Antidepressant Dispensing to US Adolescents and Young Adults: 2016–2022

We analyzed the IQVIA Longitudinal Prescription Database, a prescription-level, all-payer database that reports dispensing from most retail, mail-order, and long-term care pharmacies in the United States. The database does not capture prescription drug dispensing from pharmacies only open to patients of specific health systems, such as the Veterans Administration or Kaiser Permanente. From 2016 through 2022, the database included 92.5% to 94.4% of antidepressant prescriptions dispensed in US retail pharmacies, suggesting that coverage of this drug class was both comprehensive and stable during the study period (personal communication with Allen Campbell, IQVIA Institute, November 3, 2022). Data elements include encrypted patient identifiers, demographic characteristics (age, sex, state), days supplied, and prescription method of payment (commercial, Medicaid, Medicare, cash). Data on race and ethnicity, income, and prescription indication are not reported. As IQVIA data are deidentified, this analysis was not regulated as human subjects research by the Institutional Review Board of the University of Michigan Medical School.

Following previous studies,²⁴ we defined adolescents as ages 12 to 17 years and young adults as ages 18 to 25 years. The sample included antidepressant prescriptions dispensed to adolescents and young adults residing in 1 of the 50 US states or the District of Columbia between January 2016 and December 2022.¹⁵ We used IQVIA’s market definition to identify antidepressant prescriptions (Supplemental Table 3). This definition includes selective serotonin reuptake inhibitors, selective norepinephrine reuptake inhibitors, tricyclic antidepressants, and other antidepressants (eg, bupropion).

The outcome was the monthly antidepressant dispensing rate, defined as the monthly number of adolescents and young adults with ≥1 dispensed antidepressant prescription per 100 000 people aged 12 to 25 years. We obtained population denominators from the Census Bureau (Supplemental Table 4).

Using an interrupted time series design, we fitted linear segmented regression models assessing for abrupt level and slope changes in outcomes during March 2020.²⁵ We assessed for autocorrelation with the Cumby-Huizinga test and accounted for any autocorrelation using robust Newey-West standard errors with the appropriate number of lags.²⁶ To facilitate interpretation, we calculated the difference between the observed antidepressant dispensing rate in December 2022 and the rate predicted by the counterfactual trend (ie, the rate if pre-March 2020 trends had continued).

In subgroup analyses, we repeated analyses among populations defined by sex and age group (12–17 vs 18–25 years) and populations defined by sex and Census region. As in the overall analysis, we derived subgroup-specific population denominators from the Census Bureau. A small number of individuals with unknown sex were excluded from the subgroup analyses.

For the analyses, we used R version 4.2.1 and 2-sided hypothesis tests with α = .05. Additional details on the statistical analysis are included in the Supplemental Information.

We repeated analyses when modeling the monthly number of individuals with ≥1 dispensed antidepressant prescription, an outcome that avoids any uncertainty in the estimation of population denominators, and when modeling the monthly number of days supplied of antidepressants dispensed per 100 000 individuals, an outcome that accounts for any shifts in the mean duration of antidepressant prescriptions. To account for seasonality, we repeated analyses of the primary outcome when controlling for quarter (1 = January-March, 2 = April-June, 3 = July-September, 4 = October-December). Finally, we limited the definition of antidepressants to selective serotonin reuptake inhibitors.

From 2016 to 2022, the database included 221 268 402 antidepressant prescriptions dispensed to 18 395 915 individuals, of whom 11 836 944 (64.4%) were female, 6 508 947 (35.8%) were male, and 50 025 (0.3%) were of unknown sex. At the time of sample entry, mean (SD) age was 19.2 (3.9) years, 11 908 010 (64.7%) individuals were aged 18 to 25 years, and 7 100 766 (38.6%) individuals resided in the South, compared with 2 961 720 (16.1%) in the Northeast, 4 719 080 (25.7%) in the Midwest, and 3 614 349 (19.6%) in the West. Among the 221 268 402 dispensed antidepressant prescriptions, 148 977 404 (67.3%) were for selective serotonin reuptake inhibitors. The 3 most common medications were sertraline (53 225 721; 24.1%), fluoxetine (40 815 093; 18.5%), and escitalopram (36 169 696; 16.3%).

