Adolescent depression screening in primary care: Who is screened and who is at risk?

Abstract

Background:

Limited research has simultaneously focused on sociodemographic differences in who receives recommended adolescent depression screening in primary care and who endorses elevated depression and suicide risk on these screeners. We describe screening and risk rates in a large pediatric primary care network in the United States after the network expanded its universal depression screening guideline to cover all well-visits (i.e., annual medical checkups) for adolescents ages 12 and older.

Methods:

Between November 15, 2017 and February 1, 2020, there were 122,682 well-visits for adolescents ages 12–17 (82,531 unique patients). The Patient Health Questionnaire – Modified for Teens (PHQ-9-M) was administered to screen for depression.

Results:

A total of 99,961 PHQ-9-Ms were administered (screening rate=81.48%). The likelihood of screening was higher among adolescents who were female, 12–14 years of age at their first well-visit during the study, White, Hispanic/Latino, or publicly-insured (i.e., Medicaid-insured). Additionally, 5.92% of adolescents scored in the threshold range for depression symptoms and 7.19% endorsed suicidality. Heightened depression and suicide risk were observed among adolescents who were female, 15–17 years of age at their first well-visit during the study, Black, Hispanic/Latino, attending urban primary care practices, or Medicaid-insured. Odds of endorsing suicidality were also higher among teens who identified as other races.

Limitations:

Limitations related to data available in the electronic health record and reliance on data from a single hospital system are noted.

Conclusions:

Findings highlight misalignments in screening and risk status that are important to address to ensure more equitable screening implementation and health outcomes.

1. Introduction

1.1. Importance of identifying adolescent depression

Depression spikes during adolescence (Hankin et al., 2015) and can have substantial long-term consequences, such as adverse psychological, physical health, educational, and social outcomes into adulthood (Copeland et al., 2015; Fergusson and Woodward, 2002; Naicker et al., 2013). Moreover, depression has been deemed the single largest contributor to global disability by the World Health Organization (2017). Thus, missed opportunities for depression identification and treatment can be quite costly from both personal and public health perspectives.

1.2. What is known: depression screening in primary care

Most adolescents visit primary care for well-visits (i.e., annual medical checkups; Klein et al., 1998) and youth and families are willing to discuss mental health concerns with primary care providers (PCPs; Cheung et al., 2013), making it an opportune setting for administering routine depression screening and thus detecting risk that may otherwise go unnoticed. As such, national guidelines have recommended universal depression screening in primary care beginning at age 12 (Siu, 2016; Zuckerbrot et al., 2018). Consistent with these recommendations, pediatric primary care depression screening programs have been implemented in a number of health systems (Costello et al., 2021; Crandal et al., 2021; Davis et al., 2021; Kenny et al., 2021; Lewandowski et al., 2016).

Approaches to implementing depression screening have differed across health systems with respect to the measures used, age and type of visit at which screening is recommended, and method of administering the screener (Costello et al., 2021; Farley et al., 2020; Kemper et al., 2021; Lewandowski et al., 2016; (Zuckerbrot et al., 2007)). We focus on studies administering versions of the Patient Health Questionnaire-9 (PHQ-9; Spitzer et al., 1999) given that the PHQ-9 is commonly used in pediatric primary care to screen for depression (American Academy of Pediatrics, 2021). Studies have found screening rates as low as 11% when PHQ-9s were administered and later uploaded in the electronic health record (EHR) and as high as 82% in a study involving the Patient Health Questionnaire-2 (PHQ-2) being embedded and self-scored in the EHR, which then triggered administration of the PHQ-9 (also embedded and self-scored in the EHR) for certain adolescents (Lewandowski et al., 2016; Sudhanthar et al., 2015). Within the primary care network included in the present study, the previous system-wide guideline involved electronic depression screening at age 16 well-visits using the Patient Health Questionnaire – Modified for Teens (PHQ-9-M); a screening rate of 76.3% was reported (Farley et al., 2020). Continued transparent reporting of screening procedures and corresponding screening and symptom rates is important to determine whether recommended screenings are being carried out equitably.

