Show Summary Details
Page of

Defining Depression and Bipolar Disorder 

Defining Depression and Bipolar Disorder
Defining Depression and Bipolar Disorder
Page of

PRINTED FROM OXFORD MEDICINE ONLINE ( © Oxford University Press, 2021. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in Oxford Medicine Online for personal use (for details see Privacy Policy and Legal Notice).

date: 08 December 2021

The development of adolescent mood disorders involves a complex, multifactorial model (e.g., Akiskal & McKinney, 1975; Cicchetti, Rogosch, & Toth, 1998; Kendler, Gardner, & Prescott, 2002). No single risk factor accounts for all or even most of the variance. The most likely causal model will include individual biological and psychological diatheses that interact with various environmental stressors. There is little question that early onset is highly related to recurrence in adulthood, whether the data derive from clinical samples (Kovacs, Akiskal, Gatsonis, & Parrone, 1994), long-term population studies (Kessler and Walters, 1998), studies of high school students (Lewinsohn, Rohde, & Seeley, 1998), or studies of depressed patients (Rao et al., 1995). Over 50% of depressed adolescents had a recurrence within 5 years (Birmaher et al., 1996; Lewinsohn, Rohde, Seeley, Klein, & Gotlib, 2000), although only a small portion continues to have significant psychopathology in any one year. The few studies of depressed adolescents followed into adulthood show strong continuity between adolescent and adult depression (Frombonne, Wostear, Cooper, Harrington, & Rutter, 2001; Harrington, Fudge, Rutter, Pickles, & Hill, 1990; Weissman, Wolk, Goldstein et al., 1999; Weissman, Wolk, Wickramaratne, et al., 1999) and an increased risk of suicide attempts as well as psychiatric and medical hospitalization. Studies of prepubertal depression also show continuity into adolescence (Kovacs & Gatsonis, 1994). The most serious outcome is suicide, which is the third leading cause of death among adolescents. Other outcomes include lack of social development and skills, withdrawal from peers, poor school performance, less than optimal career and marriage choices, and substance abuse (Frost, Reinherz, Pakiz-Camra, Fiaconia, & Lekowitz, 1999; Rao et al., 1995; Weissman, Wolk, Goldstein, et al., 1999).

This chapter reviews the epidemiology and definitions of mood disorders in children and adolescence. The psychological, social, and biological factors that have been shown to increase the risk of mood disorders in children and adolescents are also discussed.


For many years, children and adolescents were thought incapable of experiencing depression, according to the psychoanalytic concept of the underdeveloped superego. Thus depression was considered “an adult disease.” However, case reports from as early as the 17th century described adolescents exhibiting symptoms resembling those observed in adults with depressive disorders. In 1975, the National Institute of Mental Health (NIMH) convened a meeting of thought leaders to discuss the incidence and diagnosis of depression among children. This meeting, followed by the publication of a book by Shulterbrant and Ruskin (1977), finally made clearer the diagnosis and the existence of depression acceptable in this population.

The last two decades have witnessed a burgeoning database on the age of onset of mood disorders. Major depressive disorder (MDD) is no longer seen primarily as a disorder of the middle-aged and elderly. Epidemiologic and clinical research from the United States and elsewhere has clearly documented that the age of first onset of major depression is commonly in adolescence and young adulthood and that prepubertal onsets, while less common, do occur. It is now clear that adolescent depression is a chronic, recurrent, and serious illness. The offspring of depressed parents, compared with children of nondepressed parents, have an over 2-to 4-fold increased risk of depression. Depressions occurring in adolescents share similar features with those of depression at other ages. Across ages, symptom patterns are similar; rates among females are higher (2-fold risk); there is high comorbidity with anxiety disorders, substance abuse, and suicidal behaviors; and high social, occupational, and educational disability can accompany depression (Angold, Costell, & Erkanli, 1999; Costello et al., 2002). In contrast, childhood MDD tends to be male predominant, mood reactive, and commonly associated with high levels of irritability and dysphoria and tends to have very heavy comorbidity with the disruptive behavior disorders (Biederman, Faraone, Mick, & Lelon, 1995; Leibenluft, Charney, Towbin, Bhangoo, & Pine, 2003).

The epidemiologic data on childhood and adolescent bipolar disorder are considerably sparser than those for MDD, in part because of the earlier erroneous belief that bipolar disorder begins in adulthood. It is also often quite difficult to assess boundaries between normal mood and mood irritability in youth, especially in community studies, and the first signs of bipolar disorder are frequently uncertain (Nottelman & Jensen, 1998). Most evidence on juvenile bipolar disorder comes from clinical samples in which efforts have been made, especially recently, to characterize early clinical presentations of bipolar disorder.

Unfortunately, until recently, persons under age 18 were excluded from epidemiological studies. Thus empirically based information on prevalence, risk factors, course, and treatment is scanty, especially for bipolar disorder. This situation is finally changing, but not rapidly enough; the consequences of mood disorders on future development in school, work, and marriage and on the next generation are often profound. This chapter will highlight the empirical basis for understanding the epidemiology, phenomenology, course, and comorbidity of MDD and bipolar disorder in youth. Because a sharp distinction between childhood and adolescent onset cannot be readily made, information on childhood (prepubertal-onset) disorder will be included when relevant.


The same criteria defined in the Diagnostic and Statistical Manual of Mental Disorders (4th ed., with text revisions) (DSM-IV-TR) (American Psychiatric Association, 1994) to diagnose MDD in adults are used to diagnose MDD in adolescents (Table 1.1). Five or more of the following symptoms must be present nearly every day during the same 2-week period to diagnose an adolescent with a major depressive episode:

  • Depressed or irritable mood most of the day

  • Markedly diminished interest or pleasure in almost all activities, most of the day

  • Significant weight loss or gain, or change in appetite; failure to gain expected weight

  • Sleep disturbance

  • Psychomotor agitation or retardation

  • Fatigue or loss of energy

  • Feelings of inappropriate guilt or hopelessness

  • Indecisiveness or diminished ability to concentrate

  • Recurrent thoughts of death or suicidal ideation, suicide attempt

Table 1.1 Symptoms of Depressive Disorders




Anxiety, anhedonia, melancholia, depressed or sad mood, irritable or cranky mood


Loss of interest in daily activities, feelings of hopelessness and helplessness, suicidal thoughts, suicidal acts or attempts


Difficulty concentrating, feelings of worthlessness, sense of guilt, low self-esteem, negative self-image, delusions or psychosis


Preference for time alone, easily angered or agitated, oppositional or defiant


Sleep disturbance, appetite change, lost or gained weight, energy loss, psychomotor agitation and retardation, lack of energy, decreased libido


Physical or bodily complaints, frequent stomachaches and headaches

At least one of the following two symptoms must be present: depressed or irritable mood, or markedly diminished interest or pleasure in almost all activities. These symptoms must cause clinically significant impairment in social, occupational, or other important areas of functioning. They cannot be due to the direct physiological effect of substance abuse or a general medical condition. Also, the symptoms should not be better accounted for by bereavement or schizoaffective disorder. A major depressive episode cannot be superimposed on schizophrenia, schizophreniform disorder, delusional disorder, or a psychotic disorder not otherwise specified.

More precisely, MDD can be rated as mild, moderate or severe; with or without psychotic symptoms; in full or partial remission. Depression should be diagnosed as chronic when the episode lasts more than 2 consecutive years. Furthermore, if loss of pleasure in almost all activities or lack of reactivity to usually pleasurable stimuli exists, the depression may be stated to have melancholic features. In addition, at least three of the following are required for melancholia:

  • Depressed mood, which must be distinctly different from one felt from death of a loved one

  • Morning depression being worse than that during the day or night

  • Waking up several hours earlier than normal

  • Evident psychomotor retardation or agitation

  • Significant weight loss or anorexia

  • Inappropriate or excessive guilt

Two of the following must be present to classify a depressive episode as having catatonic features:

  • Motor immobility, catalepsy, or stupor

  • Motor overactivity that is purposeless and not in response to external stimuli

  • Extreme negativism or mutism

  • Voluntary movement peculiarities such as posturing, grimacing, stereotypy, and mannerisms

  • Echolalia or echopraxia

It is sometimes difficult but also important to establish the seasonality of the mood disorder because a major depressive episode can present initially as seasonal affective disorder in children and adolescents. To establish the presence of a true seasonal mood disorder, there must be a regular temporal relationship between the mood disorder (depression or mania) and a particular time of the year. A full remission or switching from depression to mania must occur within that particular time of the year. The adolescent also needs to experience two episodes of mood disturbance during the last 2 years and the seasonal episodes should greatly outnumber nonseasonal episodes. Seasonal mood disorder is often missed in adolescents because it is often attributed to the stress of starting of a new school year in the fall. Postpartum depression in female adolescents is considered when the onset of depression is within 4 weeks of childbirth.

Another often undetected diagnosis in adolescents is dysthymia, which is defined in adolescents as depressed or irritable mood that must be present for a year or longer and the youth must never be symptom-free for more than 2 months. In addition, two or more of the following symptoms must be present:

  • Change in appetite

  • Change in sleep

  • Decrease in energy

  • Low self-esteem

  • Difficulty making decisions or poor concentration

  • Feelings of hopelessness

Similar to depression, dysthymia should not be diagnosed if it is a direct result of substance abuse or a general medical condition, or if it occurs during the course of a psychotic disorder such as schizophrenia. Moreover, if a major depressive episode is the first psychiatric disorder in an adolescent or the person has a history of manic, hypomanic, or mixed episodes, dysthymia should not be diagnosed.

Because dysthymia often starts in childhood, adolescence, or early adult life, it is often referred to as a “depressive personality disorder.” Dysthymic disorder is considered chronic and if the age of onset is prior to 21, it is classified as early onset. Attention-deficit hyperactivity disorder (ADHD), conduct disorder (CD) specific developmental disorder, and a chaotic home environment are some of the more frequent predisposing factors for dysthymia in children and adolescents. Kovacs and associates (1984) have reported that dysthymic children are at risk for developing depression and mania on follow-up.

