Abstract
Our purpose in this paper is to outline current research findings and directions as they pertain to adolescent mental health and mental health morbidity. We highlight three research areas of particular relevance to meeting the current challenge in detecting, treating and preventing mental health disorders in young people: (i) the scope and burden of mental health disorder; (ii) the nature of the causal pathways that lead to mental health and mental illness; and (iii) the science of prevention. While our review is centred on the mental health of young people, we should comment at the outset that we have included children in this view. Adolescence represents a central developmental transition and an opportunity to observe changes in life-course trajectory. However, young people entering the period of adolescent transition do so with, on average, 12 years of prior experience. We believe an understanding of the current research effort in adolescent mental health can more properly be achieved by acknowledging the contribution of childhood to the life-course trajectory.
The scope of child and adolescent mental health morbidity
It has now been 9 years since we first conceived of the concept of using an epidemiological survey as a vehicle to better child and adolescent mental health promotion, prevention strategies and treatment. Since then, the findings from the Western Australian Child Health Survey [1–3] have been widely used within the Western Australian health sector and, very pleasingly, across sectors such as Education, Family and Children's Services and Justice. These findings show that child and adolescent mental health morbidity in Australia is characterised by several factors: burden, increasing prevalence, downward developmental trend, persistence, unmet need and visibility.
Burden
By any standard, economic, social, or personal, the burden associated with poor mental health is very large and increasing. It has also been significantly underestimated. In 1990, of the 10 worldwide leading causes of disability, measured in years lived with a disability, five were mental health conditions. Using measures of disability-adjusted life-years Murray and Lopez [4] have shown that mental health disorders emerge as a highly significant component of global disease burden when disability, as well as death, is taken into account. Their projections show that mental health conditions could increase their share of the total global burden by almost half, from 10.5% of the total burden in 1990 to almost 15% in 2020. This is a bigger proportionate increase than that for cardiovascular diseases.
At a more local level, the burden that these disorders place upon individuals, their families and communities is very substantial. Nationally there are 219 child and adolescent psychiatric beds that generate an annual cost of $21 million [Thompson I: personal communication]. In our own state, Western Australia, the annual cost of child inpatient care for mental health disorders was $2.04 million, representing the cost of 6146 bed-days generated by 90 inpatients. (These figures do not include costs associated with Princess Margaret Hospital for Children, the major tertiary treatment facility in Western Australia. This Hospital has an eight bed inpatient ward with an annual operating cost of approximately $708 000. These data do not permit deriving a per capita cost for the median hospital stay.) This makes the average cost per occupied bed-day for a child with a mental health disorder comparable to that of an adult ($333 vs $321), and with a median length of stay of 51 days this results in a treatment cost of $16 983 per child [5].
Finally, the vast majority of the fortunate few children who receive mental health care receive it in outpatient settings. Again using data from Western Australia in 1996–1997, 1168 children attended outpatient mental health facilities and generated 7485 attendances [5]. There was a mean of 6.4 visits per episode. While estimates of the per capita outpatient cost for this year (1996–1997) were unavailable, previous estimates showed an approximate cost of $1960 per course of outpatient mental health treatment [6].
Increasing prevalence
While more apparent than before, child and adolescent mental health disorders have also experienced a net gain in prevalence. Since World War II, there has been a discontinuous increase in key morbidities that include child abuse and neglect, crime and conduct disorder, alcohol and drug abuse, depression and suicide [7]. It is also important to note that although death from injury has declined over the same period as the new morbidity has emerged, other causes of death have declined more rapidly and injury now comprises a significantly higher proportion of all childhood deaths, having increased from approximately 12% in 1968–1969 to 19% currently.
The past 20 years have seen substantial developments in international epidemiological studies of mental health and wellbeing. Many of these studies have included adolescents in the lower age range of otherwise adult samples [8 ,9] and others have specifically sampled children and/or adolescents. In concert with these international efforts are their Australian counterparts [10] including the recently completed 1998 National Survey of Child and Adolescent Mental Health, the publication of which is currently in preparation.