Table 1 reveals the annual characteristics of adolescents and young adults with ≥1 dispensed antidepressant prescription during the year, as well as the characteristics of these prescriptions. Between 2016 and 2022, the number of adolescents and young adults with ≥1 dispensed antidepressant prescription increased by 46.1%, from 4 633 433 to 6 768 106. The number of adolescents and young adults initiating antidepressant therapy (defined as having a lack of antidepressant dispensing in the previous 180 days) increased by 31.0%, from 2 750 774 to 3 603 045. In 2016, advanced practice providers, defined as nurse practitioners and physician assistants, accounted for 6 081 971 of the 25 472 600 dispensed antidepressant prescriptions (23.9%). By 2022, these providers accounted for 15 787 158 of the 38 678 393 dispensed antidepressant prescriptions (40.8%).

Table 2 reveals coefficients from segmented regression models for the monthly antidepressant dispensing rate. Between January 2016 and December 2022, this rate increased by 66.3%, from 2575.9 to 4284.8 individuals per 100 000 (Fig 1). Before March 2020, this rate increased by 17.0 (95% confidence interval [CI]: 15.2 to 18.8) per month. The outbreak was not associated with a level change (−37.4, 95% CI: −153.4 to 78.7) but was associated with a slope increase (10.8 per month, 95% CI: 4.9 to 16.7). From March 2020 onwards, the monthly antidepressant dispensing rate increased by 27.8 per month (95% CI: 22.1 to 33.4), or 63.5% higher than the rate of change before March 2020. During December 2022, the antidepressant dispensing rate was 166.7 (4.0%) higher than predicted by the counterfactual trend.

Among female adolescents, the monthly antidepressant dispensing rate increased by 17.9 per month (95% CI: 13.7 to 22.2) before March 2020. The outbreak was not associated with a level change (−188.8, 95% CI: −413.6 to 36.0) but was associated with a slope increase (23.1 per month, 95% CI: 14.3 to 32.0). From March 2020 onwards, the monthly antidepressant dispensing rate increased by 41.1 per month (95% CI: 32.9 to 49.2), or 129.6% higher than the rate of change before March 2020. Among male adolescents, the monthly antidepressant dispensing rate increased 8.7 per month (95% CI: 7.1 to 10.4) before March 2020. The outbreak was associated with a level decrease (−224.3, 95% CI: −328.2 to −120.4) but not with a slope change (1.1 per month, 95% CI: −2.3 to 4.4). In December 2022, the antidepressant dispensing rate among male adolescents was 182.2 (−7.1%) lower than predicted by the counterfactual trend, compared with 514.6 (12.7%) higher than predicted among female adolescents (Fig 2A).

Among female young adults, the monthly antidepressant dispensing rate increased by 28.6 per month (95% CI: 25.3 to 31.9) before March 2020. The outbreak was associated with a level increase (227.3, 95% CI: 6.9 to 447.6) and slope increase (16.2 per month, 95% CI: 4.6 to 27.8). From March 2020 onwards, the monthly antidepressant dispensing rate increased by 44.8 per month (95% CI: 33.3 to 56.3), or 56.6% higher than the rate of change before March 2020. Among male young adults, the monthly antidepressant dispensing rate increased 11.7 per month (95% CI: 10.5 to 13.0) before March 2020. The outbreak was not associated with a level change (17.8, 95% CI: −51.3 to 86.8) or slope change (3.7 per month, 95% CI: −0.8 to 8.2). In December 2022, the antidepressant dispensing rate among male young adults was 0.1 (0.0%) higher than predicted by the counterfactual trend, compared with 403.8 (6.2%) higher among female young adults (Fig 2B).

