4.3.1, R Foundation for Statistical Computing, Vienna, Austria) for all analyses.

RESULTS

Our study included a total of 2516 observations from 871 unique participants. The median number of follow-ups per individual respondent was 3 (interquartile range [IQR] = 2–4). The median age at baseline was 45.7 years (IQR = 35.2–56.1). Four hundred eighty-one (55.2%) participants self-identified as men, 366 (42.0%) as women and 24 (2.8%) as transgender or non-binary gender. Five hundred twenty (59.7%) self-identified as white, 304 (34.9%) as Indigenous and 40 (4.6%) as persons of colour. Additional baseline characteristics are shown in Table 1.

Table 1:

Baseline sample characteristics stratified by telemedicine addiction care engagement among 871 participants prescribed medications for substance use treatment in the past six months in Vancouver, Canada, December 2020 – May 2023.

Characteristic	Total 871 (100%)	Telemedicine Engagement		p - value
		Yes	No
		428 (49.1%)	443 (50.9%)
Age (median, Q1-Q3)	45.7 (35.2–56.1)	45.5 (35.3–56.6)	46.1 (35.1–55.9)	0.802a
Men	481 (55.2%)	227 (53.0%)	254 (57.3%)	0.202b
White	520 (59.7%)	262 (61.2%)	258 (58.2%)	0.435b
Homeless c	158 (18.1%)	59 (13.8%)	99 (22.4%)	<0.001b
Any employment c	226 (26.0%)	119 (27.8%)	107 (24.2%)	0.211b
Non-fatal overdose c	176 (20.2%)	71 (16.6%)	105 (23.7%)	0.008b
Injection substance use c	499 (57.3%)	242 (56.5%)	257 (58.0%)	0.551b
Self-reported at least daily substance use c
Non-prescribed opioid	462 (53.0%)	209 (48.8%)	253 (57.1%)	0.016b
Crystal methamphetamine	158 (18.1%)	65 (15.2%)	93 (21.0%)	0.028b
Powder/crack cocaine	87 (10.0%)	35 (8.2%)	52 (11.7%)	0.074b
Alcohol	95 (10.9%)	41 (9.6%)	54 (12.2%)	0.208b
Recent substance use medication engagement d	314 (36.1%)	192 (44.9%)	122 (27.5%)	<0.001b

Q: quartile.

^aDenotes that p-value was obtained from the Mann-Whitney U test.

^bDenotes that p-value was obtained from a Chi-square test.

^cDenotes behaviour or events occurring at any time in the six-month period of assessment.

^dDefined as having accessed medications for substance use treatment in the six months prior to the six-month period of study.

Descriptive statistics

As shown in Figure 1, in the first period of assessment, between December 2020 and May 2021, 55.6% reported using telemedicine at least once. This proportion decreased over time, and in the final period of assessment, between December 2022 and May of 2023, it had become a minority (41.8%). Additional descriptive statistics are reported in Table 2. As shown, at baseline, nearly all care visits were conducted by telephone (n=411, 96.0%), as opposed to video conferencing platform. Patients that had used telemedicine addiction care most frequently received opioids either for treatment (n=384, 89.7%) or for “risk mitigation” (n=200, 46.7%), which were not mutually exclusive. Most participants reported that when they accessed care through telemedicine, they were always or usually able to access the advice, medications or support that they were seeking (n=365, 85.3%). A narrow majority of patients indicated that they would prefer to continue to seek care through telemedicine rather than exclusively in person (n=255, 59.6%), with the most common reported reason for this preference being increased convenience (n=214, 83.9%).

Figure 1. Prevalence of telemedicine use among people prescribed medications for substance use disorder treatment in Vancouver, Canada from December 2020 to May 2023.

Table 2.

Baseline descriptive statistics on services accessed through addiction telemedicine care, patient-reported adequacy of services, and preferences for or against telemedicine among 428 participants in Vancouver, Canada, December 2020 to May 2023.

