kim

Seminar

1056 www.thelancet.com Vol 379 March 17, 2012

Lancet 2012; 379: 1056–67

Published Online February 2, 2012

DOI:10.1016/S0140- 6736(11)60871-4

Child & Adolescent Psychiatry Section, Department of

Psychological Medicine and Neurology (Prof A Thapar

FRCPsych, S Collishaw DPhil, A K Thapar PhD), and MRC

Centre for Neuropsychiatric Genetics and Genomics

(A Thapar, S Collishaw), School of Medicine, Cardiff University,

Cardiff , Wales, UK; Emotion and Development Branch,

NIMH-Intramural Research Program, Bethesda, MD, USA

(D S Pine MD); and Taff Riverside Practice, Cardiff , Wales, UK

(A K Thapar)

Correspondence to: Prof Anita Thapar, Child &

Adolescent Psychiatry Section, Department of Psychological

Medicine and Neurology, School of Medicine, Cardiff University, Heath Park, Cardiff CF14 4XN,

Wales, UK thapar@Cardiff .ac.uk

Depression in adolescence Anita Thapar, Stephan Collishaw, Daniel S Pine, Ajay K Thapar

Unipolar depressive disorder in adolescence is common worldwide but often unrecognised. The incidence, notably in girls, rises sharply after puberty and, by the end of adolescence, the 1 year prevalence rate exceeds 4%. The burden is highest in low-income and middle-income countries. Depression is associated with sub stantial present and future morbidity, and heightens suicide risk. The strongest risk factors for depression in adolescents are a family history of depression and ex posure to psychosocial stress. Inherited risks, develop mental factors, sex hormones, and psychosocial adversity interact to increase risk through hormonal factors and associated perturbed neural pathways. Although many similarities between depression in adolescence and depression in adulthood exist, in adolescents the use of antidepressants is of concern and opinions about clinical management are divided. Eff ective treatments are available, but choices are dependent on depression severity and available resources. Prevention strategies targeted at high-risk groups are promising.

Introduction Unipolar depressive disorder is a common mental health problem in adolescents worldwide,1 with an estimated 1 year prevalence of 4–5% in mid to late adolescence.2,3 Depression in adolescents is a major risk factor for suicide, the second-to-third leading cause of death in this age group,4 with more than half of adolescent suicide victims reported to have a depressive disorder at time of death.5 Depression also leads to serious social and educational impairments,6,7 and an increased rate of smoking, sub- stance misuse, and obesity.8,9 Thus, to recognise and treat this disorder is important.

Depression is defi ned as a cluster of specifi c symptoms with associated impairment. The clinical and diagnostic features of the disorder are broadly similar in adoles cents and adults (panel).10,11 The two main classifi cation systems (international classifi cation of diseases-10 [ICD-10] and the American diagnostic and statistical manual of mental disorders-IV [DSM-IV]) defi ne depres sion similarly, although DSM-IV makes one exception for children and

adolescents, whereby irritable rather than depressed mood is allowed as a core diagnostic symptom.12 Never- theless, depression in adolescents is more often missed than it is in adults,13 possibly because of the prominence of irritability, mood reactivity, and fl uctuating symptoms in adolescents. Depression can also be missed if the primary presenting problems are unexplained physical symptoms, eating disorders, anxiety, refusal to attend school, decline in academic performance, sub stance misuse, or behavioural prob lems.

In some respects depression in adolescents can be viewed as an early-onset subform of the equivalent adult disorder because of its strong links with recurrence later in life.14 The illness has similar clinical features and patterns of neural activity to that in adults, and its occurrence is also associated with a family history of the disorder.11 However, important diff erences exist between the two disorders, particularly in treatment response,

Search strategy and selection criteria

We searched PubMed (Medline and life science journals) for articles in English using the search terms “adolescent depression” and “depression” (restricted to children), combined with additional search terms, specifi cally “epidemiology”, “gender”, “puberty”, “diagnosis”, “comorbidity”, “aetiology”, “life events”, “psychosocial”, “family”, “trends”, “genetics”, “gene-environment interaction”, “temperament”, “resilience”, “dysthymic disorder”, “treatment”, “medication”, “CBT”, “prevention”, “ imaging”, “cortisol”, “longitudinal”, and “outcomes”. We also examined published reviews in the past decade, National Institute for Health and Clinical Excellence guidelines on adolescents with depression, American Academy of Child and Adolescent Psychiatry practice guidelines, and Cochrane systematic reviews. We focused on papers from the past 5 years, but also included papers published in the past 10 years (if evidence was sparse), or published at any time (if they were particularly important). Nearly all the published papers were from high-income countries.

