Trends in Use of Telemedicine for Stimulant Initiations among Children and Adults

Abstract

Objective:

To examine trends in how often stimulants are initiated via telemedicine, patient and clinician factors associated with telemedicine initiation, and the association between telemedicine initiation and receipt of follow-up care.

Methods:

Retrospective longitudinal study of national, de-identified commercial health insurance outpatient claims among children 2-17 and adults 18-64, January 2019-April 2022 from Optum Labs Data Warehouse. Regression analyses examined several outcomes: stimulant initiation, initiation via telemedicine (vs. in-person), and receipt of follow-up visit within 30 days of initiation.

Results:

The mean monthly adjusted number of initiations per 100,000 enrollees for children was similar before and during the pandemic (pre=57; during= 56) but increased for adults (pre=27; during = 33). Initiations via telemedicine peaked at 50-60% in April 2020 and dropped to 13% among children and 28% among adults in April 2022. Telemedicine initiations were more common among psychiatrists than other prescribers (OR=3.70, 95% CI=3.38-4.06 children; OR=3.02, 95% CI=2.87-3.17 adults), and less common among rural residents (OR=0.57, 95% CI=0.40-0.82 children; OR=0.75, 95% CI=0.61-0.92 adults) and residents of higher-income areas (e.g., OR=1.70, 95% CI=1.51-1.91 for quartile 4 vs. quarter 1 children; OR=1.72, 95% CI=1.61-1.84). Follow-up care was more common among those who initiated via telemedicine than in-person (OR=1.09, 95% CI=1.00-1.19 children and OR=1.61, 95% CI=1.53-1.69 adults).

Conclusions:

A sizeable fraction of initiations for children and adults occurred via telemedicine. Recently proposed rules to prohibit controlled substance prescribing without an in-person medical evaluation would require a significant change in current practice and potentially limit access to stimulant medication and related ADHD care.

Introduction

Attention-deficit hyperactivity disorder (ADHD) is a common chronic condition characterized by inattention and/or hyperactivity-impulsivity. Approximately 10% of children age 2-17^1,2 and 5% of adults^3,4 in the US have ADHD. ADHD is associated with impaired educational attainment; lower levels of employment; higher rates of accidental injury; and higher risk for developing co-occurring depression, anxiety, and substance use disorders.^3,5-8 Stimulants, such as methylphenidate or amphetamine salts, are the most commonly prescribed ADHD medications.^9,10 They are highly efficacious^10-12 and associated with a range of better outcomes (academic,¹³ accidents/injuries,^13,14 criminal justice,¹⁵ and mood and substance use disorders¹³). Professional guidelines recommend medications for children with ADHD starting age 6, and younger children if behavioral therapies alone do not adequately address their symptoms.¹⁶ In adults, medications are also a cornerstone of ADHD treatment, and are similarly effective as in children.¹⁷ Despite their efficacy, only an estimated 19% of school age children¹⁸ and 10% of adults³ with ADHD receive medication.

Telemedicine could increase access to ADHD treatment, including stimulants.¹⁹ However, due to the risk of stimulant diversion and misuse,^20-23 there is ongoing controversy about whether telemedicine initiation of stimulants should be permitted. Under the Ryan Haight Act, an in-person medical evaluation is required before a clinician can prescribe controlled substances including stimulants. This requirement was temporarily waived during the COVID-19 Public Health Emergency (PHE), and several telemedicine-only companies began offering ADHD treatment. The Department of Justice is investigating the stimulant prescribing practices of at least one such firm for possible violations of the Controlled Substances Act.²⁴ In February 2023, the Drug Enforcement Agency (DEA) proposed reinstating the in-person medical evaluation requirement once the PHE ended in May 2023.²⁵ This change would make it much more difficult or even impossible for individual clinicians and telemedicine companies to operate fully virtual care models for ADHD treatment. In May 2023, following over 38,000 public comments to the proposed rule, the DEA decided to temporarily continue the PHE flexibility, waiving the in-person requirements;²⁶ in October 2023, the DEA further extended these flexibilities through December 2024.²⁷

Recent reports have examined changes in stimulant prescribing since the pandemic began,^28-30 but no prior research examines the role that telemedicine has played in initiating stimulants, what patient and clinical factors are associated with using telemedicine to start stimulants, or the association between quality of ADHD care and stimulant initiation via telemedicine versus in-person care. Using a large national claims database, we examined trends in stimulant initiation for commercially-insured children and adults, before and during the pandemic period. We also examined factors associated with stimulant initiation delivered via telemedicine, and whether initiation via telemedicine is associated with a difference in the receipt of follow-up care.