In the main these studies document a point prevalence of mental health disorders in children and young people of the order of 16–20% between the ages of 4 and 11, rising to 20–25% between ages 12–17 and peaking at 25–40% between ages 18–24. At present there are 5.6 million children and young people under the age of 25 living throughout Australia, based on 1996 Australian Bureau of Statistics Census figures of 18.5 million people with approximately 30.7% aged 24 years or less. A total of 21%, or 3.8 million were aged under 15. So at any one time there are about 393 000 4–11-year-old children, another 370 000 12–17-year-olds and 634 000 young people aged 18–24 with identifiable mental health disorders. These figures yield a total of 1.4 million children and young people with identifiable mental health problems and an average prevalence in this developmental age range of about 23%. Quite importantly, these figures, particularly those for young adults, may still be an underestimate of the total prevalence [11].
Downward developmental trend
The increase in the population rates of mental health problems conceals a more troubling developmental trend, namely that rates of these morbidities in children and young people are approximating the adult rate at earlier developmental epochs [7]. This increase in the prevalence and change in the developmental trend of these disorders is almost certainly driven by fundamental changes in rates of biological risk exposure as well as large changes in basic economic, political and social environments. Also reflected in this trend is the fact that many of these disorders (e.g. conduct disorder) become risk exposures (or inputs) for a second stage morbidity. These second stage outcomes (e.g. juvenile delinquency) are particularly devastating. Once established, they tend to be notoriously resistant to currently available treatment and moreover participate in a chain of intergenerational transmission.
Persistence
Mental health disorders also display considerable developmental persistence. The fantasy that children ‘will grow out of it’ proves to be just that, fantasy, in light of the longitudinal evidence of persistence of disorder. For example Newman et al. have shown that 73.8% of 21-year-olds with a mental health disorder had a developmental history of mental disorder [11]. Offord et al. using longitudinal data demonstrated that 61.3% of children aged 8–12 with a diagnosed conduct disorder had had at least one of three major mental health disorders 4 years previously [12].
Unmet need
Any reading of the above characteristics of the epidemiology and burden of child and adolescent mental health disorders leads towards a commonly acknowledged reality. Namely that the gap between prevalence and treatment is vast and many of the children most in need do not find their way into any care at all, much less into specialised care. Meeting the demand for service will require not only more services, but better targeting and use of existing services. However, even with this, the demand will never be adequately met and treatment does little to address more fundamental causes. What is also required is investing in the science of preventing mental health disorders and in strategies that promote and protect good mental health and well being.
Visibility
The consequence of a higher prevalence, persistence and unmet need has been to make child and adolescent mental health disorders more visible across a wider sector of government and community activity. These sectors span criminology, education, public health, child, adolescent and adult mental health and family and children's services. Child abuse, early school drop-out, truancy, depression, alcohol and drug abuse, juvenile offending, unemployment, violence, crime and youth suicide are key outcomes of interest across these sectors. Families, schools and community leaders are now expressing concern about these problems and demanding strategies to combat them.
Comment
What can be said about the scope of child and adolescent mental health morbidity in Australia? First, we now have access to adequate population prevalence benchmarks from which to plan and monitor service development, particularly for primary mental health care. Regional requirements for fine-grained data however, represent a continued need. Most particularly, mental health information about small populations, particularly children, adolescents and their families of Aboriginal and Torres Strait Islander descent, remains critically needed. Second, we also now have a clear demonstration of the gap between prevalence and treatment. This demonstration, using local data, has been needed to influence policy, planning and political will. This gap is not likely to change and if it does, will do so only marginally. The more urgent need now is to know more about what treatments are effective for which children and what strategies can be implemented to reduce the population prevalence of mental health disorders in children and adolescents. This leads to a discussion of causal pathways.
Causal pathways
Cause and risk
Epidemiological data systems have been developed to measure the prevalence of mortality and morbidity and of health service utilisation, to predict future need and identify factors associated with disease. They have produced significant advances in our knowledge of the changing patterns of disease in response to environmental changes and have suggested underlying causes. Today, however, these systems are proving less effective in guiding the development of prevention and treatment, particularly for child and adolescent mental health disorders. Why is this so?
