COVID-19 and Adolescent Depression and Suicide Risk Screening Outcomes

Our study took place in the Children’s Hospital of Philadelphia (CHOP) primary care network, a large pediatric primary care network made up of 29 urban, suburban, and semirural practices in the greater Philadelphia region.¹⁷  The Care Network provides care for nearly 300 000 patients. Using a repeated cross-sectional design, we extracted electronic health record data from all preventive visits by adolescents aged 12 to 21 (up to the 22nd birthday) who attended a primary care visit in the CHOP primary care network between June 1, 2019, and December 31, 2020. The CHOP Institutional Review Board determined this study to be exempt from review.

To examine changes in screening, depression, and suicide risk from before the COVID-19 pandemic to during the pandemic, we defined 2 periods: June to December 2019 (prepandemic) and June to December 2020 (pandemic). These months were chosen as comparison periods because primary care visit volume had begun to normalize in June 2020 after a steep decline in March through May 2020. We assessed 3 primary outcomes in each period (overall and by month): (1) the percentage of adolescents screened for depression and suicide risk among those attending a primary care preventive visit, (2) the percentage of adolescents screened with a positive result for depression symptoms, and (3) the percentage of adolescents screened with a positive result for suicide risk. Depression and suicide risk during the 2 weeks before screening were assessed routinely through a fully automated electronic screening system by using the Patient Health Questionnaire-Modified for Teens (PHQ-9-M), which includes 9 core items and 2 supplemental items assessing suicide risk.^11,18  CHOP implemented the PHQ-9-M for routine depression and suicide risk screening after review of available tools by a committee of screening experts. The PHQ-9-M was chosen because it is brief, nonproprietary, has been validated and used previously in large health systems,^19,20  and is recommended in clinical guidelines including the Guidelines for Adolescent Depression in Primary Care (GLAD-PC)¹³  and the AAP’s Bright Futures recommendations.²¹  At CHOP, the PHQ-9-M is currently available in English and Spanish. PHQ-9-M scores of 11 to 27 indicate a positive result for moderate-to-severe symptoms of depression, whereas scores of 5 to 10 indicate borderline (ie, subthreshold) depressive symptoms. Suicide risk was assessed using 3 questions; an affirmative response to any of the 3 flagged the adolescent for suicide risk. The questions included the item on the PHQ-9-M assessing thoughts of death and/or self-harm (“Thoughts that you would be better off dead, or of hurting yourself in some way?”; on a scale ranging from 0 to 3, where a score of 1 or greater was considered endorsement of risk) and 2 supplemental yes or no questions (“Has there been a time in the past month when you have had serious thoughts about ending your life?”; “Have you ever, in your whole life, tried to kill yourself or made a suicide attempt?”) Because the item on thoughts of death or self-harm may identify adolescents with a wide range of suicidal ideation severity,¹⁸  we also examined the 2 supplemental questions separately.

We examined changes in depression and suicide risk screening outcomes during the pandemic on the basis of 4 demographic variables: sex, race and ethnicity (non-Hispanic Black, non-Hispanic white, Hispanic, Asian American, other race), payer (public, commercial, self-pay or other), and neighborhood socioeconomic status (quartiles of census tract-level median household income). Sex, race and ethnicity, and payer were extracted from the electronic health record. Patient addresses were geocoded by the health system and linked to residential census tracts. Neighborhood socioeconomic status was assessed as census tract-level median household income using data from the American Communities Survey 2014–2018 5-year estimates, and was divided into quartiles for analysis.²² 

Changes in screening, positive depression screens, and positive suicide risk screens between the prepandemic and pandemic periods were calculated overall, by month, and across subgroups on the basis of sex, race and ethnicity, insurance, and neighborhood income. We used logistic regression, followed by marginal standardization, to calculate 95% confidence intervals (CIs) around the absolute changes in each outcome. We used Poisson regression with robust variance estimates²³  to calculate prevalence ratios (PRs) to show changes in depression and suicide risk screening outcomes on a relative scale. We estimated robust standard errors accounting for the clustering of adolescents within primary care practices. To examine differences in the amount of change between demographic subgroups, we used multivariable Poisson models including the demographic variables described above, a binary indicator for time period (June–December 2020 versus June–December 2019) and interaction terms between time period and demographic variables. Statistical analyses were conducted by using Stata version 16.1 (Stata Corp, College Station, TX) and R version 3.6.1.