1.3. Disparities in screening and risk

It has become increasingly clear that widespread inequities exist in healthcare delivery and health outcomes; access to services for, and outcomes relevant to, depression are no exceptions (Thomas et al., 2011; Waitzfelder et al., 2018). Primary care has been highlighted as a crucial setting for identifying and intervening upon health and healthcare inequities given that “primary care providers treat and health inequities affect every organ, every system, every malady, in every family, and every community” (p. 851, Heintzman and Marino, 2021).

With regard to depression screening specifically, studies focused on adult patients have identified racial and ethnic disparities in screening administration (Akincigil and Matthews, 2017; Hahm et al., 2015). A study of adult primary care patients revealed that Black and Asian patients were less likely to be screened for depression compared to White patients; those identifying as Latino were more likely to be screened (Hahm et al., 2015). Moreover, for patients with moderate or severe depression, Black males and females, Asian males and females, and Latino males were less likely than White individuals to receive any mental health care (Hahm et al., 2015). Akincigil and Matthews (2017) found that Black adults were half as likely to be screened for depression as White adults. In the adolescent depression screening literature, differences in screening in primary care have been documented based on sociodemographic factors such as payor status, geographic region, patient sex, patient age, and patient race/ethnicity (Bhatta et al., 2018; Fallucco et al., 2015; Zenlea et al., 2014). For instance, Fallucco et al. (2015) found that female adolescents were more likely to be screened than male adolescents and older adolescents were more likely to be screened than younger adolescents. Given some inconsistencies in findings across studies (e.g., lower likelihood of screening among Hispanic adolescents in Zenlea et al., 2014, higher likelihood of screening for adults identifying as Latino in Hahm et al., 2015 and no association between patient ethnicity and screening in Fallucco et al., 2019), additional research is needed to more comprehensively summarize disparities in adolescent depression screening so that action can then be taken to reduce those disparities.

Additionally, relations between sociodemographic factors and depression and suicide risk have been documented (Goodman et al., 2003; Hankin et al., 2015; Lindsey et al., 2019). For instance, beginning in adolescence, depression is more prevalent in females than males and adolescent depression risk increases with age, particularly among females (Hankin et al., 2015). Indicators of lower socioeconomic status have been linked to adolescent depression (e.g., Goodman et al., 2003). Relatedly, certain subgroups of adolescents have demonstrated higher rates of depression and suicide risk on depression screeners (Davis et al., 2021; Farley et al., 2020; Mayne et al., 2021). Farley et al. (2020) found higher proportions of adolescents with elevated depression scores for those who were female, part of a racial/ethnic minoritized group, and attended urban primary care practices. Further documenting sociodemographic differences in screening status as well as depression and suicide risk can guide improvements in screening practices and facilitate identification of adolescents who may be at heightened risk for depression and suicide and would benefit from behavioral health interventions.

1.4. The present study

We provide results on the screening and risk (i.e., depression symptom elevations and suicidality) rates documented in a large pediatric primary care network at a critical juncture: the first few years after the healthcare system expanded its universal depression screening guideline to encompass all well-visits for adolescents ages 12 and older. The study has several aims: 1) to determine overall screening rates, 2) to examine whether there were disparities in receipt of screening based on sociodemographic factors (i.e., age, sex, race, ethnicity, insurance type, and practice location), 3) to determine depression and suicide risk rates, and 4) to examine sociodemographic differences in these risk rates. The aforementioned sociodemographic factors were selected given their relevance to healthcare services and outcomes broadly and depression and suicide specifically. Examining sociodemographic differences in receipt of screening as well as depression and suicide risk simultaneously can help determine whether adolescents most at risk for depression and suicide are being screened. Thus, the current study provides much needed data on ways to enhance adolescent depression screening efforts to promote equitable implementation and health outcomes.