Panel: Criteria for ICD-10 depressive episode

Core symptoms (at least two must be present) • Depressed mood present for most of the day and almost

every day • Loss of interest or pleasure in activities • Decreased energy or increased susceptibility to fatigue

Associated symptoms • Loss of confi dence or self-esteem • Unreasonable feelings of self-reproach or excessive

inappropriate guilt • Recurrent thoughts of death or suicide, or any suicidal

behaviour • Diminished ability to think or concentrate • Change in psychomotor activity, agitation, or retardation • Sleep disturbance • Change in appetite with corresponding change in weight

At least four of these symptoms must be present for 2 weeks to diagnose a mild depressive episode, six to diagnose a moderate depressive episode, or eight for a severe depressive episode.

ICD-10=international classifi cation of diseases-10.

Seminar

www.thelancet.com Vol 379 March 17, 2012 1057

with strongly divided opinions about best treatment prac- tices.10 Depression in prepubertal children is less common than depression in adolescents or adults, and seems to diff er from these disorders with respect to some causative, epidemiological, and prognostic features.15,16

We focus on unipolar depression in adolescents. When evidence is available, we focus on depressive disorder rather than its symptoms. However, in some instances, the only available data are based on studies in which depressive symptoms are reported. Such cases are noted, although there are generally strong similarities in re search fi ndings for depressive symptoms and depressive disorder.

Epidemiology The prevalence of depression in children is low (<1% in most studies)17 with no sex diff erences, and then rises substantially throughout adolescence.18 Many factors could explain the recorded post-pubertal rise in preva lence because adolescence is a developmental period charac- terised by pronounced biological and social changes.19,20 The most commonly postulated contributors are puberty and brain and cognitive maturation. They include enhanced social understanding and self-awareness,21 changes in brain circuits involved in responses to reward and danger, and increased reported stress levels, especially in girls.22,23 Median 12-month prevalence estimates in mid to late adolescence are roughly similar to those seen in adulthood (4–5%),2,3 with the cumulative probability of depression rising from around 5% in early adolescence to as high as 20% by the end of that time.24,25 However, population prevalence estimates vary widely across studies (table) and in diff erent countries,1,29,30 possibly because of methodological diff erences.

Patterns of service provision vary so much between studies and countries that comparisons of clinic-referred samples are not meaningful. Although some researchers suggest higher rates of depression in adolescents in low- income and middle-income countries,29 no consistent evidence of such systematic variation exists.1,29,30 Never- theless, because of the young age distribution in low- income and middle-income countries, the overall number of aff ected adolescents will be substantially greater in these countries than in high-income countries.1 Despite this high burden, research has focused almost exclusively on high-income countries.

One of the most robust epidemiological fi ndings is the emergence of a strong female preponderance (about 2:1) in the prevalence of depression in adolescents after puberty.31 The excess of aff ected girls is seen in epidemiological as well as clinical samples, and is robust across diff erent methods of assessment. Sex diff erences in rates of depression are therefore unlikely to be merely due to diff erences in help-seeking or reporting of symptoms. Although the reasons for this post-pubertal- onset sex diff erence are not fully understood, adolescent depression is more closely tied to female hormonal changes than to chronological age, which suggests that

depression is directly linked to pubertal changes in hormone–brain relations.32 However, hormonal changes alone rarely produce the behavioural or neural signs of depression,33 and are more likely to contribute by sensitising the brain to the harmful eff ects of stress.31,34,35 Work in animals shows that oestrogen increases the stress response in the prefrontal cortex.36 Thus, the post- pubertal sex diff erence in depression might in part result from increased exposure to stressors and hormonally- linked heightened stress sensitivity in girls. Finally, although depression is generally more common in girls, to recognise it in boys is important. Moreover, in some subgroups (eg, patients with neurodevelopmental and medical disorders) this sex diff erence might be greatly weakened, absent, or even reversed.37