Our study sample included 91 188 well visits by 68 669 adolescents, including 43 504 in the prepandemic period and 47 684 in the pandemic period. The total number of visits was higher during the pandemic period because of a local initiative to bring adolescents in for well visits where vaccines were needed in advance of the start of the school year, and because of rescheduling of adolescents whose visits were canceled during the early months of the pandemic. Adolescents’ average age was 15.2 (SD: 2.1) in the prepandemic and 15.3 (SD: 2.1) in the postpandemic period. The proportion of visits by non-Hispanic Black adolescents declined slightly during the pandemic period, as did the proportion of visits among commercially insured adolescents and those in the lowest quartile of neighborhood household income (Table 1).

The total number of adolescents screened for depression increased slightly, from 33 754 in the prepandemic period to 36 161 in the postpandemic period, reflecting the overall increase in visits during this period. However, the percentage of well visits at which a screen was completed declined slightly, from 77.6% to 75.8% (PR: 0.98, 95% CI: 0.90–1.06). The difference between 2019 and 2020 was most pronounced in June (PR: 0.86, 95% CI: 0.75–0.99; Fig 1A) and diminished over time. Screening rates declined among all demographic groups except non-Hispanic Black adolescents and those with noncommercial insurance (Table 2). However, adjusted PRs indicated there were minimal differences after accounting for demographic characteristics and clustering by practice (Table 2).

Overall, the percentage of adolescents screening positive for moderate-to-severe depressive symptoms (scores of 11–27) increased from 5.0% in the prepandemic period to 6.2% in the postpandemic period. This reflects a 24% increase in the prevalence of moderate-to-severe depressive symptoms (PR: 1.24, 95% CI: 1.15–1.34). When broken down by month, there was an increase in prevalence in the pandemic period in August, October, November, and December (Fig 1B). The amount of change differed by sex (P interaction <.001, Table 3), with an increase among female, but not male, adolescents (PR: 1.32 [1.24–1.42] vs 1.02 [0.89–1.67]). By race and ethnicity, changes were largest among non-Hispanic white (PR: 1.29, 95% CI: 1.18–1.42), non-Hispanic Black (PR: 1.24, 95% CI: 1.11–1.39), and other race adolescents (PR: 1.24, 95% CI: 1.00–1.54, P interaction = .02). Positive depression screens increased similarly among commercially and publicly insured adolescents and among adolescents of all neighborhood incomes (Table 3). The percentage of adolescents screening positive for borderline depressive symptoms increased from 15.6% to 16.7% (PR: 1.07, 95% CI: 1.02–1.13). By month, there was an increase in prevalence during the pandemic period in July and December (Fig 1C). Patterns differed by insurance type, with the largest increase among self-pay or other insurance (PR: 1.41, 95% CI: 1.06–1.87; Table 3).

The percentage of adolescents screening positive on any of the 3 suicide risk items increased from 6.1% to 7.1%, reflecting a 16% relative increase in prevalence (PR: 1.16, 95% CI: 1.08–1.26). The pandemic period had an increase in positive suicide risk screens during October, November, and December (Fig 1D). The increase in suicide risk was greater among female compared with male adolescents (PR: 1.18 [1.11–1.26] vs 1.09 [0.99–1.19], P interaction = .07) but did not differ significantly according to other sociodemographic characteristics (Table 4). For the two supplemental suicide questions, we similarly found small absolute increases from the prepandemic to pandemic periods but larger relative changes. The percentage endorsing recent suicidal thoughts increased from 1.8% to 2.2% overall (PR: 1.22, 95% CI: 1.10–1.37), with an increase relative to the prepandemic period during October and November (Fig 1E). There was a greater increase among female compared with male adolescents (PR: 1.34 [1.18–1.52] vs 1.00 [0.86–1.17], P interaction < .001). The relative increase in recent suicidal thoughts was higher among non-Hispanic white adolescents (PR: 1.45, 95% CI: 1.27–1.65) compared with other race and ethnicity groups (Table 4, P interaction = .16). Increases were similar between commercially and publicly insured adolescents, whereas the prevalence of recent suicidal thoughts decreased among adolescents with self-pay or other insurance. The percentage reporting a lifetime suicide attempt increased from 3.1% to 3.5% (PR: 1.13, 95% CI: 1.01–1.25) and did not differ substantially by month or by demographic characteristics (Fig 1F, Table 4).

