Usual Care for Adolescent Depression From Symptom Identification Through Treatment Initiation

Importance  Published guidelines describing effective adolescent depression care in primary care settings include screening, assessment, treatment initiation, and symptom monitoring. It is unclear the extent to which these steps are documented in patient health records.

Objective  To determine rates of appropriate follow-up care for adolescents with newly identified depression symptoms in 3 large health systems.

Design, Setting, and Participants  In this analysis conducted from March to September 2014, structured data retrospectively extracted from electronic health records were analyzed for 3 months following initial symptom identification to determine whether the patient was followed up and, if so, whether treatment was initiated and/or symptoms were monitored. Records were collected from 2 large health maintenance organizations in the western United States and a network of community health centers in the Northeast. The study group included adolescents (N = 4612) with newly identified depression symptoms, defined as an elevated score on the Patient Health Questionnaire (≥10) and/or a diagnosis of depression.

Main Outcomes and Measures  Rates of treatment initiation, symptom monitoring, and follow-up care documented within 3 months of initial symptom identification.

Results  Among the 4612 participants, the mean (SD) age at index event was 16.0 (2.3) years, and 3060 were female (66%). Treatment was initiated for nearly two-thirds of adolescents (79% of those with a diagnosis of major depression; n = 1023); most received psychotherapy alone or in combination with medications. However, in the 3 months following identification, 36% of adolescents received no treatment (n = 1678), 68% did not have a follow-up symptom assessment (n = 3136), and 19% did not receive any follow-up care (n = 854). Further, 40% of adolescents prescribed antidepressant medication did not have any documentation of follow-up care for 3 months (n = 356). Younger age (ages 15-17 years: odds ratio [OR], 0.78; 95% CI, 0.67-0.92 and ages 18-20 years: OR, 0.83; 95% CI, 0.70-0.99; P = .008), more severe initial symptoms (moderate: OR, 0.99; 95% CI, 0.82-1.21; moderate to severe: OR, 1.46; 95% CI, 1.19-1.80; and severe: OR, 2.14; 95% CI, 1.65-2.79; P < .001), and receiving a diagnosis (major depression/dysthymia: OR, 2.65; 95% CI, 2.20-3.20 and unspecified depression/adjustment disorder: OR, 1.75; 95% CI, 1.43-2.14; P < .001) were significantly associated with treatment initiation. Differences in rates of follow-up care were evident between sites (site 2: OR, 1.77; 95% CI, 1.45-2.16 and site 3: OR, 2.10; 95% CI, 1.72-2.57), suggesting that differences within health systems may also affect care received.

Conclusions and Relevance  Most adolescents with newly identified depression symptoms received some treatment, usually including psychotherapy, within the first 3 months after identification. However, follow-up care was low and substantial variation existed between sites. These results raise concerns about the quality of care for adolescent depression.

Major depression is a chronic, disabling condition affecting 12% of adolescents,¹ with as many as 26% of youth experiencing at least mild depressive symptoms.² Depression impairs social and academic functioning and is associated with poor long-term outcomes.^3-5 As many as 8% of youth who develop depression during adolescence complete suicide by young adulthood,³ and medical costs associated with adolescent depression are higher than those for almost any other mental health condition.^3,6

Timely initiation of effective treatment is crucial because failure to achieve remission is associated with higher likelihood of relapse, developing recurrent depression, and more impaired long-term functioning.^7,8 Early intervention for symptomatic youth may also prevent development of a depressive episode.⁹ Adolescents diagnosed as having depression who enter treatment at a younger age, earlier in the course of a depressive episode, and prior to developing chronic depression have better outcomes following acute treatment.¹⁰ However, although effective treatments exist, up to 80% of adolescents with depression do not receive appropriate care.¹¹

This gap in care and the emergence of child mental health care as a national priority¹² spurred efforts to improve care. Primary care settings have been targeted to increase early identification and access to effective care.^13,14 Depression care toolkits and practice guidelines have been disseminated to support primary care physicians (PCPs) in evidence-based identification and treatment of adolescent depression.^15-17 Previous legislation (ie, 2009 Child Health Insurance Program Reauthorization Act and 2010 Patient Protection and Affordable Care Act) emphasized adherence to effective care standards and accountability for the quality of care through the development of health care quality measures/indicators. In an effort to identify evidence-based targets for quality measures, synthesize available PCP guidelines, and inform a research agenda, Lewandowski and colleagues¹⁸ developed a care pathway describing “essential practices for adolescent depression management from screening to symptom remission.”