2. Method

2.1. Setting and screening procedures

All study procedures were approved by the healthcare organization’s Institutional Review Board. An information analyst extracted retrospective data, including sociodemographic data and PHQ-9-M depression screening results, from the EHR of a large pediatric healthcare facility and primary care provider (31 practices at the time of data collection) in the Northeastern United States. This study included adolescents ages 12–17 who were seen for primary care well-visits between November 15, 2017 and February 1, 2020. November 2017 marked the beginning of the institutional guideline to screen for depression using the PHQ-9-M at annual well-visits for adolescents ages 12 and older. Prior to November 2017, the institutional recommendation was to screen adolescents for depression at their age 16 well-visit (Farley et al., 2020). The end date was selected to be just prior to the beginning of the COVID-19 pandemic given that declines in screening rates, but increases in depression and suicide risk for certain adolescents, have been documented in this healthcare system during the pandemic (Mayne et al., 2021). Adolescents were administered the PHQ-9-M on a tablet or at a kiosk in the waiting room during their visit and results populated in the EHR for PCPs to review. The PHQ-9-M was developed based on the PHQ-9 (Spitzer et al., 1999), Revised PHQ-9 for Adolescents (Johnson et al., 2002), and the Columbia Diagnostic Interview Schedule for Children-Depression Scale (Shaffer et al., 2000). The PHQ-9-M keeps the nine core items from the PHQ-9, with slight changes to enhance its relevance to youth depression. The PHQ-9-M assesses depression symptoms in the past two weeks. The 9 core items on the PHQ-9-M are scored from 0 (“not at all”) to 3 (“nearly every day”). The PHQ-9-M includes two supplemental items that assess suicidality (“Has there been a time in the past month when you had serious thoughts about ending your life?” and “Have you ever, in your whole life, tried to kill yourself or made a suicide attempt?”). Suicide risk was flagged in the EHR if either of the dichotomous supplemental items were marked “yes” or if item 9 (“Thoughts that you would be better off dead, or of hurting yourself”) was scored a 1 or higher (i.e., “several days” or more in the past two weeks). Total PHQ-9-M scores of 0–4 were within the normal range, 5–10 were marked as subthreshold, and 11–27 were flagged as threshold, consistent with cutpoints suggested in the literature (Richardson et al., 2010). Cronbach’s alpha for the nine core items in the present study was 0.80.

2.2. Participants

Between November 15, 2017 and February 1, 2020, there was a total of 122,682 adolescent well-visits for 82,531 unique adolescents. See Table 1 for demographic information.

Table 1.

Demographic Information for Adolescents Seen for at Least one Well-Visit, Screened, Endorsing Threshold Depression Symptoms, and Endorsing Suicidality.

Variable	N (Percent of Sample Seen for Well-Visits) N = 82,531	N (Percent of Screened Sample) N = 70,988	N (Percent of Sample Endorsing Threshold Depression Symptoms) N = 4,182	N (Percent of Sample Endorsing Suicidality) N = 5,075
Sex
Male	41,810 (50.66%)	35,780 (50.40%)	1,448 (34.62%)	1,812 (35.70%)
Female	40,720 (49.34%)	35,207 (49.60%)	2,734 (65.38%)	3,263 (64.30%)
Race
White	45,913 (55.63%)	39,892 (56.20%)	1,953 (46.70%)	2,164 (42.64%)
Black	22,497 (27.26%)	19,165 (27.00%)	1,516 (36.25%)	2,048 (40.35%)
Asian	3,372 (4.09%)	2,831 (3.99%)	139 (3.32%)	180 (3.55%)
other races	10,631 (12.88%)	9,004 (12.68%)	568 (13.58%)	676 (13.32%)
Ethnicity
Hispanic/Latino	5,436 (6.59%)	4,696 (6.62%)	384 (9.18%)	426 (8.39%)
Non-Hispanic/Latino	76,808 (93.07%)	66,046 (93.04%)	3,785 (90.51%)	4,628 (91.19%)
Age at First Well-Visit
12–14 years	38,927 (47.17%)	34,488 (48.48%)	1,925 (46.03%)	2,370 (46.70%)
15–17 years	32,353 (39.20%)	27,300 (38.46%)	1,841 (44.02%)	2,227 (43.88%)
Practice Location
Urban	23,091 (27.98%)	19,441 (27.39%)	1,554 (37.16%)	1,947 (38.36%)
Nonurban	59,440 (72.02%)	51,547 (72.61%)	2,628 (62.84%)	3,128 (61.63%)
Insurance Type
Medicaid	21,304 (25.81%)	18,371 (25.88%)	1,651 (39.48%)	2,015 (39.70%)
Private	60,639 (73.47%)	52,180 (73.51%)	2,512 (60.10%)	3,029 (59.68%)