Adolescents who have dysthymic disorder and subsequently develop a major depressive episode are considered to have a “double-depression.” When dysthymia coexists with disorders such as anorexia nervosa, anxiety disorder, rheumatoid arthritis, somatization disorder, or psychoactive substance dependence, it is referred to as “secondary dysthymia.” In addition, adolescents can also exhibit atypical depressive features. Atypical features include mood reactivity with two or more of the following for a period of at least 2 weeks:

  • Significant weight gain or increase in appetite

  • Increased sleep

  • Feelings of heaviness in arms or legs

  • A pattern of long-standing rejection sensitivity that extends far beyond the mood disturbance episodes and results in significant social or occupational impairment. Atypical features are quite common among depressed adolescents.

Clinically it can be challenging to discern the difference between MDD and dysthymia in children and adolescents. However, with careful history taking with the child and the primary caregiver, this can be accomplished (Table 1.2).

Table 1.2 Comparison of Major Depressive Disorder and Dysthymic Disorder

Major Depressive Disorder

Dysthymic Disorder

Dysphoric mood

Dysphoric mood

Symptoms severe

Symptoms mild to moderate

Impaired functioning, common

Impaired functioning, less common

Psychosis may be present

No psychosis

Symptoms present every day

Symptoms usually fluctuating

Symptoms present every day for 2 weeks

Symptoms on and off for 1 year

Differentiating Prepubertal and Adolescent-Onset Major Depression

There are compelling reasons to differentiate between prepubertal-and adolescent-onset MDD (see Angold, Costello, & Worthman, 1998; Kaufman, Martin, King, & Charney, 2001). Although the frequency of MDD before puberty is not well established, it is hardly uncommon. Some estimates suggest that it may afflict as many as 2% of children at any one current period. Childhood-onset MDD tends to be male predominant, is commonly associated with irritability, and frequently is comorbid with disruptive behavior disorders (Biederman et al., 1995; Costello et al., 1996; Kessler, Foster, Webster, & House, 1992; Rutter, 1996). Some studies of prepubertally depressed children have not found continuity into adulthood (Harrington et al., 1990; Weissman, Wolk, Goldstein, et al., 1999), whereas others have documented such continuity into adolescence (Kovacs et al., 1994). Prepubertally depressed children often develop a variety of psychiatric disorders in adulthood, especially increased rates of bipolar, anxiety and substance use disorders (Kovacs, 1998, 1990, 1996; Kovacs et al., 1984; Kovacs 1998).

There is good evidence to suggest that the onset of puberty as measured by Tanner stage and hormonal levels, rather than by chronological age per se, predicts the increase in onset of MDD in girls. Angold et al. (1998) studied 4,500 boys and girls, ages 9, 11, and 13, over 3 years who were sampled from the Great Smoky Mountains region of North Carolina. At each interview, assessments of major depression and pubertal status with Tanner staging (Tanner, 1962) were undertaken. Pubertal status, not chronological age at onset, was a better predictor of the emergent preponderance of major depression in girls. Consistent with the epidemiologic data, boys had a higher rate of MDD at prepubertal Tanner Stage I, with girls increasing and surpassing boys after Tanner Stage III.



Epidemiologic data from large community surveys in the United States on the incidence of MDD among children and adolescents are sparse. This is in part due to the long-held view that MDD was rare before adulthood or was a self-limiting and normal part of growing up. In addition, there has been controversy over the means of assessing young people and over who is the best informant, the child or the parent. A few surveys of adolescents have used self-report depression symptom scales assessing 1 week to 6 months prevalence. Rates based on established adult cutoff scores for clinically significant current depression range between 20% and 30% (Offord et al., 1987; Reinherz, Giaconia, Hauf, Wasserman, & Silverman, 1999; Wickstrom, 1999). However, self-reported symptom scales do not differentiate between mild and severe mood disorders, type of mood disorder, or other psychiatric disorders. Prevalence rates with self-report scales are generally considerably higher than those found in studies using diagnostic assessments.

Published epidemiologic studies of adolescents have been limited to school districts and high schools in one community (Lewinsohn, Hops, Roberts, Seeley, & Andrews, 1993; Whittaker et al., 1990) or limited geographic areas (Cohen et al., 1993; Costello et al., 1996), or have been conducted outside the United States in Canada (Flemming, Offord, & Boyle, 1989) or New Zealand (Fergusson & Woodward, 2002). With few exceptions (Flemming et al., 1989; Lewinsohn, Hops, et al., 1993); the samples of adolescents have usually been too small, under 1,000 and usually under 500, to be reliable estimates. The diagnostic methods and age groups of the adolescents vary widely among studies. The current lifetime prevalence rates of MDD from these studies have been estimated to be about 5%. The similarity between lifetime rates in adolescents and adults suggests that a large percentage of those with major depression have onset while young.

The most comprehensive epidemiologic data in adults come from the National Comorbidity Survey (NCS; Kessler & Walters 1998), a nationally representative sample of over 8,000 persons from U.S. households ages 15 to 54 (Kessler & Walters, 1998). Although only 600 persons under age 18 were included in this sample, the rates from this U.S. population are consistent with published rates on adolescents. The lifetime prevalence for 15-to 18-year-olds was about 14% and an additional 11% were estimated to have a lifetime history of minor depression, with higher rates among females than among males. While the NCS did not sample persons under age 15, the sample was young enough so that reasonably good information from retrospective reports of age of first onset of MDD in childhood or adolescence could be obtained. Kaplan-Meier age-at-onset curves for major and minor depression in the NCS are presented in Figures 1.1 (major depression) and 1.2 (minor depression). Both curves show that meaningful risk begins in the early teens and continues to rise in a roughly linear fashion within groups of cohorts through the mid-20s (Kessler, Avenevoli, & Merikangas, 2001). The general shape of these curves is very similar to that of the onset curves reported in other epidemiologic studies of adolescent depression (e.g., Lewinsohn et al., 1998). The peak rise in rates in the late teens and early 20s is also consistent with the mean age of onset reported in the cross-national studies of adults (Weissman et al., 1996). Both curves show evidence of substantial prevalence increases in cohorts born af ter the mid-1960s. A cohort effect (that is, secular changes in rates) has also been shown in cross-national epidemiologic studies of adults, spanning a considerable older age range (Cross-National Collaborative Group, 1992; Klerman & Weissman, 1989). These statistics need to be reconciled with data derived from clinical samples documenting that more than 30% of children referred to clinical centers suffer from major depression and that in many of these cases, the disorder starts in the preschool years. Moreover, recent reports from student health services on college campuses note a marked increase in requests for counseling for depression over the last decade and list suicide as the second leading cause of death among students (Voelker, 2003). In summary, there is good evidence that the first onset of MDD is frequently in adolescence and not uncommonly in childhood, and that the rates of MDD, especially in the young, have been increasing.

Figure 1.1 Kaplan-Meier cumulative lifetime prevalence curves for major depression in the total National Comorbidity Survey sample by cohort.

Figure 1.1
Kaplan-Meier cumulative lifetime prevalence curves for major depression in the total National Comorbidity Survey sample by cohort.

Figure 1.2 Kaplan-Meier cumulative lifetime prevalence curves for minor depression in the total National Comorbidity Survey sample by cohort.

Figure 1.2
Kaplan-Meier cumulative lifetime prevalence curves for minor depression in the total National Comorbidity Survey sample by cohort.


Comorbidity with other psychiatric disorders in youth, as in adults, is the rule rather than the exception (see Angold et al., 1999 for a comprehensive review). Anxiety disorders are the most common, with over 60% of depressed adolescents having a history or a concomitant anxiety disorder. A frequent pattern of onset includes anxiety disorder, particularly phobias before puberty, with an emergence of major depression in adolescents (Pine, Cohen, Gurley, Brook, & Ma, 1998). Disruptive behavior disorders are frequent and emerge before puberty. Substance abuse in late adolescence with MDD is also common. Comorbidity with medical conditions in adolescents has been less well studied. However, a few studies have found an association between adolescent depression and obesity (Pine, Goldstein, Wolk, & Weissman, 2001); headaches (Pine, Cohen, & Brook, 1996) and asthma (Mrazek, Schuman, & Klinnert, 1998), as well as an increased risk of medical hospitalizations and accidents (Kramer et al., 1998).

Risk Factors

Information on risk factors for adolescent MDD comes both from epidemiologic and clinical studies. The two most consistent risk factors for MDD in both studies of adolescents and adults are female gender (2-to 3-fold increased risk) and a family history of MDD. The offspring of depressed parents are at 2-to 4-fold increased risk of MDD, an earlier age of onset, and recurrent episodes (Hammen, Burge, Burney, & Adrian, 1990; Weissman, Warner, Wickramaratne, Moreau, & Olfson, 1997; Weissman et al., 2004). The risk is transmitted across generations to grandchildren (Warner, Weissman, Mufson, & Wickramaratne, 1999).

Other risk factors that contribute to both the onset and recurrence of adolescent MDD are adverse family environments characterized by absence of supportive interactions; poor parental bonding; poor primary attachments; and harsh discipline (Fendrich, Warner, & Weissman, 1990; Garber & Little, 1999; Sheeber, Hops, & Davis, 2001). Separating out the effects of parental MDD from the risk factor studied is problematic, because parental MDD is frequently associated with the risk factors (e.g., divorce, poor parental bonding). One study of offspring at high and low risk of depression found that parental depression was the strongest risk factor for offspring depression, over and above other family risks, such as divorce or poor parental bonding. The rates of MDD were considerably lower in the offspring of nondepressed parents (low risk), but when MDD was present in the low-risk offspring, it was associated with poor parental bonding, parent–child conflict, and parental divorce (Fendrich et al., 1990; Nomura, Wickramaratne, Warner, Mufson, & Weissman, 2002).

Personality and Temperament

Several theorists have hypothesized a heritable trait vulnerability factor common to most, if not all, emotional disorders. This trait has been defined slightly differently and given various labels by different theorists, including harm avoidance (Cloninger, 1987), neuroticism (Eysenck, 1947), trait anxiety (Gray, 1982), behavioral inhibition (Kagan, Reznick, & Snidman, 1987), and negative affectivity (Watson & Tellegen, 1985), although the conceptual and empirical overlap among these constructs far outweighs the differences. Each implies a trait disposition to experience negative affect. The term neuroticism is often used to refer to this trait, and is consistent with the emergence of the Big Five model of personality as the dominant model of personality structure in children (e.g., Digman & Inouye, 1986; Digman & Shmelyov, 1996), adolescents (e.g., Digman, 1989; Graziano & Ward, 1992), and adults (e.g., Goldberg, 1992; McCrae & Costa, 1987).