Methods

Cohort

We used data on commercially-insured individuals from Optum Labs Data Warehouse from January 2019 through April 2022 to examine trends in stimulant initiation for children aged 2-17 and non-elderly adults aged 18-64. These data, which span all 50 states and include all care paid for by the health plan, have been previously used to describe care patterns in the U.S.^31-35 To be included in the cohort, we required continuous enrollment in medical, behavioral health, and pharmacy benefits in the observation month, the 6-months prior to that month (to identify baseline characteristics), and the month following (to assess follow-up care within 30 days).

Outcomes

We first identified enrollees who initiated a stimulant in a given month and used the enrollee-month as our unit of observation. Stimulant initiation was defined as the first stimulant prescription fill during the period (hereafter the “index fill”) with no stimulant fills in the previous 90 days. The index fill date was considered the date of initiation. We focused on the first initiation to identify new stimulant use during the period (for example, we did not want to consider a child who stopped filling stimulant prescriptions in the summer but resumed the exact same medication regimen when school resumed as an initiator in the fall).

Among stimulant initiations, we categorized whether initiation occurred via telemedicine vs. in-person and whether there was a subsequent outpatient visit with an ADHD diagnosis from the stimulant prescriber within 30 days of the index fill. To determine whether the initiation was via telemedicine or in-person, we used a published algorithm of medication initiation via telemedicine (see Supplement) that focuses on the types of visits observed in a 10-day window around the index fill.³⁶ We examined follow-up visits given this is a quality measure for children and adults with ADHD who initiate stimulants.^37,38

All analyses were conducted separately for children 2-17 and adults 18-64.

Models

To examine factors associated with stimulant initiation, we modeled monthly initiation among individuals enrolled in that month from January 2019 to April 2022. Because initiations are rare events, we used a Poisson regression model to link initiations to covariates. Models included: age category (children: 2-11 and 12-17; adults: 18-25, 26-40, and 41-64); documented sex; Census region; rural (vs. non-rural) residence; quartiles for median household income for the enrollee’s county of residence; and indicators for whether the enrollee had any claims with a substance use disorder (SUD) diagnosis, any claims with a mental health diagnosis other than ADHD, or a behavioral health specialty visit (i.e., visit with a psychiatrist, psychologist, or social worker) in the prior 180 days. (See Supplement for detail on variables.) To address temporal and seasonal trends in initiation, the models included a count variable for month, setting March 2020 to 0, negative counts from January 2019 to February 2020, and positive values from April 2020 to April 2022; a dummy variable for the COVID period (beginning April 2020); a dummy variable if the observation was during the summer months (May, June, July, August); an interaction between the month and COVID variables; and an interaction between the summer and COVID variables. We included a dummy variable for whether the initiation occurred during a summer month, anticipating that, at least for children, there might be a different stimulant initiation pattern during the summer, a time when demands on children’s attention, executive function, and behavior may be less stringent than during the school year. Our primary variables of interest in this model were the interaction terms of month and summer with the COVID pandemic dummy variable. Because of the potential modifying effects of summer and month on the association of COVID with initiations, we summarized findings as the monthly mean adjusted number of initiations per 100,000 enrollee-months during the pre-COVID period and then again during the COVID-19 pandemic. Because of the computational burden of using the full sample, we selected a 20% simple random sample of enrollees for these models.

