Update on Randomized Placebo-Controlled Trials in the Past Decade for Treatment of Major Depressive Disorder in Child and Adolescent Patients: A Systematic Review

Abstract

Objectives:

Antidepressant treatment for major depressive disorder (MDD) has been a topic of continued interest with a number of placebo randomized-controlled trials (RCTs) being published in the past decade. We review the updated literature since the 2007 Bridge meta-analysis, and reassess safety data looking at signals of treatment-emergent suicidality with the Columbia Suicide Severity Rating Scale (CSSR-S).

Methods:

PubMed literature review was performed searching for RCTs published since the 2007 article and supplemented with manual search.

Results:

Findings from seven trials (five industry sponsored, one NIMH funded, and one other) were included in this systematic review, which showed high medication and placebo response rates, with only fluoxetine and escitalopram treatment reaching statistical significance. Fluoxetine was also shown to prevent relapse of MDD with continued treatment with an odds ratio of 3.2 for prevention of relapse compared with placebo. There were no increases in treatment-emergent suicidality associated with antidepressant medication in any trial measuring suicidality systematically using the CSSR-S.

Conclusions:

Depressed pediatric patients respond similarly in these trials to antidepressant intervention and placebo, with recent studies showing that newer agents did not clearly demonstrate benefit above placebo. The evidence continues to support fluoxetine and escitalopram as first-line treatment and demonstrated effect to prevent relapse. Use of newer rating scales reveals similar rates of treatment-emergent suicidality in patients on antidepressants as placebo, in contrast to increased suicide signal suggested by past research using adverse events data. Antidepressant treatment is generally safe and well tolerated in this age group.

Introduction

Depression is a common condition that affects up to 2.8% of school-aged children and 5.6% of adolescents (Costelo et al. 2006), with a lifetime prevalence rate of up to 14.8% of a mood disorder by young adulthood (Copeland et al. 2011). Despite extensive evidence that depression affects future functioning with long-lasting deleterious effects, such as future interpersonal problems, impaired school performance, an increased risk for other mental and substance use disorders, and suicide, many pediatric patients are undertreated (Cummings and Druss 2011). This may be related to the controversial use of antidepressants in pediatric patients due to concerns about effectiveness and about treatment-emergent suicidality in this age group (Hammad et al. 2006), which has been frequently cited as a treatment barrier by primary care providers (Cheung et al. 2008b). Furthermore, the evidence base for treatment for patients in this age group has been more limited, with many adult guidelines being used to influence treatment decisions.

Bridge et al. (2007) published a meta-analysis of 27 trials assessing efficacy of antidepressant use in pediatric major depressive disorder (MDD). Their findings suggested a small but positive effect of antidepressants for pediatric patients with MDD, although there was a relatively high placebo response rate of 50%, with higher placebo response in studies with more sites. A more recent Lancet network meta-analysis interpreted results of 14 antidepressant medications studied in 34 trials as not seeming to offer a clear advantage for children and adolescents when factoring in the risk–benefit profile, based on high rates of adverse effects and increased suicide signal with active treatment (Cipriani et al. 2016). Use of newer rating scales such as the Columbia Suicide Severity Rating Scale (CSSR-S), which is designed to systematically evaluate treatment-emergent suicidality as opposed to using adverse events data, has been incorporated for safety assessment in more recent trials, and may suggest that treatment-emergent suicidality is not as significant a risk as indicated on the Food and Drug Administration (FDA)-mandated Black Box Warning label—a concept that seasoned providers are likely aware of.

We review the updated literature since publication of the 2007 Bridge et al. meta-analysis for efficacy data and reassess safety data looking at signals of treatment-emergent suicidality with the CSSR-S.

Methods

A PubMed literature review was performed searching for published randomized controlled trials (RCTs) assessing for antidepressant use in pediatric aged patients for management of MDD that were published through July 1, 2016. All antidepressant classes were included as were extension studies and trials assessing for relapse prevention. Excluded were studies of treatment-resistant depression, bipolar depression, and depression comorbid with any other major psychiatric disorder (i.e., ADHD, substance use). Search terms (variously combined) were children, child, adolescent, pediatric, MDD, depression, selective serotonin reuptake inhibitors (SSRIs), antidepressants, treatment, and medication. Additional references were manually identified from the reference lists of published articles.

Results

Seven randomized controlled trials were identified and included in this systematic review. Of the trials identified, four were acute efficacy trials assessing selegiline transdermal patch, fixed and flexible dosing duloxetine, and escitalopram. The fixed and flexible dosing duloxetine trials also had extension trials evaluating prolonged treatment, and a separate RCT was identified evaluating continued escitalopram treatment. The last two trials assessed relapse prevention after acute treatment and a relapse prevention after maintenance treatment. Trials were primarily industry sponsored, although one was NIMH funded and one was funded by the Canadian Federal Government. All seven RCTs showed high medication as well as placebo response rates. Table 1 summarizes the literature basis for this article (Atkinson et al. 2014; Emslie et al. 2014 are listed first as acute efficacy trials and repeated under the extension trials in the tables). Table 2 summarizes the efficacy data and Table 3 provides safety data.