Federal initiatives incentivizing universal use of electronic health records (EHRs), including the 2009 American Recovery and Reinvestment Act and Health Information Technology for Economic and Clinical Health Act, increased the number of opportunities to conduct comprehensive, timely evaluations of care. Incentives for meaningful use of EHRs target enhanced care coordination and reduced medical errors and provide a platform from which to conduct large-scale routine measurement of care quality. However, most of these initiatives do not include behavioral health indices.¹⁹

Little empirical research has evaluated routine care for adolescent depression. What is known suggests the need for vast improvements. Despite recommendations from several national bodies,^20-22 screening for adolescent depression in primary care is rare.²³ In pediatric primary care settings with policies specifically targeting universal screening for adolescent depression, 36% of youths exhibit significant symptoms.²³ However, only 17% to 27% of adolescents with depression receive treatment in usual care settings.^24,25 To our knowledge, no studies to date have evaluated the documented course of care for adolescents from symptom identification through treatment initiation.

The current study aimed to add to this emerging literature by evaluating the initial course of care for adolescents identified with depressive symptoms in primary care settings as it is documented in the EHR. This study examined routine care in several large health care systems to assess whether adolescents newly identified as having depression symptoms received appropriate follow-up care in the 3 months following identification. Elements of appropriate follow-up care were based on the depression care management pathway¹⁸ and included initiating antidepressant or psychotherapy treatment, having at least 1 follow-up visit, and symptom monitoring with a well-validated questionnaire.

Box Section Ref ID

Data were abstracted from EHRs of 3 large health care systems, with analysis conducted from March to September 2014. Organizations were recruited through colleagues and prior collaborations and were identified for study inclusion if they used any version of the Patient Health Questionnaire–9 (PHQ-9),²⁶ reported serving at least 500 adolescents with depression diagnoses in the previous calendar year, offered behavioral health services, and used a consistent EHR across practice settings to facilitate data collection (ie, data collected in structured fields could be queried by analysts at each site, deidentified, and transmitted to the authors). The first 3 organizations contacted that met these criteria agreed to participate. Two large health maintenance organizations in the western United States and a network of community health centers in the Northeast (identities masked per agreements) participated and were compensated for their participation. This study was approved by the Chesapeake Institutional Review Board and the institutional review boards at each site. This research relied on retrospective reviews of existing adolescents’ health records. Patient consent was not obtained as the Chesapeake Institutional Review Board and institutional review boards at the participating sites determined that this research was exempt under 45 CFR 46.101(b)(4) and granted a waiver of Health Insurance Portability and Accountability Act authorization for the collection of protected health information in accordance with 45 CFR 164.512(i)(2).

Retrospective data collection identified patients eligible for study inclusion during an intake period from January 1, 2012, to June 30, 2013. Eligible participants were adolescents aged 12 to 20 years on January 1, 2012, with at least 1 face-to-face PCP visit during the intake period and documented symptoms of depression. At the health maintenance organizations, participants were limited to those continuously enrolled during the study period.

The index event was the first evidence of newly identified symptoms of depression, defined by elevated PHQ score (≥10), new depression diagnosis, or both within the same 30-day period. To determine both history and follow-up care, data were collected for 6 months before and 3 months after the index event. Patients with evidence of bipolar, psychotic, autism spectrum, or personality disorders at any time were excluded. Because a primary aim of this study was to evaluate care for newly identified depression symptoms, patients were also excluded if there was evidence 6 months prior to the index event of depression diagnosis, antidepressants prescribed, or previous positive PHQ score. Data collected included (1) dates of administration and scores for all depression symptom questionnaires; (2) dates, Current Procedural Terminology codes, and encounter diagnoses for all face-to-face primary care or mental health visits/encounters; (3) dates and active psychopharmocologic compounds for all antidepressants prescribed or recorded on a medications list; and (4) demographic information.

The PHQ-9 is a 9-item self-report questionnaire assessing depression symptoms and severity that has been validated for use with adolescents.^26,27 Very slight modifications have been made for adolescents (eg, replacing an item about “work” to “schoolwork”), resulting in the PHQ-9 Modified for Adolescents and the PHQ-Modified for teens. All sites used at least 1 of these versions with their adolescent population, collectively referred to hereafter as PHQ. The PHQ items are based on the Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition) criteria for major depression and are rated on a 4-point scale ranging from 0 (“not at all”) to 3 (“nearly every day”). A score of 10 or higher indicates clinically significant symptoms (“moderate depression”) and is consistent with clinical cutoffs reportedly used by PCPs within each site.

