Early Detection and Prevention of Mental Health Problems: Developmental Epidemiology and Systems of Support

Abstract

This article reviews the role of developmental epidemiology in the prevention of child and adolescent mental disorders and the implications for systems of support. The article distinguishes between universal or primary prevention, which operates at the level of the whole community to limit risk exposure before the onset of symptoms, and secondary or targeted prevention, which operates by identifying those at high risk of developing a disorder. It discusses different aspects of time as it relates to risk for onset of disease, such as age at first exposure, duration of exposure, age at onset of first symptoms, and time until treatment. The study compares universal and targeted prevention, describing the systems needed to support each, and their unintended consequences.

In this article we discuss how epidemiology, the study of patterns of illness in time and space, can help to prevent mental health problems by preventing either their occurrence or their causes. Preventing something is usually easier if we can reliably detect the problem, or its cause, or both. We tend to assume, probably with reason, that it is better to detect a problem or its cause earlier rather than later in the course of the disorder and that that systems of support will be strengthened if detection occurs early. A case can be made against this argument for early intervention (think of all those tonsillectomies and umbilical hernia operations later shown to be a complete waste of resources). But until proven otherwise, epidemiology is ethically bound to operate on the assumption that early detection is a service that it can and should provide to the health care system.

TYPES OF PREVENTION

What should we be trying to prevent? Of course we want to prevent disease, but disease has many precursors, consisting both of early symptoms and of risk factors that increase the likelihood of disease. Public health has tried to create a manageable taxonomy of risk by classifying exposures according to the level of response best designed to prevent either risk or disease (Gordon, 1983; Mrazek & Haggerty, 1994; Rothman & Greenland, 1998). At the most extreme level (“indicated” prevention) we enforce treatment, or even lock people away, to prevent their harming others or themselves. Isolating or quarantining cases for the safety of the community has been used for centuries to limit the spread of infectious diseases; psychiatry is one of the few areas of chronic disease medicine that still has recourse to this type of prevention.

“Targeted” prevention projects are designed for individuals who have symptoms of a disorder or a clearly recognized risk factor (e.g., parental bereavement as a risk factor for depression; Sandler et al., 2010). “Universal” or “primary” prevention describes public health activities designed to reduce risk for the whole population In this article we present the case that most mental health problems are prevented by aspects of the way we live that have little to do with identification, treatment, or “targeted prevention” at the level of existing symptoms. Epidemiology can be used to identify ordinary everyday aspects of life, like having two parents or being born full-term, which we don’t think of as mental health interventions but which can be shown to be protective. Sometimes, however, the good-enough system of daily life fails. At that point, epidemiology can be used to identify who needs care, of what kind, and to monitor the adequacy and effectiveness of that care. Taken together, these functions of epidemiology help to answer questions about the extent to which systems of support for the basic aspects of life are cost-effective for preventing mental illness, compared with the more focused systems of care that we bring into play when things go wrong. Many of the articles in this special issue take as their focus various emotional and behavioral problems for which specialized systems of care have been devised. Here, the concern is rather how many of those problems might never have developed in the first place given a solid groundwork of support. It turns out that not only do many kinds of primary prevention avoid the costs of screening and the pain of stigma but they often prevent a wide range of health and developmental problems, in addition to mental illness, and so can be highly cost-effective.

EPIDEMIOLOGY AND PREVENTION

As its name implies, epidemiology grew out of the struggle to prevent the epidemics of infectious diseases that swept countries and continents well into the first half of the 20th century. The job for infectious disease epidemiology is to discover patterns of disease in time and space that could explain why some people become sick and others do not. Understanding patterns of transmission enables systematic interventions to break the chain—from stagnant water to malaria, or unpasteurized milk to tuberculosis. The century up to World War II was a time of many famous victories against infectious diseases.