Table 1.

Results of Literature Review

Type of study	Author	Year	Funding source	No. of sites	Active medication, daily dose	Comparator
Acute phase efficacy	DelBello et al.	2014	Industry sponsor (Somerset Pharm)	26	Selegiline transdermal patch	Placebo
	Emslie et al.	2014	Industry sponsor (Eli Lilly)	60	Duloxetine fixed dosing (60 mg vs. 30 mg)	Fluoxetine 20 mg, placebo
	Atkinson et al.	2014	Industry sponsor (Eli Lilly)	65	Duloxetine flexible dosing 60–120 mg OD	Fluoxetine 20–40 mg, placebo
	Emslie et al.	2009	Industry sponsor (Forest Labs)	40	Escitalopram 10–20 mg	Placebo
Extension trial	Emslie et al.	2014	Industry sponsor (Eli Lilly)	60	Duloxetine fixed dosing (60 mg)	Fluoxetine 20 mg
	Atkinson et al.	2014	Industry sponsor (Eli Lilly)	65	Duloxetine flexible dosing 60–120 mg OD	Fluoxetine 20–40 mg, placebo
	Findling et al.	2013	Industry sponsor (Forest Labs)	40	Escitalopram 10–20 mg	Placebo
Relapse prevention after acute treatment	Emslie et al.	2008	NIMH funded	1	Fluoxetine 10–40 mg	Placebo
Relapse prevention after maintenance treatment	Cheung et al.	2008a	CIHR (federally funded)	3	Sertraline	Placebo

CIHR, Canadian Institutes of Health Research; CSSR-S, Columbia Suicide Severity Rating Scale; OD, once daily.

Table 2.

Efficacy Data

Author	Active medications, daily dose	Categorical outcome measurement	Medication response, %	Placebo response, %	Active comparator, %	Statistical significance
DelBello et al. (2014)	Selegiline transdermal patch	Response: CGI-I ≤2	58.6	59.3	—	p = 0.8992
Emslie et al. (2014) (acute)	Duloxetine fixed dosing (60 mg vs. 30 mg)	Response: 50% improvement of CDRSR	69	60	Fluoxetine: 61	p > 0.05
Atkinson et al. (2014) (acute)	Duloxetine flexible dosing 60–120 mg OD	Response: 50% improvement of CDRSR	67	62	Fluoxetine: 63	p > 0.05
Emslie et al. (2009)	Escitalopram 10–20 mg	Response: CGI-I ≤2	64.3	52.9	—	Escitalopram>PLB ( p = 0.003)
Emslie et al. (2014) (extension)	Duloxetine fixed dosing (60 mg)	Remission: defined as CDRSR <28	81	—	Fluoxetine: 74	p > 0.05
Atkinson et al. (2014) (extension)	Duloxetine flexible dosing 60–120 mg OD	Remission: defined as CDRSR <28	41	41	Fluoxetine: 33	p > 0.05
Findling et al. (2013)	Escitalopram 10–20 mg	Response: >40% improvement of CDRSR	63.6	47.1	—	Escitalopram > PLB p = 0.003
Emslie et al. (2008)	Fluoxetine 10–40 mg	Relapse rate: CDRS >40 or clinician judgment	42	69.2	—	Fluoxetine > PLB ( p = 0.009) (OR for relapse 3.2 PLB:Med)
Cheung et al. (2008a)	Sertraline	Recurrence rates: per clinician judgment	38	0	—	p = 0.054

Bold indicates that statistical significance was reached.

CDRS-R, Child Depression Rating Scale revised; CGI-I, Clinical Global Improvement Index; OD, once daily; OR, odds ratio; PLB, placebo.

Table 3.

Safety Data

Author	Active medication, daily dose	Safety outcome	Rates of suicidal events in active medication group	Rates of suicidal events in placebo group, %	Rates of suicidal events in active comparator group, %	Statistical significance
DelBello et al. (2014)	Selegiline transdermal patch	TE-SI	2.6%	2.6	—	—
Emslie et al. (2014) (acute)	Duloxetine fixed dosing (30 mg vs. 60 mg)	CSSR-S	6.7% 60 mg, 5.2% 30 mg	9.4	Fluoxetine: 8.0	—
Atkinson et al. (2014) (acute)	Duloxetine flexible dosing 60–120 mg OD		7.1%	6.8	Fluoxetine: 8.0	—
Emslie et al. (2009)	Escitalopram 10–20 mg	MC-SSRS	9.2%	9.4	—	—
Emslie et al. (2014) (extension)	Duloxetine fixed dosing (60 mg)	CSSR-S	7.0% 60 mg, 10.3% 30 mg	7.6	Fluoxetine: 8.8	—
Atkinson et al. (2014) (extension)	Duloxetine flexible dosing 60–120 mg OD		11.1%	9.4%	Fluoxetine: 14.3	—
Findling et al. (2013)	Escitalopram 10–20 mg		14.5%	10.9	—	—
Emslie et al. (2008)	Fluoxetine 10–40 mg	Adverse effects leading to Rx discontinuation	2% (SI)	3.8 (medical AE)	—	—
Cheung et al. (2008a)	Sertraline	CASES and SI self-report	0	0	—	—