Depression diagnoses in the current study included International Classification of Diseases, Ninth Revision²⁸ codes that indicate clinically significant symptoms of depression: major depressive affective disorder, other and unspecified depressive disorders, dysthymia, and adjustment disorders that reflect significant depressive symptoms (ie, with depressed mood; with mixed anxiety and depressed mood; with mixed disturbance of emotions and conduct; and prolonged depressive reaction).

Demographic information was collected from EHR data fields. Variables included age, sex, race, ethnicity, insurance status at index event, presence of chronic medical conditions (eg, asthma, obesity, or other physical health conditions) and co-occurring behavioral health diagnoses.

Antidepressant medication treatment was determined by an indication of an antidepressant on the patient’s medication list or in a specific field for prescription information within the EHR.

Any visit including a Current Procedural Terminology code for psychotherapy was identified.

Standard descriptive statistics were used to summarize demographic and clinical characteristics overall and across sites. Rates of treatment initiation within 3 months were compared across levels of demographic and clinical characteristics, with statistical hypothesis tests conducted by χ² tests. Logistic regression models were used to compare the odds of any treatment initiation within 3 months across levels of demographic and clinical characteristics and across sites. Univariate and bivariate analyses were conducted to describe the sample and compare differences across organizations and patient characteristics. Logistic regression analyses were performed to assess the impact of demographic and clinical characteristics on treatment initiation. Demographic (age, sex, race, and ethnicity) and clinical (index event, symptom severity, and diagnosis) characteristics were included in the regression models if they significantly contributed to the model (ie, P < .05). Data were analyzed using IBM SPSS Statistics version 21.²⁹

A total of 4612 adolescents (67% female) were identified with depression symptoms. Significant differences across sites were evident for all baseline variables except patient sex (Table 1). Most adolescents were white (63%) and not identified as Hispanic/Latino (74%), although race/ethnicity data were frequently missing (21%). Sites 2 and 3 primarily served patients with commercial insurance/self-pay (95% [n = 1669] and 99% [n = 1989], respectively), while most patients at site 1 had Medicaid/Children’s Health Insurance Program insurance (72%; n = 617). Nearly two-thirds of patients had comorbid behavioral health diagnoses, and chronic physical health conditions were also common, at 41% (n = 347) at site 1 and 70% (n = 1232) at site 2 (not reported at site 3).

Most patients were identified at the index event through a positive PHQ score alone (site 1 = 57% [n = 487] and site 3 = 59% [n = 1187]), or a new depression diagnosis alone (site 2 = 51% [n = 889]). Diagnoses at the index event were most often major depression (47%; n = 1295) or other/unspecified depression (24%; n = 654). The average PHQ score at the index event indicated moderately severe symptoms (mean [SD], 15.3 [4.1]), although this was significantly lower at site 1 (mean [SD], 14.0 [3.6]), consistent with moderate symptoms. Adolescents identified with a positive PHQ score only had a significantly lower score (mean [SD], 15.5 [3.9]) than those with both a positive PHQ score and a new diagnosis (mean [SD], 16.4 [4.2]). Among adolescents with a positive PHQ score only, 12% (n = 227) received a depression diagnosis during the 3-month follow-up period, and an additional 20% (n = 378) received another behavioral health diagnosis (eg, anxiety disorders).

Regardless of other follow-up care, documentation of symptom monitoring (ie, PHQ administered) in the 3-month follow-up period was present for only 32% of adolescents (n = 1486). Significant differences between sites indicated that patients at site 1 were less likely to have a PHQ score recorded (6%) than sites 2 (31%) or 3 (44%) (P < .001).

Sixty-four percent of adolescents initiated treatment within the follow-up period; 19% (n = 891) received antidepressants only, 29% (n = 1315) received psychotherapy only, and 16% (n = 728) received combined treatment (Table 2). Older adolescents were significantly more likely to receive treatment and were more than 4 times more likely to receive only antidepressants. Rates and types of treatment were similar for males and females. Adolescents with a positive PHQ score only at the index event were least likely to receive any treatment (54%). Higher initial PHQ scores (ie, more severe symptoms) and diagnoses of major depression were associated with significantly higher rates of receiving treatment.

