Receipt of Evidence-Based Pharmacotherapy and Psychotherapy Among Children and Adolescents With New Diagnoses of Depression

Depressive illness of children and adolescents (youths) is a major public health concern. A nationally representative U.S. study from 2010 reported a lifetime prevalence of major depressive disorder or dysthymia for approximately 11.7% of adolescents (15.9% for females and 7.7% for males). Depression is a critical factor for suicide, which is the third leading cause of death among teenagers in the United States (1). Approximately 60% of depressed youths report having thought about suicide, and 30% attempt it (2).

Because recommendations for the treatment of depression are based in part on safety and efficacy studies conducted with adults, clinical practice guidelines for the treatment of depression in youth suggest that treatment modalities should be conservative (3). The effects of psychotropic agents on brain development have not been adequately delineated for humans (4,5), and the therapeutic effects of antidepressants vary significantly across age groups (6). Although untreated depression has substantial long-term consequences, unsafe and ineffective depression treatment also can have adverse outcomes. Therefore, clinical practice guidelines are promulgated for the best benefit-to-risk ratio for youths with depression.

Although there is considerable research targeting depression treatment in both the public and private sectors of the U.S. health care system, there is a lack of data regarding how treatment varies for the latter. Compared with the public sector, the private sector is characterized by more complex systems of insurance, fewer incentives to use clinical practice guidelines, and higher likelihood of receiving non–evidence-based medications (7).

In light of these factors, we sought to investigate the patterns of combination treatment (psychotherapy plus medication), psychotherapy, and antidepressant and antipsychotic treatment across provider specialty, insurance type, regional location, disease complexity and severity, and individual characteristics among privately insured youths with a depression diagnosis. The aim of the study was to determine the extent to which depression treatment conforms to the extant evidence base and to identify factors contributing to observed variations in treatment modalities.

Methods

Data

We conducted a retrospective claims analysis of children and adolescents (six to 17 years of age) at the time that a claim appeared. Annual data from 2008–2010 were derived over 24 months from the Truven Analytics database, a nationally representative data sample of individuals with employer-provided health insurance in the United States, representing the general medical experience of approximately 50 million covered lives. The database includes private-sector health information of approximately 100 payers throughout the country. This study was approved by the Boston Medical Center Institutional Review Board.

Inclusion and Exclusion Criteria

Table 1 summarizes different treatment options for youths with depression. “Combination” treatment was identified by the Treatment for Adolescents With Depression Study (TADS; 8,9) as the most effective treatment to hasten remission of depressive symptoms while offering the most favorable risk-benefit profile (10,11). Combination treatment also was found in the Treatment of Resistant Depression in Adolescents (TORDIA) study to be the most effective treatment for depression that is resistant to selective serotonin reuptake inhibitors (SSRIs; 12). Fluoxetine, the only antidepressant approved by the U.S. Food and Drug Administration (FDA) for the treatment of depression among youths, has a consistently favorable risk-benefit profile based on available evidence (13–15) and is considered “first line” treatment. The remaining SSRIs (except fluoxetine) are considered “second line” by most clinical guidelines (15–18), but most of the evidence supporting their use remains equivocal. The remaining drug types, described as non–evidence based, have two subcategories: newer antidepressants—such as noradrenergic and specific serotonergic antidepressants (NaSSAs), norepinephrine and dopamine reuptake inhibitors (NDRIs), and serotonin and norepinephrine reuptake inhibitors (SNRIs)—and second-generation antipsychotics. These categories are not supported by sound empirical evidence of safety and efficacy for the treatment of depressed youths. Finally, the psychotherapy-alone treatment option excluded any medication use.

Children and adolescents were identified as having a depressive disorder if they met three criteria: if they received at least two outpatient diagnoses of depression, because such a method offers higher sensitivity than just one outpatient diagnosis (19) or one medication claim plus one diagnosis of depression; if they had a six-month “clean period” without any diagnosis of depression or medication utilization; and if they had at least 12 months of continuous insurance coverage after the date of diagnosis. This methodology is sufficiently accurate and has shown high positive predictive value (20). Exclusions were children and adolescents with a diagnosis of bipolar (296.0x, 296.4x) or schizophrenia (295.6x) disorders and those with mixed disorders (depression and psychosis, or depression and bipolar disorder) because they may appropriately receive second-generation antipsychotic treatment for these illnesses, thus making it difficult to distinguish between antipsychotic medication used for treatment of psychotic symptoms versus depressive symptoms. Our final sample consisted of 61,599 youths who met our inclusion criteria.