Diagnostic interview

Sample size

Prevalence % (SE* or 95% CI†) Incidence

Current Lifetime

National comorbidity survey26

15–24 years (Data for 15–16 shown) CIDI 1769 7‡ (1·8*) 14·6 (2·6) ··

Female CIDI ·· 12·4 23·4 ··

Male CIDI ·· 1·5 5·7 ··

Great Smoky Mountains study27

11 years CAPA 901 1·9§ (1·0–3·5†) ·· ··

13 years CAPA 833 2·6§ (1·3–5·3†) ·· ··

15 years CAPA 1136 3·7§ (2·3–6·0†) ·· ··

16 years CAPA 1101 3·1§ (1·8–5·3†) ·· ··

Oregon adolescent depression project6

14–18 years K-SADS 1709 2·9¶ 20·4 7·8

Female K-SADS ·· 3·4 24·8 10·4

Male K-SADS ·· 2·0 11·6 4·8

British child and adolescent mental health survey28

11–12 years DAWBA 1901 0·4‡ (0·14*) ·· ··

13–15 years DAWBA 2624 1·9‡ (0·28*) ·· ··

Dunedin longitudinal study25

11 years DISC-C 653 1·1|| 1·1 1·07

Female DISC-C ·· 0·3 ·· 0·3

Male DISC-C ·· 1·8 ·· 1·8

13 years DISC-C 2·1|| 3·2 2·1

Female DISC-C ·· 2·2 ·· 2·2

Male DISC-C ·· 2·1 ·· 2·1

15 years DISC-C 2·8|| 5·7 2·5

Female DISC-C ·· 4·4 ·· 4·4

Male DISC-C ·· 1·2 ·· 0·6

18 years DIS 16·8|| 20·7 15·0

Female DIS ·· 23·2 ·· 20·7

Male DIS ·· 10·8 ·· 9·6

CIDI=composite international diagnostic interview. CAPA=child and adolescent psychiatric assessment. K-SADS=schedule for aff ective disorders and schizophrenia for school-age children. DAWBA=development and well-being assessment. DISC-C=diagnostic interview schedule for children-child version. DIS=diagnostic interview schedule. ‡Current=28–30 days. §Current=3 months (any depressive disorder). ¶Current=point. ||Current=1 year. Timeframes refer to period over which symptoms of depression were assessed.

Table: Prevalence, incidence, and sex diff erences (when reported) for major depressive disorder from large population cohort studies of adolescents by age (years)

Seminar

1058 www.thelancet.com Vol 379 March 17, 2012

Changes in taxonomy and assessment methods pre clude strong conclusions about whether or not the true prevalence of depressive disorder in adolescents has changed over time.38 Certainly, the use of depression services and recognition of the disorder have increased, as has treatment in many countries.39 Youth suicide rates have fl uctuated over the same period.40 One meta-analysis found no evidence of an increase in rates of depressive disorder in adolescents between 1965 and 1995.3 However, comparisons of identical symptom screens across repeated epidemio- logical cohorts have shown evidence of increasing rates of depression and anxiety symptoms in some countries.38

Clinical outcomes of depression in adolescents Longitudinal studies of community and clinic-based population samples suggest that 60–90% of episodes of depression in adolescents remit within a year.41,42 How- ever, in follow-up studies 50–70% of patients who remit develop subsequent depressive episodes within 5 years.41,43 Clinic-referred patients generally fare less well than those treated in the community.6 Moreover, in adults very few individuals show complete symptomatic and functional recovery between depressive episodes, with most reporting residual symptoms or impairment.44,45

Finally, depression in adolescence also predicts a range of mental health disorders in adult life—notably, anxiety disorders, substance-related disorders, and bipolar dis- order,46–48 as well as suicidal behaviour, unemployment,48 and physical health problems.49 Thus, an episode of depression during adolescence often heralds a chronic or relapsing disorder, and forecasts a broad range of psychosocial diffi culties and ill health.

Cause and pathogenenesis Because of the clinically heterogeneous and diverse causes of the illness, to understand the pathogenesis of depression in adolescents is challenging. Like many other common health disorders, several risk factors interact to increase the risk of depression in a probabilistic way. To assess the contribution of any single risk factor in isolation and to identify crucial developmental periods when exposure is especially risky is diffi cult because many individual, family, and social risks are strongly correlated and relate to continuing and later adversities.