Treatment initiation was significantly associated with site and several patient characteristics (Table 3; R² = 0.13). Preliminary analyses indicated that patient sex was not significantly associated with treatment initiation; therefore, it was excluded from the model. Compared with site 1, sites 2 and 3 were twice as likely to initiate treatment. Patient race/ethnicity (white) and younger age (ages 15-17 years: odds ratio [OR], 0.78; 95% CI, 0.67-0.92 and ages 18-20 years: OR, 0.83; 95% CI, 0.70-0.99; P < .001) were significantly associated with treatment initiation. Adolescents who did not receive a PHQ at the index event and those with scores in the moderate range were equally likely to initiate treatment, while those with more severe scores initiated treatment 1.5 to 2.2 times more often (moderate: OR, 0.99; 95% CI, 0.82-1.21; moderate to severe: OR, 1.46; 95% CI, 1.19-1.80; and severe: OR, 2.14; 95% CI, 1.65-2.79; P < .001). Receiving a diagnosis was significantly associated with treatment initiation diagnosis (major depression/dysthymia: OR, 2.65; 95% CI, 2.20-3.20 and unspecified depression/adjustment disorder: OR, 1.75; 95% CI, 1.43-2.14; P < .001).

As shown in the Figure, 19% of adolescents with newly identified depression symptoms did not receive any follow-up visit. Treatment was initiated for 78% of adolescents (n = 2934) who did receive a follow-up visit. Among adolescents prescribed antidepressants only (ie, did not receive therapy), 40% did not have another visit recorded in the 3-month follow-up period.

Rates of appropriate follow-up care for adolescents with newly identified depression symptoms in 3 large health systems were examined via structured data retrospectively extracted from EHRs. Treatment was initiated for nearly two-thirds of adolescents (n = 2934), most of whom received psychotherapy. Nearly 80% of youth with diagnoses of major depression initiated treatment. Regardless of treatment initiation, two-thirds of adolescents did not have further symptom monitoring with a PHQ. Nineteen percent of adolescents (n = 854) identified with clinically significant depression symptoms, and 40% of adolescents (n = 356) prescribed antidepressants had no documentation of follow-up care. Differences in rates of follow-up care were evident across sites, suggesting differences within health systems may affect care.

Appropriately identifying symptoms and diagnoses are initial steps in the adolescent depression care pathway¹⁸ and are essential to providing adequate follow-up care and treatment.³⁰ Essential to assessment is measuring symptom severity³¹ and 70% of adolescents received a symptom-based questionnaire (PHQ) at the index event. Symptom assessment is critical to guiding treatment decisions for depressed adolescents because initial PHQ scores have been associated with symptom severity up to 6 months later.³²

The presence of both an elevated PHQ score and a new diagnosis at the index event is arguably representative of higher quality of care than either alone, as it suggests assessment to confirm a diagnosis. A modest percentage of adolescents fell into this category (13%-37%). Variability across sites was evident for diagnoses conferred; site 1 reported higher rates of dysthymia and other/unspecified depression than expected.⁵ This may reflect PCPs’ lack of diagnostic confidence or use of these diagnoses “in lieu of,” perhaps owing to concerns of stigma associated with major depression. Importantly, neither a PHQ score nor a recorded diagnosis confirm that an adequate clinical assessment was conducted. It may be the case that some PCPs inadvisably recorded a diagnosis based solely on the positive screen, which has implications for assumptions of appropriate treatment and follow-up care.

Regardless of diagnosis, 19% of adolescents identified with clinically significant depression symptoms received no care for the following 3 months. Follow-up contact and further assessment following a positive screen are recommended even for mild symptoms by current guidelines and best practices.¹⁸ These findings raise concerns that many adolescents with depression receive an unacceptable level of care, particularly striking because more than half of adolescent suicide completers have chronic, unremitted depression.^33,34

Also concerning was the lack of follow-up care after prescribing an antidepressant. Current black box warnings highlight the risk for increased suicidality for youth prescribed antidepressants and recommend patients are “monitored appropriately and observed closely…especially during the initial few months.”³⁵ Yet, nearly half of adolescents prescribed an antidepressant did not have a visit in the 3 months following prescription.