Longitudinal studies have shown that neuroticism predicts later negative affect and symptoms of emotional distress (Costa & McCrae, 1980; Larson, 1992; Levenson, Aldwin, Bosse, & Spiro, 1988), even after controlling for initial symptom levels (Gershuny & Sher, 1998; Jorm, Christensen, Henderson, & Jacomb, 2000). Clark, Watson, and Mineka (1994) reviewed several longitudinal studies showing that neuroticism predicts both subsequent diagnoses and chronicity of major depression. Since this review, studies reported by Hayward, Killen, Kraemer, and Taylor (2000), Kendler and colleagues (Kendler, Kessler, Neale, Heath, & Eaves, 1993; Kendler et al., 2002; Roberts & Kendler, 1999), and Krueger et al. (1996) have each obtained results consistent with the conclusions of Clark et al. (1994). For example, in a large adult female twin sample, Kendler et al. (1993) found that neuroticism predicted the onset of MDD over a 1-year period, and recently, Kendler et al. (2002; Kendler, Kuhn, & Prescott, 2004) tested a multifactorial model and showed that, after stressful life events, neuroticism was the strongest predictor of the onset of major depression.

The relation between neuroticism and depression may vary somewhat by age. Hirschfeld et al. (1989) found that whereas among 31-to 41-year-old individuals neurotic-like characteristics of decreased emotional strength, increased interpersonal dependency, and increased thoughtfulness predicted the first onset of depression, this was not the case for younger 17-to 30-year-old individuals. Similarly, Rohde, Lewinsohn, and Seeley (1990) found that adult participants who experienced a first episode of MDD had exhibited elevated levels of dependent traits 2 to 3 years earlier. Rohde, Lewinsohn, and Seeley (1994), however, found no differences with regard to prior levels of dependency between adolescents who later developed a first MDD and adolescents who were depression-free during a 1-year follow-up period.

In contrast, studies using other measures of neurotic-like traits in children have found evidence of a link with vulnerability for depression. Elevated levels of behavioral inhibition have been observed in laboratory tasks with young offspring of depressed parents (Kochanska & Kuczynski, 1991; Rosenbaum et al., 2000). Caspi, Moffit, Newman, and Silva (1996) reported that children who had been rated as inhibited, socially reticent, and easily upset at age 3 had elevated rates of depressive disorders at age 21. Similarly, van Os, Jones, Lewis, Wadsworth, and Murray (1997) found that physicians' ratings of behavioral apathy at ages 6, 7, and 11 predicted both adolescent mood disorders and chronic depression in middle adulthood. Finally, Gjerde (1995) reported that gender may moderate the relation between temperament and mood disorders. Whereas females with higher levels of chronic depression during young adulthood had been described as shy and withdrawn at 3 to 4 years of age, males with chronic depression exhibited higher levels of undercontrolled behaviors as young children. Thus, there is some evidence of an association between neurotic-like traits during childhood and subsequent depression, though it may depend on gender as well as how these traits are measured.

Neuroticism also has been found to be a risk factor for other forms of psychopathology, however, and thus it is not specific to mood disorders. For example, neuroticism has been shown to be a risk factor in the development of posttraumatic stress disorder (PTSD) following trauma (Breslau & Davis, 1992; Breslau, Davis, & Andreski, 1995; Helzer, Robins, & McEvoy, 1987). Behaviorally inhibited children are at greater risk for the development of multiple phobias and various anxiety disorders in later childhood (Biederman et al., 1990; Hirshfeld et al., 1992) and social phobias in adolescence. Hayward et al. (2000) also found that neuroticism predicted the development of panic attacks in a 4-year prospective study in adolescents. Thus, neuroticism appears to be a significant predictor of depression, though it might not be a specific vulnerability marker. Moreover, it is still difficult to distinguish among common cause, precursor, and predisposition models of the relation between neuroticism and depression (Klein, Durbin, Shankman, & Santiago, 2002).

Cognitive Vulnerability

According to cognitive theories of depression (Abramson, Metalsky, & Alloy, 1989; Abramson, Seligman, & Teasdale, 1978; Beck, 1967), depressed individuals have more negative beliefs about themselves, the world, and their future, and tend to make global, stable, and internal attributions for negative events. These negative cognitions are expected to be both concurrently associated with depression and to contribute to the onset and exacerbation of depressive symptoms. Cognitive theories of depression are inherently concordant with diathesis-stress theories. When confronted with stressful life events, individuals who have such negative cognitive tendencies will appraise the stressors and their consequences negatively, and hence are more likely to become depressed than are individuals who do not have such cognitive styles.

Several types of cognitions have been proposed to be related to depression, including low self-esteem, negative automatic thoughts, dysfunctional attitudes, and cognitive distortions (Beck, 1967); self-control (Rehm, 1977); control-related beliefs and self-efficacy (Bandura, 1977); depressive attributional style (Abramson et al., 1978); hopelessness (Abramson et al., 1989); and a ruminative response style (Nolen-Hoeksema, 2000). Cross-sectional studies with clinic and community samples of children have consistently shown a significant relation between negative cognitions, particularly low self-esteem and a pessimistic attributional style, and depression (Garber & Hilsman, 1992). Meta-analyses of studies reporting on attributional style and depression have demonstrated moderate to large effect sizes in cross-sectional studies suggesting a strong concurrent association between negative attributional style and higher levels of depressive symptoms in children and adolescents (Gladstone & Kaslow, 1995; Joiner & Wagner, 1995).

Longitudinal investigations of the role of cognitions in the prediction of childhood depression have yielded varying results. Global self-worth (Allgood-Merton, Lewinsohn, & Hops, 1990; Garber, Martin, & Keiley, 2002; Vitaro, Pelletier, Gagnon, & Baron, 1995) and perceived self-competence in specific domains (Hoffman, Cole, Martin, Tram, & Seroczynski, 2001; Vitaro et al., 1995) have predicted child and adolescent depressive symptoms (e.g., Allgood-Merton et al., 1990; Vitaro et al., 1995) and diagnoses (Garber, Martin, & Keiley, 2002), controlling for prior levels of depression. However, these same cognitive constructs also failed to predict depressive symptoms (Dubois, Felner, Brand, & George, 1999) and onset of new episodes. However, in one of these null studies, participants were selected from a drug and alcohol treatment clinic. The mean depression score in this sample was lower at the second assessment. Treatment procedures may have reduced depression levels during the assessment interval, making it difficult to predict maintenance or exacerbation of depression.

Attributional style generally has been investigated in the context of stress, though several studies have tested main-effects models or reported main effects in the absence of interactions. Significant prospective relations have been observed between attributional style and later depressive symptoms in children and young adolescents (Nolen-Hoeksema, Girgus, & Seligman, 1986; 1992; Panak & Garber, 1992), though a few studies have failed to find this relationship. In a longitudinal study of the developmental trajectories of negative attributions and depressive symptoms, Garber, Keiley, and Martin (2002) showed that attributional styles that were increasingly negative across time were associated with significantly higher initial levels and increasing growth of depressive symptoms during adolescence.

Prospective studies in children and adolescents have also found support for the cognitive diathesis–stress model of depression (Dixon & Ahrens, 1992; Hilsman & Garber, 1995; Lewinsohn, Joiner, & Rohde, 2001; Nolen-Hoeksema et al., 1992; Panak & Garber, 1992). Using different stressors (grades, peer rejection, and school transition) and different time periods, Garber and colleagues showed in three (Dixon & Ahrens, 1992; Panak & Garber, 1992; Robinson, Garber, & Hilsman, 1995) different short-term longitudinal studies that cognitions (attributions, self-worth) measured before the stressors occurred moderated the effect of the stressors on depressive symptoms in children. Among children who experienced high levels of stress, the relation between negative cognitions about the self or causes of events and depressive symptoms was stronger than in those without such negative cognitions. Lewinsohn et al. (2001) found that among adolescents who had experienced negative life events, intermediate levels of dysfunctional attitudes predicted the onset of depressive disorders a year later.

Developmental theorists (Nolen-Hoeksema et al., 1992; Weisz, Southam-Gero, & McCarty, 2001) have suggested that negative cognitions emerge over time and that their relation with depression becomes stronger with development. For example, in a longitudinal study of children in grades 3 through 8, Nolen-Hoeksema et al. (1992) showed that attributional style alone and in conjunction with stress significantly predicted depressive symptoms in the older but not in the younger children. Similarly, in a cross-sectional comparison of children in grades 4, 6, and 8, Turner and Cole (1994) found that negative cognitions contributed to the prediction of depressive symptoms for the oldest children, but not for the two younger groups. Thus, the relation between the cognition–stress interaction and depressive symptoms appears to be increasing from middle childhood to early adolescence.

If negative cognitions contribute to the development of mood disorders, then “high-risk” offspring of depressed parents should be more likely to exhibit a cognitive vulnerability than children whose parents have not experienced mood disorders. Indeed, children of depressed mothers report significantly lower perceived self-worth and a more depressive attributional style than do children of well mothers (Garber & Robinson, 1997). Thus, children who are at risk for depression, but who have not yet experienced depression themselves, have been found to report a more negative cognitive style that likely represents a vulnerability to later depression.

In summary, correlational, predictive, and offspring studies have provided evidence that there is a cognitive style that represents a vulnerability to depression in children. This cognitive style involves beliefs about the self and explanations about the causes of negative events. Future studies need to examine the development of this cognitive vulnerability over time, and whether it needs to be primed in children (Ingram, Miranda, & Segal, 1998).