Demographic information in this table is presented at the patient level. There were a total of 122,682 well-visits for adolescents ages 12–17 years during the study period; 70,988 unique adolescents were screened and 70,590 of those adolescents had total scores available for subsequent analyses. Percentages not adding up to 100% are due to rounding and/or missing demographic information.

2.3. Analytic plan

We present the PHQ-9-M screening rate both as 1) the total number of PHQ-9-M screening administrations compared to the total number of well-visits and 2) the number of unique patients screened compared to the number of unique patients seen for well-visits. When adolescents were screened multiple times using the PHQ-9-M, their first PHQ-9-M administration was used in analyses involving depression scores and suicide risk.

Depression symptom risk (i.e., threshold symptom levels) and suicide risk rates based on endorsing item 9 or either of the two supplemental suicide items are presented as the number of adolescents in each risk category compared to the total number of adolescents with completed PHQ-9-Ms.

To determine if there were sociodemographic differences between those who were screened vs. not screened with the PHQ-9-M among adolescents with well-visits during the study period, we ran a binary logistic regression model. Screening status (i.e., screened vs. not screened) was the dependent variable and sex (males compared to females), age at first well-visit (12–14 years of age or 15–17 years of age), race (Black, Asian, White, or other races), ethnicity (Hispanic/Latino or Non-Hispanic/Latino), practice location (urban, meaning a large central metro area, or nonurban, meaning sites located outside of the larger metro area; Ingram & Franco, 2013), and insurance type (Medicaid, meaning public insurance, compared to private insurance, meaning insurance provided through a private entity) were entered simultaneously as predictors. Similarly, we examined sociodemographic characteristics as predictors of depression symptom and suicide risk levels using separate binary logistic regression models. In one model, the dependent variable was depression risk category. Adolescents with threshold depression symptoms were considered high risk and were compared to adolescents with normal and subthreshold symptoms. Endorsement of any suicidality across item 9 and the two supplemental items was entered as the dependent variable in the other logistic regression model; adolescents endorsing no suicidality were compared to those endorsing at least one of the three risk items. Odds ratios were calculated as a measure of effect size along with 95% confidence intervals.

3. Results

3.1. Overall screening rates

Between November 15, 2017 and February 1, 2020, there was a total of 122,682 adolescent well-visits and a total of 99,961 PHQ-9-M administrations (a screening rate of 81.48%; see Fig. 1). Of the 82,531 unique adolescents who had at least one well-visit during the study period, 70,988 of these patients were screened at least once (a screening rate of 86.01%). Of the 70,988 unique patients who were screened during the study period, 70,590 (99.44%) had completed PHQ-9-Ms (i. e., available total scores) that were included in the analyses involving depression and suicide risk. Many of the 70,988 adolescents (n = 27,565; 38.83%) were screened more than once during this period, in line with the recommendation for annual depression screening during well-visit appointments. The average number of PHQ-9-Ms administered per adolescent during this time was 1.51 (SD = 0.59), similar to the average number of well-visits 1.49 (SD = 0.54).