AE, adverse events; CASES, Common Adverse and Side Effects Scale; CSSR-S, Columbia Suicide Severity Rating Scale; MC-SSRS, Columbia Suicide Severity Rating Scale; OD, once daily; SI, suicidal ideation; TE-SI, treatment-emergent suicidal ideation.

Acute efficacy trials

The four acute efficacy trials showed response rates similar to placebo, with only escitalopram treatment reaching statistical significance. In the selegiline patch trial, 308 adolescent subjects with moderate-to-severe depression were randomized with 70% completing the trial. Patients had a significant decline from baseline on their Child Depression Rating Scale-Revised (CDRS-R) total score with a mean reduction of 21.4 ± 16.6 for active treatment and 21.5 ± 16.5 for placebo. Response rates, as defined by a Clinical Global Impressions-Improvement (CGI-I) of one or two at the study end, were similar at 58.6% versus 59.3%; however, these were without statistical significance. The selegiline patch was felt to generally be safe and well tolerated in the adolescent sample (DelBello et al. 2014).

Emslie et al. (2014) showed inconclusive results in a trial of fixed dosing duloxetine (60, 30 mg) compared with an active control (fluoxetine) in a cohort of children and adolescents as neither the investigation drug nor the active control separated from placebo on the CDRS-R at 10 weeks. The authors discuss that “this was the first randomized control study to fail to detect a statistically significant difference between placebo and an active control with [previously] known efficacy in pediatric MDD, rendering results inconclusive as opposed to negative,” with high placebo response explicating the lack of separation.

This was similarly true for a trial assessing flexible dosing duloxetine (60–120 mg daily), where neither the active drug nor the active control separated from placebo. Stratifying the treatment groups into children and adolescent age ranges did not reveal statistically meaningful differential responses, which is important as subgroup analysis in previous trials showed positive response in the adolescent age group while child response did not separate from placebo (Atkinson et al. 2014).

Escitalopram was studied in 312 adolescent patients for 8 weeks, with a reduction in mean CDRS-R score of −22.1 versus −18.8 (p = 0.002) seen in the escitalopram group relative to placebo; these differences between active treatment and placebo are clinically relevant differences. The effect size of 0.27 compared with placebo was similar to an effect size of 0.25 determined for an adolescent sample in two combined double-blind trials of fluoxetine (0.39) and similar to the known escitalopram effect size in depressed adults (0.31). The largest drug–placebo difference was in the fourth week of treatment, with 85% of the eventual placebo responders responding by week 4 compared with 69% of placebo eventual responders (Emslie et al. 2009).

Extension studies

The fixed and flexible dosing duloxetine studies were extended by an additional treatment time of 26 weeks. During the extension treatment, all patients received either duloxetine (60–120 mg) or fluoxetine after a 2-week double-blind tapering period, with both studies using flexible duloxetine dosing. For patients initially randomized to duloxetine 60 mg, there was no statistically significant probability of remission at 36 weeks (81%) compared with those randomized initially to fluoxetine 20 mg (74%) (Emslie et al. 2014). For patients initially randomized to flexible dosing duloxetine, the probability of remission at 36 weeks was 72% and 84% for patients initially randomized to fluoxetine. There were no statistically significant differences in the probability of remission at any time during the 36-week study between patients treated with flexible dosing duloxetine or randomized to fluoxetine (Atkinson et al. 2014).

Adolescents who completed the escitalopram lead-in study were enrolled in a 16–26-week trial and maintained the same lead-in randomization and dosage during the extension. After lead-in, 165 patients were entered into the extension trial with roughly half of patients completing treatment. CDRS-R total score improvement was significantly greater for escitalopram than for placebo (p = 0.005) with a mean change in CDRS-R total score from baseline to 24-week endpoint of −26.0 for patients who received open-label escitalopram. Response rates, defined as a ≥40% reduction in CDRS-R from baseline and CGI-I ≤2, were significantly higher for escitalopram than for placebo. Remission rates, defined as a CDRS-R of ≤28, were 50.6% and 35.7% (p = 0.002) for escitalopram and placebo, respectively. Continued active treatment showed ongoing response in the subset of patients who were initial responders at the end of week 8.