Analysis

Medication data were extracted from the outpatient pharmacy claims that showed the national drug code. At least one claim for antidepressant or second-generation antipsychotic medication was defined as “any medication use.” Providers were classified into two mutually exclusive categories: mental health specialists, represented by psychiatrists, psychologists, psychiatric nurses, and child psychiatrists, and primary care providers, represented by internists, pediatricians, family physicians, nurse practitioners, and general practitioners. Psychotherapy utilization was identified by Current Procedure Terminology codes (90804–90809, 90853, and 90847). Private insurance was classified into four groups (Table 2): those enrolled in a health maintenance organization (HMO), point-of-service plan (POS), or preferred provider organization (PPO) were considered to be managed care clients. The others were receiving fee-for-service coverage. We controlled for demographic characteristics, including age (childhood, six to 11 years; early adolescence, 12–14 years; and late adolescence, 15–17 years), gender, U.S. geographic region (West, Midwest, South, and Northeast), and psychiatric comorbidities (anxiety, ICD-9 code 300.0; posttraumatic distress disorder [PTSD], ICD-9 code 309.81; attention-deficit hyperactivity disorder [ADHD], ICD-9 code 314.01; and suicidal ideation, ICD-9 code V62.84). Having more than one mental disorder has been identified in the literature as a proxy of illness complexity among depressed individuals (6).

Descriptive statistics are presented for the entire study population as well for strata defined by provider specialty, insurance type, and regional and individual variables. The percentages of youths receiving each type of treatment were calculated. Prevalence estimates used either of two denominators: the total number of youths with depression was the denominator for conditions including psychotherapy, and the total number of youths receiving at least one medication was the denominator for those receiving pharmacologic treatment. Multiple logistic regression models with stable standard errors accounting for the complex survey design of the sample and allowed the assessment of the independent effects of provider specialty, insurance type, illness severity, and regional and individual characteristics on the likelihood that depressed youths would receive one of the six depression treatments. Results are presented as adjusted odds ratios with 95% confidence intervals.

Interaction was examined between the provider specialty and the other covariates. Interaction terms were ruled out from the model if they were not significant or not policy relevant. Collinearity was examined among all covariates, with the variance inflation factor not exceeding 10. The expected performance and stability of the models were measured by the c statistic and the Hosmer-Lemeshow test, respectively. Using the SAS Raking Macro, developed by Izrael and colleagues (21), we weighted the data to represent the privately insured population in the United States. The accuracy of the estimates between medication groups and the predictors of interest were tested with Bonferroni methods and the bootstrap inference (10,000 bootstrap samples) (22). All analyses were run with SAS software, release 9.3 (23).

Results

Table 3 reports the prevalence and the unweighted univariate predictors of type of treatment. Nearly a quarter (24.8%) of all depressed youths received medication monotherapy. Of those, 23.5% received first-line medication (fluoxetine), 50.6% received second-line medications (other SSRIs), 9.2% received newer antidepressants (SNRIs, NDRIs, and NaSSAs), and 15.6% received second-generation antipsychotics. Combining all SSRIs (first and second line) accounted for 74.1% of medication monotherapy treatment. Together, non–evidence-based choices (SNRIs, NDRIs, NaSSAs, and second-generation antipsychotics) accounted for 24.8% of medication monotherapy treatment. About 1% of the medication monotherapy sample reported use of tricyclic antidepressants and monoamine oxidase inhibitors (data not shown). Of all youths with depression, 33.6% received psychotherapy alone and 2.7% received combination treatment (fluoxetine plus psychotherapy).

For the unweighted univariate analysis, six characteristics were assessed (provider specialty, private insurance, age group, gender, region, and psychiatric comorbidities). Overall, compared with mental health specialists, primary care providers reported an increased likelihood of choosing fluoxetine alone (p<.001) but were less likely to choose combination treatment, psychotherapy alone, and second-generation antipsychotics (p<.001 for each category). For insurance type, psychotherapy alone was less likely to be used in fee-for-service care compared with HMO care (p<.001). On the other hand, newer antidepressant treatment (p<.001), second-generation antipsychotics (p<.001) and fluoxetine alone (p<.001) were more frequently used under fee-for-service arrangements, compared with HMOs.