Survey content	N (%)
Substances and care types provided to patients accessing addiction telemedicine care
Medications for treatment - opioids	384 (89.7%)
Medications for risk mitigation - opioids	200 (46.7%)
Medications for treatment - stimulants	10 (2.3%)
Medications for risk mitigation - stimulants	15 (3.5%)
Medications for treatment - alcohol	5 (1.2%)
Medications for risk mitigation - alcohol	2 (0.5%)
Medications for treatment - benzodiazepines	3 (0.7%)
Medications for risk mitigation - benzodiazepines	4 (0.9%)
Medications for treatment - nicotine	1 (0.2%)
Medications for risk mitigation - nicotine	2 (0.5%)
Medications for treatment - other	2 (0.5%)
Medications for risk mitigation - other	2 (0.5%)
Platforms used to engage with addiction medicine care
Telephone	411 (96.0%)
Videoconferencing platform	31 (7.2%)
Proportion of time able to access advice/medication/support sought through telemedicine care
Always (100% of the time)	316 (73.8%)
Usually (75% of the time or more)	49 (11.5%)
Sometimes (26 to 74% of the time or more)	36 (8.4%)
Occasionally (25% of the time or less)	18 (4.2%)
Never	6 (1.4%)
Expressed participant preference for continued telemedicine care instead of in-person clinic visits
Yes	255 (59.6%)
No	144 (33.6%)
Reasons cited for preference for telemedicine care among “Yes” respondents
More convenient	214 (83.9%)
Less stigmatizing	8 (3.1%)
No wait times	42 (16.5%)
More comfortable at home	81 (31.8%)
Challenges with travel/mobility	49 (19.2%)
Other1	44 (10.3%)

¹“Other” responses were written as text below the above options. The majority of responses roughly mapped to existing potential answers, with 6 describing challenges with travel or mobility, 2 describing a less-stigmatizing or more honest environment, 1 describing a preference against going outside, and 1 describing shorter wait times. Other themes included avoiding people, triggers, or risks near the clinic (8 respondents), negative perceptions of the care team or care environment (9 respondents), avoiding COVID-19 (12 respondents), and anxiety (2 respondents). The final two respondents described conflict with employment, and needing to engage in pet care.

GEE analyses results

The results of bivariate and multivariable GEE analyses for Models 1–3 are shown in Table 3. For Model 1 (Engagement with Telemedicine Care), in the multivariable analyses, the factors negatively associated with engagement with telemedicine addiction care were daily opioid use (Adjusted Odds Ratio [AOR] = 0.78, 95% CI 0.64 – 0.94), homelessness (AOR = 0.71, 95% CI 0.54–0.92) and follow-up period (AOR = 0.86, 95% CI 0.81–0.91), with telemedicine engagement becoming less common in later periods of assessment (Figure 1). Those participants who had been prescribed medications for substance use treatment in the six months before the period of assessment were more likely to have engaged with telemedicine during the period of assessment (AOR 1.53, 95% CI 1.11–2.11).

Table 3.

Bivariate and multivariable GEE analyses of factors associated with telemedicine use among people prescribed medications for substance use disorder treatment in Vancouver, Canada from December 2020 to May 2023.

Variable	Telemedicine use1 (yes vs. no)		Expressed telemedicine as meeting care needs2 (always, usually or sometimes vs. occasionally or never)		Expressed preference for telemedicine3 (yes vs. no)
	OR (95% CI)	AOR (95% CI)	OR (95% CI)	AOR (95% CI)	OR (95% CI)	AOR (95% CI)
Age	1.00		1.00		0.98††	0.98††
(per year older)	(0.99–1.01)		(0.97 – 1.02)		(0.97 – 0.99)	(0.97 – 0.99)
Gender *	0.82†	0.83	0.88		0.79
(men vs women/other)	(0.67 – 1.00)	(0.66 – 1.05)	(0.50 – 1.57)		(0.59 – 1.05)
Ethnicity/ancestry	1.12		0.58	0.58	1.03
(white vs BIPOC)	(0.90 – 1.39)		(0.31 – 1.07)	(0.31 – 1.07)	(0.77 – 1.38)
Homeless *	0.67††	0.71††	1.38		0.78
(yes vs. no)	(0.54 – 0.83)	(0.54 – 0.92)	(0.55 – 3.42)		(0.55 – 1.11)
Any employment *	1.20†	1.17	0.87		1.01
(yes vs no)	(0.99 – 1.45)	(0.94 – 1.46)	(0.48 – 1.59)		(0.77 – 1.33)
Non-prescribed opioid use *	0.83†	0.78†	0.67		1.34†
(≥ daily vs < daily)	(0.70 – 0.98)	(0.64 – 0.94)	(0.39 – 1.17)		(1.04 – 1.71)
Crystal methamphetamine use *	0.89		1.15		0.95
(≥ daily vs < daily)	(0.71 – 1.11)		(0.53 – 2.49)		(0.67 – 1.34)
Powder/crack cocaine use *	0.87		1.09		0.89
(≥ daily vs < daily)	(0.67 – 1.14)		(0.41 – 2.94)		(0.60 – 1.34)
Alcohol use *	0.82		1.09		1.48
(≥ daily vs < daily)	(0.62 – 1.10)		(0.41 – 2.95)		(0.92 – 2.38)
Recent substance use medication engagement 4	1.48†	1.53†	0.77		1.32
(yes vs. no)	(1.08 – 2.02)	(1.11 – 2.11)	(0.20 – 2.90)		(0.76 – 2.28)
Non-fatal overdose *	0.95		0.87		0.82
(yes vs. no)	(0.78 – 1.16)		(0.43 – 1.74)		(0.60 – 1.12)
Intravenous drug use *	0.96		0.87		0.94
(yes vs. no)	(0.80 – 1.15)		(0.50 – 1.49)		(0.73 – 1.23)
Follow-up period *	0.89††	0.86††	0.97		1.06
(Assessed per 6 months passed)	(0.84 – 0.93)	(0.81 – 0.91)	(0.81 – 1.18)		(0.97 – 1.15)