The authors analyzed week 8 CDRS-R total score change stratifying whether patients did or did not enter the extension trial to investigate for potential introduction of bias; there was no evidence to support a potential bias of improvement in patients who received active treatment and decided to enter the extension trial with similar CDRS-R changes in patients who did not enter, compared with a greater CDRS-R change in placebo-treated patients who received extended treatment (Findling et al. 2013).

Relapse prevention

In order to assess relapse frequency and time to relapse of MDD after initial treatment (12 weeks of open fluoxetine), Emslie et al. performed an extension trial in 2008. They randomized children and adolescents who experienced a reduced CGI-I score of 1 or 2 and a decrease of at least 50% in the CDRS-R to receive fluoxetine and placebo for an additional 6 months. A total of 102 children and adolescents were randomized, with significant reduction in relapse rates in the fluoxetine group (42%) compared with the placebo group (69.2%) (p = 0.0009). Use of a stricter definition of relapse of CDRS-R ≥40 continued to show more frequent relapse in the placebo group than in the fluoxetine group (48.1%, 22.0%, p = 0.007). Time to relapse was also prolonged in the active treatment group, 14 median weeks to full relapse in placebo-treated subjects compared with >24 weeks for fluoxetine. This study demonstrated that the odds of relapse were 3.2 times more likely in the placebo group than in patients actively treated with fluoxetine, with similar rates persisting using the stricter definition of relapse. These data confirm that treatment continuation is required beyond the initial remission of symptoms to prevent relapse and support treatment continuation for 6–9 months in children and adolescents per recommendations from adult guidelines (Emslie et al. 2008).

Cheung et al. (2008a) looked at adolescents who responded initially to open-label treatment with 12 weeks of sertraline and who further did not relapse with open-continuation treatment for an additional 24 weeks; this was the first trial assessing maintenance treatment in pediatric MDD. They randomized 22 subjects to continued sertraline treatment for 52 weeks to assess the value of continued antidepressant use in a maintenance phase after acute response is achieved. Although adolescents treated with sertraline remained well compared with those on placebo (38% vs. 0%), this did not reach statistical significance, likely due to the study being underpowered secondary to a small sample size and poor sample retention. Although not conclusive, the results suggest that maintenance treatment with sertraline may be beneficial to subjects with MDD (Cheung et al. 2008a), which is important particularly as sertraline treatment did not separate from placebo in two acute efficacy trials.

Safety measures

Of the available trials, there were no increases in treatment-emergent suicidality associated with antidepressant medication in any trial measuring suicidality systematically using the CSSR-S. Specifically, the transdermal selegiline patch had equal rates of suicidal events as placebo at a rate of 2.6% (DelBello et al. 2014).

The acute fixed (Emslie et al. 2014) and flexible dosing (Atkinson et al. 2014) duloxetine trials showed small although insignificant differences in suicide rates compared with active fluoxetine treatment. There was a small differential in the rates of treatment-emergent suicidality associated with higher duloxetine dosing, with a nonsignificant increase of 1.5% in the 60 mg duloxetine dose compared with the 30 mg dose during acute treatment; however, the opposite was seen with chronic treatment, and the extension trial showed a 10.3% rate of worsening of suicidal thinking at the 30 mg dose and 7.0% rate in the 60 mg dose. Importantly, this is in comparison with a placebo rate of 9.4% and 6.8% in the acute studies and 7.6% and 9.4% in the extension studies, with placebo showing higher rates than acute duloxetine fixed dosing treatment (Atkinson et al. 2014; Emslie et al. 2014).

There were no statistically significant suicidal signal increases in acute escitalopram treatment compared with placebo (9.4% vs. 9.2%) (Emslie et al. 2009), and although this range widened during the extension study with an overall increased incidence in any suicidal behavior or ideation of 10.9% in the placebo group and 14.5% in the active treatment group, this also did not reach significance (Findling et al. 2013).

The relapse trials did not utilize the CSSR-S to detect rates of suicide, but rather employed safety outcomes evaluating adverse effects leading to medication discontinuation and self-report. The fluoxetine study that showed a significant reduction in relapse compared with placebo also demonstrated that placebo had higher rates of suicidal events at 3.8% than active treatment at 2%. This did not reach statistical significance (Emslie et al. 2008). The Cheung et al. (2008a) sertraline maintenance treatment trial showed nil reports of suicidality with either active treatment or placebo, which was captured exclusively through spontaneous reporting by subjects.

Adverse events not related to emergent suicidality were also assessed in each study and are summarized here. A summary of the most frequent adverse effects from escitalopram was headache, nausea, insomnia, vomiting, influenza-like symptoms, diarrhea, and urinary tract infection, the majority of which were mild/moderate and may not have been related to the study drug although they did seem to occur more frequently than with placebo treatment (Findling et al. 2013; Emslie et al. 2015). Discontinuation rates of escitalopram were low generally due to adverse effects at 2.6% compared with 0.6% for placebo, with low rates of serious adverse events of 2.6% with escitalopram and 1.3% with placebo (Emslie et al. 2009).