The challenge for the past half century, however, has been how to prevent not infectious or communicable diseases so much as chronic or episodic or noncommunicable diseases, mental illnesses among them. In the case of infectious diseases we now generally know the germs that cause them and can employ a fairly simple two-pronged attack: to exterminate the germs and to eliminate the environments in which they breed. So the old “systems of care” for infectious diseases—the plague hospitals and tuberculosis infirmaries—have dwindled to a small fraction of the health care system, and funding goes almost entirely to early universal prevention via inoculation against these diseases. Much of the cost of primary prevention is borne by nonmedical services like the Environmental Protection Agency, the schools, and the local water boards.

At present, the picture for mental health is very different. We are still at the stage of discovering the “patterns of disease in time and space” that may help to identify major risk exposures. On the whole, the “one bug–one drug” model that defeated many infectious diseases does not fit mental illnesses (syphilis, caused by the spirochete Treponema pallidum, is no longer treated as a psychiatric disorder). This has the corollary that experimental methods such as case-control designs for identifying causal factors rarely work either. Even if we have a causal hypothesis we cannot, for example, randomly assign newborn infants to nurturing versus nonnurturing mothers. To chart a course between true causes and confounders, epidemiologists often have to rely on observational methods or at best quasi-experiments (Cook & Shadish, 1994).

TYPES OF PREVENTION FOR CHRONIC DISEASE

In line with the well-established sequence of prevention levels, the sequence of interventions runs from health promotion, through universal prevention, to selective (high-risk) prevention, to indicated prevention (Weisz, Sandler, Durlak, & Anton, 2005). Different writers have slightly different emphases, but the general principles are similar. Briefly, health promotion enables people to increase control over their health and its determinants (World Health Organization, 2005); primary or universal prevention aims to enable people to escape disease by either avoiding exposure to risk factors or being able to resist it after exposure; secondary or targeted prevention aims to minimize harm from disease exposure by preventing full-blown disease; and tertiary prevention aims to restore functioning and reduce disease-related complications, relapse, or spread to others. A key difference among them is that different levels of prevention have different target populations. Health promotion and universal prevention focus on the entire population, either throughout the life course (e.g., food safety regulations) or at a particular developmental stage (e.g., safe car seats for infants). Targeted prevention has at its focus those identified as at risk for a disease, either because of symptoms (e.g., depression identified by screening) or because of some kind of risk exposure (e.g., parental bereavement). Indicated prevention or “Treatment as Prevention” as advocated for the prevention of HIV/AIDS (http://www.who.int/hiv/pub/mtct/programmatic_update_tasp/) looks to clinical treatment both to reduce relapse rates and to prevent the spread of the disease to others (e.g., antiretroviral treatment for HIV/AIDS). This review concentrates on the pros and cons of the more common universal versus targeted prevention strategies, although we need to remember that psychiatry still use tertiary “treatment as prevention” in the form of incarceration and compulsory hospitalization of children and adolescents who might harm themselves or others.

Preventing Symptoms or Preventing Exposure

As noted in the discussion of infectious disease epidemiology, once a causal pathway to disease has been identified, preventive strategies may focus on exposure to risk, or on dealing with the first symptoms of disease (or both). The former (universal prevention) covers everyone in the population, whereas the latter (targeted prevention) is restricted to high-risk or potential cases. Arguments about the cost-effectiveness of each type of prevention strategy have been made for decades (Mrazek & Haggerty, 1994). It is likely that the balance depends on the type of illness and the population in question (Eaton et al., 2002). For example, evidence that low birth weight is a strong predictor of female adolescent depression (Bohnert & Breslau, 2008; Costello, Worthman, Erkknali, & Angold, 2007; but see Vasiliadis et al., 2008) supports the case for universal prevention, whereas prevention of depression following bereavement makes more sense only after bereavement has occurred (Haine, Ayers, Sandler, & Wolchik, 2008).