Fig. 1.

Study Flow for Adolescents Included in the Current Analyses.

3.2. Associations between sociodemographic factors and screening status

Demographics for adolescents screened are presented in Table 1 and logistic regression results are detailed in Table 2. The odds of being screened with the PHQ-9-M during the study period were higher for females than males OR = 1.09, 95% CI [1.04, 1.14], p < .001. Adolescents who were 15–17 years of age at the first well-visit during the study period were less likely to be screened during the study window compared to adolescents who were 12–14 years of age at the first well-visit, OR = 0.69, 95% CI [.66, 0.72], p < .001. Compared to White adolescents, adolescents who identified as Black, OR = 0.91, 95% CI [.86, 0.98], p < .01, Asian, OR = 0.77, 95% CI [.69, 0.85], p < .001, or other races (e.g., American Indian or Alaska Native, multiple races), OR = 0.80, 95% CI [.74, 0.86], p < .001 had lower odds of being screened. Adolescents who identified as Hispanic/Latino had higher odds of being screened compared to adolescents who were Non-Hispanic/Latino, OR = 1.13, 95% CI [1.03, 1.25], p < .05. Adolescents with private insurance were less likely to be screened than those with Medicaid, OR = 0.92, 95% CI [.87, 0.98], p < .01. Practice location (i.e., urban or nonurban) did not predict screening status.

Table 2.

Logistic Regression Analyses of Sociodemographic Factors Associated with Screening Status, Depression Risk, and Suicidality.

	Screening Status		Depression Symptoms		Suicidality
	OR	CI	OR	CI	OR	CI
Demographics
Sex (male = ref)	1.09***	[1.04, 1.14]	2.07***	[1.93, 2.23]	1.91***	[1.79, 2.03]
Age (12–14 years = ref)	.69***	[.66, 0.72]	1.24***	[1.16, 1.32]	1.22***	[1.14, 1.29]
Black (White = ref)	.91**	[.86, 0.98]	1.13*	[1.02, 1.24]	1.53***	[1.40, 1.67]
Asian (White = ref)	.77***	[.69, 0.85]	.92	[.76, 1.11]	1.11	[.94, 1.32]
other races (White = ref)	.80***	[.74, 0.86]	1.04	[.93, 1.17]	1.16**	[1.05, 1.29]
Hispanic (no = ref)	1.13*	[1.03, 1.25]	1.32***	[1.16, 1.51]	1.29***	[1.14, 1.45]
Practice Location (urban = ref)	1.05	[.99, 1.11]	.79***	[.73, 0.87]	.87**	[.81, 0.95]
Insurance (Medicaid=ref)	.92**	[.87, 0.98]	.59***	[.54, 0.64]	.62***	[.58, 0.67]
	χ2(8) = 355.84***		χ2(8) = 921.68***		χ2(8) = 1120.41***
	n = 70,443		n = 60,780		n = 60,780

^*p < .05.;

^**p <0.01.;

^***p < .001.

OR = odds ratio. CI = 95% confidence interval. Ref = reference group. n = sample size for each analysis.

3.3. Depression and suicide risk rates

Of the 70,988 unique patients who were screened during the study period, 70,590 had completed PHQ-9-Ms that were able to be analyzed for depression symptom and suicide risk elevations. Among those 70,590 adolescents, 5.92% (n = 4,182) were in the threshold range for depression symptoms. For suicide risk, 7.19% (n = 5,075) of adolescents endorsed suicidality on item 9 and/or one of the two supplemental suicide items; 4.61% (n = 3,254) endorsed a score of 1 or higher on item 9, 2.43% (n = 1,715) endorsed serious suicidal ideation in the past month, and 3.16% (n = 2,234) endorsed a lifetime suicide attempt.