Then, focusing on initiations that occurred April 2020 through April 2022, we used logistic regression to identify patient characteristics associated with telemedicine initiation versus in-person. We focused on the COVID period for this analysis because telemedicine was rarely used before that period (less than 1% of stimulant initiations). Models included the same patient-level characteristics described above, replacing the behavioral health specialty visit indicator with an indicator of whether initiation was conducted by a psychiatrist (vs. a primary care physician or other specialist). The models also included the summer variable plus linear and quadratic month counts.

Finally, as an estimate of care quality, among stimulant initiations during the pandemic period we estimated a logistic regression model to identify factors associated with receiving a follow-up visit within 30 days, adapted from the Healthcare Effectiveness Data and Information Set (HEDIS) measure monitoring follow-up care in ADHD.^37,38 The model included the same variables used in the logistic model of telemedicine initiation described above, as well as a dummy variable indicating whether the initiation occurred via telemedicine (our primary variable of interest).

This study was deemed exempt by the Harvard Institutional Review Board and follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines.³⁹ Analyses were conducting using SAS for Windows Version 9.4.

Results

In the 20% random sample, there were 9,914,590 enrollee-months among children 2-17 (representing 535,629 unique children) and 40,308,509 enrollee-months among adults 18-64 (2,116,160 unique adults) from January 2019 to April 2022 that met study criteria (Table 1). Among stimulant initiations by children in the COVID period (April 2020 to April 2022) (N=17,560), just over half (52.9%) were for children under age 12, 59.8% were male, and 16.0% were via a psychiatrist (versus a primary care physician such as a pediatrician or other type of specialist). Among the initiations by adults (N=42,527), 28.4% were 18-25, 44.8% were 26-40, and 26.8% were 41-64. Less than half (43.3%) were for male patients, approximately one-quarter (22.7%) were conducted by a psychiatrist, and 4.0% were for patients with co-occurring SUD.

Table 1:

Characteristics of all enrollee-months in cohort and of COVID-era stimulant initiations

	All enrollee-months that meet inclusion criteria from January 2019 through April 2022				Stimulant initiations from April 2020 through April 2022
	Age 2-17 (N=9,914,590)		Age 18-64 (N=40,308,509)		Age 2-17 (N=17,560)		Age 18-64 (N=42,527)
	N	%	N	%	N	%	N	%
Age group
2-11	5,844,836	59.0	--	--	9,288	52.9
12-17	4,069,754	41.0	--	--	8,272	47.1
18-25	--	--	5,814,605	14.4	--	--	12,085	28.4
26-40	--	--	12,341,259	30.6	--	--	19,046	44.8
41-64	--	--	22,152,645	55.0	--	--	11,396	26.8
Sex
Female	5,066,511	51.1	20,236,578	50.2	7,057	40.2	24,099	56.7
Male	4,843,488	48.9	20,058,471	49.8	10,503	59.8	18,428	43.3
Region
Midwest	2,901,437	29.3	10,979,177	27.2	5,726	32.6	11,737	27.6
West	2,058,728	20.8	7,503,892	18.6	2,872	16.4	8,006	18.8
South	3,935,257	39.7	17,291,402	42.9	7,364	41.9	18,263	42.9
East	1,019,168	10.3	4,534,038	11.2	1,598	9.1	4,521	10.6
Rural residence	196,221	2.0	911,464	2.3	308	1.8	638	1.5
Co-occurring SUD					115	0.7	1,719	4.0
Initiation by psychiatrist					2,810	16.0	9,672	22.7

Note: The enrollee-months data presented in the first panel cover a 20% sample of all enrollee-months with medical, behavioral health, and pharmacy benefits for six months before the observation month, the observation month itself, and one month after. This 20% random sample includes 535,629 unique children 2-17 and 2,116, 160 unique adults 18-64. The second panel includes information on people who initiated a stimulant during the COVID period after telemedicine became commonly used (April 1, 2020 through April 30, 2022).

SUD = substance use disorder.