Distal risks such as inherited factors and adversity in early life might directly and indirectly predispose to depression. Such risks could mediate their eff ects through temperament and personality attributes (nega- tive emotionality, decreased positive emotionality and attentional control, behavioural inhibition, and neur- oticism) and cognition.50 However, consistent evidence is sparse as to whether such characteristics are risk factors, mediators, or consequences of depression.

Distal factors, together with hormonal and maturational changes can also alter individual sensitivity to stressors. These risk factors in turn aff ect proximal biological risk mechanisms such as changes to the neuroendocrine

system and brain structure and function, although no one pathway has been identifi ed as causal. Prevention and treatments have been targeted at reduction of early and later adversities, modifi cation of ways of thinking and feeling, and reduction of core depression symptoms by alteration of biological pathways with medication.

Familial and genetic risk Off spring of parents who have depression face three to four times increased rates of depression compared with off spring of healthy parents.51 Both genes and non- inherited factors contribute to this risk.51,52 Although exposures to maternal depression in the prenatal and postnatal periods have been regarded as especially important,53 unlike in animals, evidence that risks to off spring are restricted to these crucial periods is inconsistent. Longitudinal research,54,55 studies of chil- dren who are genetically unrelated to their mothers,52,56 and treatment trials of maternal depression57 suggest that the risk eff ects of early maternal depression are mediated through exposure to later, continued maternal symptoms or associated psychosocial adversity.58 Al though the importance of paternal mental health is now starting to be recognised,59 the long-term eff ects on depression during adolescence have yet to be inves tigated.

Inherited factors also play a part. Most twin studies show that depression becomes increasingly heritable from childhood (zero to low heritability) to late adolescence (modest heritability, around 30–50%).60 The heritability rates during late adolescence are similar to those seen in adult life.60,61 Inherited liability for depression manifests in diff erent ways and at diff erent times. Family and twin studies60 suggest that anxiety and depression share inherited liability, but anxiety in childhood tends to precede later depression during adolescence. Some genetic risks are shared with disruptive behaviour problems.62,63 Several workers have suggested that adolescent-onset depression, like other early-onset forms of illness, such as breast cancer, represents a more strongly genetic subform of depres- sion,64 but consistent evidence is scarce.65

Some studies have focused on identifying specifi c genes that increase susceptibility for unipolar depression. So far, no replicated, signifi cant fi ndings have emerged.66,67 These negative fi ndings for unipolar depression contrast with those for other psychiatric disorders such as schizophrenia and bipolar disorder, perhaps because genes operate in more complex ways for depressive disorder than we currently understand, or because opti- mum genetic research strategies have not been used.

Psychosocial risk factors Much research has been done into the association between depression and environmental factors such as exposures to acute stressful events (eg, personal injury, be reavement)68,69 and chronic adversity (eg, maltreat ment, family discord, bullying by peers, poverty, physical

Seminar

www.thelancet.com Vol 379 March 17, 2012 1059

illness). Nevertheless, such exposures do not always lead to the development of depression in adolescents, although those at high genetic risk seem to be especially susceptible to the eff ects of such stressors.70 Stressful life events seem more strongly associated with fi rst onset rather than recurrence of depression,71 and risk is considerably greater in girls,29,72 and in adolescents who have multiple negative life events than those exposed to one event.71 Chronic, severe stressors that aff ect relationships seem most important. Negative family relationships,73,74 peer victimisation through bullying,75 and maltreatment are common risks for depression.

Some children exposed to the traumatic events of war, torture, displacement, orphanhood, or HIV infection76,77 might develop long-term psychiatric disorders, especially if there is a family history of such disorders, or they are exposed to multiple traumas.78 Although recorded associ- ations between psychosocial stressors and depres sion could show the eff ects of unmeasured confounders and reverse causation, fi nd ings with diff erent research designs consistently show genuine causal risk eff ects.

