4.2. Screening instruments
4.2.1. Introduction
Epidemiological studies have shown that many young people and some children in the community who are depressed remain undetected (Angold & Costello, 2001). Even in child mental health clinics depressive signs and symptoms may be missed through cursory inquiry or greater attention being paid to other concurrent difficulties in the child or family. As a consequence efforts have been made to develop instruments that are capable of detecting clinically depressed children and young people in different community and clinical settings (Dierker et al., 2001; Pavuluri & Birmaher, 2004).
4.2.2. Principles of detection
To date most instruments developed for the purpose of detecting depression in young people have been focused on either detecting a given disorder according to operationally defined criteria or characterising a defined set of signs and symptoms according to a given content. Criterion validity refers to the ability of the instrument to ‘find’ the cases of interest in the population being examined. Content validity refers to the ability of the instrument to characterise the symptoms that occur within the disorder. These key issues are not the same and it cannot be assumed that they will always work together in producing the ‘best instrument’. For example criterion validity for DSM-IV major depression may be best achieved by merely asking a few questions knowing that if these are answered ‘yes’ there is a very good chance the young person is currently clinically depressed. In contrast, content validity requires asking many questions to determine the full form of symptoms, which would include uncommon and common symptoms, many of which might only be weakly associated with the disorder. If we want to find individuals with a predefined syndrome of major depression we need instruments that are focused on criterion validity. If the task is to determine the range of depressive symptoms in the community at large we need instruments that are focused on content validity.
4.2.3. Who should be asked?
There is general agreement in child mental health research that both criterion and content validity for common behavioural and emotional problems in children and adolescents are best achieved by asking both a parent and the child about their current symptoms and problems and combining the two sets of answers to achieve a best estimate of detection. For depression it appears that parent reports alone are likely to miss clinically depressed offspring (Pavuluri & Birmaher, 2004). In contrast, child and adolescent reports are likely to include individuals who are not depressed. Thus parents appear to under report and children over report depressive signs and symptoms. Other potential reporters (teachers, siblings and peers) appear somewhat more like parents except in the case of close confidants who may report more like the index child.
4.2.4. What should be asked?
This depends on the prevailing set of definitions for depression. Signs and symptoms based on existing syndrome definitions (DSM-IV and ICD-10) are generally seen as the most efficient way of detecting people with disorders. Inclusion of items considered i) important by some clinicians but not in the agreed definition, or ii) providing more detail of a given construct that is considered key needs to be very carefully considered. Seldom is greater detail or wider coverage likely to improve on the ability of the instrument to detect real cases. Often this is an attempt to deal with worries about content when the focus is really on criterion validity. For example some authorities believe that physical signs and symptoms are too important to cover in just one or two questions. Others may express concerns about the lack of detail about the items asking about current depressive thinking. Invariably the key items in an instrument are already closely associated with the additional items and lengthening the instrument to include more content will not improve the ability to find individuals who meet criterion.
4.2.5. Purpose of detection
The purpose of clinical detection is to identify from within a group of individuals those who have the disorder of interest (depressive symptomatology). Is screening an attempt to detect all forms of clinical depression or just a particular type? Is it trying to find individuals who are currently depressed, recently depressed or depressed at previous points in time? Are there special requirements that must be incorporated such as culture, language and ethnicity or features in the child such as age, educational ability or gender? As yet these factors are seldom taken into account in instrument development.
4.2.6. Pragmatics of screening
The likelihood of instruments being acceptable to the population of interest must be considered. This attention to the ecological validity is of great importance. The length of the screen, complexity of instructions, method of completion and presentation (e.g. paper and pencil, handheld computer, via the web) all influence the extent to which a screening instrument will be completed fully, reliably and by as many respondents as possible.