Common to all definitions of stress is a focus on environmental conditions that threaten to harm the biological or psychological well-being of the individual (Grant et al., 2003). Stress may occur either as an acute event or as chronic adversity, and as a major life event or as minor events with accumulated effects (either additive or multiplicative) (Grant et al, 2003). Stressful events may be normative (e.g., school transition) or patho logical (e.g., abuse) and may be independent of, or directly related to and thus dependent on, an individual's actions. Objective environmental consequences of a stressor (i.e., can be reliably rated by objective observers) are hypothesized to have a direct effect on the development of depression. The subjective threat of a stressor involves individuals' appraisals of an event as stressful, which then may impact their psychological well-being (Lazarus, Delongis, Folkman, & Gruen, 1985). Finally, there may be specificity in the relation between stress and psychopathology such that certain subdomains of stressors may be more highly related to depression than others (Beck, 1967; Grant et al., 2003).

Stress plays a prominent role in most theories of depression, and a clear empirical link exists between stressful life events and depression in children and adolescents (Compas, Grant, & Ey, 1994). In infants, depressive symptoms have been associated with stressful life circumstances and often are responsive to changes in the environment (Moreau, 1996). One stressor particularly linked with depression in infants is separation. Spitz and Wolf (1946) noted that a common feature in depressed infants ages 6 to 8 months is separation from the mother. Separation in young children has been found to be associated with grief responses characterized by negative changes in sleep patterns, activity, heart rate, temperature, monoamine systems, immune function, and endocrine function (Kalin & Carnes, 1984). Spitz noted the phenomenon of hospitalism, referring to evidence that infants subjected to long hospital stays experienced a number of psychological difficulties. Longer and more frequent hospital stays and earlier age of entering the hospital were associated with more depressive symptoms in infants (Moreau, 1996).

In school-aged children, cross-sectional studies using either life events checklists or interview methods consistently have shown that depressive symptoms and disorders are significantly associated with both minor and major undesirable life events in children, particularly cumulative or chronic stressors, and negative life events are more prevalent among depressed than nondepressed children (Goodyer, Wright, & Altham, 1988). Cross-sectional studies, however, are not informative about the direction of the relation between stress and depression. Given the association between dependent stressors and depression (Garber, Martin, & Keiley, 2002), it is possible that depression contributes to the occurrence of stressors. Depressed individuals have been found to generate many of the stressors they encounter, and these stressors then serve to exacerbate and maintain the depressive symptoms (Bennett, Pendley, & Bates, 1995).

Laboratory animal studies have shown that antenatal stress impacts the developing physiology of the fetus and later physiological and behavioral outcomes in the offspring of stressed rat and primate mothers. These studies are reviewed in part in this section and in further detail later in this chapter. Henry, Kabbaj, Simon, Le Moal, and Maccari (1994) showed that prenatally stressed rat pups had an elevated corticosterone response to novel environments and reduced hippocampal corticosteroid receptor density, suggesting that prenatal stress may affect the neurobiological development of systems associated with depression. Behaviorally, rat pups stressed in utero had greater distress and defensive behavior (Takahashi, Baker, & Kalin, 1990) and reduced environmental exploration when exposed to aversive or stressful conditions (Poltyrev, Keshet, Kay, & Weinstock, 1996).

Prepartum exposure to stress also may result in hyperresponsiveness to later stressors. Clarke and associates (Clarke & Schneider, 1993; Clarke, Wittwer, Abbott, & Schneider, 1994) randomly assigned pregnant rhesus monkeys to stress and control conditions. The prenatally stressed offspring were less likely than control offspring to play and explore the environment and were more likely to engage in clinging, which is associated with distress in primates. Clarke and Schneider suggested that hypothalamic–pituitary–adrenal (HPA) axis activity was implicated in the hyperresponsiveness to later environmental stressors of prenatally stressed rhesus monkeys.

Thus, these data from animal models and additional data (see below) indicate that stress that occurs as early as at conception likely contributes to an increased vulnerability to depression. In human infants, stress during pregnancy is associated with negative outcomes for offspring (Lou et al., 1994). Although the mechanisms by which stress impacts the developing fetus are still unknown, it is hypothesized that fetal neurophysiological development may be sensitive to the intrauterine hormonal environment, and neurophysiological vulnerability (e.g., HPA axis dysregulation) may make these offspring more sensitive to stress and thereby predispose them to depression as they mature. Recently, Stowe and colleagues (unpublished observations) noted that infants of women depressed during pregnancy or postpartum had significantly greater salivary cortisol responses to a standardized stressor than offspring of normal control women or women with a past history of depression who were not depressed during the index pregnancy/postpartum period.

Longitudinal studies in which stressors are assessed prior to the onset of symptoms can be informative about the temporal relation between stress and depression. Prospective studies have found that stress predicts depressive symptoms, controlling for prior symptom levels in children (Goodyer, Herbert, & Altham, 1998) and adolescents (Allgood-Merten et al., 1990). The relationship is stronger when children's self-reports are used than when parents' reports of children's depressive symptoms are used (Stanger, McConaughy, & Achenbach, 1992).

Fewer studies have examined the contribution of negative life events to the onset of depressive disorders in children. Stress has predicted the onset of depressive symptoms in previously asymptomatic children (Aseltine, Gore, & Colten, 1994) and the onset of clinically significant depressive episodes, controlling for prior symptom levels in samples comprised of both children and adolescents (Hammen, 1991) and adolescents alone (Garber, Keiley, et al., 2002). Only three of these studies (Aseltine et al., 1994; Garber & Kaminski, 2000; Monroe, Rohde, Seeley & Lewinsohn, 1999) controlled for lifetime history of MDD to rule out the possibility that earlier depressive disorder contributed to onset.

Reports of stressful life events have been shown to increase for both boys and girls from childhood through adolescence, with increases being greater for girls (Ge, Longer, Lorenz, & Simons, 1994), paralleling increases in rates of depression for boys and girls (Hankin et al., 1998). However, few studies have found that gender moderates the relationship between stress and depression. Cohen (1987) reported that negative events predicted depressive symptoms in girls who had experienced minimal positive events in the same time interval, and Ge et al. (1994) showed that growth of stressful life events over time predicted growth in depressive symptoms for girls but not boys.

Although no one specific type of stressful event invariably leads to depression in children and adolescents, certain stressors consistently have been found to be associated with depression. Childhood abuse or maltreatment is an especially robust predictor of depression (Andrews, 1995; Bifulco, Brown, & Adler, 1991; Trad, 1994), and this is particularly true for women (Weiss, Longhurst, & Mazure, 1999). Sexual assault during childhood or adulthood has been found to increase the risk of depression by 2.4 in women (Burnam, Stein, Golding, Siegel, Sorenson, Forsythe, et al., 1988). Poverty also has been shown to contribute to vulnerability to depression (Bruce, Takeuchi, & Leaf, 1991; Grant et al., 2003; McLoyd, 1998). For example, the rates of depression among low-income mothers are about twice as high as in the general population (Bassuk, Buckner, Perloff, & Bassuk, 1998). Recently, Caspi et al. (2003) elegantly demonstrated the relationship between a genetic variable, polymorphism of the serotonin transporter (SERT), and the development of depression after exposure to child abuse. Individuals with the l/l form of the SERT were immune to the depressogenic affects of child abuse, whereas those with s/s allele were highly likely to develop depression after child abuse.

Events such as disappointments, loss, separation, and interpersonal conflict or rejection are also linked with depression (Aseltine et al., 1994; Goodyer, Herbert, Tamplin, & Altham, 2000; Panak & Garber, 1992). This is especially probable for individuals who tend to be more socially dependent or sociotropic. According to the specific vulnerability hypothesis (Beck, 1983, Blatt, Quinlan, Chevron, McDonald, & Zuroff, 1982), individuals whose self-esteem is derived from interpersonal relationships (sociotropy) are at increased risk for depression when they experience stressors within the social domain; in contrast, those who derive their self-worth from achievement-related goals are at greater risk for depression when they encounter occupational failure. Studies investigating this specific vulnerability hypothesis in children have been supportive (Garber & Kaminski, 2000).

In summary, a clear link exists between stress and depression. But by what mechanisms does stress increase an individual's vulnerability to depression? Although stressors often precede mood disorders, not all individuals exposed to stressors become depressed. Thus, there is not a perfect correspondence between exposure to negative life events and the onset of depressive symptoms. How individuals interpret and respond to events and how resilient they are also differentiates who does and does not become depressed. Some of the individual variability is due to differences in appraisals of the meaning of the events with regard to the self and future.

Interpersonal Relationships

Interpersonal perspectives on depression emphasize the importance of the social environment and the development of secure attachments. Vulnerability to depression presumably arises in early family environments in which the children's needs for security, comfort, and acceptance are not met. Bowlby (1980) argued that children with caretakers who are consistently accessible and supportive will develop cognitive representations, or “working models,” of the self and others as positive and trustworthy. In contrast, caretakers who are unresponsive or inconsistent will produce insecure attachments leading to working models that include abandonment, self-criticism, and excessive dependency. Such working models may contribute to the development of negative cognitions about self and others, and presumably increase individuals' vulnerability to depression, particularly when exposed to new interpersonal stressors.

Reviews of the literature on the relation between the family environment and depression (Beardslee, Versage, & Gladstone, 1998; Rapee, 1997) indicate that families of depressed individuals are characterized by problems with attachment, communication, conflict, cohesion, and social support, as well as poor childrearing practices. Security in attachments helps infants cope with the environment and a lack of such attachments may lead infants to seek protection by withdrawing from the environment altogether (Bowlby, 1980; Trad, 1994). Two-year-old children with secure attachments have been found to be more cooperative, persistent, and enthusiastic, show more positive affect, and function better overall than those with insecure attachments (Matas, Arend, & Sroufe, 1978). In adolescents, depression has been linked with less secure attachments to parents (Kenny, Moilanen, Lomax, & Brabeck, 1993). Moreover, adolescents undergoing stressful life events are more likely to become depressed if they had insecure attachments to their parents than adolescents with more secure attachments (e.g., Kobak, Cole, Ferenz-Gillies, Fleming, & Gamble, 1993).

Beyond attachment, other kinds of dysfunctional family patterns have been found to be associated with depression in children (Rapee, 1997). Serious abuse and neglect interfere with normal expressions of infants' emotions and lead to avoidant or resistant attachments, especially if the mother is the perpetrator of the abuse. Maltreatment also leads to withdrawal behaviors in infants and self-esteem deficits later in childhood (Trad, 1987). The parent–infant relationship is inevitably worsened from such abuse, which in turn puts the infant in higher danger of being abused again (Trad, 1987).