Among female adolescents and young adults in each Census region, the outbreak was not associated with a level change in the monthly antidepressant dispensing rate but was associated with a slope increase. In December 2022, the antidepressant dispensing rate was 3.3% to 10.1% higher than predicted by the counterfactual trend depending on the region, with the smallest increase occurring in the Midwest and the largest increase occurring in the South.

Among male adolescents and young adults, the outbreak was associated with level decreases in the antidepressant dispensing rate in every region, except the West, but was only associated with slope increases in the Northeast and South. In December 2022, the antidepressant dispensing rate was −6.0% to −0.1% lower than predicted by the counterfactual trend depending on the region, with the greatest decrease occurring in the Midwest and the smallest decrease occurring in the South (Fig 3).

In the main analysis, the COVID-19 outbreak was not associated with a slope change in the monthly antidepressant dispensing rate among male young adults. In contrast, the outbreak was associated with a slope increase in this population when modeling the monthly number of individuals with ≥1 antidepressant prescription, when modeling the monthly number of days supplied of antidepressants dispensed per 100 000 people, and when controlling for season. However, in all sensitivity analyses, the outcome decreased after the outbreak among male adolescents but increased among female adolescents and young adults (Supplemental Tables 5–9).

In this analysis of data from an all-payer, national prescription dispensing database, the monthly antidepressant dispensing rate among US adolescents and young adults increased by 66.3% between January 2016 and December 2022. This rate was increasing before the COVID-19 outbreak in March 2020 but increased 63.5% faster after the outbreak. Subgroup analyses revealed that the increase in antidepressant dispensing after the outbreak was driven by females in all Census regions and occurred despite a decline in dispensing to male adolescents.

Our findings are consistent with a national study revealing an increase in antidepressant dispensing among US adolescents aged 12 to 17 years from March 2020 to September 2020 compared with the same period in 2019.²² Findings differ from a national study revealing minimal changes in the number of antidepressant prescriptions dispensed to US individuals aged 0 to 19 years from March 2020 to December 2020 compared with the same period in 2019 and with a national study revealing a decline in new starts of antidepressants among individuals aged 0 to 18 years from March 2020 to August 2020, perhaps owing to different age groups and time periods.^21,23 Our analysis builds on these 3 studies by using data through 2022, thus allowing for the assessment of changes in antidepressant dispensing over a longer period after the COVID-19 outbreak. We also used several years of data before 2020, which facilitated the establishment of the counterfactual trend.

Although our study does not allow for the identification of the mechanism of the observed increase in antidepressant dispensing to adolescents and young adults, multiple factors are likely involved. The increase may partly reflect a greater need for antidepressants, given evidence that the prevalence of depression and anxiety increased among adolescents and young adults after the outbreak.^11,–14 Changes in access and treatment patterns may also have played a role. For example, for patients facing long waitlists for psychotherapy after the outbreak, initiating or continuing antidepressant therapy may have been more practical than relying on a therapy-only approach. Additionally, the shift toward telehealth for mental health care delivery may have increased access to clinicians who could prescribe antidepressants, at least among individuals without barriers to telehealth use.²⁷ 

Subgroup analyses revealed marked heterogeneity in changes in antidepressant dispensing after the outbreak by sex and age group. Among female adolescents, a population with one of the highest rates of depression and anxiety of any demographic subgroup even before the COVID-19 outbreak^1,3, the monthly antidepressant dispensing rate increased 129.6% faster from March 2020 through December 2022 compared with January 2016 through February 2020. This finding aligns with studies revealing marked increases in mental health symptoms, emergency department visits for mental health conditions, and hospitalizations for these conditions among female adolescents after the outbreak.^15,28,–31 Collectively, these studies, along with ours, are consistent with the notion that the COVID-19 pandemic exacerbated a preexisting mental health crisis among female adolescents.³² 