Gene–environment interplay Inherited factors seem to contribute to depression in adolescents in two ways—not only by directly increasing risk, but also indirectly through gene–environment interplay, specifi cally by increasing sensitivity to adversity (gene–environment interaction) and by increasing the probability of exposure to risky environments (gene– environment correlation).60,79,80 Several twin and family studies suggest that adolescents (especially girls) at high inherited and familial risk of depression show increased sensitivity to psychosocial risk factors (gene–environment interaction), such as stressful life events81 and family adversity,60,80 and are the ones most likely to be exposed to such risks.80

Do specifi c gene variants moderate the eff ects of adversity on brain function and clinical outcome? Many reports suggest that a variant (5-HTTLPR) in the sero- tonin transporter gene might increase risk of depression, but only in the presence of adverse life stressors or early maltreatment.82,83 The fi ndings are less robust in adoles- cent boys than in girls.83 Importantly, this same gene variant has also been reported to aff ect fear-related and danger-related brain circuitry that is altered in depression; specifi cally it is associated with increased amygdala activation in healthy individuals in response to viewing fearful faces.84 However, such fi ndings in relation to depression and depression-related brain mechanisms seem to vary not only by genotype, but also by age, sex, and severity of symptoms, and are also reliant on good quality measures of adversity and depression. Although one meta-analysis failed to support the 5-HTTLPR gene-environment interaction fi nding,85 it was criticised on many grounds,83,86 and a re- analysis of data showed sig nifi cant evidence of the importance of gene-environment interactions, especially

in the context of maltreatment or a medical illness.87 Thus, although thought to be controversial by some researchers,85 these fi ndings are promising.

Brain and neuroendocrine mechanisms Both inherited factors and psychosocial stressors can be thought of as distal risk factors, which determine the risk for depression by shaping more proximal biological mechanisms such as activity in underlying neural circuits and endocrine systems.88 The similarity of fi ndings in adolescents and adults suggests that core biological features are shared.

Work in animals not only has shown that the immature, adolescent brain is more susceptible than the adult brain to the eff ects of these genetic and environmental infl uences,89,90 but has also identifi ed specifi c brain regions where changes in circuitry occur.90 Moreover, these regions seem to be the same ones implicated in human brain imaging studies. Two interrelated neural circuits and associated modulatory systems have been closely linked to risk for depression, both in adults and in adolescents. These circuits are active in the response to danger and learning about rewards.91,92

One circuit connects the amygdala to the hippocampus and ventral expanses of the prefrontal cortex (PFC) and is linked to hypothalamic-pituitary-adrenal (HPA) axis activ- ity. Activity in this circuit consistently seems raised in patients with major depression, and both established and novel treatments for depression could work through reduction of its activity.93 Some of these neural changes are also seen in healthy, at-risk individuals, such as in those with particular temperament types (behaviourally inhibited)94,95 and those at high familial and genetic risk for depression.94 Perturbations in this circuit link depression to stress-related enhancements in HPA-stress systems, such as higher than expected cortisol concentrations,96,97 and activity in the serotonergic system.

Genetic factors, psychosocial stress, sex hormones, and development have also been linked to changing activity in this circuit,84,94 with evidence that this circuit matures after adolescence. High concentrations of sex steroid receptors have been identifi ed within this circuit,23 and might provide a biological mechanism for why girls have higher risk of depression than boys.

The other key circuit implicated in depression encompasses the striatum and its connection to both the PFC and ventral dopamine-based systems. Like the fi rst circuit, this one also continues to mature through adolescence. Sex diff erences emerge in both circuits.98 Research into this reward circuit implies that reduced activity is linked with expression of and risk for depression. Reduced striatal and PFC activity during tasks involving rewards has been recorded both in individuals with major depression and in those with depressed parents.99 Both inherited factors and stress-related perturbations seem to contribute to these changes.100

Seminar

1060 www.thelancet.com Vol 379 March 17, 2012

Resilience Many children who are at high-risk for depression through familial predisposition and exposure to adversity do not develop the disorder.101 Research into depression resilience has the potential to identify targets for prevention of depression. Individual factors that have been reported to protect against the development of depression in high-risk adolescents include inherited factors and high intelligence, as well as potentially modifi able factors such as emotion-regulation capacities, coping mechanisms, and thinking styles.102–105 The most consistent fi ndings relate to the protective eff ects of

good quality interpersonal relation ships, suggesting that improvement of these relationships could be a target for prevention of depression. Children with a high familial risk of depression tend to have better mental health if their relationships with their parents are characterised by warmth, acceptance, low hostility, and low parental control.102,105 The broader social context also needs to be considered because quality of peer support seems especially predictive of resilience to depression in the context of child maltreatment and maternal depression.103–105

However, factors that enhance resilience to depression vary considerably between diff erent groups of children and across diverse risk contexts, which means that prevention strategies will probably need to be carefully tailored for maximum benefi t.101 Also, studies have tended to focus on correlates of resilience, and underlying mechanisms remain poorly understood. Whether or not targeted interventions aimed at promoting children’s resilience to depression prove successful needs to be assessed.