3.4. Associations between sociodemographic factors and depression and suicide risk

Demographics for adolescents scoring in the threshold range for depression symptoms as well as those endorsing any suicidality are presented in Table 1. Logistic regression results are detailed in Table 2. In the model predicting depression risk status, females had higher odds of elevated depression symptoms (i.e., scores of 11–27) compared to males, OR = 2.07, 95% CI [1.93, 2.22], p < .001. Adolescents who were 15–17 years of age at their first well-visit in the study window were more likely to have elevated depression symptoms compared to adolescents who were 12–14 years of age at their first well-visit, OR = 1.24, 95% CI [1.16, 1.32], p < .001. Odds of depression symptom elevations were higher for Black adolescents compared to White adolescents, OR = 1.13, 95% CI [1.02, 1.24], p < .05. The difference in depression risk status between Asian and White adolescents was not significant, OR = 0.92, 95% CI [.76, 1.11], p = .40. Similarly, the difference in depression risk status between adolescents identifying as other races (e.g., American Indian or Alaska Native, multiple races) compared to White adolescents was nonsignificant, OR = 1.04, 95% CI [.93, 1.17], p = .47. Hispanic/Latino adolescents had higher odds of elevated depression symptoms than Non-Hispanic/Latino adolescents, OR = 1.32, 95% CI [1.16, 1.51], p < .001. Odds of depression symptom elevations were higher among adolescents seen in urban primary care practices than those seen in nonurban practices, OR = 0.79, 95% CI [.73, 0.87], p < .001, and among adolescents with Medicaid compared to those who had private insurance, OR = 0.59, 95% CI [.54, 0.64], p < .001.

Results for the logistic regression model with suicidality as the dependent variable followed a similar pattern. Females were more likely to endorse suicidality than males, OR = 1.91, 95% CI [1.79, 0.2.03], p < .001. Adolescents ages 15–17 at the time of their first well-visit in the study period had higher odds of suicidality than adolescents who were 12–14 years of age at the fist well-visit, OR = 1.22, 95% CI [1.14, 1.29], p < .001. Black adolescents, OR = 1.53, 95% CI [1.40, 1.67], p < .001, as well as adolescents identifying as other races, OR = 1.16, 95% CI [1.05, 1.29], p < .01, were more likely to endorse suicidality than White adolescents. There was no significant difference in the likelihood of endorsing suicidality between Asian and White adolescents, OR = 1.11, 95% CI [.94, 1.32], p = .22. Adolescents who were Hispanic/Latino had higher odds of endorsing suicidality than adolescents who were Non-Hispanic/Latino, OR = 1.29, 95% CI [1.14, 1.45], p < .001. Odds of suicidality were greater among adolescents seen at urban practices, OR = 0.87, 95% CI [.81, 0.95], p < .01, and those with Medicaid insurance, OR = 0.62, 95% CI [.58, 0.67], p < .001, than adolescents seen in nonurban practices and those with private insurance.

4. Discussion

We examined screening, depression symptom, and suicide risk rates, and corresponding sociodemographic differences in each, among adolescents seen for well-visits in pediatric primary care. Results demonstrate the sustainment of a high overall screening rate and highlight areas for improvement to ensure equitable screening implementation and health outcomes.

The depression screening rate observed in the present study, whether at the encounter- or patient-level (81.48% and 86.01%, respectively), was high, suggesting strong fidelity to the health system’s screening guideline. In addition to being greater than, or commensurate with, screening rates reported in the literature (e.g., 11%, 82%; Lewandowski et al., 2016; Sudhanthar et al., 2015), it is slightly above the rate previously reported in the same healthcare system (76.3%; n = 7016 adolescents) when depression screening was recommended for age 16 well-visits only (Farley et al., 2020). Thus, the current study, which included 70,988 unique patients screened, displays the considerable increase in reach of adolescent depression screening upon the expansion in ages covered in the healthcare system’s guideline. Additionally, demonstrating that the health system’s high screening rate was maintained over time is crucial as sustainability of health programs is often overlooked in the implementation process (Proctor et al., 2015; Stirman et al., 2012). While feedback on specific factors that facilitated screening implementation was not collected, it is likely that the existing infrastructure (e.g., tablet-administered screening and EHR-integrated results, clinician education) helped make the transition to age 12 and older screening successful. Additional research, particularly involving qualitative data collection, will be important to not only further understand factors that facilitate universal depression screening but also to identify barriers that need to be addressed to ensure screening takes place at all adolescent well-visits.

