Prescribing for depressed adolescents: Office decision-making in the face of limited research evidence


  • Michael Gordon,

    Corresponding author
    1. Child and Adolescent Stream, Early in Life Mental Health Service, Monash Medical Centre, Melbourne, Victoria, Australia
    • Correspondence: Dr Michael Gordon, Early in Life Mental Health Service, Monash Medical Centre, 246 Clayton Rd, Clayton, Vic. 3168, Australia. Fax: 03 9594 6333; email:

    Search for more papers by this author
  • Glenn A Melvin

    1. Department of Psychiatry, Centre for Developmental Psychiatry and Psychology, Southern Clinical School, Monash University, Melbourne, Victoria, Australia
    Search for more papers by this author

  • Conflict of interest: Glenn Melvin has received competitive research funding from the National Health & Medical Research Council and beyondblue: the National Depression Initiative. Dr. Gordon has been involved in a comparative treatment trial for adolescent depression in which the sertraline was provided free of charge for the study.

Major depressive disorder (MDD) is a relatively common, disabling, potentially life-threatening medical condition.[1] MDD is diagnosed when there are five or more symptoms occurring over 2 or more weeks.[2] Adult and adolescent symptoms of MDD are almost the same. However, irritable mood and lack of weight gain can qualify as depressive symptoms in adolescents.[2] The lifetime prevalence of unipolar depression (MDD or dysthymic disorder) in adolescence is 11.7%; three-quarters of this percentage suffer severe impairment.[3] Adolescent MDD has very high levels of comorbidity with other psychiatric disorders.[4] Depression is the condition most commonly implicated in adolescent and adult suicide.[5] Adolescent depression runs a remitting and relapsing course, often recurring later in adolescence or adulthood[6, 7] (see Fig. 1). Of those adolescents who recover from their depression, about 50% will have another depressive episode in the next 5 years.[6, 8] In summary, adolescent MDD is a serious and sometimes chronic medical condition that requires assertive treatment.

Figure 1.

Course of adolescent major depression. *No depressive symptoms for over 2 months.

A number of antidepressants have come onto the market since 1987 that provide fewer side effects at therapeutic dose and in overdose than the older tricyclic antidepressants.[9, 10] These newer agents include the selective serotonin re-uptake inhibitors (SSRIs; e.g. fluoxetine, sertraline, citalopram), the serotonin–norepinephrine re-uptake inhibitors (e.g. venlafaxine, duloxetine, desvenlafaxine) and the noradrenergic and specific serotonergic antidepressants (e.g. mirtazapine). The prescribing of antidepressants to children and adolescents has increased from 1996 to 2007.[11] Overall, 19% of Australian paediatricians and 68% of child psychiatrists self-report prescribing SSRIs.[12] However, this is very unlikely to be specifically for major depression, as paediatricians prescribe across a broad variety of diagnoses and at times for off-label conditions.

The SSRIs are a class of antidepressant that have a high affinity for blocking the presynaptic serotonin transport proteins but not the other monoamine transmembrane transport proteins.[13] SSRIs differ in their effects through secondary binding to non-serotonergic receptors,[14] the antidepressant effects of their metabolites, their excretion half-lives and their effect on the hepatic cytochrome P450 system.[15] Despite their differences, the available SSRIs are all effective treatments for moderate adult depression[16] and display similar overall side effect profiles.[15]

In 2004, evidence emerged from antidepressant trials conducted by pharmaceutical companies that there was an increase in suicidal thoughts and behaviours in children and adolescents given antidepressants. In October 2004, the U.S. Food and Drug Administration (FDA) ordered that pharmaceutical companies add a ‘black box’ warning on all antidepressant medication warning of the heightened risk of suicidality.[17] There have been several key guidelines, meta-analyses and papers published since the FDA black box warning that have added further information in an attempt to clarify the role of antidepressants in the treatment of depressed adolescents. Despite the increasing amount of information available, it is our contention that the available literature remains limited and at times contradictory about prescribing. A selective review of the key papers is provided with a discussion on the implication for practice-based management of depressed adolescents. The association of antidepressants and suicidal behaviour in adolescents is discussed elsewhere.[18]