Treatment early in a depressive episode, particularly for younger adolescents, may improve outcomes,^9,10,36 so finding that treatment was initiated for most of the adolescents was encouraging. Although youth with more severe symptoms based on PHQ score were more likely to receive treatment, 22% of adolescents endorsing severe symptoms remained untreated. Younger adolescents were more likely to receive psychotherapy and much less likely to receive medication alone. This is consistent with current recommendations, which encourage supportive counseling, monitoring, and/or psychotherapy as first-line treatment, particularly for younger adolescents or youth with mild symptoms.¹⁸

Half of adolescents with only a positive PHQ score at the index event initiated treatment, 29% of whom received antidepressants. A small percentage of adolescents identified with a positive PHQ score only received a depression or other behavioral health diagnosis later in the follow-up period, possibly reflecting initial observation or further assessment. The remaining youths treated without a diagnosis may reflect PCPs’ reluctance to diagnose, although they recognized the need to provide care.

Regardless of treatment initiation, symptom monitoring is recommended.^21,22 A study of a collaborative care model, including regular symptom monitoring, found significantly better depression outcomes for adolescents.²⁴ In the current study, rates of symptom monitoring were modest (32%). Significant variability across sites may represent the impact of organizational policy on care. Site 3 reported policies requiring PHQs be administered for antidepressant refills, possibly driving the noticeably higher rate of symptom monitoring at the site (44%). In contrast, site 1, where rates of symptom monitoring were 6%, reported policies encouraging universal screening but not symptom monitoring.

The primary limitation of this and other studies relying on medical record (EHR) data are that conclusions depend on how information is gathered and recorded. Distinctions between lack of appropriate follow-up care or failure to document care events cannot be made. Alternative appropriate PCP behaviors (eg, specialist referrals and telephone follow-up contact) or relevant patient behaviors (treatment dropout/refusal, electing to receive care elsewhere, change in insurance or location, and barriers to access including transportation and insurance coverage) were not evaluated in the current study because such information was rarely documented in structured data fields. Future studies should seek to accurately assess the extent to which PCPs conduct accurate clinical assessments that inform appropriate follow-up care.

These challenges to documentation are common to EHR-based research evaluations of mental health care quality.³⁷ However, similar documentation is required for quality measurement, quality-improvement initiatives, and continuity of care. Thus, limitations of data availability are not only problematic for research, but also reflect limitations in what members of the clinical team are able to learn about the patient. Therefore, limitations of EHR documentation are themselves likely contributors to poor quality of care.

Finally, it is unclear the extent to which findings from the current study are generalizable beyond the settings in which data were collected. Differences in outcomes and potential health care disparities between site 1 (serving predominantly racial/ethnic minority Medicaid recipients) and sites 2 and 3 (serving primarily white, privately insured youths) were not explored in the current study; and, in fact, high rates of missing data (eg, race/ethnicity) hinder further analyses. Moreover, the participating sites are highly regarded health care institutions. They are often looked to as leaders in cutting-edge care that routinely use quality-improvement initiatives focused on adolescent behavioral health care. Thus, results from the current study, discouraging as they are, may overstate the quality of care in other settings.

Current standards of care recommend that adolescents identified with depression symptoms receive further assessment, initiate antidepressant medication and/or psychotherapy treatment, and are monitored for changes in symptoms, especially following an antidepressant prescription. Evidence from this study suggests that quality of care in routine practice diverges from these standards. Given the negative outcomes associated with untreated adolescent depression,^3-5 greater attention to improving adherence to quality standards is warranted.

Corresponding Author: Briannon C. O’Connor, PhD, Coordinated Care Services Inc, 1099 Jay St, Rochester, NY 14611 (briannon.oconnor@gmail.com).

Accepted for Publication: November 7, 2015.

Published Online: February 1, 2016. doi:10.1001/jamapediatrics.2015.4158.

Author Contributions: Drs O’Connor and Scholle had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: O’Connor, Lewandowski, Rodriguez, Gardner, Hoagwood, Scholle.

Acquisition, analysis, or interpretation of data: O’Connor, Lewandowski, Tinoco, Gardner, Hoagwood, Scholle.

Drafting of the manuscript: O’Connor, Gardner, Hoagwood.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: O’Connor.

Obtained funding: Gardner, Hoagwood, Scholle.

Administrative, technical, or material support: Lewandowski, Rodriguez, Tinoco, Hoagwood, Scholle.

Study supervision: Lewandowski, Gardner, Hoagwood, Scholle.

Conflict of Interest Disclosures: None reported.

Funding/Support: This project was supported by grant U18HS020503 (principal investigator, Dr Scholle) from the Agency for Healthcare Research and Quality and Centers for Medicare & Medicaid Services.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the Agency for Healthcare Research and Quality.