Table 4 reports the independent predictors from the multivariable logistic regression. Overall, provider specialty was the most consistent factor predicting treatment of depressed youths. After analyses controlled for demographic characteristics, comorbidities, and illness severity, mental health specialists compared with primary care providers were more likely to choose combination treatment (odds ratio [OR]=8.61), newer antidepressants (OR=1.44), second-generation antipsychotics (OR=2.24), and psychotherapy alone (OR=5.90). First-line treatment was not used differently across specialties.

Compared with HMOs, the odds that PPOs and POSs would use newer antidepressant options were significantly higher (OR=1.44 and OR=1.65, respectively), whereas second-generation antipsychotics were more likely to be used under fee-for-service coverage (OR=1.81). In addition, PPOs were more likely to use fluoxetine alone (OR=1.23) and less likely to use psychotherapy alone (OR=.92). Combination treatment and second-line antidepressant treatment were similarly used across all plan types.

Medication-only and combination regimens were less likely to be used with children compared with adolescents (except for first-line antidepressant medication and second-generation antipsychotics), whereas psychotherapy alone was more likely to be used with children (OR=1.61) than with adolescents. The odds of receiving second-line (OR=.66) and newer antidepressants (OR=.21) were lower for children compared with older adolescents, and males were more likely than females to receive newer antidepressants (OR=1.23) and second-generation antipsychotics (OR=1.54). For regional differences, youths in the Northeast were more likely than youths in the West and South to receive combination treatment, whereas psychotherapy alone was more likely to be used in the West (OR=1.22) than in the Northeast.

Considering illness severity, we found that compared with youths with major depressive disorder, youths with dysthymia were less likely to receive combination treatment (OR=.50), fluoxetine alone (OR=.81), non–evidence-based medication (OR=.83), or psychotherapy alone (OR=.65).

Discussion

The results from this study have important implications for policy makers, physicians, and patients. Nationally, two-fifths of privately insured youths with depression did not receive any depression-specific treatment, and less than 3% received combination treatment. Of youths treated with medication, one-quarter received medications with little to no empirical support, and an additional one-half received medications with uncertain benefits. As such, provider practice for the treatment of depression among privately insured youths does not appear to be in alignment with the current evidence base.

Our findings corroborate previously published studies, signaling low utilization rates of combination treatment (16). This finding reflects the well-documented difficulties associated with the delivery of psychotherapy in primary care settings (24). Provider specialty has been recognized in this study as well as in previous research as a key determinant of treatment selection for younger populations (25). Current health care delivery migration to primary care settings in the United States (26) and the escalating shortages of mental health specialists all around the country (27), especially for pediatric patients, require primary care providers to treat patients that historically were managed by mental health specialists. However, primary care providers are typically overburdened with additional demands to provide mental health care with fewer resources, fewer support staff, additional budget constraints, and limited training compared with specialists’ circumstances (28,29). The Medical Outcomes Study reported that patients seen by psychiatrists had better functional outcomes (30), and Callahan and colleagues (31) found that less than half of primary care providers prescribed a treatment for their patients with depression, even though the clinicians had proper diagnostic scales and treatment algorithms. Furthermore, medication alone for patients with depression is not always effective (especially at follow-up) (32), and current reimbursement incentives for primary care providers to offer psychotherapy as an alternative may not be worth the time and intensive labor that psychotherapy requires (30).

Insurance coverage was significantly associated with four out of six treatment options, suggesting an association between less restrictive private insurance arrangements with higher rates of nonrecommended medication options. Our data suggest that prescription of non–evidence-based antidepressants and second-generation antipsychotics was more prevalent in fee-for-service and PPO insurance arrangements, the latter being the more flexible managed care option in terms of specialists’ availability, noncapitated payment arrangements, and no prior approval needed for prescribing specific medications. Compared with more flexible insurance arrangements, managed care’s use of formularies may predispose providers to choose recommended treatments more often. For example, DeBar and colleagues (33) reported that compared with adolescents covered by PPOs, adolescent members of HMOs were more likely to receive treatment via treatment-concordant protocols and prescription formularies.

In terms of age differences, second-line antidepressants were more likely used to treat adolescents compared with children; this finding may derive from providers’ awareness of FDA approval of escitalopram to treat major depressive disorder beginning at age 12. In general, the age-difference findings reflect a more conservative approach to the treatment of depression for younger compared with older youths.