Infectious disease epidemiology has mainly concentrated on universal approaches to prevention: for example, ensuring that children are inoculated; cleaning up water, food, and air; and setting safety standards for cars and toys. Even the chronic diseases that are the main causes of mortality today are being approached using universal prevention methods: exercise, diet, smoking reduction.

What are the universal prevention programs that would prevent mental illness? Interesting to note, the work to evaluate these has hardly begun (Durlak, Weissberg, Dymnicki, Taylor, & Schellinger, 2011; Koenen et al., 2009). Lists of exposures that might respond to universal interventions have been proposed (Eaton et al., 2002), but the evidence for them is based on common sense more than on research, and there are few tests of the relative efficacy of universal versus targeted versions of the same or similar interventions. Several studies have suggested that, just as environmental effects on psychopathology may have a stronger effect at the most stressed end of the distribution (Lee, Kosterman, McCarty, Hill, & Hawkins, 2012), an intervention to relieve a particular stress may also be most effective at that level (Costello, Compton, Keeler, & Angold, 2003). For example, an income supplement led to lower levels of behavioral problems in children whose families were moved out of poverty by the supplement but had no effect on the (already low) levels of behavioral problems in those who were never poor (Costello et al., 2003). The types of exposure that might respond to primary prevention include poor diet, lack of physical exercise (Koenen et al., 2009), poverty (Lee et al., 2012), and low birth weight (Bohnert & Breslau, 2008; Sonuga-Barke & Halperin, 2010). All of these exposures are also associated with a range of other chronic diseases. This raises questions such as, Does a particular preventive intervention prevent more than one disorder? How does this affect the cost–benefit calculations for the intervention? In the example of the income supplement described earlier, subsequent studies showed that supplementing family income reduced substance use and abuse several years later, when the children had left home (Costello, Erkanli, Copeland, & Angold, 2010). It also increased the chance of completing high school, reduced minor crime (Akee, Costello, Copeland, Keeler, & Angold, 2010), and slowed the rate of obesity problems (Akee, Simeonova, Copeland, Angold, & Costello, 2013). Thus, a single intervention applied to a whole population had a wide range of positive effects.

An advantage of universal prevention is precisely that it is universal; there is no need to screen or select and therefore there are no screening costs, as there often are for secondary prevention. Another advantage is that it avoids the stigma associated with being picked out for “anger management class” or “fat camp.” For tertiary prevention, someone—parent or teacher usually—has to identify a problem or symptom and seek help, which brings its own problems. For example, in a study with more than 20,000 assessments of children made by the children themselves and a parent (Costello et al., 1996), when a child reported one of nine symptoms of depression, the parent reported the same symptom between 0% and 29% of the time. This means that at best 71% of child-reported symptoms were missed; in the case of cognitive problems, psychomotor agitation, and anhedonia, the parent never identified cases reported by the child. Given that even the most highly developed instruments to identify child and adolescent psychopathology, whether screening questionnaires or interviews, are of only moderate test–retest and interrater reliability (Angold et al., 2012), the probability that the children in need of care will be correctly identified is low. Unfortunately, primary care physicians have an even worse record for case-identification (Dulcan et al., 1990).

If one is making a case for primary or universal prevention, one has to address the question of systems of support. It is a waste of effort, as well as arguably unethical, to identify cases for which no treatment is available. The standard systems of care for children (medical and psychiatric specialty services, educational, juvenile justice, and social services; Burns et al., 1995)) cannot provide income supplements or a healthy diet, although they can certainly do much to reduce low birth weight and improve facilities for exercise. But, perhaps more important, primary prevention of this kind can help to bring different branches of medicine and service providers together to share resources and lobby together for more. For example, reducing the proportion of low birth weight children also lightens the burden of special needs children on educational services. On the other hand, it can often happen that an intervention puts demands on one part of the system (e.g., improving antenatal care) but brings benefits only later to another part of the system (e.g., reducing special education service needs). So, institutional pressures can work against otherwise rational reallocation of resources.

