Impact of COVID-19 on Stimulant Use Disorder Treatment: A National Cohort Study in the Veterans Health Administration

Abstract

Objective

Rates of stimulant overdose have increased dramatically, which may have been exacerbated by treatment disruptions during the pandemic, but no recent studies have examined use of stimulant use disorder (StUD) treatment.

Methods

In this retrospective cohort study (March 2018 to February 2022) of national Veterans Health Administration patients, we use an interrupted time-series analysis to examine the impact of COVID-19 (starting in March 2020) on treatment use for StUD.

Results

The number of patients receiving StUD care was increasing pre-COVID (22,640 to 23,020, February 2018-February 2020) but dropped post-pandemic to 18,578 in February 2022. The monthly number of patients receiving StUD care increased by 34.6 (95% CI, 1.1 to 68.0; P=0.04) patients per month before March 2020, decreased by 2803.3 patients (95% CI, −3912.3 to −1694.3; P<0.001) in March 2020, and, accounting for pre-COVID trends, further decreased by 85.85 (95% CI,−148.9.2 to −23.0; P=0.01) patients per month after March 2020.

Conclusions

Care for StUD drastically declined during the COVID-19 pandemic and has yet to show signs of a return to pre-pandemic levels despite surging rates of stimulant-involved overdose deaths and a critical need to engage people with StUD in care.

Introduction

Stimulant overdose rates have increased over 300% in the US since 2013, contributing to the 4th wave of the overdose epidemic (Mattson et al., 2021). These increases are driven by increases in methamphetamine and cocaine use alongside opioids (Coughlin et al., 2022; Han et al., 2021; Mustaquim et al., 2021). The resulting stimulant-related impacts, including hospitalizations, and overdose deaths highlight the increased need for engaging those with stimulant use disorder (StUD) in care.

At the same time, the COVID-19 pandemic has caused monumental shifts in healthcare needs, utilization, and delivery modalities (Moynihan et al., 2021), particularly for patient populations facing social and health disadvantage (Baggio et al., 2021; Moynihan et al., 2021). Rising negative outcomes associated with stimulant use may be due to increasing stimulant use and/or decreases in treatment. Although rates of stimulant use, stimulant use disorders, and stimulant-involved overdose death are increasing, including in veterans served by the Veterans Health Administration (VHA) (Coughlin et al., 2022; Han et al., 2021; Kariisa et al., 2019; Mustaquim et al., 2021; Warfield et al., 2022), it is unknown how use of treatment has changed for people with StUD.

To our knowledge, no prior studies have examined StUD treatment during the most recent wave of stimulant-involved overdose deaths, including the impacts of COVID-19 and subsequent telehealth policies on provision of care for StUD. Thus, the purpose of the current study is to examine StUD care, and the impact of COVID-19 and related policies, across all outpatient and residential treatment settings in the VHA, the largest addiction treatment provider in the US.

Methods

We used interrupted time series analysis (ITSA) to examine changes in the monthly number of patients receiving StUD treatment in the VHA after (March 2020 to February 2022) compared to before (March 2018 to February 2020) COVID-19 policy changes. This study was deemed exempt by the VA Ann Arbor Healthcare System Institutional Review Board.

Cohort

Data were obtained from the VHA Corporate Data Warehouse, a repository of electronic health record data. We defined monthly cohorts of adult patients (18 years or older) with at least one outpatient or residential encounter with a StUD diagnosis.

Outcomes

The primary outcome was the number of patients who received treatment for StUD each month. In an effort to capture all treatment for people with StUD, we included any visits for substance use disorder or other mental health treatment (see Supplementary Materials Appendix 1 for included clinic stop codes and CPT codes) with a primary or secondary StUD diagnosis. We also described the modality (in-person, telephone, video) of care received. Demographic characteristics (age, sex, race/ethnicity, rural/urban status), comorbid other substance use disorders (SUDs), type of StUD (cocaine, methamphetamine, or both), and modality of care (telephone, video, and in-person) were summarized for each patient who received StUD care during the study period.

Statistical Analysis

We examined the monthly number of patients who received treatment for StUD over the course of the study period using ITSA (see Supplementary Materials Appendix 2). To examine the trend level change, we fit a 2-phase interrupted time series regression model. This method estimated the immediate change and change in the monthly trend of patients receiving StUD care from pre- to post-COVID-19, adjusting for pre-COVID-19 numbers of patients receiving care and trends to evaluate the impact of the pandemic on care receipt. Patients who were receiving care for a StUD before and after March 2020 were compared on demographic and clinical characteristics using generalized estimating equations with an indicator for the pandemic period to account for correlation of patients who received StUD care in both the pre- and post-pandemic periods. As a secondary analysis we conducted another ITSA to estimate the change in proportion of patients with an inpatient or outpatient StUD diagnosis (primary or secondary diagnosis; defined as rolling monthly cohorts with a past-year diagnosis) who received StUD care.

Results

In the two-year period before the pandemic (March 2018-February 2020), 111,121 unique patients nationally within the VHA received StUD care. In the two years following the onset of the pandemic (March 2020-February 2022), the number of patients decreased 13.6% to 95,988 patients (See Supplementary Table 1 for demographic and clinical characteristics).

ITSA showed a precipitous decrease in the number of visits related to StUD at the start of the COVID-19 pandemic (p<0.001; see Table 1 and Supplemental Figure 1). Before March 2020, patients receiving care for StUD increased by 34.6 (95% CI, 1.1 to 68.0; P=0.04) patients per month. There was an immediate decline of 2,803.3 (95% CI, −3912.3 to −1694.3; P<0.001) patients in the first month of the pandemic (March 2020), followed by a further decline of 85.95 (95% CI,−148.9.2 to −23.0; P=0.01) patients per month, after adjusting for pre-pandemic trends. Most StUD care was in specialty SUD clinics (67.9% of all visits).