Antidepressant Efficacy versus Effectiveness

Do antidepressant drugs work at all in adolescent depression? There is universal agreement that tricyclic antidepressants have no role in the treatment of adolescent depression. A Cochrane meta-analysis found that there was no difference in the rate of recovery in children and adolescents with depressive disorder between those given tricyclic antidepressants and those given placebo treatments.[19]

The randomised controlled trials (RCTs) published in English peer-reviewed journals for the treatment of adolescent or combined adolescent/child depression (n = 14) are shown in Table 1. The majority of the studies were industry-funded and were conducted in North America. Unpublished and non-English-language studies on nefazodone, citalopram, paroxetine, venlafaxine and mirtazapine have been noted elsewhere.[34-37] Fluoxetine is the drug with the most consistent positive response, in three out of the four studies. On the basis of evidence of efficacy from RCTs, the FDA has approved both fluoxetine, in 2003,[38] and escitalopram (the active enantiomer of citalopram), in 2009, for use in adolescent depression.[39] While seven studies show superiority of drug over placebo, the reported findings from some of the studies have been queried since publication. A study of sertraline by Wagner et al. (2003) that reported positive results had in fact pooled two studies with the same methodology.[25] When each of the sertraline studies is taken on its own there is no significance in either study.[18] Jureidini (2008) noted that the significant difference between paroxetine and placebo reported by Keller et al. (2001) was found using measures not cited in the original protocol (Hamilton Rating Scale for Depression depressed mood item, Kiddie Schedule for Affective Disorders and Schizophrenia – Lifetime depressed mood item, Clinical Global Impression).[40]

Table 1. Randomised controlled trials of newer antidepressants (non-tricyclic, non-monoamine oxidase inhibitors)
StudyAge (years)nDuration (weeks)Drug and doseFinding
  1. †Equals signs indicate no difference in efficacy; greater than signs relate to a significantly greater reported efficacy for the condition to the right of the sign as compared with that on the left. CBT, cognitive–behavioural therapy.
Simeon et al. (1990)[20]13–18408Fluoxetine 20–60 mg daily vs. placeboFluoxetine = placebo
Emslie et al. (1997)[21]7–17968Fluoxetine 20 mg daily vs. placeboFluoxetine > placebo
Mandoki et al. (1997)[22]8–17406Venlafaxine 37.5–75 mg daily vs. placeboVenlafaxine = placebo
Keller et al. (2001)[23]12–182758Paroxetine 20–40 mg vs. imipramine 300 mg daily vs. placeboParoxetine > placebo
Emslie et al. (2002)[24]7–172198Fluoxetine 10 to 20 mg daily vs. placeboFluoxetine > placebo
Wagner et al. (2003)[25]6–1737610Sertraline 50 to 200 mg daily vs. placeboSertraline > placebo
Wagner et al. (2004)[26]7–171788Citalopram 20 to 40 mg daily vs. placeboCitalopram > placebo
Emslie et al. (2006)[27]7–172068Paroxetine 10 to 50 mg daily vs. placeboParoxetine = placebo
Berard et al. (2006)[28]13–1828612Paroxetine 20–40 mg daily vs. placeboParoxetine = placebo
Von Knorring et al. (2006)[29]13–1824412Citalopram 10–40 mg daily vs. placeboCitalopram = placebo
Wagner et al. (2006)[30]6–172688Escitalopram 10–20 mg vs. placeboEscitalopram = placebo
TADS (2004)[31]12–1743912Fluoxetine and CBT 10–40 mg daily vs. fluoxetine only vs. CBT vs. placeboCombination > fluoxetine > CBT > placebo
Emslie et al. (2007)[32]7–173678Venlafaxine extended-release 37.5–225 mg daily vs. placeboVenlafaxine extended-release = placebo
Emslie et al. (2009)[33]12–173128Escitalopram 10–20 mg vs. placeboEscitalopram > placebo

It is possible that the outcomes reported from each of the RCTs may be influenced by the variables related to the study design, medication compliance, placebo response rate and the nature of the cohort recruited.[37, 41-43] The differences between the studies restrict direct comparison of their reported findings. The limitations of single RCTs point us to systematic reviews with meta-analyses that have the capacity to pool data and address the problem of small sample sizes.[44]