RISK, EXPOSURE, AND THE MEANING OF TIME

Many questions about early detection and prevention can be answered only by methods that take into account temporal characteristics of risk factors, including age at onset and the “dose” or level of exposure over time. To be most effective, systems of support need to be sensitive both to children’s developmental needs and to the developmental course of risk exposure.

Age at first exposure, time since first exposure, duration of exposure, and intensity of exposure are all interrelated aspects of timing that may have different implications for prevention. In addition, every attempt to prevent illness is an implicit or explicit test of a causal hypothesis, and the timing of preventive interventions adds another level of causal questioning. The kinds of questions we are thinking of include the following:

• Does physical abuse by parental figures cause psychiatric disorders in children? Is a single blow a sufficient cause or does abuse have to go on for a period of time, or happen at a certain level of severity, before it constitutes a risk factor? Are children of different ages or developmental stages differentially vulnerable to physical abuse as a risk factor? What risks are associated with removing children of different ages from home because of physical abuse?

• Why are depressive disorders rare in both prepubertal girls and boys but much more common in postpubertal girls? What causes the observed sex difference to develop? Is it associated with hormonal, morphological, or social changes occurring around puberty? Why is earlier-than-average maturation apparently a positive event for boys but a negative one, associated with increased risk of behavioral and school problems, for girls?

The answers to such questions imply different assumptions about the causal role of timing, intensity, and duration. Here we offer a few examples of the impact on psychiatric disorder of different aspects of development: (a) age at first exposure, (b) time since first exposure, (c) duration of exposure, and (d) intensity of exposure. We then consider the implications for support systems.

The importance of age at first exposure has been studied most intensively of all the aspects of risk over time in child psychopathology because of the theoretical importance attached to early experiences in the Freudian and other psychodynamic models of development. For example, researchers investigating the role of attachment in children’s development have concentrated on the very early months and years of life as the crucial period during which the inability to form one or more such relationships may have damaging effects that last into childhood and perhaps even into adulthood (Sroufe, 1988). The critical date of onset of risk appears to occur after 6 months, but the duration of the risk period is not yet clear. Hay (1985) presented evidence that maternal depression, which presumably interferes with mothers’ ability to form normal relationships with their infants, affects motor development if it occurs during the 1st year of life, and language development but not motor development if it occurs during the 2nd year of life. This is a case where age at first exposure appears to interact with the developmental processes most salient at a particular age. In another example of the importance of timing, Rutter (1985) pointed out that once children have achieved urinary continence at around age 2, there is a period of risk for relapse into incontinence that appears to coincide with starting school. Once this period of risk is over, the chance of developing enuresis is very slight. In this case, age at exposure is clearly the critical developmental risk factor, because it is very rare for a parallel increase in functional enuresis to occur at later times of stress, such as moving to middle or high school, and there is no delay between the stress and the symptoms.

Timing of exposure has rarely been treated separately in studies of child psychopathology. Brown and Harris (1978), in their work on the social origins of depression, argued that women who lost their mother in the first decade of life were more vulnerable as adults to depressive episodes in the face of severe life events. However, theirs was a retrospective study that did not address the question of whether these women were also at greater risk of depressive episodes during later childhood and adolescence. It is not clear whether the crucial factor was the length of time since exposure to the risk factor of the mother’s death, or the age of the child at the time of exposure, or some combination of the two.

Timing of puberty has emerged as an important aspect of risk in relation to both depression and behavioral problems. In a longitudinal study that measured not only age at menarche but also morphological development, Tanner staging (Marshall & Tanner, 1969), and levels of gonadal and steroidal hormones, it was clear that it was high levels of estrogen and testosterone, not timing of puberty, that predicted adolescent depression (Angold, Costello, Erkanli, & Worthman, 1999). However, there are many studies showing that girls who are early in developing the morphological signs of puberty, indexed by Tanner stage or menarche, are at risk for behavioral problems, especially if they have unsupportive families (Costello, Sung, Worthman, & Angold, 2007; Ge, Brody, Conger, & Murry, 2002; Ge, Conger, & Elder, 1996; Magnusson, Stattin, & Allen, 1985).