The selegiline trial had a much higher reported rate of adverse effects overall, at 62.5% with active treatment compared with 57.5% with placebo, accounted for largely by reports of local site reactions (24.3% selegiline vs. 21.8% active treatment) occurring most commonly, and similar rates of other constitutional symptoms (headache and nausea). Treatment groups did not differ on any laboratory parameters, vital signs, or electrocardiogram findings, and no hypertensive crises were reported in the trial (DelBello et al. 2014).

In the Emslie et al. trials comparing fixed and flexible duloxetine dosing with active fluoxetine treatment and placebo control, total adverse events and discontinuation separated significantly only for the duloxetine 60 mg group during acute treatment. Otherwise, there were no significant differences between the groups with serious or total adverse events or discontinuation in the acute trial. During the extension study, the proportion of patients with one adverse event was similar for patients randomized initially to duloxetine and fluoxetine, and lower than in patients transitioned from placebo to duloxetine (63.9%, 62.0%, and 72.1%). The most frequently reported side effects were headache and nausea. There were no significant EKG or laboratory abnormalities observed during the study, with resolution of threefold elevation of alanine aminotransferase in one fluoxetine and one duloxetine patient (Atkinson et al. 2014; Emslie et al. 2014).

Extended fluoxetine treatment for relapse prevention showed similar rates of adverse events relative to placebo without discontinuation secondary to physical adverse events (Emslie et al. 2008). The most common acute adverse effects in the sertraline study included fatigue, drowsiness, irritability, agitation, anxiety, and inner tension, affecting more than half of active treatment and placebo recipients. Maintenance treatment with sertraline showed a numeric 5% difference from placebo with ongoing symptoms as mentioned, with sweating, lightheadedness, and sexual dysfunction added, however, there were a number of adverse effects that occurred numerically 5% more frequently with placebo treatment. These high rates of placebo adverse effects included rash, hypersomnia, anxiety, inner tension, dry mouth, nausea/vomiting, diarrhea, runny nose, weight loss, decreased/increased appetite, abnormal movements, and cough. No significant differences were found using statistical comparisons of the two groups (Cheung et al. 2008a).

Discussion

Compared with adults, children and adolescents with MDD continue to be underdiagnosed and undertreated (Olfson et al. 2014). Studies have suggested that this may be secondary to a differential developmental presentation of depression in children and adolescents with rather undifferentiated depressive symptoms—for example, unhappiness, irritability, aggressive behaviors, and school refusal (Thapar et al. 2012). The DSM includes criteria changes for children and adolescents such as the presence of irritability as an alternative to a depressed mood for this age group. Furthermore, comorbid conditions (such as anxiety, substance use disorder, or a conduct disorder) are common in this age group. This leads to uncertainty and possible confounding with diagnosis, with the possibility that the inclusion of prepubertal children in these studies may have resulted in the inclusion of participants with all forms of unhappiness and causes of juvenile irritability.

In addition, children and adolescents with comorbid psychiatric conditions, and those at heightened risk for suicide, are frequently excluded from trials. This suggests that industry-sponsored trials, especially, where there is a high push to rapidly recruit large number of patients, may not be representative of the population who present to clinical service. Furthermore, sample and methodological differences between industry-supported trials and those funded by the NIMH may actually contribute to high rates of placebo response and reduced signal of medication effect (Walkup 2017).

It has been well documented that research assessing antidepressant effectiveness in adults as well as younger populations can be complicated by high placebo response rates. Industry-sponsored trials in youth populations have had placebo response rates in the 50%–60% range compared with 55%–65% in active drug treatment, which represents only an average 8% group difference. This is in contrast to the NIMH studies, wherein there are more substantial between-group differences of 25%–30%, in large part owing to lower placebo response rates in the 30%–35% range; this is more consistent with placebo response rates in adult subjects (Walkup 2017).

There are various hypotheses attempting to explain high placebo response rates and relatively low active treatment response rates, including sample differences and measurement error associated with the pressure to recruit a large patient sample in a timely manner. This is made possible by the inclusion of potentially clinically subthreshold depressive illnesses and utilization of large number of sites, wherein investigators may have limited experience with pediatric depression or clinical trials, and with recruitment of only a few patients per site. Walkup (2017) argues that significant placebo rates raise the question of whether a representative population was enrolled for the condition under study, which is further challenged by evidence wherein the inclusion of young children in industry-sponsored trials has been associated with increased placebo response rates.

Bridge et al. in 2009 examined characteristics of placebo response rates and determined that the single best predictor of high placebo response rates was the number of study sites. Although it has been well documented that placebo rates increase proportionally to the number of study sites, this and industry sponsorship cannot exclusively account for the limited separation of active treatment from placebo in the more recent placebo RCTs. For example, Emslie et al.'s (2009) escitalopram trial showed superior acute efficacy and Findling et al.'s (2013) escitalopram extension trial both had 40 study sites and were industry sponsored. The fluoxetine trial evaluating continued treatment for relapse prevention, however, was NIMH funded and there was only one study site (Emslie et al. 2008).