Psychometrics
Instruments must show reliability generally through test-retest on the same population at intervals between 1 and 4 weeks apart. If data recorded are not consistent then the instrument is unreliable and cannot be used. The type of statistic used depends on whether the reliability of items, their total scale score or sub-scales, a categorical threshold or specific diagnosis, is being measured. The internal consistency of the instrument refers to the extent to which different items measure the same overt construct (e.g. negative thoughts or physical changes). Instrument length can be considerably shortened by reducing the number of items required through these methods to ensure that key areas are covered by as few items as is statistically possible. Validity of the instrument refers to the extent that it is measuring what it purports to measure. There are a number of forms of validity that require different statistical methods. First, items in the instrument should be seen to be measuring the construct of interest (face validity); new instruments can be compared with existing ones known to be valid (concurrent validity); a new instrument can be assessed against a different form of measure already in use as a gold standard e.g. questionnaire for depression against clinical diagnosis by interview (criterion validity); an instrument can be used to determine a given outcome, such as response to treatment or the risk of recurrence (predictive validity); finally a measure can be used to determine change in severity or nature of depression over time (sensitivity to change).
From the public health perspective it is essential to establish how good the instrument is at doing the job it is intended for. The sensitivity of an instrument refers to the proportion of true cases in the population correctly identified by the tests. An instrument that detects a low percentage of depressed cases will not be very helpful in determining the numbers of children who should receive a known effective treatment, as many individuals who should receive the intervention will not do so. This would make for poor planning and underestimating the prevalence of the disorder and the cost of treatments to the community. As the sensitivity of an instrument increases the number of ‘false negatives’ it detects will decrease (i.e. the number of cases the instrument says are depressed who are in fact well).
The specificity of an instrument refers to the proportion of well individuals correctly identified by the test. This is important so that well individuals are not given treatments or other interventions they do not need. As the specificity of an instrument increases the number of ‘false positives’ will decrease (i.e. the number of well individuals who are said to be depressed).
Instruments with low sensitivity and specificity are very unhelpful screening instruments. They will fail to identify the depressed population with sufficient validity.
There are a number of statistical procedures for determining sensitivity and specificity of which the AUC (receiver operating characteristic) is the most valid as it displays the trade off between sensitivity and specificity at all possible scores available to the instrument. This is displayed as a figure between 0 and 1. An instrument's diagnostic accuracy is considered as follows: AUC ≤0.7, low; 0.7–0.9, moderate, >0.9 high (Henderson, 1993).
Sensitivity and specificity do not address differences in the prevalence of depression in different populations. To address this, the positive predictive value (PPV) and negative predictive value (NPV) should be considered. The PPV of a test is the probability that the patient has depression when restricted to those patients who test positive. The NPV of a test is the probability that the patient will not have depression when restricted to all patients who test negative. PPV depends crucially on the prevalence of depression in the population screened (i.e. it will be higher in specialist clinics than in the community for the same instrument; also see ).
Estimated proportion of subjects with depression at each age according to level of self-report depression scores. The legend indicates three levels of MFQ self-report scores, 20, 30 and 50, from a possible range of 0 to 66, obtained from 1056 girls aged (more...)
4.2.7. Self-rated depression scales as screens
The commonest method used for detecting clinical depression is to ask the child to complete a questionnaire that asks them to record how they have been feeling and thinking recently – often over the past week or 2 weeks. To date, most screening instruments have been about current depression. In addition the focus has in the main been on determining the presence or absence of major depression. There are six available instruments with psychometric data (see Appendix J for further details).
The Beck Depression Inventory (BDI) is a commonly used scale in adult studies, especially when measuring mild/moderate depression (Beck et al., 1961). In adolescents however it is not clear that the BDI is truly measuring depression (LeBlanc et al., 2002). The reading level and response format may present problems for young adolescents and those with low literacy skills. The scale is sensitive to change in depressed young adult patients (Reynolds & Coates, 1986). The sensitivity and specificity of the scale are not particularly good in adolescents (Roberts et al., 1991). Several authors have suggested that rather than clinical depressive disorders the scale measures dissatisfaction and demoralization, non-clinical low mood and anxiety (Brooks & Kutcher 2001).
The Children's Depression Inventory (CDI) is specifically aimed at children under 12 (Kovacs, 1992). The instrument is a modified version of the original BDI developed originally for children under the age of 8. The reliability and internal consistency data are not particularly satisfactory and no single cut-off score works well in both clinical and community settings (Asarnow & Carlson 1985; Stark et al., 1987; Kovacs, 1992). There is evidence for sensitivity to change but there are serious concerns that the instrument does not discriminate adequately between depressed and non-depressed children (Stark et al., 1987; Meyer et al., 1989; Fine et al., 1991; Stark & Laurent, 2001). The instrument may be better as a continuous measure of current dysphoric mood than as a screen for the presence or absence of depression.