Trends in Stimulant Initiations and Factors Associated with Initiation

In unadjusted analyses, the percentage of children with a stimulant initiation was 0.06% in January 2019 and 0.07% in April 2022 (Figure 1). The percentage of adults who initiated stimulants increased from 0.03% in January 2019 to 0.04% in April 2022. Initiation exhibited a pattern of seasonal variation (e.g., lower in the summer months when school is out) in children, whereas in adults this pattern was not evident.

Figure 1:

Percentage of child (age 2-17) and adult (age 18-64) enrollees who initiate stimulants, January 2019-April 2022

Using the Poisson regression model, among adults, the risk of stimulant initiation was higher for those with a behavioral health specialty visit in the past six months (rate ratio (RR)=2.49, 95% CI=2.37-2.60), males (vs. females)=1.15, 95% CI=1.11-1.19), those with a recent non-ADHD mental health diagnosis (RR=6.50, 95% CI=6.20-6.82), and those residing in higher income areas (vs. areas in the lowest income quartile (e.g., for quartile 4 vs. quartile 1, RR=1.28, 95% CI=1.21-1.35) (table available online). These patterns were similar among children. Regardless of age group, initiations decreased in the summer relative to non-summer months (e.g., among adults, RR=.93, 95% CI=.87-1.00).

In both adults and children, a small monthly decline in initiations was observed (RR=0.99, 95% CI=0.98-1.00 for children and RR=0.96, 95% CI=0.96-0.97 for adults) in the pre-pandemic period (Table 2). However, during the pandemic, a decline in initiations (RR=0.86, 95% CI=0.77-0.97) for children and an increase for adults (RR=1.09, 95% CI=1.00-1.19) relative to the pre-pandemic period were observed, although the rate of monthly initiations increased in both groups. Over the entire observation period, the mean monthly adjusted number of stimulant initiations per 100,000 enrollees for children aged 2-17 remained approximately the same before and during the pandemic (pre=57; during= 56), whereas initiations increased in adults (pre=27; during = 33) (data not shown).

Table 2:

Factors associated with initiating stimulants

	Age 2-17		Age 18-64
	Rate Ratio	Wald 95% Confidence Limits	Rate Ratio	Wald 95% Confidence Limits
Age group (Reference: ages 2-11 for the Age 2-17 cohort; ages 18-25 for the Age 18-64 cohort)
12-17	0.88	0.83, 0.92	--	--
26-40	--	--	0.77	0.73, 0.80
41-64	--	--	0.27	0.26, 0.29
Male	1.61	1.52, 1.69	1.15	1.11, 1.19
Rural residence	1.03	0.85, 1.26	0.88	0.76,1.02
Income quartile (reference:Q1)
Q2	1.05	0.97, 1.13	1.17	1.11, 1.24
Q3	1.08	1.00, 1.16	1.32	1.26, 1.40
Q4	1.19	1.11, 1.28	1.28	1.21, 1.35
SUD diagnosis past 6 months	1.11	0.83, 1.46	0.83	0.77, 0.91
Non-ADHD MH diagnosis in past 6 months	5.10	4.71, 5.51	6.50	6.20, 6.82
Visit with behavioral health specialist in past 6 months	3.34	3.08, 3.63	2.49	2.37, 2.60
Month	0.99	0.98, 0.99	0.96	0.96, 0.97
COVID	0.86	0.77, 0.97	1.09	1.00, 1.19
Summer	0.63	0.57, 0.70	0.893	0.87, 0.99
Month*COVID	1.03	1.02, 1.04	1.06	1.05, 1.07
Summer*COVID	1.00	0.88, 1.14	1.04	0.95, 1.13

Note: These models were estimated on a 20% random sample of the enrolled population due to constraints of the virtual research data platform and the size of the enrolled population. Parameter estimates from the Poisson models were exponentiated to produce rate ratios. Models also included dummies for Census region.