Detection and diagnosis A diagnostic approach is needed when decisions about treatment of adolescents with depression have to be made. This approach allows clinicians to apply evidence from treatment trials, make decisions on the risk-benefi t ratio for medication, and rationalise referral to expensive specialist resources (if available). Depression can also be thought of as lying along a continuum.106 Subsyndromal depression in which adolescents have high levels of symptoms that do not meet the diagnostic threshold for depressive disorder predicts concurrent impairment and future risk.46,106 Longitudinal studies of adolescents with sub-syndromal depression show that they are at increased risk of later full-blown depressive disorder.43,46,107 Thus, there are important reasons for paying attention to sub- syndromal depression, and targeting individuals with these symp toms for prevention, low-risk intervention strategies, and lifestyle changes.108

Since depression is under-diagnosed, professionals caring for adolescents need to be aware of the possi- bility of this diagnosis, particularly in high-risk groups. Figure 1 shows a suggested process for detecting adoles- cents with depression in primary care, child health clinics, schools, and social care settings. WHO has launched an important initiative to improve the worldwide diagnosis and management of certain key conditions, and a strategy for screening and management of depression has been identifi ed.109 Targeted screening of high-risk individuals rather than universal screening of the general population is recom mended,110 but the value of screening in diff erent settings has yet to be critically assessed.

Questionnaires provide an economical and quick method of screening. The two-item patient health questionnaire (PHQ-2)111 is a brief measure widely used to screen for depression in adults in primary care, and

Figure 1: Detection of depression in adolescents in non-specialist settings *If patient scores <2, generally no further action is needed.

High-risk groups Parent or sibling with depression Exposure to chronic adversity—eg, history of abuse, bullying Previous episode of depression Disruptive behaviour, substance abuse Other psychiatric symptoms (eg, anxiety) Learning problems Deliberate self-harm/suicide attempt Frequent general practitioner or hospital attender Chronic physical health problem, especially neurological problems Multiple physical symptoms (eg, abdominal pain)

High level of suspicion, especially if in high-risk group or if parent or guardian has concerns

Administer two-item patient health questionnaire (PHQ-2)

If uncertain about validity

Score ≥2*

Administer detailed questionnaire assessment (eg, short mood and feelings questionnaire [sMFQ], children’s depression inventory (CDI), or nine-item patient health questionnaire [PHQ-9])

Score above cutoff point (>11 for sMFQ, ≥11 for CDI, or ≥10 for PHQ-9) or suicidal intent

Clinical assessment ideally by trained clinician (eg, primary care physician) to identify symptoms of depression, suicidal thoughts, impairment at home, school, and other settings, family history and exposure to psychosocial adversities, history from other informants, mental state assessment; check for comorbidities

High questionnaire scores but not clinically depressed based on clinical assessment

1 Monitor severity of symptoms, suicidal ideation and social and educational functioning in primary care or child health clinic

2 Provide non-specialist-psychosocial intervention (see figure 2)

Clinically depressed

Uncomplicated mild depression—see figure 2 Moderate to severe depression, or complications—see figure 2

Seminar

www.thelancet.com Vol 379 March 17, 2012 1061

studies have shown it also works well for adolescents.112,113 The full 25-item strength and diffi culties questionnaire (SDQ) provides additional screens for attention defi cit hyperactivity disorder (ADHD) and disruptive behaviour symptoms.114 Other measures focus on depression alone and include the short mood and feelings questionnaire,115 and the children’s depression inventory.116 These ques- tionnaires can be used as an additional screen in adolescents if there is some uncertainty about the reliability of the responses to the PHQ-2 or SDQ items, and as an adjunct to monitor change (fi gure 1).