The origins of epidemiology lie in the investigation of infectious epidemics. Infectious agents, once discovered came to be viewed as the sole, and sufficient, cause of epidemics. This resulted in the dominance of the belief in ‘single sufficient causes’. Ironically, the incidence of infectious diseases started to fall before the introduction of interventions that targeted the infectious agents themselves. This decline in incidence reflected social measures such as provision of clean water, less crowding, better nutrition, improved literacy and education. The fact that epidemics of infectious disease could be controlled by addressing such social factors demonstrated that the social factors lay on causal pathways to infection that were very complex.
Many diseases have been successfully eradicated or treated by addressing a single factor, encouraging the expectation that a ‘miracle cure’ or ‘magic bullet’ will be found for every health problem. However, the heterogeneity and causal complexity of mental health disorders limits the usefulness of collecting disease outcome data in the absence of detailed information of the many factors that may lie on the causal pathways to these newer disorders.
It is now recognised that most of the diseases that represent a significant burden to human populations are multifactorial. Many are caused by the joint action and interaction of genes and environment, this is what is meant by a ‘complex’ disease. Mental health disorders are some of the most prominent of these ‘diseases’. Consequently, it is seldom possible to identify a single principle cause analogous to an infectious agent. In view of this, an aetiological determinant is properly viewed as being any factor that modifies (as distinct from being merely associated with) the risk of disease [13]. In mental health, this approach to causation is crucial for two reasons.
First, the determinants that predict persistence of a disorder (prognostic variables) are not the same as those that predict onset (risk variables) [12]. The former determinants are critical to treatment while the latter are critical to prevention. This distinction is of central concern to those who wish to advance the prevention of mental health disorders. Risk variables may no longer be current by the time the patient sees a mental health practitioner and their control (if still current) may be irrelevant to treatment. This is one of the reasons why mental health practitioners are reluctant to engage in prevention activities. A practitionerled field will understandably approach disorder from the standpoint of the determinants that predict persistence of the disorder and its responsiveness to treatment, which are not necessarily the same as those predicting onset, which have to be addressed for prevention. If prevention intervention involves the manipulation of factors that mental health practitioners are neither equipped nor in a position to influence, it may be quite inappropriate that they should be expected to do so. This is particularly true if they are already overburdened with patient treatment; they are not in contact with persons until the first manifestation of disease (by which time it is too late to consider prevention) and our structure of practitioner recompense is based on a per capita basis, rather than on maintaining the health of the population. The whole point of prevention is that it is invisible. In practical terms, as soon as a person becomes a patient, prevention has failed.
There are of course, many other factors that prevent the mental health field from engaging in prevention. The prevention evidence for mental health is only now being consolidated and assessed; the infrastructure to support prevention intervention, policy development, training and evaluation is generally lacking; and there are long-standing organisational cultures in health and mental health that impose significant challenges to implementing prevention at any level.
Second, normally a finding that a risk factor is only a weak cause of a disorder has resulted in little if any effort being spent either in determining the nature of the association or in attempting to prevent it. However, as Doll points out, if a large population of individuals is exposed to a weak causal factor, then preventing or interrupting the exposure to this risk factor can result in valuable reductions in health and social burden [14]. Mental health disorders are complex disorders, that is, they involve the complex interplay of a number of causal factors. Very importantly, large populations are being exposed to multiple risks that have weak causal associations to the development of these disorders. A critical feature of this pattern of risk exposure is that multiple risks have a cumulative effect on outcome. A consequence of this pattern of exposure is that a large number of individuals exposed to a small risk may generate many more cases than a small number exposed to a high risk [15]. Furthermore, preventive efforts that secure a large benefit for the community bring relatively little benefit to each participating individual. In other words, the benefits of mental health prevention are seen and best understood by examining the effects on whole populations, not at the individual level. An understanding of these causal features underscores the current emerging interest by other fields, notably public health, in population approaches to mental health prevention.