Two main parenting dimensions particularly associated with depression in children are acceptance/rejection and psychological control/autonomy (Barber, 1996). In retrospective studies, currently depressed adults recalled their parents as having been critical, rejecting, controlling, and intrusive (Parker, 1993). Currently depressed children have described their parents as authoritarian, controlling, rejecting, and unavailable (Stein et al., 2000), and they tend to perceive their families to be less cohesive and more conflictual than do nondepressed youth (Walker, Garber, & Greene, 1993; although see Asarnow, Carlson, & Guthrie, 1987, for contrary findings). Mothers of depressed children similarly describe themselves as more rejecting, less communicative, and less affectionate than mothers of both normal and psychiatric controls (Puig-Antich et al., 1985a). In observational studies, mothers of depressed children have been described as being less rewarding (Cole & Rehm, 1986) and more dominant and controlling than mothers of nondepressed children.

Several longitudinal studies have found a significant relation between the family environment and subsequent depressive symptoms (e.g., Barber, 1996; Sheeber, Hops, Alpert, Davis, & Andrews, 1997), whereas others have reported null findings (Burge et al., 1997). Barber (1996) showed that children's ratings of parents' psychologically controlling behavior predicted their depressive symptoms, controlling for prior levels of depression, although children's prior depressive symptoms also predicted their ratings of their parents' behavior. Burt, Cohen, and Bjorck (1988) found for girls' ratings of family expressiveness predicted depression after controlling for prior depressive symptoms. Other studies have shown that adolescents' reports of family adaptability and cohesion (Garrison, Jackson, Marsteller, McKeown, & Addy, 1990) and perceptions of family support (McFarlane, Bellissimo, & Norman, 1995) contribute to adolescent depressive symptoms, controlling for prior symptom levels. In addition, maternal hostile child-rearing attitudes have been found to significantly predict increases in children's de-pressive symptoms (Katainen, Raikkonen, Keskivaara, & Keltikangas-Jarvinen, 1999). Using observational data of parental warmth, hostility, and disciplinary skills, Ge et al. (1994) reported that increases in adolescent internalizing symptoms were predicted by lower levels of parental warmth and higher levels of maternal hostility. In this same sample, Rueter, Scaramella, Wallace, and Conger (1999) found that escalating parent–adolescent conflict predicted increases in adolescent internalizing symptoms, which in turn increased the risk of the onset of internalizing disorders.

Depressed children also have significant peer difficulties and social skills deficits (Altmann & Gotlib, 1988). Self-reported depression significantly correlates with teachers' reports of peer rejection in children (Rudolph, Hammen, & Burge, 1994). In laboratory studies, children with depressive symptoms were rated by their peers more negatively than were children without symptoms (Peterson, Mullins, & Ridley-Johnson, 1985). French, Conrad, and Turner (1995) noted that rejection by peers predicted higher levels of self-reported depressive symptoms among antisocial, but not among non-antisocial youth. Panak and Garber (1992) found a significant relation between peer-rated rejection and self-reported depression, and this relation was mediated by perceived rejection. Kistner, Balthazor, Risi, and Burton (1999) similarly found that perceived rejection predicted increases in depressive symptoms during middle childhood. Finally, in a longitudinal study of children in sixth grade, Nolan, Flynn, and Garber (2003) found that a composite measure of rejection by peers, family, and teachers significantly predicted depressive symptoms across 3 years. Thus, depression in children is associated with high levels of interpersonal conflict and rejection from various members in their social domain.

Finally, relationships between depressed parents and their children have also consistently been found to be disrupted. Depressed parents report more conflict and less coherence in their families (Billings & Moos, 1983), are less involved and affectionate with their children, and experience poorer communication in parent–child relationships than nondepressed parents (Weissman, Paykel, Siegel, & Klerman, 1971). Moreover, depressed mothers tend to feel more hostile toward their children and less positive and competent about their parenting than do well mothers (Webster-Stratton & Hammond, 1988).

Observations of depressed mothers interacting with their children reveal that these mothers are more negative (Lovejoy, 1991), more controlling (Kochanska, Kuczynski, Radke-Yarrow, & Welsh, 1987), and less responsive and affectively involved (Cohn & Tronick, 1989) and use less productive communications (Gordon et al., 1989). Depressed mothers spend less time talking to and touching their infants and show more negative affect in their interactions with their infants who themselves show less positive affect, less activity, and more frequent protests (Field, 1995). Parental depression can also lead to disturbed attachment behavior and an inability by the infant to regulate emotions, thereby putting the infant at greater risk for developing depression (Gaensbauer, Harmon, Cytryn, & McKnew, 1984). Offspring of depressed parents have more insecure attachments than do offspring of well mothers (DeMulder & Radke-Yarrow, 1991; Teti, Gelfand, Messinger, & Isabella, 1995). Moreover, insecurely attached offspring of depressed mothers tend to have difficulties in their relationships with peers (Rubin, Booth, Zahn-Waxler, Cummings, & Wilkinson, 1991). Finally, negative reciprocal interaction patterns have been observed between depressed mothers and their children (Radke-Yarrow, Nottelman, Martinez, Fox, & Belmont, 1992).

In summary, two important findings emerge regarding the link between interpersonal vulnerability and depression. First, families with a depressed member tend to be characterized by less support and more conflict, and such family dysfunction increases children's risk of developing depression. Second, depressed individuals are themselves more interpersonally difficult, which results in greater problems in their social network. Thus, the link between interpersonal vulnerability and depression likely is bidirectional (Gotlib & Hammen, 1992). Longitudinal studies examining the contribution of family dysfunction, parent–child conflict, peer difficulties, and interpersonal rejection to increases in and maintenance of depressive symptoms in children have shown both that social problems temporally precede depression, and that depression contributes to interpersonal difficulties. Moreover, interpersonal difficulties appear to persist after depressive symptoms have remitted (Puig-Antich et al. 1985b). In addition, social adversities such as persistent poor friendships, low involvement of fathers, negative attitudes by family members, and stressful family environments can contribute to the maintenance or relapse of depressive disorders in youth (e.g., Asarnow, Goldstein, Tompson, & Guthrie, 1993).

The interpersonal environment clearly is an important and sometimes stressful context in which children develop schema about themselves and others, which can then serve as a vulnerability to depression. In addition, children's own reactions to these environments can exacerbate and perpetuate negative social exchanges, which furthers the interpersonal vicious cycle, thereby resulting in more rejection and depression. Thus, a transactional model of mutual influence probably best characterizes the association between depressed individuals and their social environment.


The view that mania in younger people is extremely rare or nonexistent has been increasingly challenged by many case reports and by large-scale community surveys of adults; for example, Akiskal et al. (1985), in a case series of adolescent relatives of “classic” adult bipolar patients, found that despite frank symptoms of depression and mania and frequent mental health contacts, none of these youth had been diagnosed with an affective disorder. Weller, Weller, Tucker, and Fristad (1986) reviewed over 200 articles published between the years of 1809 and 1982 and identified 157 cases that would likely be considered manic by modern standards. However, approximately half of those subjects retrospectively diagnosed as manic according to DSM-III criteria were not considered so at the time of referral. More recently, Wozniak et al. (1999) reported that 16% of psychiatrically referred prepubertal children satisfied diagnostic criteria for bipolar disorder. Biederman et al. (1996) reported that a sizeable minority of children with ADHD has bipolar disorder. These reports suggested that mania in children/adolescents (pediatric mania) may not be rare, but difficult to diagnose. Despite continued debate and controversy over the validity of the diagnosis of mania in children (Biederman, 1998; Klein, Pine, & Klein, 1998), there is a growing consensus that many seriously disturbed children are afflicted with severe affective dysregulation and high levels of agitation, aggression, and dyscontrol that may be early bipolar disorder. These issues have received increased scientific attention, as evidenced by the scheduling of two NIMH workshops on bipolar disorder in children and adolescents and in exhaustive reviews that have supported the validity of the disorder in youth (Faedda et al., 1995; Geller and Luby, 1997; Weller, Weller, & Fristad, 1995). The NIMH Strategic Research Plan for Mood Disorder Research recommended the establishment of a multisite network program on pediatric onset bipolar disorder (Costello et al., 2002).

Agreement on what constitutes the first presentation of bipolar illness is critical for epidemiologic studies to obtain the true age of onset and estimate of prevalence and risk. The following questions need to be addressed: Does pediatric bipolar disorder differ from the adult form? What are the early signs and symptoms? What is the relationship of ADHD and other disruptive disorders to juvenile onset bipolar disorder? Answers to these questions are complicated by the uncertainty about the duration of a manic episode; that stems from the fact that youth more frequently report manic symptoms that persist only a few hours or days (Carlson and Kelly, 1998; Geller et al., 1995), and therefore do not fulfill adult criteria.

Differentiating Prepubertal and Adolescent Onset

In April 2000, the NIMH convened a meeting of experts to discuss diagnostic issues impeding research on children/adolescents bipolar disorder. There was general agreement that a diagnosis of bipolar disorder, using DSM criteria, is possible in prepubertal children. Children fall into two categories: (1) those who clearly have a bipolar disorder (because they meet DSM-IV criteria for bipolar I [the classic form of the illness characterized by recurring episodes of mania and depression] or bipolar II (A form of the illness characterized by milder episodes of hypomania that alternate with depression) and (2) those who may be bipolar but do not completely fit the adult phenotype defined in DSM-IV but suffer from severe mood disturbances and symptoms of bipolar disorder and are highly impaired.

Currently, severely ill children with an impaired mood disturbance, but not meeting full DSM-IV criteria for bipolar I or II, are not included in research studies of bipolar disorder because of the perceived uncertainty of their diagnosis. It was recommended that “bipolar-NOS” (not otherwise specified) could be used as a “working diagnosis” for advancing research on this broader phenotype, as long as the children are well described (with particular attention to symptoms of ADHD). Because available diagnostic instruments may not generate a reliable and replicable diagnosis of bipolar-NOS, it was recommended that careful assessment include all of the behaviors that are impairing (Leibenluft et al., 2003).