An early meta-analysis (n = 5 RCTs) reported a small effect size for antidepressants over placebo, with suggested negligible clinical benefit.[45] Tsapakis et al. (n = 29 RCTs including tricyclics) reported an overall number needed to treat (NNT) of 9, with a higher response rate in adolescents compared with children.[46] Hetrick, McKenzie and Merry (n = 12 RCTs)[47] and Whittington et al. (n = 5 RCTs)[48] reported a modest benefit for fluoxetine over placebo. In summary, these meta-analyses found small effect sizes that have been interpreted as reflecting limited to negligible clinical improvement. The most recent Cochrane review (n = 19 RCTs) of 3335 participants treated with newer antidepressants (paroxetine, sertraline, fluoxetine, citalopram, escitalopram, mirtazapine, venlafaxine)[35] found a small statistically significant difference in response/remission favouring antidepressants over placebo, but this finding was qualified as being not clinically meaningful.[35] Other authors came to different conclusions. Bridge et al. (n = 15 RCTs) found in their meta-analysis that the NNT for antidepressants in children and adolescent MDD was 10 but advocated for ‘the cautious and well monitored’ use of antidepressants.[37]

Other techniques have been used to assess antidepressant efficacy. Gibbons et al. (2012) pooled all the data from each participant in the fluoxetine RCT studies (n = 708) over the first 6 weeks of treatment, arguing this approach was superior to a meta-analysis as it utilised all the data available.[49] Gibbons et al. found that using fluoxetine compared with placebo, the NNT for response to treatment was 4.[49] Higher response and remission rates were reported for fluoxetine in depressed youth than adults.[49]

The overall findings suggest that there is evidence for the efficacy of fluoxetine and less evidence for that of other antidepressants in the short term. The various complex analyses, in the main, have been confusing, as different authors have argued for exclusion of selected RCTs from analysis depending on their orientation.[46, 50, 51] Further, meta-analysis itself is not an exact science and is open to some interpretation depending on which studies are included, the quality of the studies and the nature of the statistical analysis (fixed-effects vs. random-effects regression).[37, 52]

Antidepressants and Cognitive–Behavioural Therapy

Some studies that have compared combined cognitive–behavioural therapy (CBT) and antidepressant medication treatments to monotherapeutic approaches for adolescent depression have shown differences in the relative efficacy of CBT, antidepressants or combination after 3 months.[31, 53] However, there is convergence in efficacy across all the treatments (medication vs. CBT vs. combination) in the subsequent months[53, 54] and years.[8] In a meta-analysis of five RCTs, antidepressants (predominantly fluoxetine) have been shown to be equivalent to a combination of antidepressant and CBT at 26 and 36 weeks' follow-up.[55] These findings challenge the assumption that two treatments are ultimately better than one.

Switching Antidepressants

In contrast, for adolescents with MDD who do not respond to the initial SSRI, changing to another antidepressant (SSRI or venlafaxine), especially in conjunction with CBT, has been shown to result in an improvement in depressive symptoms in half of patients.[56] In this treatment-resistant group the change to a combination of CBT and antidepressant or another antidepressant led to 38.9% achieving remission at 24 weeks.[57]

Relapse Prevention

RCTs of antidepressant efficacy have typically aimed to examine short-term outcome (2 to 3 months). Adolescent depressive disorder runs a relapsing, remitting course into early adulthood.[6] There have been two placebo-controlled studies of children and adolescents with MDD looking at relapse prevention.[58, 59] Over 32 weeks' (n = 40) and 6 months' (n = 102) follow-up of patients who remitted from MDD, fewer patients relapsed on fluoxetine than placebo, and those on placebo relapsed more quickly than those on fluoxetine,[58, 59] providing preliminary evidence for the effectiveness of fluoxetine in reducing risk of MDD relapse. Kennard et al. conducted a pilot study of adolescents with MDD (n = 46) who had responded to fluoxetine, randomising them to CBT and fluoxetine or fluoxetine only over the next 6 months.[60] Those adolescents who were given CBT and fluoxetine were eight times less likely to relapse than those who received medication alone.[60] There is early evidence that in the continuation phase of treatment (6 to 12 months after the acute phase), there is a role for antidepressant and possibly CBT prophylaxis to prevent relapse.