The Mood and Feelings Questionnaire (MFQ) is aimed for use with children and young people aged 8-17 years. It has a parent and a child form and good diagnostic validity and some predictive validity has been established (Wood et al., 1995; Kent et al., 1997). The scale has been used in both epidemiological and clinical studies (Costello & Angold, 1988; Messer & Gross, 1995; Messer et al., 1995; Goodyer et al., 1996; Goodyer et al., 2000a; Angold et al., 2002). There is normative data showing that the probability of being clinically depressed varies with age and sex (Angold & Rutter 1992; Cooper & Goodyer 1993; Goodyer & Cooper 1993; Angold et al., 2002). A score of 50 or more (scale range 0–66) in a 13-year-old girl indicates a 30% probability of being clinically depressed compared with 68% for the same score in a 16-year-old girl (see ). There is acceptable case detection ability (AUC ranging from 0.75–0.85) in clinical settings with a cut off score of ≥27. The instrument does not assess suicidal ideation. There is adequate diagnostic validity for depressed patients but modest epidemiological data on validity of case detection in the community.
The Reynolds Adolescent Depression Scale (RADS) is specifically for adolescents aged 13–18 years (Reynolds, 1987). It has well documented reliability and validity and normative data obtained from school settings in the manual but there are few independent studies reported using this measure. What data there are (including unpublished reports cited in the manual) suggest that the scale has a rather high false negative rate (30%) at the suggested cut-off score for clinical depression and is not particularly effective at detecting change (Radloff, 1977; Brooks & Kutcher, 2001).
The Center for Epidemiological Studies – Depression Scale (CES-D) was developed for use in community studies of adults and subsequently used in adolescents (Radloff, 1977). The overall view is that this scale does not have any clear strengths and many weaknesses when used with adolescents (Garrison et al., 1991; Olsson & von-Knorring, 1997; Brooks &Kutcher, 2001). In the younger age group this scale measures general non-clinical emotional turmoil rather than depression.
The Kutcher Adolescent Depression Scale (KADS) shows good reliability and validity and promising sensitivity and specificity (AUC 0.89). There is a very brief 6 item and a longer 16 item version. The brief screen may be effective in ruling out depression in community samples and appears better than the BDI (LeBlanc et al., 2002).
There are other depression instruments in the literature but the above have the most evidence base on which to form a judgement regarding reliability, validity and clinical utility. However the Birleson Depression Inventory deserves mention (Birleson et al., 1987). This has been used in studies of anxiety, post-traumatic stress disorder and depressive conditions (Kashani et al., 1989; Yule et al., 1990; Yule & Udwin, 1991). The psychometric properties of the scale are not well described but clinical use suggests that there are very similar component properties to the Beck Depression Inventory and the MFQ.
4.2.8. Interviewer-based instruments
There are four instruments available for assessing the diagnosis of depressive syndromes using direct face-to-face interview procedures (see Appendix J for further details).
The Schedule for Affective Disorders and Schizophrenia for School-Age Children (K-SADS) is a well used reliable and valid procedure for diagnostic assessment of depression including severity of current episode (Kaufman et al., 1997). It is interviewer led, very time consuming and designed to be used by trained individuals with some clinical experience. It is intended for use in participants aged 6–17 years. Originally focused on patients with current psychiatric disorder the most frequently used version has been used in community studies and can assess current and past lifetime episodes according to DSM criteria. Reliability is acceptable but studies are few (Kaufman et al., 1997). There is reasonable evidence for validity including the severity scales, predicting depression over time and diagnosing comorbid disorders with depression (Ambrosini et al., 1989; McCauley et al., 1988; McGee & Williams 1988; Herbert et al., 1996). The instrument is strong in the assessment of depression and good for detailed psychopathology evaluations. However, because it is time consuming it is not ideal for everyday clinic use and it is an inefficient means of assessing change in symptom severity. A brief screening version has been used in one community research project which may prelude a more flexible screening tool in future studies, particularly in combination with a self-report instrument (Goodyer et al., 2000a).