The shared and non-shared environment
Mental health researchers working with causal frameworks through the 1990s have witnessed a rapid growth in studies of the genetic and biological influences that interact with specific environments to determine mental health outcomes [16]. This form of research may involve two quite different classes of variable. On the one hand, it may focus solely on observed variables. Typical examples of such variables include: (i) the set of questionnaire-based responses; or (ii) the record of the full genotype for each individual in a data set at a sequenced genetic locus. On the other hand, research may also be based upon unobserved (or unobservable: latent) variables. For example, one may be interested in the identity unknown and causal impact of as yet unknown genes or unknown environmental determinants.
The ability to work with ‘unobserved’ variables, generally by modelling their effect on the correlation of residual outcomes within natural groups in a data set (such as families), is extremely useful. However, it is essential that the limitations of such work are recognised and that its proper role is understood. This can nicely be illustrated from the perspective of genetic research. Gene discovery is extremely expensive, and it is therefore sensible to precede any molecular work with a study of the familial aggregation of a phenotype of interest. The presence of such aggregation does not prove that important genetic determinants exist, but in its absence it would be unwise to commit extensive resources to a search for genes. Similarly, if the data structure is supportive, an attempt to resolve residual genetic effects from environmental effects is a worthwhile pointer to ongoing research directions. For example, if the data are consistent with the presence of unobserved genetic effects, it is more likely to be worthwhile committing resources to a search for them. These preliminary analyses can all be undertaken without knowing a single genotype. Researchers simply use the pattern of observed phenotypes and ask the key questions: (i) Is there any familial aggregation? and (ii) Does the pattern of the aggregation suggest the effect of unmodelled genes or an unmodelled environment? This is important knowledge, but it is fundamentally impossible to move on from here to localise or identify a single important gene variant unless one has access to some observed genotypes.
Likewise, it is a truism that even if the pattern of aggregation in a data set is known to be consistent with unknown environmental determinants, this says no more than that it may be worthwhile to investigate the environment more comprehensively. No substantive hypotheses can be tested unless one takes the fundamental step of obtaining ‘observed’ environmental data. Aetiological research based on latent variables is complementary, not an alternative, to research based on observed variables. The latter is expensive and often socially or biologically invasive and should, whenever possible, be preceded by the former. However, a proper understanding of the complex causal pathways underlying mental health and disease will never be attained without a comprehensive attempt to actually measure the key personal, familial and social variables of relevance.
A second key concept in the investigation of the aetiological determinants of mental health and disease is that of the causal sequence. If A causes B causes C causes disease D, simplistic approaches to analysis such as regressing (in a multiple or logistic regression model) D on A, B and C can be exceedingly misleading. For example, if A is the fundamental causal determinant and it is modifiable at a population level it is critical that any analysis identifies A as a key determinant. But, A is more distant from D than is C and if, for example, A is measured with greater error, it is quite possible for C to take up all the ‘effect’ of the causal pathway in a conventional regression model and for A to appear to be completely unimportant.
The first step in ensuring a rational approach to analysis is to recognise the concept of causal sequences. A second key step is to consider where in a causal sequence a particular variable may lie. Just as molecular biologists and genetic epidemiologists are working on the candidate genes for mental health disorders, our particular interest is in the specification of those variables in the environments of children and adolescents that can be considered as the candidate environmental exposures in the causal pathway of mental health or mental illness.
Because the current research focus has been on a search for genes, the environmental components have been statistically manipulated rather than directly measured. It is important to appreciate that the statistical techniques used by behavioural geneticists may be used independently of actually measuring anything in the external world of the child [17]. If the intent of gene-environment research is to document interactions between the environment on one hand and genes on the other (i.e. what aspects of environment are capable of switching which genes on and off and to what effect?), then real world measures of the environment that reflect the circumstances in which children currently live their lives are critically needed. Once again, the selection of which of those aspects of the environment to measure should be guided by hypotheses about those environmental determinants (i.e. risks) that are likely to be on the causal pathway in any gene-environment interaction.