Sometimes duration of exposure may be a more powerful predictor of later harm than timing. For example, Copeland, Wolke, Angold, and Costello (2013) found that the number of years over which children reported being bullied was a stronger predictor of poor adult outcomes than the timing of the experience (e.g., childhood, adolescence, or both.) In a longitudinal study from New Zealand, Moffitt (1990) found that children identified at age 13 as both delinquent and hyperactive had experienced significantly more family adversity (poverty, poor maternal education and mental health), consistently from the age of 7, than children who were only delinquent or only hyperactive at age 13. The most striking increase in the antisocial behavior of ADD+delinquent boys diagnosed at age 13 occurred between the ages of 5 and 7, when they attained a mean antisocial rating that was not reached by other delinquent boys until 6 years later. School entry and reading failure coincided temporally with this exacerbation of antisocial behavior. These data suggest that the problem behavior of this group, despite being generally persistent, is responsive to experience. The data also reveal a key point of vulnerability that could be a target for intervention: reading readiness.

Another example comes from the Great Smoky Mountains Study. Children who had been assessed over an 8-year period were classified into four groups on the basis of their body mass index (a ratio of weight to height) at each assessment: no obesity (72.8%), childhood-only obesity (5.1%), adolescent-only obesity (7.5%), and chronic obesity (14.8%). Only the chronically obese group was at increased risk of psychiatric disorder (Mustillo et al., 2003). This is an example of duration of exposure as the key risk characteristic. It is also an example of the impact of intensity of exposure because no effects were found of overweight that fell just below the threshold of obesity.

Intensity of exposure to lead (Needleman & Bellinger, 1991) provides an example of a definite dose–response relationship. Needleman and Bellinger (1991) divided dentine lead levels into six classes, from less than 5.1 parts per million to more than 27 parts per million, and showed a highly significant effect of dose on teachers’ ratings of children’s functioning on a wide range of intellectual and behavioral tasks. Another aspect of intensity is the number of different risk factors to which a child is exposed (Seifer, Sameroff, Baldwin, & Balwin, 1989). Most children appear to be able to cope with a single adverse circumstance, but rates of psychopathology rise sharply in children exposed to several adverse circumstances or events (Seifer et al., 1989). However, Rutter (1985) and others have pointed out that children exposed to one risk factor are at increased risk of exposure to others (e.g., no father in the home and poverty) and that the dose–response relationship to an increasing number of different risk factors is not a simple linear one

These examples show that it is possible to design studies that at least begin to allow us to tease out the respective roles played by time since first exposure, age or developmental stage at first exposure, duration of exposure, and intensity of exposure. Multistage models of risk, which have been developed to address the complexities of causality in chronic disease, are one way of putting the pieces together. Several such models have been proposed, particularly in the context of carcinogenesis (Peto, 1984), and have been reviewed in terms of developmental psychopathology (Pickles, 1993). Statistical techniques for exploring causality in such multistage models have made great strides recently (Robins, 1997). The challenge is to incorporate all the various aspects of risk into a single model and distinguish the ones that carry the tune from those that are just noise.

EARLY DETECTION OF MENTAL HEALTH PROBLEMS AND SYSTEMS OF SUPPORT

Is early detection in the form of detection of the first symptoms likely to be beneficial? Do our treatments work well when employed with symptomatic but subsyndromal cases? Is a system of support that focuses on early intervention a good use of resources?

Figure 1 shows the mean age at onset of a range of disorders, using repeated assessments up to age 21 in a representative population sample of children (National Institute of Medicine, 2009). The mean age at onset of the first symptom, in children who would eventually get a diagnosis of that disorder, preceded the full diagnosis by about two years. Sometimes the first symptom and full disorder occurred almost simultaneously (attention deficit/hyperactivity disorder, anxiety disorders), whereas in other disorders (oppositional defiant disorder, conduct disorder, depression) there was a time gap of about three years. This implies that for many psychiatric disorders there is time to identify early symptoms and intervene.