The Diagnostic Interview Schedule for Children (DISC) is a highly structured interview that is respondent-based (Costello et al., 1985; Edelbrock et al., 1985). Its strengths are that it can be given by non-clinical personnel in community settings after a few days' training. The psychometric properties are suspect for depression with high estimates of depression obtained with its use in epidemiological and clinical studies. There are however considerable data with this instrument on a range of psychiatric disorders using both a full and acute down screening version (Shaffer, 1988; Fisher et al., 1993; Shaffer et al., 1993; Schwab-Stone et al., 1996; Shaffer et al., 1996; Shaffer et al., 2000; Lucas et al., 2001). The instrument is not particularly suited to assess change in symptoms.
The Diagnostic Interview for Children and Adolescents-Revised (DICA-R) is also a respondent-based interview with somewhat better features than the DISC that result in quite good validity for clinical depression diagnoses in those detected (Herjanic & Reich 1982; Reich et al., 1982). The psychometric properties are acceptable to good with more recent versions to be used with both child and parent (Welner et al., 1987; Reich, 2000a; Reich, 2000b). The instrument can be used in both community and clinical populations but its relationship to DSM-IV diagnosis is unclear. There appears to be, however, a potential tendency to under diagnose depression in adolescents whilst over-diagnosing externalising disorders although there is modest data overall in this regard.
The Child and Adolescent Psychiatric Assessment (CAPA) is a detailed interviewer led instrument that is good at delineating clinical depression and other diagnoses (Angold & Costello, 2000). Lay interviewers can use it with training. The CAPA incorporates interviewer and respondent-based approaches. There is an extensive glossary for interviewers that details operational definitions for symptoms, distress ratings and symptom frequencies which interviewers make. There is a child and a parent version. The instrument is highly reliable for diagnosing depression (Angold & Costello, 1995). There is as yet no concurrent validity study for the instrument but considerable support for its concurrent validity has accumulated from other sources, twins, sex differences and family studies (Angold et al., 1996; Costello et al., 1996a, 1996b; Angold et al., 1998). The strengths of the instrument are in the highly reliable diagnosis of depression in the 9- to 16-year-old population of both sexes; that lay persons can use it after 2–4 weeks training; and the extensive glossary for standard coding within and between interviewers. In its current form the CAPA is intended as a research tool. It has the required clinical framework and potential to be modified and considered for use in clinical practise subject to the necessary field trials.
The Development and Well-Being Assessment (DAWBA) is a novel package of questionnaires, interviews, and rating techniques designed to generate ICD-10 and DSM-IV psychiatric diagnoses on 5–16-year-olds. Non-clinical interviewers administer a structured interview to parents about psychiatric symptoms and resultant impact. When definite symptoms are identified by the structured questions, interviewers use open-ended questions and supplementary prompts to get parents to describe the problems in their own words. These descriptions are transcribed verbatim by the interviewers but are not rated by them. A similar interview is administered to 11–16-year-olds. The DAWBA has a growing track record as an epidemiological measure in UK populations, and may prove to be of general clinical value too (Goodman et al., 2000). Further examination of this instrument for use in clinical practise with depressed patients could be considered but its specific value over and above the CAPA and the K-SADS for the diagnosis of depression is not known.
4.2.9. Clinical summary
4.2.9.1. Self-report questionnaires
There are a number self-report measures available for screening in community and clinical populations. There is very little comparative data between the available questionnaires. With the exception of the MFQ there are no developmental sensitivity data. The evidence suggests that a self-report questionnaire approach for diagnostic screening of depression in pre-pubertal children is not advised. For adolescents the MFQ is amongst the most studied and the most robust.
4.2.9.2. Interviews
The interview measures available were not designed to act as screens. Their current form makes them unlikely to enhance the screening process. A direct screen interview would be highly desirable for certain settings such as in residential care, with learning disabled patients or others with special needs limiting the use of self-reports. Again, computer-based interview procedures have yet to be made available.