Pediatric mania tends to be chronic and continuous rather than episodic and acute (Carl-son, 1983; Carlson, 1984; Feinstein and Wol-pert, 1973; McGlashan, 1988). In a recent review of the past 10 years of research on pediatric mania, Geller and Luby (1997) concluded that childhood-onset mania is a nonepisodic, chronic, rapid-cycling, mixed manic state. Wozniak, Biederman, Kiely, et al. (1995) also found that the overwhelming majority of 43 children from an outpatient psychopharmacology clinic who met diagnostic criteria for mania on structured diagnostic interview had a chronic and mixed presentation. Carlson, Bromet, and Sievers (2000) reported that early-onset manics were more likely to have comorbid behavior disorders in childhood and, compared to adult-onset cases of mania, to have fewer episodes of remission over a 2-year period. A review by McElroy et al. (1997) described “mixed mania,” which affects 20%–30% of adults with mania. Subjects with mixed mania tend to have a chronic course, absence of discrete episodes, onset of the disorder in childhood and adolescence, a high rate of suicide, poor response to treatment, and an early history of neuropsychological deficits highly suggestive of ADHD. Determination of the adult course of pediatric mania awaits data from longitudinal studies with larger samples. Presently, the same criteria defined in the DSM-IV-TR to diagnose bipolar disorder in adults are used for adolescents, as follows:

  • A. A distinct period of abnormally and persistently elevated, expansive, or irritable mood, lasting at least 1 week (or any duration if hospitalization is necessary)

  • B. During the period of mood disturbance, three (or more) of the following symptoms have persisted (four if the mood is only irritable) and have been present to a significant degree:

    • Inflated self-esteem or grandiosity

    • Decreased need for sleep

    • More talkative than usual or pressure to keep talking

    • Flight of ideas or subjective experience that thoughts are racing

    • Distractibility

    • Increase in goal-directed activity

    • Excessive involvement in pleasurable activities that have a high potential for painful consequences

In addition, the symptoms do not meet the criteria of a mixed mood episode, in which both criteria for a manic episode and for MDD (except for duration) are met with symptoms nearly every day during at least a 1-week period. For hypomania, the elevated or irritable mood lasts for 4 days. The mood disturbance causes significant impairment and is not better accounted for by other psychiatric disorders or medical conditions. The DSM IV-TR notes that adolescents are more likely to exhibit psychotic features and may be associated with school difficulties, substance abuse, and antisocial behavior. It also notes that a significant minority of adolescents have a long-standing pattern of behavior problems that may represent an extended prodomal phase to bipolar disorder or, in fact, may represent a distinct disorder.



Epidemiological data on juvenile bipolar disorder must be seen within the context of these diagnostic uncertainties. The 1980s Epidemiologic Catchment Area study (ECA), based on over 18,000 adults ages 18 and over for five U.S. communities, provided the first community-based data using modern diagnostic criteria and the first epidemiologic clue about the youth onset of bipolar disorder (Robins & Price, 1991) (Table 1.3). The lifetime prevalence of bipolar disorder was about 1/100, with few gender differences in rates and an overall median age of onset of 18 years.

Table 1.3 Lifetime and Annual Prevalence/100 (SE) and Incidence Rate per 100 of Bipolar I Disorder in Five U.S. Sites


New Haven, CT

Baltimore, MD

St. Louis, MO

Durham, NC

Los Angeles, CA


Number in sample







Response rate (%)







Female (%)







Lifetime Rate/100


1.2 (.21)

0.6 (.20)

1.0 (.20)

0.3 (.14)

0.6 (.16)

0.8 (.09)


1.0 (.27)

0.7 (.22)

1.1 (.30)

0.1 (.11)

0.6 (.23)

0.7 (.12)


1.3 (.30)

0.5 (.24)

1.0 (.27)

0.6 (.25)

0.5 (.21)

0.9 (.12)








Annual Rate/100


1.0 (.16)

0.5 (.14)

1.0 (.20)

0.2 (.08)

0.4 (.13)

0.6 (.07)


0.9 (.22)

0.5 (.21)

1.0 (.30)


0.3 (.15)

0.5 (.09)


1.1 (.23)

0.5 (.18)

1.0 (.27)

0.4 (.16)

0.5 (.20)

0.7 (.09)








Annual Incidence/100


0.6 (.12)

0.2 (.09)

0.1 (.07)

0.3 (.10)

0.5 (.15)

0.4 (.05)


0.4 (.15)



0.1 (.09)

0.5 (.21)

0.2 (.06)


0.7 (.19)

0.4 (1.6)

0.2 (.13)

0.5 (.18)

0.5 (.20)

0.5 (.08)








Age of Onset (years)


21.3 (1.6)

22.3 (3.1)

14.6 (1.6)

20.7 (3.6)

16.1 (1.0)

18.6 (0.9)








Data from Epidemiologic Catchment Area study; age = 18+ years.

The 1990 National Comorbidity Survey (NCS) included a representative national sample in the United States of over 8,000 subjects ages 15 to 54 (Kessler et al., 1994) and provided the best currently available epidemiologic information. The younger age included in the NCS was based on the ECA findings that many psychiatric disorders have a young age of onset. The overall lifetime prevalence of bipolar I disorder in the full sample was 1.7% (median age of onset, 21 years); in the sample ages 15 to 17 it was 1.3% with equal gender rates. Both the ECA and the NCS suggested that onset of bipolar disorder in adolescence and childhood is common.

In the 1990s, the Cross-National Collaborative Group was formed to directly compare rates and risk of psychiatric disorders by standardizing analysis to overcome the problem of disparate data among studies. Seven countries (U.S., Canada, Puerto Rico, Germany, Taiwan, Korea, and New Zealand) provided data on bipolar disorder (Table 1.4). The lifetime prevalence rates for bipolar I ranged from 0.3% in Taiwan to 1.5% in New Zealand (Weissman et al., 1996), with equal gender ratios across sites (with the exception of Korea) and median ages of onset of 18 to 25 years.

Table 1.4 Lifetime Prevalence/100 (SE) Bipolar I Disorder in Cross-National Sample Lifetime Prevalence/100 (SE)





M/F Ratio

Age of Onset (Years)

Median Age of Onset (Years)

United States

0.8 (.14)

1.0 (.15)


18.1 (.68)


ECA, 1980

0.9 (.10)

NCS, 1990a

1.7 (.21)

1.6 (.29)

1.8 (.30)


24.0 (.92)


Edmonton, Canada

0.6 (.16)

0.7 (.25)

0.5 (.21)


17.1 (1.12)


Puerto Rico

0.6 (.23)

0.8 (.38)

0.5 (.27)


27.2 (3.40)


West Germanyb

0.5 (.37)

1.0 (.71)




0.3 (.06)

0.3 (.09)

0.3 (.07)


22.5 (1.90)



0.4 (.09)

0.6 (.16)

0.2 (.09)


23.0 (2.54)


Christchurch, New Zealand

1.5 (.36)

1.7 (.56)

1.2 (.45)


18.2 (5.90)


ECA, Epidemiologic Catchment Study.

a Age 18–54 years in National Comorbidity Survey (NCS).

b Only one case of bipolar disorder was reported from West Germany.

Efforts have been made to study children and adolescents directly in epidemiologic studies. Table 1.5 presents the prevalence rates from these efforts. There is considerable variability across these studies in the age ranges, diagnostic methods, and diagnostic criteria. All of these differences affect prevalence rates and may account for discordant findings.

Table 1.5 Prevalence Rates/100 of Bipolar Disorder from Studies in Children and Adolescents


Country (Year)

Sample Size

Age Range (Years)

Diagnostic Method


Time Period


Shaffer, unpublished

USA (1992)





6 months


Lewinsohn, et al., 1995

Western Oregon (1987)







Wittchen, et al., 1988

Bavaria, Germany (1988)









Aalto-Setälä et al., 2001

Helsinki, Finland (2000)






1 month

1 month



BPD, bipolar disorder; BPI, bipolar I; BPII, bipolar II; CIDI, Composite International Diagnostic Interview; DiSC, Diagnostic Interview Schedule for Children; KSADS, Kiddie Schizophrenia and Affective Disorders Schedule; NOS, not otherwise specified; SCAN, Schedules for Clinical Assessment in Neuropsychiatry.

The Methodological Epidemiologic Catchment Area study (MECA) was conducted in a population-based sample of children and adolescents (Shaffer, unpublished observations). Probability household samples of 1,285 youth aged 9 to 17 years were selected at four sites (Atlanta, Georgia; New Haven, Connecticut; Westchester County, New York; and Puerto Rico; see Lahey et al., 1996 for methods). Data were collected for both children and their parents. The current (6-month) prevalence for mania was 1.2% and for hypomania was 0.6% (unpublished); 4.5% of youth with suicide ideation and 7.1% with suicide attempts had mania, compared to 0.9% of those who had neither suicide ideation nor attempts (Gould et al., 1998). Significant associations between bipolar disorder and risk for suicide in youth has been previously reported by Brent et al. (1988), highlighting the long-held notion that bipolar disorder in youth is associated with not only high levels of morbidity but also mortality.

Lewinsohn, Klein, and Seeley (1995) studied 1,709 adolescents (14 to 18 years) randomly selected from nine senior high schools, representative of urban and rural districts in western Oregon. The lifetime prevalence of bipolar disorder (primarily bipolar disorder II and cyclothymia) was approximately 1.0%. An additional 5.7% of the sample reported experiencing a distinct period of abnormally elevated mood, expansive or irritable, although they never met the criteria for bipolar disorder per se. Both groups exhibited significant functional impairment and high rates of comorbidity (anxiety and disruptive behavior), suicide attempts, and mental health service utilization. The prevalence, age of onset, phenomenology, and course of bipolar disorder did not differ by gender. In a follow-up study, they found that less than 1.0% of adolescents with depression switched to bipolar disorder by age 24 (Lewinsohn, Rohde, Seeley, Klein, & Gotlib, 2000). The bipolar disorder and subsyndromal bipolar disorder subgroups both had elevated rates of antisocial and other personality symptoms. Both groups showed significant impairment in psychosocial functioning and had higher mental health treatment utilization. In general, adolescents with bipolar disorder showed significant continuity across developmental periods and adverse outcomes during young adulthood. Adolescent subsyndromal bipolar disorder was also associated with adverse outcomes in adulthood, but not with increased incidence of bipolar disorder, which questions the relationship of subsyndromal bipolar disorder in children and adolescents to true adulthood bipolar disorder.