In this analysis of depression and suicide risk screening among adolescents in pediatric primary care, we found that the percentage of adolescents screened initially declined but quickly returned to prepandemic levels. The percentage of positive depression screens increased in the pandemic period, particularly among female, non-Hispanic Black, and non-Hispanic white adolescents. Suicide risk–positive screens also increased, especially among female adolescents. By month, increases in positive screens relative to the prepandemic period tended to be largest later in the year (October–December). Although the absolute magnitude of the changes was modest, reflecting the low prevalence of positive screens overall, the relative increase in the percentage of adolescents screening positive for moderate-to-severe depressive symptoms or recent suicidal thoughts was substantial.

The initial decrease in screening prevalence during the pandemic was likely due to infection prevention efforts at practice sites that resulted in reduced tablet use for screening and efforts to room patients as quickly as possible. These workflow changes varied by network site, as demonstrated by the attenuation of sociodemographic differences after accounting for clustering by practice. These patterns suggest that urban practices caring for predominantly lower income, publicly insured, and non-Hispanic Black adolescents may have experienced less of a decline in screening during the early months of the pandemic. However, for depression and suicide risk outcomes, accounting for practice-level differences did not eliminate differences among demographic subgroups, indicating that results were more consistent across practices.

The changes in suicide risk we observed in primary care were smaller than those reported in a recent analysis of ED-based suicide risk screens.¹⁰  This may reflect the fact that the PHQ-9-M is routinely administered at preventive visits where adolescents are not specifically seeking care for depression or suicide risk. Adolescents with the most severe suicide risk concerns may have sought care at the ED, and the primary care population may reflect an adolescent sample with fewer social and medical needs. However, authors of a recent study of largely low-income adolescents in an urban primary care network in Ohio also reported higher suicide risk prevalence compared with our network, although they did not find increases in depression or suicide risk during the pandemic.²⁴  Our larger study population in the Philadelphia area included adolescents from urban, suburban, and semirural areas as well as a wide range of socioeconomic backgrounds, which may contribute to the differences in our findings. However, in the primary care context, even a small increase in adolescents screening positive for suicide risk may have a substantial impact on practice, given the amount of work required to safely engage at-risk adolescents in appropriate services. The mental health system is already overburdened, and many children’s hospitals have been overwhelmed by mental health complaints during the pandemic.^3,25  Our findings suggest that female adolescents in particular may be at heightened risk of suicidal ideation during the pandemic, which may have exacerbated national prepandemic trends.^2,26,27 

As with suicide risk, the most consistent increase in depressive symptoms in our analysis was among female adolescents. This also aligns with national prepandemic data²  and suggests that the pandemic may have exacerbated existing patterns, at least in the Philadelphia region. Social isolation and loneliness are associated with increased risk of depression in children and adolescents,²⁸  and school closures and other COVID-19 mitigation measures have restricted social interactions among adolescents. Adverse effects of social media (eg, bullying) have been implicated as potential contributors to the greater burden of depression among girls,²  and these effects may have been magnified during the pandemic as in-person social interactions have declined or shifted online. The increases in depressive symptoms among non-Hispanic Black and non-Hispanic white adolescents during the pandemic are also concerning. National prevalence estimates suggest depression rates are higher among non-Hispanic white and Hispanic adolescents compared with Black and Asian American adolescents.²⁹  However, Black and Hispanic adolescents may be less likely to receive a depression diagnosis and treatment when needed.¹⁵  Black communities have experienced disproportionate health and economic impacts of COVID-19 as a result of structural racism,^30,31  which may explain the increase in depressive symptoms in Black adolescents and suggest growing unmet mental health needs. In addition, persistent school closures and stricter social distancing policies in urban areas may have disproportionately increased isolation among non-Hispanic Black adolescents. Although COVID-19 has also heavily impacted Hispanic communities,³²  and anti-Asian American violence and rhetoric has risen during the pandemic,³³  we did not identify increases in moderate-to-severe depressive symptoms among Hispanic and Asian American adolescents. This is in contrast to national survey data indicating heightened rates of suicidal ideation and attempts among Hispanic youth.³⁴  However, because there were relatively few Hispanic and Asian American adolescents in our study population, and these groups are not monolithic and represent many nationalities, these findings should be interpreted with caution. Because we lack details regarding potential protective factors (eg, parenting practices, peer supports), the reason for these patterns remains unclear and warrants further investigation. Rates of borderline depressive symptoms, however, increased by a similar amount among Hispanic and Asian American compared with non-Hispanic white adolescents. Closer attention to adolescents with borderline depressive symptoms is needed, because these adolescents may be at risk for more severe depression and suicide in the future.