Overall universal screening for depressive disorders in the community at large is not recommended. In addition there is no evidence to screen very high-risk groups (e.g. looked after children, asylum seekers and refugees, and those with exposure to multiple risk). Current available tools, both self-report and interviewer-based instruments are potentially important adjuncts in the detection of depressive diagnoses in symptomatic individuals and those of concern to child professionals.
4.2.9.3. Conclusion
Depressive disorders in children of primary school age are unlikely to be detected using paper and pencil tests. There is insufficient use of computer technology and more child friendly methods of assessing current mood and feelings. Pictorial (drawings and art work) and interactive methods should be examined for future use. In primary care, any screening instrument should be user-friendly for the health worker. Computer versions could assist by automatically scoring and even recommending action on the basis of resultant scores. This may improve take-up of screening devices in schools, other community settings and even busy clinical services looking to improve the time it takes to determine the needs and assessment pathway of new referrals.
Despite these caveats the evidence is that child mental health policies have been influenced by the findings using the available instruments. For example screening programmes have been utilised in schools as part of intervention programmes for depressed children and young people (Andrews et al., 2002; Burns et al., 2002). Clinical services can now consider the use of self-reports as adjuncts to the standard clinical process. Uptake is probably influenced by a large perceived increase in workload compared with the small clinical gain over standard clinical procedures.
There is a reason to be optimistic that a second generation of screening devices could be used in primary care and clinical services. The biggest obstacles will be delivery by an overworked professional workforce poorly trained and/or supported in computer aided assessment tools. Computerised devices that allow item responses to be converted into scale scores and into written advice options is likely to greatly enhance their use. This technology could be applied to questionnaire, interview and non-verbal data from drawings and art work. Pilot studies are required in primary and secondary care settings on computer-aided devices as an adjunct to assessment of face-to-face interviews. These could be usefully be carried out in schools and clinics.
4.3. Risk factors
4.3.1. Introduction
4.3.1.1. What is meant by risk
Risk is the degree to which the likelihood of a given adverse outcome will occur following exposure to a defined toxic agent. The relative importance of exposure is estimated by the probability of the outcome occurring in a given population compared with the level of occurrence in a non-exposed population. Risks for depression occur from a variety of sources both within and external to the child. For example, individuals may be born with genes that render them susceptible to depression, acquire lesions such as head injury that alter their ability to control mood, suffer infections that result in altered brain metabolism, be exposed to chronic family discord or to negative peer group environments that alter the development of emotional processing and self-percept (Goodyer, 2001). In addition, there may be risks associated with poor housing or living in a violent or dangerous society. Almost all research concurs that the onset of clinical depressions occur as a consequence of multiple rather than single risk effects that are frequently not independent of each other (Kraemer et al., 1997). There is however less agreement about how risks exert their effects over time or what they do to the individual to bring about psychiatric signs and symptoms and functional impairments.
4.3.1.2. Relations between risks over time and the magnitude of their effects
The size of the association between a risk factor and onset of disorder indicates its potency and is the maximal discrepancy achievable between depressed and not depressed groups exposed and unexposed to risk. For example exposure to severely and personally disappointing life events in adolescents occurring in the month prior to onset of depression is estimated to increase the risk for depression about nine times over not being exposed (Goodyer et al., 2000b). These estimates regarding one type of risk can be misleading as seldom are all the known adversities measured in one study. When measuring a range of possible risks in the same study we need to know three things: i) if risks that occur at a distance in time (i.e. months and years previously) influence the occurrence and the effects of more recent adversities such as acute personally undesirable life events; ii) if these distal processes themselves increase the liability for depression regardless of proximal risks and; iii) if there is some form of combined effect arising from exposure or possession of risks occurring distally and proximally not explained by one set or the other.
For example 60% of all adolescents with depression are exposed to acutely disappointing life events in the month prior to the onset of the disorder but more than 90% are already exposed to two or more previous ongoing risks either in their social environment or within themselves (Goodyer et al., 2000a; Goodyer, 2001). The impact of the recent adversity can only best be appreciated by taking into account the contribution of past risks on both the liability for the recent event and the onset of disorder. Current evidence from adult studies shows that there are very likely to be multiple risk pathways that may lead to the emergence of depressive illnesses (Kraemer et al., 1997). These involve genetic predispositions, different types of adversities occurring during the first two decades of life and acute personally disappointing life events not a consequence solely of past difficulties in the weeks prior to onset (Kendler et al., 2002). Adolescents at high risk for depression are exposed to, or possess on average, three psychosocial risks in the 12 months before follow-up (Goodyer et al., 2000b). Around 1 in 5 of this high psychosocial risk population will get depressed over the ensuing 12 months. Thus even amongst those at very high risk a significant number do not immediately become depressed. The presence of an acute event considerably increases this liability.