Depression and the Family

Depression in children and adolescents is associated with parental depression.[61] This association is likely to be mediated by genetic susceptibility; exposure to parental depressive emotions, cognitions and behaviours; and increased familial stresses. Intervention studies to treat depressed parents have largely focused on maternal depression.[62] For depressed adolescents whose mothers are concurrently depressed, treating the mother's depression has been shown to improve internalising disorders in their offspring.[63] This improvement continues at 12-month follow-up.[64] These studies, taken together, point to the importance of looking at the whole family system and not just treating the adolescent in isolation.


There are several clinical practice guidelines that incorporate these research findings and expert opinion to assist paediatricians in the management of the depressed adolescent. The Guidelines for Adolescent Depression in Primary Care treatment (GLAD-PC) were published in 2007,[65] the beyondblue Clinical Practice Guidelines were released in 2010,[66] and the United Kingdom National Institute for Health and Clinical Excellence (NICE) guidelines were released in 2005[67] and updated in 2011 without change.

The NICE guidelines recommend that children and adolescents with moderate to severe depression should be offered, as first-line treatment, CBT, interpersonal therapy or short-term family therapy of at least 3 months' duration. Antidepressants should only be prescribed if the adolescent has moderate to severe depression, is assessed by a Child and Adolescent Mental Health Services tier 2 or 3 professional and is receiving concurrent psychological treatment. The beyondblue guidelines recommend that fluoxetine should be considered for reduction of depressive symptoms in the short term where psychological treatment has failed, is inaccessible or is refused.[66] The GLAD-PC guidelines recommend using an SSRI for moderate to severe depression, outlining dosing regiments for six different SSRIs.[65]

The NICE guidelines appear to be the most conservative in their recommendations and the GLAD-PC the most liberal. The recommendation for the use of antidepressants in the short term by the beyondblue guidelines is in line with the evidence base, which, in the main, has investigated short-term outcomes of depression.

Applying Research Findings in the Office

A number of factors limit the clinician's ability to apply RCT findings in his or her practice.[68] The RCTs described use placebos, are short-term and are not representative of those depressed adolescents who present in the clinician's office. Hetrick, McKenzie and Merry reported that the children and adolescents in RCTs were unlike those who presented for treatment, who were suffering with lower levels of comorbidity and suicidality.[47] The paediatrician often sees patients who would be excluded from most of the adolescent depression RCTs. For instance, there is scant information about the assessment of use of antidepressants in those with MDD and developmental disability,[69] leaving them therapeutic orphans. Further, the clinician does not have the option of using a placebo. If the likelihood of a short-term response is 61% on an antidepressant,[37] where does that leave the prescribing clinician? It is very likely that many clinicians will consider trialling an SSRI.

Seligman (1998) has made the point that the aims of clinical trials overlap with those of the practicing clinician, but they are not the same thing. Seligman notes that clinical trials are largely short-term, single-intervention studies, usually focusing on a single outcome (e.g. reduction in depressive symptoms).[70] By contrast, clinical work is, in the main, a long-term engagement, involving multiple treatments that are self-correcting.[70] The decision to medicate, while guided by the available research, is also directed by the adolescent's severity of symptoms, unique strengths and vulnerabilities, family factors, access to mental health services and individual wishes of the patient and his or her parents. The paediatrician is well placed to integrate all these factors.

Summary: What Can the Clinician Take from the Research?

The treatment guidelines for adolescent depression suggest that the role of antidepressants is as a second-line treatment for moderate to severe depression as part of stepped treatment where there has been inadequate response to psychological therapies. Of the antidepressants, fluoxetine has the strongest evidence of efficacy; however, the evidence is largely short-term. There is preliminary evidence to suggest that fluoxetine can prevent relapse. Tricyclics have no role in the treatment of adolescent depression. CBT may be added to an antidepressant; however, evidence of an additive effect is limited. Effectiveness studies are largely unavailable, and treatment response rates may vary in clinical practice. If an adolescent and his or her mother are both depressed, then both are likely to benefit if the mother is the one who takes and responds to an antidepressant.