In fact, industry sponsorship and disparity in protocol likely do not sufficiently explain the high placebo response rates. Large scale NIMH-funded trials (TADS, CAMS, TORDIA, TASA, and POTS) actually borrowed published implementation protocols from the Emslie et al. initial industry-sponsored fluoxetine study, and the placebo response rates in these studies were generally low and did not limit differentiation from active treatment. Walkup (2017) suggests that it is the rigor associated with NIMH sponsorship that makes a significant difference in study results.

Factors include the active involvement of experts to assure fidelity and accuracy in identification and recruitment of subjects with the studied disorder, and frequent reviews of recruited patients for “caseness” of participants to ensure internal validity of the study population. Industry study clinical research sites have more experience with clinical trial implementation and the study teams aim for enhancement of internal validity through required development of mastery of the protocol by evaluators, supported by extensive oversight by independent review boards of study protocols to monitor consistency of implementation cross sites and fidelity to study procedures.

Furthermore, NIMH-funded studies had additional goals of establishing public health effect size and inform providers of best practices, above identification of a signal of efficacy; these studies employed numerous other quality indicators, such as, use of research techniques (comparative treatment trials with ecologically valid comparison and control groups) and collection of data that could be used in mediator and moderator analyses, with the intention of reducing placebo response rates. These factors likely improve internal validity of the NIMH-sponsored trials and may account for uniformly low placebo rates, and may be difficult for industry-sponsored studies to replicate due to time pressures to recruit the sample and complete the execution of the study. Inclusion of these factors, which shift focus from recruitment of high sample size, and targeting the objective of limiting placebo rate, may improve the quality and, therefore, outcomes of industry-sponsored trials (Walkup 2017).

In addition, publicly funded research may allow for more extensive secondary analysis, and to examine population characteristics of the sample. In fact, a recent reanalysis of the DelBello et al. (2014) selegiline patch trial allowed for the development of a composite scoring system based on aggregate patient data to predict the probability of placebo response. Examination of population-level characteristics through such a prediction system could allow for incorporation into the research design and treatment strategy for youth who are more likely to experience placebo response, to better objectify the efficacy and placebo responses (Nakonezny et al. 2015).

There is emerging evidence that subject expectation additionally influences placebo rates and treatment outcomes, both for effectiveness and for development of adverse effects. Negative effects of placebo, clinically known as nocebo effects, occur in up to 25% of patients, and may lead to earlier discontinuation and study attrition rates (Hauser et al. 2012). Rutherford et al. (2017) demonstrated a differential response rate in patients with high expectation of receiving the active study drug, improving on average 6 Hamilton depression scale points more than those who were receiving the same active treatment though they were aware of the possibility of being treated with placebo. They established that expectancy has a powerful influence on outcomes and, therefore, placebo effects.

Efficacy is one of the major outcomes that is routinely evaluated through experimental studies; however, trials also provide information about safety and tolerability. Even without positive efficacy results, the publication of youth tolerance for active medication treatment compared with placebo or active comparator provides important information for providers. Walkup has suggested that instead of concluding that trials with high placebo response rate and limited between-group differences be considered “negative trials” or that they failed to demonstrate efficacy, rather they should be considered “failed” trials due to substantial methodological problems. Some of these have been previously listed, including selection of young prepubertal children and low sample sizes in large studies with multiple sites. Walkup argues that in contrast to industry trials, those funded by the NIMH have many strengths, with lower placebo response rates and meaningful between-group differences, which support antidepressant efficacy. He posits that NIMH-funded studies should be weighted preferentially when reviewing the literature, adding that overinclusion of poor quality industry-sponsored trials leads to incorrect conclusions and distorted interpretations about effectiveness of antidepressants and does not adequately represent the available data (Walkup 2017).

This is of particular relevance, as safety concerns regarding increased suicide signal in depressed youth treated with antidepressants lead to the development of the FDA Black Box warning in October 2004. The impact of this was the reduction of antidepressant prescriptions by >50% between 2003 and 2005 for pediatric depression. Simultaneously, the suicide rate among Americans under the age of 19 years rose by 18%, which represented the most dramatic 1 year change since the government started collecting suicide statistics in 1979.

The concern for treatment-emergent suicidality has been cited as one of the major contributors to provider hesitance to use this class of medications (Cheung et al. 2008b). Of the included trials, the acute and extension trials exclusively utilized the CSSR-S to evaluate for treatment-emergent suicidality. Their research findings revealed similar rates of suicidal events in active treatment, comparator groups when utilized, and placebo rates, with none of the trials separating statistically from placebo. This suggests that the utilization of newer rating scales, which have been systematically designed to evaluate for treatment emergence of suicidal thinking and behavior, and validated, may lead to more rigorous assessment and more accurate capture of serious suicide risk and further suggests that antidepressant treatment is generally safe and well tolerated.