4.3.2. Typology of risk
Environmental risks are invariably classified by their:
personal characteristics (e.g. accident, illness including post-infective mood states, financial etc.)
latent psychological process inferred from these (e.g. disappointment, danger)
personal focus (self, parent, friend, etc.)
origin (self-induced, independent of self)
time of onset and (less frequently) offset giving duration of exposure
locus of control (uncontrollable by self, controllable)
age and developmental stage of exposure (prematurity, infancy, childhood adolescence, pre- or post-puberty).
Unfortunately there is no agreed standard definition for classifying risks and most studies use widely different methods and classification processes.
4.3.3. Social risks
Social adversities that are most associated with the onset of depression are those that are outside the child's control, occur as unpredictable happenings in the daily environment and recur over time. They mainly arise within family relationships or within friendships and are largely interpersonal in nature (Rueter et al., 1999; Goodyer, 2001).
4.3.3.1. Family risks
The most common group of adversities to occur within the family, which are relational in origin and produce negative effects on the child, arise from dysfunctions between two or more people. Perhaps the commonest of these are marital discord and emotional difficulties between one parent and the child, although parental psychopathology may underlie a significant proportion of these (Hammen & Brennan, 2003; Hammen et al., 2004).
The impact of events within the family on the child, such as physical maltreatment, are also associated with the onset of depression, but the onset appears often to be at a considerable distance in time from such abuse events (Jaffee et al., 2002). However both violence and sexual abuse to the child by parents, siblings or strangers are associated with depression, as are severe acute family difficulties such as sudden death, serious physical illness in a close relative or sudden separation of parents.
In contrast, unhappy marriages, parents being away from home due to work, low income, poor housing and living in a deprived neighbourhood occurring singly are not strongly associated with clinical depressive onsets in young people. Overall mild ongoing dysfunctions in family life do not appear on their own to be markedly associated with the subsequent onset of clinical depression (Tamplin et al., 1998; Tamplin & Goodyer, 2001). However, persistent family disagreement through early adolescence does increase the general level of low mood and depressive symptoms over time (years) and it is this rising level of non-clinical negative mood and thoughts that is associated with the onset of later clinical depression in older adolescents (Rueter et al., 1999).
Those children with higher IQ, better family functioning, closer parental monitoring, more adults in the household, and higher educational aspiration are less likely to show depression in the presence of elevated psychosocial risk (Tiet et al., 1998). In the absence of these protective or buffering factors the risk for both emotional and behavioural difficulties arises when children and young people are exposed to adversities. The more the family environment is chronically emotionally neglectful, involves chronic marital discord and a lack of authoritative parenting (the ability to be firm and clear within a positive emotional environment), the greater the risks for psychiatric disorders in general including personality difficulties in young adult life. Psychiatric disorder in a parent is another high-risk adversity for the child, with parental history of recurrent depression over the lifetime of the child strongly associated with depression in the offspring (Hammen et al., 1990; Hammen & Brennan, 2003).
Depression runs in families with a resultant increased risk for depression in the offspring of adults with a history of depression. Adults with a history of depression may also have a dual diagnosis such as substance misuse or alcoholism (Stallings et al., 1997). Thus there is an increased association between depression, substance misuse and alcoholism in parents and psychiatric disorders in offspring. Children and adolescents within such families may therefore be at risk for a range of undesirable outcomes including depression, behaviour disorders and substance misuse.