The move to universal screening and maintenance of a high screening rate within the healthcare system included in this study provided a novel and important opportunity to examine sociodemographic differences in screening. While overall screening rates were high at the patient and encounter-level, screening rates varied based on a number of sociodemographic factors. Specifically, likelihood of screening was higher among adolescents who were female, between 12 and 14 years of age at their first well-visit in the study window, White, Hispanic/Latino, or Medicaid-insured. Although some of our findings are consistent with prior literature on depression screening among adolescents and adults, including the greater likelihood of screening among females and lower likelihood of screening among adolescents from certain racial/ethnic minoritized backgrounds (Fallucco et al., 2015; Hahm et al., 2015), other findings were not, such as the higher likelihood of screening for adolescents who were younger at their first well-visit. Consistencies in findings between the current study and the extant literature likely point to larger disparities that need to be addressed. The difference in screening based on payor status may be a result of Medicaid’s mandate for depression screening and willingness to pay for this screening. Thus, this finding points to the need for private insurance to also cover screening to ensure all adolescents are screened routinely for depression. While universal screening has the potential to promote equity, we cannot assume that guidelines advocating for universal screening alone eliminate disparities (Roberts and Nuru-Jeter, 2012). From a health equity perspective, it is critical to gain a detailed understanding of barriers and facilitators to equitable implementation (e.g., racism, structural discrimination) to identify how best to intervene to optimize uptake and sustainment of health services (Shelton et al., 2020; Woodward et al., 2021).

It is challenging to compare depression and suicide risk rates detected in the current study to prior adolescent depression screening studies given differences in methodologies (e.g., screening measures administered, cutoffs used for the same measure, populations sampled from, screening reach). However, the risk rates observed in the current study are lower than some other studies (Costello et al., 2021; Lewandowski et al., 2016; Kemper et al., 2021). It makes sense that studies conducted with adolescents who are more likely to experience elevated depression and suicide risk, such as patient populations who are predominantly Medicaid-insured and identify as being from a racial or ethnic minoritized group (Kemper et al., 2021), would observe higher rates of depression and suicide risk than this study. In comparison to data from the healthcare system included in the current study when the age 16 screening guideline was in effect, depression and suicide risk rates are slightly lower. This is likely due to the inclusion of younger adolescents, as evidenced by the present results demonstrating adolescents ages 12–14 years at the time of their first well-visit in the study window were at lower risk for depression and suicide than those 15–17 years of age.

Similar to screening rates, we identified sociodemographic differences in depression and suicide risk rates. These sociodemographic differences in depression and suicide risk largely map onto the broader literature (Hankin et al., 2015; Lindsey et al., 2019; Miller and Taylor, 2012; Nock et al., 2013; Silva and Van Orden, 2018). Additionally, the findings are relatively consistent across depression and suicide risk. These results are particularly informative when interpreted in conjunction with the data on screening status, especially when the two were misaligned. For example, the fact that the odds of being screened were lower for Black adolescents and adolescents who were 15–17 years of age at their first well-visit but these adolescents were more likely to endorse depression and suicide risk highlights critical opportunities for risk identification that may be missed when routine depression screening is not conducted. Identifying and working to close gaps in screening, particularly for higher-risk groups, is especially important at this stage so that any additional scale up efforts can serve to decrease rather than augment disparities.