Although the use of the CSSR-S has been applied to more recent trials, previous studies that have utilized a systematic assessment of suicidality have not yielded an excess signal for active treatment. Hammad et al. (2006) performed a meta-analysis of 17 individual antidepressant drug trials used in pediatric psychiatric diagnoses to evaluate for suicide. Although the bulk of the data analyzed in this report utilized spontaneously reported adverse events, a secondary analysis specifically focused on systematically collected depression rating scale suicide item scores.

Their findings showed that few of these trials had data “suggestive of excess worsening or emergence of suicidality for drug compared with placebo” and overall “no signal for excess suicidality for drug” when the rating scale suicidality items were analyzed, contrasting with the findings utilizing adverse events data. Their analysis of suicide item score revealed a relative risk for worsening suicidality of 0.92 and for emergence of suicidality of 0.93, neither of which reached statistical significance. They reached the conclusion that there does not seem to be an increased risk of treatment-emergent suicidality with use of active drug in pediatric MDD when the assessment is systematic rather than the previously more commonly used metric of spontaneously reported and nonsystematically collected adverse events (Hammad et al. 2006).

Clinical treatment recommendations, by nature, need to balance risk and benefit. A report from American Academy of Child and Adolescent Psychiatry (AACAP) suggests that depression contributes to >500,000 suicide attempts by children and adolescents annually. The average length of an untreated depressive episode in children and adolescents is often >12 months, and untreated depression predisposes to recurrent or chronic episodes. Over the long run, children and adolescents with depressive illness are likely to suffer considerable disability and impairment, high rates of comorbid disorders with poor academic functioning, difficulties in peer and family relationships, increases in substance use, legal problems, and attempted and completed suicide. These concerns highlight the importance of early recognition, diagnosis and active treatment, and the need for further research on interventions that potentially can be effectively and safely used with depressed youth (Zhou et al. 2015).

Conclusions

There continues to be controversy regarding the effectiveness of antidepressants in depressed youth. In the past decade, there have been seven placebo RCTs, with recent studies showing that newer agents, such as duloxetine and the selegiline transdermal patch, which are FDA approved for the treatment of adults, did not clearly demonstrate benefit above and beyond placebo. Evidence is emerging that methodological weaknesses may contribute to the persistent findings of high placebo response rates and limited differentiation of active treatment, especially in industry-supported research. The totality of available evidence continues to support the use of fluoxetine and escitalopram as first-line medication treatment for youth with depressive disorders and demonstrated effect to prevent relapse. Specifically, escitalopram showed significant reduction in depression in the acute and extension trial, and fluoxetine was demonstrated to prevent relapse 3.2 times more than placebo with long-term use.

In terms of safety, in the more recent studies, the systematic assessment of treatment-emergent suicidality utilizing newer rating scales specifically developed for this purpose reveals similar rates of treatment-emergent suicidality in patients on antidepressants as those on placebo, in contrast to increased suicide signal suggested by past research using adverse events data. Notably, there are relatively few studies using this methodology and higher sensitivity meta-analyses are not yet available, suggesting that such results must be interpreted with caution. However, taken in sum, the available evidence suggests that antidepressant treatment is generally safe and well tolerated in this age group and that the benefits of antidepressant treatment outweigh the risk of suicidal ideation/suicide attempts. Ongoing evaluation regarding the safety and efficacy of antidepressants in pediatric populations should continue to be a research priority.

Clinical Significance

In the last decade, there have been 7 RCT's assessing new medications for the treatment of pediatric MDD. Methodological differences may account for the high placebo rates in trials and the lack of statistically significant separation of active treatment and these may be thought of as “negative trials” rather than “failed trials.” The evidence continues to support the use of escitalopram and fluoxetine as first-line treatment, and that treatment continuation to prevent relapse is required. Importantly, antidepressants are relatively safely used with depressed youth and the risk of suicidal emergence is less than previously suggested.

Footnotes

Disclosures

B.W., research contract with Actavis Pharma, Inc.; research contract with Pfizer through Duke Clinical Research Institute; research contract with Baystate Health. M.J.I. has no competing financial interests.

References

Atkinson

, Prakash

, Zhang

, Pangallo

, Bangs

, Emslie

, March

: A double-blind efficacy and safety study of duloxetine flexible dosing in children and adolescents with major depressive disorder. J Child Adolesc Psychopharm, 24:18–19, 2014.

Bridge

, Birmaher

, Iyengar

, Barbe

, Brent

: Placebo response in randomized controlled trials of antidepressants for pediatric major depressive disorder. Am J Psychiatry, 2009:166, 42–49.