In addition to the primary care setting, screening for mental health concerns routinely occurs in schools. With the transition to virtual school during the pandemic, primary care may be the only screening opportunity available to many adolescents. The rise in mental health concerns suggests that additional resources may be needed to support clinicians in conducting depression and suicide screening. As one strategy, telehealth has been rapidly adopted during the pandemic to support access to the medical home in the context of physical distancing policies³⁵  and could be used to expand screening and support follow-up of adolescents with mental health concerns, as highlighted in the recent AAP “Interim Guidance on Supporting the Emotional and Behavioral Health Needs of Children, Adolescents, and Families During the COVID-19 Pandemic.”³  Effective telehealth interventions, however, must take into account the digital divide and the often disparate rates of uptake of these services.^36–38  In addition, the integration of social workers and co-located mental health providers into pediatric primary care can facilitate more successful mental health treatment and follow-up.³⁹  In our network, half of the practices had integrated behavioral health and only 6 had embedded social workers during the study period. However, many primary care practices nationwide do not have co-located or integrated behavioral health,⁴⁰  indicating an opportunity to better support screening and follow-up of adolescents with mental health concerns. Finally, patient-facing strategies, such as educating families to ask their pediatricians about depression screening and resources, could be implemented alongside clinician and systems-focused changes. Advocacy on the part of health systems is needed for ongoing billing reform to support these changes.

A strength of this study is the large and diverse population of adolescents seen in pediatric primary care spanning urban, suburban, and semirural geographic areas. This study was also subject to several limitations. First, the use of routinely collected electronic health record data is inherently limited to adolescents who attend primary care visits. Rates of primary care visit attendance tend to be lower among adolescents than younger children,⁴¹  and patterns may not reflect depressive symptoms or suicide risk among adolescents who do not receive primary care. Adolescents with the most severe symptoms might have bypassed primary care and sought care at the ED for acute concerns. It is also possible that disparities in rates of routine preventive visit attendance may have masked larger differences in depression and suicide risk outcomes by race and ethnicity, because non-Hispanic Black adolescents have lower attendance at preventive visits compared with non-Hispanic white adolescents in several practices in our network with high levels of concentrated poverty. In addition, the COVID-19 pandemic impacted primary care visit volumes. We attempted to account for this by comparing the June to December period after visit patterns had stabilized, but this issue may have introduced bias if adolescents with mental health concerns were differentially more (or less) likely to attend visits during the pandemic. It is also possible, given the lower screening rates during the pandemic, that clinicians were selectively screening adolescent patients they perceive to be higher risk. Despite this possibility, screening activities in the network during the pandemic were driven largely by practices’ ability to safely sanitize tablets used for screening. Also, results may not generalize to other geographic regions. Because of limitations in the electronic health record data, we were unable to examine patterns among lesbian, gay, bisexual, transgender, and queer/questioning (LGBTQ) adolescents or according to language preference. Also, given the nature of the PHQ-9-M, interpretations are restricted to depressive symptoms, rather than diagnostic rates of depression. Additional research examining changes in diagnostic rates during the COVID-19 pandemic would be beneficial. Also, we did not have access to details on follow-up and interventions that were put in place for teenagers who screened positive for depression or suicide risk, although this has been examined in other studies.^11,18,24  Studies in which researchers examine patterns of follow-up care provided after positive screens during COVID-19 will be an important next step, given high variability in follow-up after primary care depression and suicide risk screening in general.^11,18  Finally, future work should be conducted to examine longitudinal, individual-level changes in depressive symptoms and suicide risk in addition to population-level patterns.

Positive screens for depressive symptoms and suicide risk increased by a small but significant amount during the COVID-19 pandemic. Increases were most pronounced among female adolescents for both depression and suicide risk screening, with some indication of increases among non-Hispanic white and non-Hispanic Black adolescents. Given these patterns, pediatricians are encouraged to consistently screen adolescents for depression and link identified adolescents to treatment. Additionally, advocacy and effort on the part of primary care practices is needed to ensure adolescents are screened and linked to treatment and that existing gaps between the identification of depression symptoms and/or suicide risk and initiating (and sustaining) treatment are filled, including known barriers to receiving treatment faced by non-Hispanic Black adolescents.

AAP

American Academy of Pediatrics

CHOP

Children’s Hospital of Philadelphia

CI

confidence interval

COVID-19

coronavirus disease 2019

ED

emergency department

PHQ-9-M

Patient Health Questionnaire-Modified for Teens

PR

prevalence ratio