In sharp contrast to female adolescents, the monthly antidepressant dispensing rate among male adolescents declined abruptly during March 2020 and did not recover afterward. One potential explanation for this finding is that the mental health of male adolescents improved during the pandemic, thus decreasing their need for antidepressants. However, this explanation is not supported by data revealing that emergency department visits for suspected suicide attempts in male adolescents were higher in early 2021 compared with early 2019,¹⁵ or by data revealing that the proportion of male high school students reporting persistent feelings of sadness or hopelessness increased between 2019 and 2021.⁴ A more plausible explanation is that male adolescents received care for mental health symptoms less often after the outbreak despite a lack of a decrease in the frequency of these symptoms. In support of this possibility, an analysis of national commercial claims data from 2016 to 2021 revealed that dispensing of medications for attention-deficit hyperactivity disorder declined after the outbreak among males aged 15 to 24 years.³³ Future research should investigate if the decrease in antidepressant dispensing among male adolescents after the outbreak reflects underuse of these medications relative to the level of need.

Our study provides important information on changes in mental health utilization patterns among young adults after the COVID-19 outbreak. As with adolescents, changes varied markedly by sex, with little change in antidepressant dispensing to males and an increase among females. Although this increase was smaller compared with the increase among female adolescents, findings suggest mental health may have similarly worsened among female young adults after the outbreak, highlighting the importance of implementing interventions to promote mental health and improve access to mental health care in this population. For example, expanding insurance coverage may be one particularly important such intervention, given that young adults have the highest rates of uninsurance of any age group.³⁴ 

This study has several strengths, including the use of timely national data over a 7-year period and the use of patient demographic data to explore heterogeneity in changes by sex, age, and region. However, the study had limitations. First, as with most pharmacy dispensing databases, the IQVIA database does not include information on prescription indication. Consequently, we could not determine if antidepressants were written to treat depression, anxiety, or other conditions, nor whether their use was guideline-concordant.^35,–37 Second, the IQVIA database does not include information on whether antidepressants were prescribed during in-person versus telehealth visits. Third, the IQVIA database only includes information on prescription method of payment, which does not always equate to insurance type (eg, privately insured patients who pay for prescriptions with cash). As such, we could not examine whether changes in the monthly antidepressant dispensing rate after March 2020 varied according to whether patients were privately or publicly insured. Fourth, it was beyond the study’s scope to examine changes in the concurrent use of multiple antidepressants after the COVID-19 outbreak, changes in the concurrent use of other psychotropic drugs, or changes in patterns of antidepressant initiation and retention in therapy.

Finally, it is unclear whether findings can be generalized beyond 2022. The number of mental health-related emergency department visits among US adolescents was lower in fall 2022 compared with fall 2021,³⁸ and the end of the Medicaid continuous enrollment provision on March 31, 2023 has resulted in large-scale Medicaid disenrollment, potentially disrupting antidepressant therapy among Medicaid-insured adolescents and young adults.^39,40 Thus, the increase in antidepressant dispensing to adolescents and young adults observed in this study could slow over time.

Antidepressant dispensing to adolescents and young adults was rising before the COVID-19 outbreak and rose even faster afterward. Future research should investigate the degree to which this increase was driven by changes in mental health, changes in access to mental health care, and changes in treatment patterns for mental health conditions. Future research should also investigate which interventions can best promote the mental health of adolescents and young adults, thus mitigating the deleterious implications of poor mental health both in the short term and long term.

The authors would like to acknowledge Dr Michael Tang, MD, MBA, for his feedback on this manuscript. He was not compensated for these contributions. This study was presented at the American Academy of Pediatrics Presidential Plenary during the 2023 Pediatric Academic Societies meeting in Washington, DC on May 1, 2023.

CI

confidence interval

COVID-19

coronavirus disease 2019