Bridge

, Iyengar

, Salary

, Barbe

, Birmaher Pincus

, Ren

, Brent

, et al.: Clinical response and risk for reported suicidal ideation and suicide attempts in pediatric antidepressant treatment: A meta-analysis of randomized controlled trials. JAMA, 297:1683–1696, 2007.

Cheung

, Kusumakar

, Kutcher

: Maintenance study for adolescent depression. J Child Adolesc Psychopharm, 18:389–394, 2008a.

Cheung

, Sacks

, Dewa

, Pong

, Levitt

: Pediatric prescribing practices and the FDA Black-box warning on antidepressants. J Dev Behav Pediatr, 29:213–215, 2008b.

Cipriani

, Zhou

, Del Giovane

, Hetrick

, Qin

, Whittington

, Coghill

, Zhang

, Hazell

, Leucht

, Cuijpers

, Pu

, Cohen

, Ravindran

, Liu

, Michael

, Yang

, Liu

, Xie

: Comparative efficacy and tolerability of antidepressants for major depressive disorder in children and adolescents: A network meta-analysis. Lancet, 388:881–890, 2016.

Copeland

, Shanahan

, Costello

, Angold

: Cumulative prevalence of psychiatric disorders by young adulthood: A prospective cohort analysis from the Great Smoky Mountains Study. J Am Acad Child Adolesc Psychiatry, 2011:50, 2525–2561.

Costelo

, Erkanli

, Angold

: Is there an epidemic of child or adolescent depression?. J Child Psychol Psychiatry, 2006:7, 1263–1871.

Cummings

, Druss

: Racial/ethnic differences in mental health service use among adolescents with major depression. J Am Acad Child Adolesc Psychiatry, 2011:50, 160–170.

10.

DelBello

, Hochadel

, Portland

, Azzaro

, Katic

, Khan

, Emslie

: A double-blind, placebo-controlled study of selegiline transdermal system in depressed adolescents. J Child Adolesc Psychopharm, 24:311–317, 2014.

11.

Emslie

, Kennard

, Mayes

, Nightingale-Teresi

, Carmody

, Hughes

, Rush

, Tao

, Rintelmann

: Fluoxetine versus placebo in preventing relapse of major depression in children and adolescents. Am J Psychiatry, 165:459–467, 2008.

12.

Emslie

, Prakash

, Zhang

, Pangallo

, Bangs

, March

: A double-blind efficacy and safety study of duloxetine fixed doses in children and adolescents with major depressive disorder. J Child Adolesc Psychopharm, 24:170–179, 2014.

13.

Emslie

, Ventura

, Korotzer

, Tourkodimitris

: Escitalopram in the treatment of adolescent depression: A randomized placebo-controlled multisite trial. J Am Acad Child Adolesc Psychiatry, 48:721–729, 2009.

14.

Emslie

, Wells

, Prakash

, Zhang

, Pangallo

, Bangs

, March

: Acute and longer-term safety results from a pooled analysis of duloxetine studies for the treatment of children and adolescents with major depressive disorder. J Child Adolesc Psychopharm, 25:293–305, 2015.

15.

Findling

, Robb

, Bose

: Escitalopram in the treatment of adolescent depression: A randomized, double-blind, placebo-controlled extension trial. J Child Adolesc Psychopharm, 23:468–480, 2013.

16.

Hammad

, Laughren

, Racoosin

: Suicidality in pediatric patients treated with antidepressant drugs. Arch Gen Psychiatry, 2006:63, 332–339.

17.

Hauser

, Hansen

, Enck

: Nocebo phenomena in medicine: Their relevance in everyday clinical practice. Dtsch Arztebl Int, 2012; 109:459–465.

18.

Nakonezny

, Mayes

, Byerly

, Emslie

: Predicting placebo response in adolescents with major depressive disorder: The Adolescent Placebo Impact Composite Score (APICS). J Psychiatr Res, 2015; 68:346–353.

19.

Olfson

, Blanco

, Wang

, Laje

, Correll

: National trends in the mental health care of children, adolescents, and adults by office-based physicians. JAMA Psychiatry, 2014:71, 81–90.

20.

Rutherford

, Wall

, Brown

, Choo

, Wager

, Peterson

, Chung

, Kirsch

, Roose

: Patient expectancy as a mediator of placebo effects in antidepressant clinical trials. Am J Psychiatry, 2017:174, 135–142.

21.

Thapar

, Collishaw

, Pine

, Thapar

: Depression in adolescence. Lancet, 2012:379, 1056–1067.

22.

Walkup

: Antidepressant efficacy for depression in children and adolescents: industry- and NIMH-funded studies. Am J Psychiatry, 2017:174, 430–437.

23.

Zhou

, Qin

, Whittington

, Cohen

, Liu

, Del Giovane

, Michael

, Zhang

, Xie

: Comparative efficacy and tolerability of first-generation and newer-generation antidepressant medications for depressive disorders in children and adolescents: Study protocol for a systematic review and network meta-analysis. BMJ Open, 2015:5, e007768.