AOR: Adjusted odds ratio. BIPOC: Black, Indigenous or person of color. CI: Confidence interval. DTES: Downtown Eastside. OR: Odds ratio.

¹Defined as having accessed medications for substance use care prescribed via telemedicine in the six-month period of assessment.

²Defined as the proportion of the time participants reported they were able to obtain the advice/medication/support that they were seeking by telemedicine: always (100% of the time), usually (75% of the time or more), sometimes (26–74% of the time or more), occasionally (25% of the time or less) or never (0% of the time).

³Defined as expressing that, when possible, participants would prefer to be able to seek care through telemedicine over exclusively visiting a clinic in person.

⁴Defined as having accessed medications for substance use treatment in the six months prior to the six-month period of study.

^*Denotes behaviour or events occurring in the six-month period of assessment.

^†Denotes p<0.05

^††Denotes p<0.01.

In the multivariable analyses for Model 2, the expressed adequacy of telemedicine care was not associated with any variables. In the adjusted Model 3, the only significant association found with an expressed preference for telemedicine addiction care was younger age (AOR = 0.98, 95% CI 0.97–0.99).

DISCUSSION

In BC, telemedicine-based substance use care has become common, and although telemedicine use decreased significantly during the period of our study, it has remained a significant part of the treatment landscape for many participants. We found that those reporting daily opioid use and homelessness were less likely to be using telemedicine. Those who had not been on medications the last time they were surveyed were less likely to be using telemedicine, relative to in-person care, when they were surveyed again. Medications prescribed were almost exclusively opioids.

At the time of writing, this is one of the first prospective cohort studies evaluating patient experiences of telemedicine substance use care in Canada. Similar studies have been conducted in the United States,^17–22 though these have assessed telemedicine services using clinician health data²³ or by assessing program availability by posing as prospective clients.²⁴ The VIDUS, ACCESS, and ARYS cohorts follow patients at high risk of substance use-related morbidity and mortality but were not sampled specifically to assess telemedicine-related factors. Therefore, we can gather a relatively naturalistic assessment of the state of telemedicine-based substance use care for similar populations. A large proportion of participants were accessing telemedicine, far in excess of the proportions suggested in previous US-based studies conducted during the same time period,²³ which might reflect at least in part different regulations regarding prescribing controlled substances via telemedicine in Canada compared to the US, where such medications are subject to the Ryan-Haight Act. Notably, telephone calls were the near-exclusive modality of contact in our study, presumably driven by patient access barriers to other modalities and/or a rapid transition to telemedicine in the context of the pandemic, without time for the adoption of new technologies.²⁵ The dominance of telephone-based services may have limitations for patients needing visual assessments (compared to video conferencing software), though standardized best practices in the provision of substance use care via telemedicine have not yet been established. In our clinical experience, at the time of the study, the decision to engage in telemedicine-based substance care was typically initiated by the care provider or clinic policies, rather than the patient.