Heightened depression and suicide risk among adolescents who were female, 15–17 years of age at their first well-visit in the study window, Black, Hispanic/Latino, attending urban primary care practices, and Medicaid-insured, as well as elevated suicide risk for adolescents identifying as other races, points to the need to ensure that culturally-relevant interventions addressing suicide and depression risk are especially accessible to these youth. Depression screening in the absence of linkages to high-quality care is likely counterproductive (Mitchell et al., 2016). A recent randomized clinical trial indicated that universal school-based depression screening facilitated treatment initiation (Sekhar et al., 2021), suggesting implementing universal depression screening may be one way to promote linkages to care. Integrated behavioral health services in primary care offices also offer an important avenue for linking screened youth to care and such services have the potential to reduce health disparities (Njoroge et al., 2016). Additionally, single-session interventions meant to bridge the gap between primary care depression screening and longer-term behavioral health services may be a promising approach for not only reducing short-term depressive symptoms but also increasing treatment-seeking behaviors (Schleider et al., 2020).

5. Limitations

The present study has the potential to make important contributions to the literature, but limitations are acknowledged. There are inherent limits to data that are available, and therefore extractable, from the EHR. For instance, based on the data available we were unable to conduct more fine-grained analyses (e.g., comparisons involving gender identity in addition to biological sex). Relatedly, based on the nature of the EHR data, we lacked information on why screening did not occur during certain visits. It would be beneficial to unpack the reasons for missed screening, and triangulating other sources of data, such as qualitative interviews, with EHR results will be important for doing so. Additionally, chart-stimulated recall, in which clinicians reviews patients’ charts to prompt recall of specific clinical encounters during brief, structured interviews (Goulet et al., 2007; Jennett and Affleck, 1998), may elicit additional information on clinical decision-making. We relied on adolescents’ initial PHQ-9-M scores in the study window for analyses involving sociodemographic predictors of depression and suicide risk; examining longitudinal trajectories of risk and predictors of those trajectories in our future work will be important.

We acknowledge that some of the effects sizes for sociodemographic differences in screening and risk rates were small and may not have been statistically significant with a smaller sample size. However, many of the findings were consistent with the broader literature and we argue that small effects can have large implications for disparities when considered on a hospital system-wide or even a national scale. As universal depression screening in primary care continues to become more common, it is important to identify and address these disparities now to reduce inequities. Furthermore, identifying modifiable factors (e.g., clinician attitudes, availability of interpreter services and screeners in different languages, community resources) that may be contributing to the current findings will be important for promoting equity. While the sample size for the current study was large and represented over 30 primary care practices, we focused on a single hospital system. Thus, our findings may not generalize to other systems. Additionally, while we were able to compare screening status as well as depression and suicide risk among adolescents from a number of racial and ethnic backgrounds, the percentages of Asian and Hispanic adolescents in the current sample were relatively small, as were the percentages of individuals identifying as part of each racial group referred to in analyses as other races. Future research with greater representation from these racial and ethnic groups will be important to amplify the experiences of youth from historically marginalized backgrounds and help us to further assess and refine universal depression screening practices.

6. Conclusion

Taken together, results from the current study support the possibility of sustaining high screening rates over time following expansion of universal adolescent depression screening guidelines in a large pediatric primary care network. Differences in screening and risk status by sociodemographic factors demonstrate that even in a healthcare system where the vast majority of adolescents are receiving routine, preventative screening, more can be done to maximize equity and to ensure that the most at-risk adolescents are being screened and linked to care. It is imperative that we partner with patients, caregivers, clinicians, and healthcare system leaders to ensure we are providing screening and follow-up services that meet the needs and preferences of the diverse adolescents served in primary care.

Abbreviations:

PCPs

primary care providers

PHQ-9

patient health questionnaire-9

PHQ-9-M

patient health questionnaire – modified for teens

EHR

electronic health record