4.3.3.2. Friendship risks
Non-family-based adversities are also associated with the onset of depression and other psychopathologies in young people. Children with poor friendships, characterised by low numbers of friends, infrequent contact and no intimate relations, are more likely to develop depression as well as deviant behaviours and increased social isolation from the desired peer network (Goodyer et al., 1990; Cairns et al., 1995; Bukowski et al., 1996; Hartup, 1996). This appears to be independent of family strengths and weaknesses. The most potent form of acute negative life event is that of a recent (last few weeks) severe personal disappointment (i.e. the failure of a previously held belief in an expected outcome) with a close friend (Goodyer et al., 2000b). When recent personal disappointments with a close friend arises its effects as a risk factor is particularly large in those with previous psychosocial risks (Goodyer et al., 2000b). Depression is markedly increased in the presence of multiple adverse experiences involving both longstanding family and more recent friendship events and difficulties. Under these social conditions, the child may not perceive an emotionally supportive relationship in their social world.
4.3.4. Individual risks
4.3.4.1. Genetic risks
Current evidence suggests that while genetic factors appear somewhat less important in child onset depression, there are genetic contributions to adolescent depression (Rice et al., 2002). The environmental processes may be similar in nature but the implication is that these are sufficient to cause depression in the pre-pubertal child but insufficient in the post-pubertal adolescent. The studies on which this review is based are twin samples in which depressive symptoms are the outcome rather than clinical disorders. It is not clear if genetic factors are low in pre-pubertal children with clinical depressions, which are rare in this population (Angold & Costello, 2001). The precise genes involved remain unknown. In addition, the genetic risks may not act directly to produce the disorder, but act through increasing the liability for other risks in the environment. There may be a complex patterning of gene-environment interactions combining to cause depressions in the post-pubertal depressed adolescent (Caspi et al., 2003b). In contrast, direct associations (and therefore effects) of single genes with depression are uncommon (Henderson et al., 2000; Zill et al., 2002).
4.3.4.2. Temperament
Children and adolescents (as well as adults) with a highly emotional temperamental style (react quickly to everyday events, easily brought to tears, easily soothed) are more likely to be depressed than those low in these behavioural characteristics (Goodyer et al., 1993; Hodgins & Ellenbogen, 2003; McWilliams, 2003). Although this is true for both sexes, more girls than boys have this temperament and this may be one component that differentially increases the risk for depression in females over males. The evidence suggests that there are genetic influences on individual variations in temperament (Eley et al., 2003; Sen et al., 2004). The relationships between temperament and personality development over time suggests that there is coherence across time between the commonest used in both terms although the precise definitions appear to be somewhat different (Caspi et al., 2003a; Shiner et al., 2003). The precise relations between personality and later depression remain unclear but neuroticism shows an important but complex relationship with depressive onset (Kendler et al., 2004).
4.3.4.3. Cognitions
As well as emotional styles, there are thinking styles that increase the liability for depression. High levels of particular types of self-critical thoughts known as global self-devaluations (thinking of oneself as abandoned, a failure, feeble, incapable, a loser, a mess, pathetic, pitiful, rejected, stupid, unlovable, unwanted, useless, worthless), if present at times of low mood are significantly associated with clinical depression (Teasdale & Cox, 2001). A ruminative style, in which young people dwell or even perseverate on a particular thought, also increases the risk for depression. Adults and adolescents with both global self-devaluations and a ruminative style have markedly increased risk for depression (Alloy et al., 1999). Ruminating lowers mood and increases memory difficulties in adolescents (Park et al., 2004).
4.3.4.4. Physiological risks
Studies of physiological factors as risk components for depression in young people are relatively new and few have been published. There is some evidence that both the monoamines and glucocorticoids are implicated in the biology of depression in children and adolescents (Birmaher & Heydl, 2001). Children with a positive family history of depression show abnormalities of serotonin function even when well, suggesting serotonin vulnerability for subsequent affective disorders. Increased cortisol and a second adrenal steroid dehydroepiandrosterone (DHEA) are both elevated and predict the onset of depression in a subset of adolescents at high psychosocial risk for depression (Goodyer et al., 2000a). Elevated cortisol levels may themselves arise in part from interpersonal difficulties in early parenting related to maternal depression (Halligan et al., 2004). High risk children and young people with no history of prior depression but with a positive family history for the disorder have also been shown to have abnormalities in sleep architecture associated with subtle changes in cortisol secretion (Dahl et al., 1996; Feder et al., 2004). Overall the evidence suggests biological vulnerabilities in both the serotonin and the adrenal steroid systems. These are likely to be brought about by a combination of genetic and environmental influences.