MH= mental health SUD = substance use disorder

Behavioral health specialist = psychiatrist, psychologist, clinical social worker

Trends in Percentage of Initiations Conducted Via Telemedicine and Factors Associated with Telemedicine Initiation

Among adults who initiated stimulants, the telemedicine percentage reached a peak of 53.8% in April 2020 and then dropped almost in half to 28.3% by April 2022 (Figure 2). In contrast, among children, the unadjusted percentage of initiations via telemedicine reached a peak of 56.6% in April 2020, and then began a steady decline, dropping to a low of just 13.3% by April 2022. During the COVID-19 period, a much larger share of stimulant initiations conducted by psychiatrists (vs. non-psychiatrists) were performed via telemedicine. Psychiatrists used telemedicine for 55.2% of adult stimulation initiations versus only 26.6% for non-psychiatrists. For child initiations, almost half (48.5%) conducted by psychiatrists were via telemedicine while only 17.4% of those conducted by non-psychiatrists were via telemedicine.

Figure 2:

Percentage of stimulant medication initiations conducted via telemedicine for children 2-17 and adults 18-64, January 2020 through April 2022

Note: Results from logistic regression model of receipt of a stimulant initiation via telemedicine are available in Exhibit A.2. The vertical line indicates the initiation of the COVID-19 public health emergency declaration.

In regression analyses, among adults, the odds of a telemedicine initiation were lower for those living in rural (vs. non-rural) areas (odds ratio (OR)= 0.75, 95% CI=0.61-0.92) and higher among those whose initiations were conducted by a psychiatrist vs. another type of prescriber (OR= 3.02, 95% CI=2.87-3.17), those with a co-occurring non-ADHD mental health condition (OR=1.70, 95% CI=1.62-1.77), and those who lived in a higher income area (e.g., OR=1.72, 95% CI=1.61-1.84 for quartile 4 vs. quarter 1 ) (table available online). These patterns were similar for children. Among adults, those 26-40 had higher odds of a telemedicine initiation (OR=1.14, 95% CI=1.08-1.20) and those 41-64 had lower odds (OR= 0.84, 95% CI=0.79-0.89) relative to those 18-25. Also, male adults had lower odds of a telemedicine initiation relative to female adults (OR=0.87, 95% CI=0.83-0.91).

Trends in Outpatient Follow-Up After Stimulant Initiation and Factors Associated with a Follow-up Visit

In unadjusted analyses, on average, from January 2019 through April 2022 only one in four (26.5%) stimulant initiations among children 2-17 and one in five (21.7%) among adults 18-64 received a follow-up visit from the prescriber within 30 days (figure available online). In regression analyses, the likelihood of a follow-up visit was positively associated with a telemedicine initiation (relative to an in-person initiation) for both children (1.09, 95% CI=1.00-1.19) and adults (1.61, 95% CI=1.53-1.69) (table available online). Odds of follow-up care were also higher among those whose initiation was conducted by a psychiatrist (1.33, 95% CI=1.21-1.46 children; 1.45, 95% CI=1.38-1.53 adults) and those with a co-occurring mental health diagnosis other than ADHD (1.17, 95% CI=1.09-1.25 children; 1.19, 95% CI=1.13-1.25 adults).

Discussion

In this commercially-insured population, we found notably different trends in stimulant initiation among adults and children. Pre-pandemic, monthly stimulant initiations were declining regardless of age. During the pandemic, among children, the mean monthly adjusted number of initiations per enrollee was stable compared to pre-pandemic, while initiations increased among adults. An increase in stimulant prescriptions fills during the pandemic among adults has been seen in recent research by the Centers for Disease Control and Prevention and news reports.^28-30 Our research extends this prior work by focusing on new stimulant starts, which is currently a focus of policy discussions, and adjusting for patient clinical and demographic characteristics.

There was a sudden peak early in the pandemic in telemedicine use to start stimulants, followed by a slow decline; by spring 2022, telemedicine was still used for a substantial minority of stimulant starts (13% children, 28% adults). Telemedicine initiation offered a way for adults and children to access ADHD treatment early in the pandemic when many clinics closed or curtailed in-person visits, and it may have helped reduce barriers to accessing care even after in-person visits became more of an option. The drop in telemedicine initiation during the pandemic across children and adults is not surprising given the large share of initiations conducted by primary care physicians, who rapidly returned to delivering most visits in-person.^40-42 The larger drop for children vs. adults may be driven by several factors. ADHD is more commonly treated by psychiatrists than primary care providers among adults relative to children,^43-46 and psychiatrists have been more likely to use telemedicine since the pandemic’s onset. Telemedicine may also be a less effective modality for young children.⁴⁷ Finally, the dataset may include some care provided by telemedicine-only companies, which deliver fully virtual care models and tend to focus on adult populations.