Our current research seeks to understand the causal basis of child mental health disorders and their relationship to the environment by describing and measuring distal and proximal risk exposures (Table 1). We have chosen to call these effects ‘exposures’. Alternately they could be called ‘risks’ depending on the outcome that they are associated with. Our choice of the word ‘exposure’ here stresses our preference for a population perspective (rather than an individual or case perspective) in considering these effects. In this model, distal risk exposures are those that influence and characterise large populations. They operate at some distance from the immediate outcome (i.e. mental health disorder) of interest, but through their action morbidity may be potentiated. Ecologically, distal exposures are more likely to be used in descriptions of communities or nations and are of particular interest to social science. Thus, communities and populations may be described by their exposure to market deregulation, regional levels of poverty, or high rates of family re-formation. In contrast, proximal exposures are those that are closer to individuals and families. While these exposures have a substantial causal relationship to the health outcome of interest, mechanisms that influence them may overlap with contexts well beyond the control of the immediate family or individual.
Distal and proximal exposures associated with poor mental health outcomes
Taken together, these exposures are contributing to a high level of burden of mental health disorders across the Australian population. The time trends in the growth of these disorders suggest that significant factors in biological, psychological, social and ecological settings have independently and jointly contributed to the increase in the burden of mental health disorders. More particularly, this growth in mental health disorders has also occurred during a period of post-War prosperity and security in Australia's economic policy and fiscal environment. So the establishment of national economic prosperity and security is not a sufficient condition to ensure low rates of mental health disorders.
Comment
As with the previous section, we can ask here, what is causal pathway research contributing to our understanding of child and adolescent mental health problems and more importantly what remains to be done?
Causal pathway research demands the use of increasingly sophisticated paradigms that distinguish the determinants of the persistence of disorder from the determinants of the onset of disorder. Such research is inherently longitudinal and frequently requires access to population samples. Much of the current contribution of this research is focused on the search for genes related to mental health disorder. However, this will inevitably lead to the need for a better understanding and measurement of those features of the child's environment that interact with genes.
Research aimed at understanding these mechanisms represents a critical research investment. Such research should seek to define and measure the developmental timing, onset and offset of these exposures to better determine the mechanisms by which they operate. These measures will then need to be applied in appropriately drawn longitudinal samples that balance opportunities to observe gene-environment interactions.
The science of prevention
Prevention refers to those interventions that occur before the initial onset of a clinically diagnosable disorder. These intervention strategies may be universal, selective or indicated and they all seek to reduce new cases of disorder. In Australia, the Commonwealth Government under its National Mental Health Strategy has recently launched a mental health promotion and prevention national action plan to advance mental health prevention initiatives [18]. Additionally, the Australian Attorney-General's Department has released a comprehensive analysis of the developmental opportunities for preventing early onset of criminal behaviour in young people that has considerable significance for mental health practitioners [20].
Preventing mental health disorders in children and young people
To date the largest review of preventive intervention in the mental health field has been conducted by the US Institute of Medicine (IOM) [19]. This report concluded that, while more research was needed to show that preventive interventions can reduce the incidence of diagnosed mental disorders, there is robust evidence to demonstrate that risk factors associated with disorder onset can be reduced, as can sub-clinical symptom levels. The IOM report noted that the level of emerging evidence in preventive intervention was impressive and that the field was hampered by sporadic effort and under-funding, part of which reflects the reliance of this research area upon expensive and long prospective longitudinal studies to assess distal outcomes of intervention.
There is now an emerging body of evidence that supports the efficacy of early intervention in reducing later mental health burdens. This includes findings from large-scale, well-conducted longitudinal studies of randomised control trials, as well as the results of the implementation of universal, selective and indicated interventions that have operated outside of the traditional fold of mental health treatment research. These studies show an influence on mental health outcomes through interventions that build capacity and resilience (e.g. educational infant day care with selected populations), modify risks known to cause high mental health morbidity (e.g. with the use of parenting programs) or through the combined action of these (e.g. school programs that control aggression) (see Table 2). While more research is needed to understand their specific action, the efficacy of these programs is based in their capacity to reduce the more generic or non-specific burden of mental health disorders. They frequently reduce comorbidity by jointly reducing symptoms associated with delinquency, attention deficit hyperactivity disorder, depression and anxiety [19,20]. Our current research with children and young people seeks to know more about how these mechanisms work.