A study from Bavaria, Germany, on a sample of 3,021 adolescents and young adults age 14 to 24 found a lifetime prevalence of 1.4% for bipolar I disorder and 0.4% for bipolar II disorder (Wittchen, Nelson, & Lachner, 1998). Ninety-four percent of the youth with either bipolar disorder considered themselves socially and economically impaired.

A 5-year follow-up of over 500 students of ages 20–24 years in Helsinki, Finland, found a 1-month prevalence rate of bipolar I of 0.2%. However, when bipolar II and bipolar-NOS were added, the prevalence rates were 0.9/100. In summary, the epidemiologic studies of adolescents and adults show an early age of onset of bipolar disorder and prevalence in adolescents close to what is found in studies of adults.


With Attention Deficit Hyperactivity Disorder (ADHD)

The symptomatic overlap of childhood mania with ADHD is one major source of diagnostic controversy. Rates of ADHD range from 60% to 90% in pediatric patients with mania (Borchardt & Bernstein, 1995; Geller et al., 1995; West, McElroy, Strakowski, Keck, & McConville, 1995; Wozniak, Biederman, Kiely, et al., 1995). Although the rates of ADHD in samples of youth with mania are universally high, the age at onset modifies the risk for comorbid ADHD. For example, while Wozniak, Biederman, Kiely, et al. (1995) found that 90% of children with mania also had ADHD, West et al. (1995) reported that only 57% of adolescents with mania had comorbid ADHD. Examining further developmental aspects of pediatric mania, Faraone, Biederman, Wozniak, et al., (1997) found that adolescents with childhood-onset mania had the same rates of comorbid ADHD as manic children (90%), and that both these groups had higher rates of ADHD than adolescents with adolescent-onset mania (60%). Most recently, Sachs, Baldassano, Truman, & Guille (2000) reported that among adults with bipolar disorder, a history of comorbid ADHD was only evident in those subjects with onset of bipolar disorder before 19 years of age. The mean onset of bipolar disorder in those with a history of childhood ADHD was 12.1 years (Sachs et al., 2000). Similarly, Chang, Steiner, and Ketter (2000) studied the offspring of patients with bipolar disorder and found that 88% of manic children had comorbid ADHD and that the onset of mania in adults with bipolar disorder and a history of ADHD was 11.3 years of age. These findings suggest that age of onset of mania, rather than chronological age at presentation, may be the critical developmental variable that identifies a highly virulent form of the disorder that is heavily comorbid with ADHD.

Although ADHD has a much earlier onset than pediatric mania, the symptomatic and syndromatic overlap between pediatric mania and ADHD raises a fundamental question—namely, do children presenting with symptoms suggestive of mania and ADHD have ADHD, mania, or both? One method to address these uncertainties has been to examine the transmission of comorbid disorders in families (Faraone & Tsuang 1995; Faraone, Tsuang, & Tsuang, 1999). If ADHD and mania are associated because of shared familial etiologic factors, then family studies should find mania in families of ADHD patients and ADHD in families of manic patients. Studies that examined rates of ADHD among the offspring of adults with bipolar disorder all found higher rates of ADHD among these children than in controls (Faraone, Biederman, Mennin, & Russell, 1998).

Wozniak, Biederman, Mundy, et al. (1995) used familial risk analysis to examine the association between ADHD and mania within families of manic children. They found that relatives of children with mania were at high risk for ADHD that was indistinguishable from the risk in relatives of children with ADHD and no mania. However, mania and the comorbid condition of mania plus ADHD selectively aggregated among relatives of manic youth compared with those with ADHD and comparison children (Wozniak, Biederman, Mundy, et al., 1995). Almost identical findings were obtained in two independently defined family studies of ADHD probands with and without comorbid mania (Faraone, et al., 1998; Faraone, Biederman & Monuteaux, 2001). Taken together, this pattern of transmission in families suggests that mania in children might be a familiarly distinct subtype of either bipolar disorder or ADHD.

With Conduct Disorder (CD)

Like ADHD, CD is also strongly associated with pediatric mania. This has been observed separately in studies of children with CD, ADHD, and mania. Kovacs and Pollak (1995) reported a 69% rate of CD in a referred sample of manic youth and found that the presence of comorbid CD heralded a more complicated course of mania. Kutcher et al. (1989) found that 42% of hospitalized youths with mania had comorbid CD. The Zurich longitudinal study found that hypomanic cases had more disciplinary difficulties at school and more thefts during their juvenile years than did other children (Wicki & Angst, 1991). These reports are consistent with the well-documented comorbidity between CD and major depression (Angold and Costello 1993), because pediatric depression often presages mania (Geller, Fox, & Clark, 1994; Strober and Carlson, 1982). It is not surprising that mania and CD are also frequently comorbid.

Biederman et al. (1999) (Biederman, Faraone, Hatch, et al., 1997) investigated the overlap be tween mania and CD in a consecutive sample of youth referred for treatment. They found a striking similarity in the features and age of onset of mania, regardless of comorbid CD. Whether or not CD was present, mania presented with a predominantly irritable mood, a chronic course, and was mixed with symptoms of major depression. Only two manic symptoms differed between those with versus without CD: “physical restlessness” and “poor judgment” were more common in the mania-with-CD group than in the mania-only group. Similarly, there were few differences in the frequency of CD symptoms between CD youth with and without comorbid mania. Nevertheless, it is important to assess for both diagnoses because delinquency and mania require different treatment strategies.

With Anxiety Disorders

Although anxiety is frequently overlooked in studies of mania, pediatric studies of youth with panic disorder and youth with mania document a bidirectional overlap. Biederman, Faraone, Marrs, et al. (1997) found that subjects with panic disorder and agoraphobia had very high rates of mania (52% and 31%), which were greater than those observed among psychiatric controls (15%). Wozniak, Biederman, Kiely, et al. (1995) and Bowen, South, & Hawkes (1994) reported significantly more panic and other anxiety disorders in children with mania. These findings indicate that mania at any age is frequently comorbid with severe anxiety that requires additional clinical and scientific scrutiny.

With Substance Use Disorders (SUD)

There is an extensive and bidirectional overlap between mania and substance use disorders (SUD) in youth (Biederman, Wilens, et al., 1997; West et al., 1996; Wilens, Bierderman, Abrantes, Spencer, & Thomas, 1997b) as well as in adults. Juvenile-onset mania may be a risk factor for SUD. For example, a prospective study of children and adolescents with and without ADHD found that early-onset mania was a risk factor for SUD independently of ADHD (Biederman, Wilens, et al., 1997). Similarly, controlled studies in adults show that mania is often an antecedent and is strongly associated with SUD (Wilens et al., 1997b). Mania has also been shown to be overrepresented among youth with SUD (West et al., 1996; Wilens, Biederman, Abrantes, Spencer, & Thomas, 1997a).

Wilens et al. (1999) found that mania significantly increased the risk for SUD independently of conduct disorder. Furthermore, they reported that the risk for SUD was carried by those subjects with an adolescent-onset form of mania. While this may be consistent with the notion that, like adults, adolescents self-medicate manic symptoms with substances of abuse (Khantzian, 1997), it is also consistent with the hypothesis that child and adolescent onset mania are etiologically distinct forms of the disorder with different risk profiles and natural courses.

Risk Factors

Given the paucity of epidemiologic studies of bipolar disorder in youth and the fact that the relatively low lifetime prevalence (about 1%–2%) would require very large samples, risk factors have not been clearly identified from community-based studies. The most consistent risk factor for bipolar disorder is family history. While studies vary considerably in methods, sample size, and controls, they consistently show offspring of adult bipolar patients as having an increased risk, over 3-fold, of bipolar disorder as well as mood disorders, compared to offspring of controls (Carlson and Weintraub, 1993; Decina et al., 1983; Gershon et al., 1985; Hammen et al., 1990; Klein, Depue, & Slater, 1985; Nurnberger et al., 1988; Todd et al., 1996). Because most of the offspring have not lived through or even entered the age of risk, an accurate estimate of prevalence rates is not possible. However, it is important to note that although the risk to offspring of bipolar patients is increased, the majority of high-risk offspring will have neither a diagnosable bipolar illness nor any other mood disorder.


Although there has been an exponential increase in our understanding of the pathophysiology of major depression in adults, we are largely dependent on extrapolations from these findings to inform us on the biology of childhood and adolescent depression (Coyle et al., 2003). In the last two decades, considerably more research on the biology of depression in children has been forthcoming. Moreover, the neurobiology of bipolar disorder even in adults remains poorly understood (Berns & Nemeroff, 2003; Wang & Nemeroff, 2003), though it represents an active avenue of investigation. Not surprisingly, our understanding of the biology of childhood bipolar disorder is virtually nonexistent.


As noted above, the age of onset of serious mood disorders is now known to occur in children and adolescents and to persist into adulthood. Thus, much of the information that has accrued concerning the pathophysiology of mood disorders in adulthood would appear to be applicable to the childhood onset mood disorders. One important difference, however, is the apparent lack of efficacy of tricyclic antidepressants in youth as compared to that of the selective serotonin reuptake inhibitors (SSRIs), though controversy surrounds the latter contention; clearly both classes are effective in adults (Wagner & Ambrosini, 2001).

Although heritable factors appear to be the most consistent predictors of risk for major depression in children, environmental factors also play an important role (Rice, Harold, & Thapar, 2002). Especially with regard to bipolar disorder, twin, family and adoption studies have shown that heritable factors are substantial predictors of risk (Smoller and Finn, 2003). In the case of bipolar disorder in the parents, the risk extends to major depressive disorder and early onset of mood symptoms in their children (Mortensen, Pedersen, Melbye, Mors, & Ewald, 2003). No specific genes have yet been identified that are unequivocally associated with mood disorders, regardless of age of onset. However, linkage studies, especially in bipolar disorder, have identified a number of regions on the human genome associated with high heritable risk for affective disorders (Schulze and McMahon, 2003; Tsuang, Taylor, & Faraone, 2004; Table 1.6). Given the evidence for interactions between risk genes and environment, and the important role of early trauma (Heim, Newport, Bonsall, Miller, & Nemeroff, 2001; Nemeroff, 2004), the risk for early-onset depression may be magnified by the impact of a parent with depression (e.g., postpartum disorder) on a genetically vulnerable child (Weinfeld, Stroufe, & Egeland, 2000).