Our study findings provide unique insights on telemedicine-based substance use care. There is little previous evidence on the availability of telemedicine to PWUD generally, though evidence does exist describing the proportion of telemedicine care occurring in certain programs and on the barriers patients accessing telemedicine report, such as lack of privacy and inconsistent access to or familiarity with the necessary technology.²⁶ In our study, telemedicine was perceived as useful to patients. Most respondents that had accessed telemedicine at least once during the period of assessment (n=365, 85.3%) reported that they could access the care they were seeking this way, though a preference for telemedicine was by no means universal. While previous studies have generally centred on the evaluation of specific substance use treatment programs, our results are generally consistent with previous research indicating that telemedicine-based substance use care is acceptable to most patients who access it, though factors such as privacy, modality, rapport-building, interruptions, ease of use, and the proportion of telemedicine vs. in-person care have been identified as important in previous research.^{8,17–19,21,26–30} Also consistent with previous findings,⁸ the most cited reasons to prefer telemedicine were convenience, time to access care, mobility concerns, and comfort. While our study did not explore reasons for the observed declines in telemedicine use over time, we speculate that the resumption of many in-person activities in general around 2022–2023 after the COVID-19 pandemic period might have also reduced telemedicine use despite the high stated patient preference.

Crucially, our findings suggest that populations facing structural marginalization and those with markers of higher intensity substance use were less likely to access telemedicine. It is likely that care providers are more likely to offer telemedicine to patients whom they are feel are stabilized in their treatment regimen, with less complex health needs. It may also be possible that people experiencing structural marginalization, including homelessness, face barriers in using telemedicine care, as it requires access to a device, privacy, and internet or a telephone line. It may also be related to participant preferences, with some patients reporting that telemedicine-based services were not adequate to meet their needs. Other studies have raised concern that a pivot to telemedicine-based care might exacerbate existing inequities, with rural and remote communities, gender and racialized minorities, and older adults being less likely to access telemedicine.⁵ While our study did not identify an independent contribution to the likelihood of telemedicine engagement based on ethnicity, gender, or age, it did identify that those using opioids daily, who had not recently been prescribed medications, and/or who were experiencing homelessness were less likely to have accessed telemedicine. This presents an opportunity to focus on engaging these individuals in care, perhaps identified as at higher risk of substance use-related morbidity and mortality than others.^31,32 Low-threshold adaptions, such as those involving support of harm reduction service providers,³³ may be needed to mitigate the inequity in access to addiction medicine.

Our study has limitations. As this study relies on self-report data, reporting bias may be present. In addition, the question about telemedicine use in our questionnaire referred to being prescribed a medication for substance use care by a physician via telemedicine. We cannot exclude the possibility that participants who were prescribed medications by nurse practitioners or nurses via telemedicine may have answered “no.” Therefore, the prevalence of telemedicine use might have been underestimated. A more objective way of ascertaining telemedicine use, such as via linkages to health records databases, could mitigate this bias. The generalizability of our findings is limited by the fact that our study population primarily includes Vancouver’s DTES and the Greater Vancouver area. Studies conducted using these data are not well-positioned to make inferences about the nature of telemedicine in other communities of interest for the development of new substance use treatment programs, such as in rural and remote communities and, more generally, communities in which there are no prescribers offering substance use care or regulations for prescribing controlled substances via telemedicine are different. Further, because our study population was characterized by high rates of structural marginalization, some of our findings may not be generalizable to other patients who have not experienced such marginalization and have stable access to a telemedicine device. As participants were assessed every six months, we have limited capacity to assess the duration of time participants remained connected to care without interruptions. Patients who were prescribed a medication for their substance use in the past six months were all asked the same questions about their experience with telemedicine, regardless of discontinuing the prescriptions for substance use care within the same six-month period, though some participants may have accessed telemedicine only once, and others may have been reliably connected to telemedicine-based prescribing throughout the study period.

In summary, using cohort data collected between December 2020 and May 2023, we saw relatively common use of telemedicine for substance use care, mostly focused on providing treatment and harm reduction for those who use opioids. Participants experiencing structural marginalization, including homelessness, using opioids daily, and not having previously accessed medications for substance use treatment, were less likely to access telemedicine relative to exclusive in-person care. The proportion of participants engaged with telemedicine decreased over the period of study but remained a significant portion of the substance use treatment landscape. Future study is warranted to examine telemedicine’s effectiveness in comparison with exclusive in-person care (including understanding who, when, and how often in-person visits are still beneficial), determine how telemedicine can be safely and appropriately integrated into substance use care, and develop means of improving equity in access for the treatment of substance use disorders, though the rapidly evolving practice and evidence base for risk mitigation prescribing precludes such evaluations of telemedicine use in the context of risk mitigation prescribing at this time.