Table 1.

Interrupted time-series analysis of patients receiving care for stimulant use disorder.

Factor	Estimate	95% Confidence interval	P value
Monthly No. of patients receiving stimulant use care in March 2018	22755.04	(22334.63, 23175.45)	<0.001
Monthly trend prior to March 2020 (change in patients/month)	34.55	(1.10, 68.00)	0.04
Immediate change in No. of patients receiving care in March 2020	−2803.32	(−3912.31, −1694.34)	<0.001
Monthly trend after policy change in March 2020 (compared with prior to COVID-19)	−85.95	(−148.87, −23.03)	0.01
Monthly change in No. of patients receiving care after March 2020	−51.30	(−109.24, 6.45)	0.08

Compared to patients who received care for StUD before the pandemic, those who received care after the pandemic were largely similar in demographic and clinical characteristics but were more likely to receive the care via phone and video (see Supplemental Table 1 and Figure 1). During the two year time-period following the onset of the pandemic, 71.0%, 41.0%, and 69.6% of patients received StUD care via telephone, video, and in person, respectively where modalities are non-exclusive.

Figure 1.

Trends in number of patients in the Veterans Health Administration receiving care for stimulant use disorder before and after COVID-19-related policy changes (March 2018-February 2022).

The secondary ITSA on the proportion of people with a past-year StUD diagnosis who received StUD care also showed a significant immediate decline in March 2020 (see Supplemental Table 2). However, the trend in proportion of patients with a StUD diagnosis who were receiving StUD care in the following two years was not significantly different from the pre-pandemic trend (slope change=0.05, 95% CI, −0.02 to 0.13; P=0.16).

Discussion

In this national retrospective cohort study of VHA patients, we found that StUD care decreased considerably at the onset of the COVID-19 pandemic and, concerningly, has not shown signs of recovery to pre-pandemic levels. In light of rapidly increasing stimulant overdose rates (Coughlin et al., 2022; Mattson et al., 2021), the drop in stimulant-related treatment is concerning given the extensive and increasing need for these services.

Unlike some SUDs (e.g., opioid use disorder, alcohol use disorder) where there are pharmacological treatments available, all current empirically-supported treatments for StUD are behavioral, and contingency management (CM) has the strongest support.(Ronsley et al., 2020) CM reinforces a target behavior, such as abstinence from stimulants, via rewards (e.g., prizes, money). The largest scale implementation of CM is within the VHA system (DePhilippis et al., 2018); however, COVID-19 may have disrupted CM and other StUD care.

Care for other SUDs in the VHA through the pandemic showed varying trends in care delivery. For example, opioid use disorder medication treatment increased with loosening prescribing restrictions through a transition to predominantly telephone and video care delivery (Frost et al., 2022; Lin et al., 2022). Alternatively, alcohol use disorder care declined at the onset of the pandemic (Perumalswami et al., 2023). In comparison, the current study shows evidence that the number of patients receiving StUD care declined at the start of the pandemic and continued to decline throughout the pandemic. Even as the trends in the number of patients receiving StUD care decreased, the proportion of those diagnosed with a StUD who received care showed a slight increase. This may be because fewer people were presenting to health systems during the pandemic (Moynihan et al., 2021), limiting the opportunity for StUD detection despite documented increases in StUD diagnosis leading into the pandemic (Warfield et al., 2022). Thus, a measure of how many people were diagnosed with a StUD is not a measure of true prevalence nor of need for StUD care, and is likely confounded with the measure of care delivery. However, the overall decreases seem to more closely parallel the decreases in alcohol use disorder care.

In addition, there was an uptick in telehealth delivery of care for StUD. However, a large portion of care remained in-person following the onset of the pandemic, with over a quarter of patients with StUD treated in residential settings, rather than in outpatient settings via telehealth. Post-pandemic care delivery modality (i.e., in person vs. telehealth) may be related to challenges in treating StUD compared to other SUDs. Due to lack of effective treatment options and patient complexity, including high prevalence of comorbid conditions, high prevalence of housing instability, presence of other mental health symptoms related to stimulant use (e.g. psychosis), providers may be more likely to refer patients to a residential treatment, which is less amenable to telehealth.

These findings should be considered in the context of limitations. First, ITSA is a quasi-experimental design commonly used to assess the longitudinal effects of healthcare policies or other major changes. However, it is not possible to determine if observed changes are due to pandemic policies, overall impacts of the pandemic, or other factors occurring at the same time. Second, the current study is in the national VHA health system, which may have limited generalizability to other patient populations. Finally, the actual number of patients with a need for StUD treatment could not be measured. Future work is needed to more clearly illustrate the need for StUD care within the VHA to be able to accurately identify the extent of the treatment gap.

In summary, in the national VHA healthcare system, we see a considerable, clinically meaningful drop in StUD care at the onset of the pandemic that has not shown signs of recovery in the two-year period following the start of the pandemic. In the context of increasing rates of StUDs, consequences (e.g., hospitalizations), and overdose deaths, the delivery of StUD care is paramount. Future work on care delivery models to increase delivery of StUD and policy to increase provision of empirically-supported StUD care (e.g., CM) both within and external to the VHA are necessary to curb current trends in stimulant use and consequences effectively.

Figure 2.

Patients receiving stimulant use disorder care by modality (in person, video, phone) by month.