Table 1.6 Genomic Regions Associated with High Heritable Risk for Affective Disorders

Genomic Location

Principle Report

Independent Confirmations



Berrettini et al., 1994 and 1997

Stine et al., 1995; Nothen et al., 1999; Turecki et al., 1999

Paternal parent-of-origin effect; see Schwab et al., 1998


Straub et al., 1994

Detera-Wadleigh et al., 1996; Smyth et al., 1996; Kwok et al., 1999; Morissette et al., 1999


Kelsoe et al., 2001

Detera-Wadleigh et al., 1997 and 1999

Velocardiofacial syndrome region; possible overlap with a schizophrenia locus


Stine et al., 1995

McInnes et al., 1996; McMahon et al., 1997; De Bruyn et al., 1996

See Freimer et al., 1996


Morissette et al., 1999

Ewald et al., 1998; Detera-Wadleigh et al., 1999

Principal report in a Canadian isolate


Blackwood et al., 1996

Ewald et al., 1998; Nothen et al., 1997; Detera-Wadleigh et al., 1999

See Ginns et al., 1998

Early Life Stress

In unipolar major depression, environmental influences are clearly substantial. Consistent with predications from early psychoanalytic models, losses early in life, maternal deprivation, and physical and sexual abuse appear to be major risk factors when controlled for heritable risks. Recent research has highlighted the seminal importance of early psychological insults contributing to the neurobiological processes thought to underlie the pathophysiology of major depressive disorder. The pronounced effects of early life stress (ELS) throughout the lifespan are believed to be mediated by the substantial plasticity of the developing central nervous system (CNS) as a function of experience. It has been proposed that stress and emotional trauma during development permanently shape the brain regions that mediate stress and emotion, leading to altered emotional processing and heightened stress responsiveness, which in the genetically vulnerable individual may then evolve into psychiatric disorders, such as depression.

Hypothalamic–pituitary–adrenal axis (HPA)

The system that has been most closely scrutinized in mood disorders is the HPA axis, with evidence of its hyperactivity in adult patients with unipolar depression and bipolar patients in the mixed state (Evans & Nemeroff, 1983; Swann et al., 1992). Upon stress exposure, neurons in the hypothalamic paraventricular nucleus (PVN) secrete corticotropin-releasing factor (CRF) into the hypothalamic-hypophyseal portal circulation, which stimulates the production and release of adrenocorticotropin (ACTH) from the anterior pituitary. Adrenocorticotropin in turn stimulates release of glucocorticoids from the adrenal cortex. Glucocorticoids have marked effects on metabolism, immune function, and the brain, adjusting physiological functions and behavior in response to the stressor. Glucocorticoids exert negative feedback control on the HPA axis by regulating hippocampal and PVN neurons. Persistent glucocorticoid exposure exerts adverse effects on hippocampal neurons, in-cluding reduction in dendritic branching, loss of dendritic spines, and impairment of neurogenesis. Such damage might progressively reduce inhibitory control of the HPA axis. Corticotropin-releasing factor neurons integrate information relevant to stress not only at the hypothalamic PVN but also in a widespread circuitry throughout the limbic system and brain stem. Direct CNS administration of CRF to laboratory animals produces endocrine, autonomic, and behavioral responses that parallel signs of stress, depression, and anxiety, including loss of appetite, sleep disruption, decreased sexual behavior, despair, increased motor activity, neophobia, and enhanced startle reactivity.

Laboratory animal studies have provided direct evidence that ELS indeed leads to heightened stress reactivity and alterations in the aforementioned neural circuits that persist into adulthood. For example, adult rats separated from their dams for 180 minutes/day on postnatal days 2–14 exhibit up to 3-fold increases in ACTH and corticosterone responses to a variety of psychological stressors compared with control rats (Ladd et al., 2000; Plotsky and Meaney, 1993). Maternally separated rats also develop marked behavioral changes, including increased anxiety-like behavior, anhedonia, alcohol preference, sleep disruption, decreased appetite, and cognitive impairment. Subsequent studies revealed multiple CNS changes that likely underlie physiologic and behavioral sensitization to stress after maternal separation or lack of maternal care. These findings include increased activity (increased CRF mRNA expression) and sensitization of CRF neurons in hypothalamic and limbic regions, decreased glucocorticoid receptor density in the hippocampus and prefrontal cor tex, increased mineralocorticoid receptor in the hippocampus, decreased mossy fiber development and neurogenesis in the hippocampus, as well as alterations in norepinephrine, GABA, and other systems (Heim, Plotsky, & Nemeroff, 2004; Ladd, Huot, Thrivikraman, Nemeroff, & Plotsky, 2004). Behavioral sensitization to fear stimuli have been observed in nonhuman primates reared by mothers exposed to unpredictable conditions with respect to food access over 3 months (Coplan et al., 1996, 2001). Taken together, ELS induces manifold changes in multiple neurocircuits involved in neuroendocrine, autonomic, and behavioral responses to stress. If similar changes also occur in humans exposed to ELS, these changes might indeed confer an enhanced risk for depression.

As noted earlier, several clinical studies have evaluated the long-term consequences of ELS in adult humans. In astonishing parallel to findings in rodents and nonhuman primates, women abused as children, including those with and those without current depression, exhibit greater plasma ACTH responses in a standardized stress paradigm than controls. The increase was more pronounced in abused women with current depression, and these women also showed greater cortisol and heart rate responses than controls (Heim et al., 2000). Several studies have reported similar neuroendocrine and neurochemical changes in abused children, which are reviewed in detail elsewhere (Heim & Nemeroff, 2001).

Magnetic Resonance Imaging

Among depressed patients, magnetic resonance imaging (MRI) analysis has revealed decreased hippocampal volumes only in adult women with ELS (Vythilingam et al., 2002). Because hippocampal volume loss is not observed in abused children or young adults (Teicher, 2002) (although corpus callosum, amygdala, and cortical development seems to be impaired), some have suggested that repeated bursts of cortisol secretion over the course of time may eventually result in smaller hippocampi. Enhanced CRF secretion during development may also contribute to progressive hippocampal volume loss (Brunson, Eghbal-Ahmadi, Bender, Chen, & Baram, 2001). The fact that adult patients with major depression exhibit HPA axis hyperactivity and profound CRF hypersecretion as evidenced in studies of cerebrospinal fluid (CSF) and postmortem tissue (Flores, Alvarado, Wong, Licinio, & Flockhart, 2004; Merali et al., 2004) and that these findings are also observed after ELS raises the question as to whether the HPA axis in general and the CRF system in particular play a role in the pathogenesis of childhood mood disorders.

Steingard and colleagues (2002) found significant reductions in frontal lobe volume and increased ventricular volume in a large cohort of children and adolescents with depressive disorders. In resolving white and gray matter, they found significant reductions in frontal white matter in adolescents with major depression. Furthermore, orbitofrontal choline levels are elevated in depressed adolescents (Steingard et al., 2000), consistent with findings in adults. A single photon emission computed tomography (SPECT) study found a significant elevation of the density of serotonin transporters in the hypothalamus and midbrain, but no change in the dopamine transporter in children and adolescents with major depression. Chang and colleagues (2004) used functional magnetic resonance imaging (fMRI) to compare 12 children and adolescents with bipolar disorder who had at least one parent with bipolar disorder and 10 age-and IQ-matched healthy male controls. Significant differences in brain activation patterns in prefrontal areas were noted in the bipolar subjects, compared to patterns in the controls, when performing both cognitive and affective tasks. Brain areas affected included the anterior cingulate cortical and dorsolateral prefrontal cortex. This same group (Chang et al., 2003) had previously reported a reduction in dorsolateral and prefrontal cortical N-acetyl-aspartate concentrations, a marker of neuronal integrity, in children with bipolar disorder, as measured by magnetic resonance spectroscopy (MRS). Children during a manic episode were also reported to exhibit increased anterior cingulate concentrations of myo-inositol (Davanzo et al., 2001).


Gender is well known to be an important but poorly understood factor influencing the risk of MDD. The prevalence of MDD, while equal between boys and girls prior to puberty, doubles in young women after puberty. This increase in females has been hypothesized to be secondary to hormonal changes occurring during puberty. These endocrine changes surely influence brain function, but the attendant social and psychological factors of puberty cannot be ignored. Nevertheless, twin studies suggest that the impact of genetic risk factors become more prominent as girls pass through puberty and enter adolescence (Silberg et al., 1999).

An early and logical focus of biological studies of child and adolescent mood disorders was neuroendocrinologic, motivated by the findings described above of significant HPA dysregulation in a sizeable proportion of adults with MDD. However, the results of a number of studies have found rare and only modest abnormalities in 24-hour cortisol secretion, the dexamethasone suppression test, and the CRF stimulation test. Recently, Feder and colleages (2004) reported that of 86 children (depressed, anxious, or normal) tested, those children who exhibited an abnormally elevated cortisol concentration during the evening and an abnormal delay in the rise of cortisol during the night, whether depressed or not as children, exhibited depression as adults.

Puig-Antich and colleagues (1985a,b) observed a blunted growth hormone (GH) response in adolescents with MDD when challenged with insulin or growth hormone releasing hormone (GHRH), a finding previously reported in adults with major depression. However, the link between this abnormality and the underlying pathophysiology of depression remains obscure.


Accurate epidemiologic data are useful for determining the magnitude of the problem, identifying risk factors, monitoring changes of rates (epidemics), and identifying the underserved. Accurate estimates rest on accurate diagnosis. The explosive developments in neurosciences, genetics, and neuroimaging will undoubtedly help advance our pathophysiological understanding of these complex mood disorders afflicting the young. Such advances can help shed light on etiology, identify dysfunctional brain circuits and, most importantly, define subtypes of the disorder. This will undoubtedly lead to improved treatment of afflicted youth and their families. Longitudinal studies focusing on the biology and treatment response of childhood and adolescent mood disorders are sorely needed.