For both children and adults, non-rural residents and those who live in higher-income counties were more likely to receive initiations via telemedicine. These results are consistent with other prior literature and point to additional concerns about disparities in access to telemedicine among those who reside in rural areas or areas with lower household income.^36,48,49

The recent investigation by the Department of Justice into at least one direct-to-consumer telehealth company that prescribes stimulants highlights how policy makers have to balance public safety with access to care.⁵⁰ The DEA’s previously proposed rules would no longer allow initiation of stimulants via telemedicine (i.e., without an in-person medical evaluation).⁵¹ Our results highlight that the proposed rule could have important implications for access to stimulants given that one in seven stimulant initiations for children and one in four for adults were delivered via telemedicine as recently as spring 2022. Our finding that follow-up care within 30 days was more likely when the initiation was conducted via telemedicine (vs. in-person) suggests that telemedicine may increase opportunities for patients to maintain closer follow-up. This is consistent with prior research that describes patients reporting improved satisfaction due to increased convenience or timeliness of their care via telemedicine.^52-54 Instead of a blanket ban on stimulant prescribing, federal policy could instead focus on identifying and curtailing telehealth companies or other providers that are inappropriately prescribing these medications absent appropriate clinical assessment of patients, and by developing advertising regulation that addresses the promotion of prescription drugs via social media. ^55-57

There are several limitations to our analysis. Using claims data, we are unable to determine whether stimulant initiation was clinically appropriate or whether it led to diversion. Quality of care can be difficult to discern with claims data in general, and claims lack information that clinicians might use in monitoring stimulant side effects and outcomes (e.g., data on weight and height for children). Further, while follow-up within 30 days of initiating stimulants is a NCQA-endorsed quality measure for children and adolescents,³⁷ the similar measure in adults is in an earlier process of development and validation.³⁸ We are unable to distinguish initiations delivered by fully virtual, “direct to consumer” telehealth providers from those delivered by providers who maintain a “brick and mortar” practice setting but also deliver telehealth. Additionally, our findings use data from one large national commercial insurer; they may not be generalizable to enrollees of other commercial insurers, Medicaid or Medicare beneficiaries, or individuals who are uninsured. We are also unable to examine stimulant initiations by race or ethnicity categories, because this information is unavailable in commercial insurance claims. Finally, we are unable to include new stimulant fills for which there are no associated outpatient visits, which would occur if an individual fills the prescription using their pharmacy benefit but does not submit the outpatient visit for insurance reimbursement. Of note, stimulant shortages were first reported in late July 2022 and again by the FDA in October 2022;^58,59 given that our study period ended in April 2022, it is unlikely these shortages affected our results.

Conclusion

Stimulant initiation increased during the COVID period among adults relative to the pre-pandemic period, with a substantial fraction of initiations for both children and adults occurring via telemedicine. Telemedicine initiation of stimulants was associated with higher rates of follow-up within 30-days among adults and children. Future research is needed to understand the extent to which patterns of stimulant initiation and follow-up care observed since the onset of the pandemic represent high-quality care.

Highlights.

• Compared to pre-pandemic, stimulant initiations remained the same for children during the pandemic but increased among adults.

• Telemedicine initiations of stimulants account for a sizeable minority of both adult and child initiations, are more commonly conducted by psychiatrists (vs. other prescribers), and are associated with greater odds of receiving follow-up care.

• Telemedicine initiations were less common among adults and children residing in rural (compared to urban) and lower-income areas.

• Recently proposed rules to prohibit controlled substance prescribing without an in-person medical evaluation would require a significant change in current practice and potentially limit access to stimulant medication and related ADHD care.