For individuals who score above the cutoff scores a full clinical assessment should be undertaken to reliably diagnose depression. This assessment consists of sensitive, empathic questioning about symptoms of depression (panel), their duration, severity, and any associated impairment through a direct interview with the adolescent. It might be necessary to interview them separately from accompanying adults for part of the consultation. Information provided by a parent or guardian is important because diagnoses based on the accounts of several informants show greater reliability and validity.117 The risk of suicide and how the adolescent functions at home and school should be assessed. Possible warning signs include a decline in academic achievement or engagement in leisure activities, changes in peer and family relationships, and social withdrawal. The best strategy for clinical assessment by non-specialists has yet to be determined. Training programmes to improve detection and diagnosis of depression have not proved consistently successful. An alternative strategy would be to assess the use of standardised structured diagnostic interviews such as the freely available development and well being assessment.118

Diff erential diagnoses and comorbidities Two diagnoses in classifi cation systems that are characterised by prominent depressive symptoms and associated impairment but that do not meet the criteria to diagnose major depression are adjustment disorder and dysthymic disorder. Research into the validity and treatment of these disorders is sparse. Adjustment disorder is short-lived, arises within 3 months of the onset of a stressor and does not persist longer than 6 months after the desistance of the stressor. Dysthymic disorder describes a pattern of chronic symptoms of depression that are present for most of the time on most days with a minimum duration of 1 year for children and adolescents. Sometimes individuals with dysthymic disorder have superimposed episodes of major depression onto their usual symptoms.

Although bipolar disorder and schizophrenia are much less common in adolescents than is unipolar depression (prevalence is <0·1–1% for both),119–121 they are important diff erential diagnoses to be remembered. Both disorders can be preceded or accompanied by depression or symptoms of depression, but they have other key features—such as episodic elated or irritable mood or enhanced energy in bipolar disorder—that do not occur

in major depression. Treatment is very diff erent for the three conditions. Patients with bipolar disorder can also present, like patients with depression, with agitation and irritability. Chronic non-episodic irritability which is a core feature of a recently defi ned syndrome of severe mood dysregulation is not the same as bipolar disorder, but seems to be a strong predictor of future depression.122

Two-thirds of adolescents with depression have at least one comorbid psychiatric disorder and 10–15% have two or more comorbidities.28 Adolescents with depression are six to 12 times more likely to have anxiety, four to 11 times more likely to have a disruptive behaviour disorder, and three to six times more likely to have a substance misuse problem than are adolescents who are not depressed, with adjustment for multiple comorbidities.123 For example, almost 20% of adolescents with depression also meet diagnostic criteria for generalised anxiety disorder,124 with reported lifetime rates of 50–70%.125 Depression can also complicate eating disorders, autistic spectrum disorders, and ADHD.

Comorbidity is especially increased in adolescents with severe depression,126 and predicts severe impairment, poor long-term outcome,126 and complicates treatment.127 The high level of comorbidity can be explained in part by shared risk factors for multiple disorders, but could also arise if comorbid disorders are risks or consequences of depressive disorder. When a patient with a depressive disorder also has another psychiatric disorder, the depression should be treated in its own right, with the understanding that additional interventions will be needed to address the specifi c comorbidity. Patients with medical illnesses such as diabetes or HIV/AIDS can have comorbid depression. The value of distinguishing between primary and secondary depres sion in the context of another disorder is not clear because there is growing evidence that the relation between medical illness and depression is often bi-directional,128 and treatment of depression is still needed. The extent to which diff erent comorbidities might be associated with heterogeneity in cause, prognosis, and treatment remains unclear.

Treatment Treatment data in adolescents have been reviewed (fi gure 2).10,129 Three important issues are highlighted. First, treatment choices are not the same in adolescents as in adults. Second, best treatment practice is contro- versial because accepted practice and clinical guidelines vary in diff erent countries, and because of concerns about the use of antidepressant drugs in patients younger than 18 years, with some recommendations based on consensus rather than on evidence. Third, the evidence relates to the short-term eff ectiveness of psychological treatments and medication. Evidence for the long-term benefi ts of treatment to rates of recurrence and for the eff ectiveness of non-specialist interventions is scarce.

The two best studied psychological treatments are cognitive behavioural therapy (CBT) and interpersonal