Effects of illness complexity on prescribing practices indicated an increased likelihood of second-generation antipsychotic prescription for youths with depression plus either PTSD or suicidal ideation, even though second-generation antipsychotics lack empirical support for these indications. Although much less research has focused on treatments for the milder forms of depression (including unspecified depression and dysthymia), current guidelines (34,35) suggest that treatment of mild depression should begin with psychotherapy. Nonetheless, the vast majority of youths in this study with a diagnosis of dysthymia received monotherapy (one medication).

Several limitations of this study should be noted. First, we used claims data and may have underidentified patients with depression by excluding those who had an outpatient visit in which a depression diagnosis was not recorded. Second, not all possible comorbid diagnoses were assessed. It is possible that less frequent comorbid conditions, such as disruptive behavior disorders, have an impact on choice of medication, including second-generation antipsychotics. However, their low frequency in the data did not allow separate analysis. Third, even though we tried to identify a depression-free period by excluding those with a prior medication or depression diagnosis before the index date, we recognize that a fraction of our sample may not represent first visits because some may have received treatment prior to our designated clean period. In addition, our definition of combination therapy as the most desirable treatment choice is potentially more applicable to youths with depression nonresistant to SSRIs. Randomized clinical trials are needed to guide future clinical algorithms for youths with other types of resistant depression (such as psychotherapy resistance and SNRI resistance). Finally, primary care providers may be more willing than mental health professionals to use adjustment disorder diagnosis rather than depressive diagnosis for reimbursement purposes. Hence, we may be underestimating the frequency of youths being treated for depression in primary care providers’ settings.

Mental health treatment recommendations and primary care providers’ busy schedules seem to move in opposite directions. Even though recent meta-analytic evidence positions psychotherapy as equal to or possibly greater in effectiveness than antidepressant medication for the treatment of depression of youths (36), less than 10% of primary care providers in the private sector delivered psychotherapy in 2009. [A table provided in the online supplement to this article provides further detail.] The collaborative care model for the treatment of depression, shown in over 70 randomized trials to be efficacious in primary care settings for the treatment of depressed adults (37), now has empirical support for enhancing access to evidence-based treatments (including psychotherapy) and improving clinical outcomes for depressed youths (38,39). The collaborative model incorporates (alongside medication management as indicated) evidence-based psychotherapy delivered by colocated mental health clinicians (predominantly master’s level), as well as care management to facilitate patient engagement, treatment adherence, and referral to specialty care if needed (38,39). However, the implementation of psychotherapy in primary care settings remains contingent on proper training. For example, cognitive-behavioral therapy (CBT) has been taught in experiential instruction and residential programs and has been effective in increasing providers’ ability to use CBT as a treatment after one year of training (40). However, for primary care providers, a one-year training program is likely to be too much of a time commitment, and shorter CBT training alternatives should be sought. A 2010 study developed a formalized, multimodal, brief psychotherapy training approach for mental health practitioners in primary care. The workshop lasted 1.5 days and significantly improved both knowledge and self-reported abilities, and the effect was maintained at three-month follow-up (40).

The use of mental treatment networks (internet-based) may be another feasible approach. For example, for persons between the ages of 14 and 21, the CATCH-IT initiative (Competent Adulthood Transition with Cognitive-Behavioral and Interpersonal Training) is a low-cost preventive intervention that is fairly easy to disseminate in primary care settings and offers resources on building competencies to reduce current and future depressive symptoms (41). Such modalities may be particularly effective for rural or socioeconomically marginalized clinics.

Conclusions

Recognizing that the role of primary care providers in identifying and treating children and adolescents with emotional, behavioral, and developmental disorders will continue to increase in the context of a nationwide shortage of mental health specialists, both the American Academy of Pediatrics and the American Academy of Child and Adolescent Psychiatry (42) have advocated the integration of mental health care into the pediatric medical home. One-quarter of youths seen in the primary care setting and about one-half of all pediatric office visits involve behavioral, emotional, developmental, psychosocial, or educational concerns (42). As a result, primary care providers are on the front line in the effort to identify, diagnose, and treat psychiatric disorders of youths. The results of this study highlight the increasing need for more research on clinician judgment and decision making in the application of guideline-concordant treatment for depression. They also underscore the need to use innovative health care delivery strategies to improve access to safe and effective treatment for youths experiencing depression.