4.3.5. Very high-risk groups
Within the child and adolescent population at large there are known groups at very high risk for mental health difficulties including depression. These are already the focus of policy review and include looked after children, refugees, the homeless and asylum seekers. Children and adolescent offenders, particularly those in secure institutions, are particularly at risk for mental difficulties. The known numbers of successful suicides in young offenders strongly indicates high levels of depression that currently may not be adequately assessed or managed. It is unclear if ethnicity exerts a specific risk for depression above and beyond the known increase in social, behavioural and emotional difficulties for selected populations (e.g. Afro-Caribbean). Maltreatment as risk has already been mentioned but ‘Hidden maltreatment’ should be considered in children with adolescents with unexplained mood disorders with no family history of depression and an absence of other overt social adversities.
4.3.6. Special risk groups
Finally there are some families and individuals who have a known set of risk actors whose precise theoretical mechanism (vulnerability, activating or formation) is unclear. These include children with a physical or a learning disability. Disabled children are more at risk for mental illness and behavioural problems including depressive disorders, compared with the population at large (Dekker et al., 2002; Goodman, 1998; Martinez & Semrud-Clikeman, 2004). Because of their visible handicaps, challenging behaviours and/or their more overt educational difficulties, mood disorders may be easily missed in such individuals. Likewise adolescents with complex endocrine diseases, adverse reactions to drug treatments, pervasive developmental disorders, autism and Asperger's syndrome are at risk greater than would be expected by chance or by the effects of being physically or developmentally impaired. Clinical services may need to consider depressive disorders in these adolescents when social withdrawal and/or irritability are presenting features or there is a persistent exaggeration of their obsessional habits and mannerisms suggesting a mood disorder.
4.3.7. Summary
Risks for depression are multiple in origins and may be correlated with each other. Single risks resulting in the onset of clinically meaningful depression are rare.
The majority of first depressive episodes arise in adolescents compared with children and in the presence of at least two and invariably three long-standing psychosocial risks.
Acute life events are key destabilising elements in those already at high psychosocial risk evoking relatively sudden onset in about 50% to 70% of cases. The other third to a half appears to arise more slowly through chronic persisting interpersonal difficulties.
Genetically mediated factors via the serotonin and adrenal steroid systems may be important features in determining potency of social adversities.
The intermediate psychological vulnerabilities for adolescents between physiology and the social environment are a high level of global self-devaluative thinking at times of low mood in combination with a ruminative thinking style.
There is increasing evidence that the pattern and potency of risks varies with development, severity and number of episodes of depression. The physiological risks for recurrence appear to be greater with an increasing number of past depressive episodes suggesting an effect of depression on brain function.
4.3.8. Risk classification
It is critical to remember when looking at this list that the specificity of individual risk factors to the onset of depressive disorders is low to moderate, with the exception of those starred * where specificity is high.
4.3.8.1. Probable vulnerability factors
These increase the general liability to but seldom directly provoke disorder:
Presence of short arm serotonin promoter gene
Elevated morning cortisol levels
Acquired fetal infections
Maltreatment or emotional neglect through infancy
Maternal postnatal depression
Parental history of depressive disorder*
Brain illnesses in childhood including trauma and infection
Being female*
Being post-pubertal*
Divorced parents
Chronic parental psychiatric illness.
4.3.8.2. Probable activating factors
These are directly implicated in the onset of depressions and in the presence of vulnerability factors their effects can be large:
Personally undesirable life events resulting in permanent change of interpersonal relationships in friends or family*
Acute brain illnesses
Community disasters such as war, famine and infections
Personal assault.
4.3.8.3. Formation factors
These are responsible for the clinical characteristics of the depressive state:
4.3.8.4. Known risk factors whose precise role is currently unclear
These may be vulnerability, activating or formation factors but currently available information does not permit the classification of their role:
4.3.8.5. Protective factors
These reduce the likelihood of depression in the presence of vulnerability and activating factors:
A good sense of humour
Positive friendship networks
Close relationship with one or more family member
Socially valued personal achievements
High normal intelligence.