Mental health and wellbeing: indicators, mediators and significant others/settings
Mental health promotion
Some mention here should be made of mental health promotion as a preventive strategy. The IOM report has been criticised for its unduly narrow conceptualisation of prevention to the exclusion of strategies that seek to promote mental health and wellbeing [21]. The IOM view of prevention is driven by a focus on illness rather than on the enhancement of mental wellbeing or the maintenance of ‘wellness’. Thus, major health promotion strategies in areas such as childhood and adolescent resiliency and competency, which are critical to mental health risk modification, are not seen to be germane to the mental health prevention agenda.
The point to stress here is that the distinction between prevention and promotion is scientifically unsound, it artificially segregates theories and activities that are known to be on the causal pathway of mental health disorder and it alienates providers and communities. In Australia, where there is a vigorous health promotion field, there is an opportunity to create an intervention setting in mental health that integrates prevention and promotion through the fairbrokerage of good theory, science and practice. It is essential that this be done as it will bridge the gap between the science and practice of treatment, on one hand, and the science and practice of prevention and promotion, on the other.
Comment
What types of mental health problems in children and adolescents can specifically be prevented and what are the opportunities for research?
First, anxiety, depression, and conduct disorder (delinquency and aggression) are high-prevalence conditions which represent some of the most burdensome of mental health disorders in children and adolescents. They also represent some of the most potentially preventable mental health conditions [18–20]. Continued investment in research, especially of appropriate universal, selected or indicated trials that recruit children prior to a formal onset of one or more of these disorders is needed. Particular attention also needs to be paid to investing in evaluations of such programs when implemented in real-world settings and taking into account those opportunities that arise when embedding such evaluations within communities.
In addition to preventive interventions specifically targeting a reduction in the incidence of new cases of anxiety, depression and conduct disorder, Table 2 also illustrates some of the key indicators of mental health in children and adolescents. These indicators include such characteristics as positive self-worth, self-efficacy, attachment, engagement and life satisfaction. We still know relatively little about the developmental trajectories of these characteristics. Nor do we know much about how the mechanisms of known mediating factors such as good speech and language, pro-social and life skills, and coping ability, act to protect children and young people from subsequent mental disorder. These represent significant opportunities for mental health research.
Conclusion
Twenty-one per cent of the Australian population is under 15 and at least one in five may have a mental health disorder. By any standard, financial, social or personal, the burdens are large, and probably growing.
Research is needed in three areas: (i) epidemiological estimates of the burden of mental health disorders, particularly for indigenous children and young people; (ii) studies of gene-environment interaction are now underway, however, these require better measurement, in longitudinal studies, of those proximal exposures of the child's environment that are known to be on the causal pathway to disorder; (iii) universal, selective and indicated prevention trials and evaluations are needed; these should be directed at burdensome child disorders that are known to be responsive to prevention: Notably, anxiety, depression and conduct disorder.
Finally, our epidemiological perspective of the determinants of mental health problems in children is much clearer now than even 20 years ago. We know that more determinants of child mental health problems lie outside the mental health sector. Many determinants of persistence are also beyond the reach of clinicians. The implications are straightforward: preventive intervention and promotion will entail effective collaboration at national, state and local levels between health, welfare and education sectors. These must be informed by epidemiology and knowledge of the causal pathways of mental disorders. Intervention must also improve the interface between scientific knowledge, and policy and praxis. This will require a vision of the urgency, costs and consequences of mental disorders in young people coupled with effective leadership and political resolve.
Footnotes
Acknowledgements
We thank Anwen Williams and Jenny Kurinczuk for their input, the Western Australian Health Promotion Foundation, Rotary Health Research Fund and Ian Thompson of the Commonwealth Department of Health and Family Services for figures on the costs of mental health inpatient care.