FIGURE 1 Age at onset of first symptom and of full psychiatric disorder, by age 21. Data from the Great Smoky Mountains Study. Note. ADHD = attention deficit/hyperactivity disorder; ODD = oppositional defiance disorder; CD = conduct disorder. Reproduced with permission from the National Institute of Medicine (2009). Preventing Mental, Emotional, and Behavioral Disorders Among Young People: Progress and Possibilities. Committee on Prevention of Mental Disorders and Substance Abuse Among Children, Youth, and Young Adults: Research Advances and Promising Interventions. Washington, DC: The National Academies Press. © [Rightsholder]. Reproduced by permission of National Academy of Medicine. Permission to reuse must be obtained from the rightsholder.

It is also the case that many children who experience one or two symptoms never develop a full disorder. In the study from which the figure comes, for example, there were 10 times as many observations including one or more symptoms of depression as there were cases of depression (N = 3,924 with one or more symptoms; N = 278 with a full diagnosis). But the mean age at onset of the first symptom was almost two years earlier in the children who eventually developed a full diagnosis (12.5 vs. 14.4). This suggests that really early intervention (e.g., when prepubertal children showed even one symptom of depression) could identify the children most likely to go on to a full diagnosis.

EARLY DETECTION, PREVENTION, AND THE SERVICE SYSTEM

Universal and targeted prevention make very different demands on those whose job it is to reduce the rate of onset of mental disorders. They also identify different kinds of people, with different skills, as the key interveners.

Targeted Prevention and the Service System

In the past two decades a whole new prevention industry has emerged in the United States, complete with master’s, doctoral, and postdoctoral training programs; professional societies and journals; and expert lobbyists at state and federal levels. This is understandable, because targeted prevention is a complicated and expensive activity. First, it is necessary to decide what disorders or disabilities are to be prevented. This requires a taxonomy agreed to and shared across the prevention community. Second, prevention providers need to work toward one or more recognized prevention programs or strategies, a process that requires controlled trials, tests of generalizability from one site to another, and all the accoutrements of clinical research. In fact, targeted prevention may be even more complex than clinical treatment because it requires the interveners to go out and find subjects with whom to work, rather than waiting for them to walk through the clinic door. So, targeted prevention requires methods for identifying potential cases. People normally in contact with children in the community—parents and teachers most obviously—need to be trained and helped to identify subjects for intervention, and if necessary “screening” measures must be developed and tested. “Screening” can mean case identification, if the signs are clear enough, but more often the term is used to mean picking possible cases out of a general population, with the concomitant risk of false positives and false negatives. Targeted prevention is often advocated as being less expensive to implement than universal prevention, but it is important to bear in mind that screening implies at least some level of involvement with everyone in the target population, which entails its own costs.

Next, to be successful, targeted prevention mandates a service system that can provide effective interventions, available on a scale that matches the need identified in the screening phase. It is unethical to arouse the awareness of needs that cannot be met.

The processes involved in targeted prevention are listed here at some length because of the widely held assumption that targeted intervention is more efficient and cost-effective than primary or universal intervention. This may be true, but targeted prevention carries very considerable costs that will have to be paid by someone. Also to be considered are the costs to those identified as being “labelled,” correctly or incorrectly, and the potential damage to those incorrectly missed in the screening stage. In sum, “targeting” is neither cost-free nor entirely benign.

Universal Prevention and the Service System

Primary or universal prevention programs generally need a different set of service systems, most of which have not been established, or exist in bits and pieces across the field of public health. As with targeted prevention, it is helpful to have a clearly defined target, which means a taxonomy, such as those of the Diagnostic and Statistical Manual of Mental Disorders (5th ed.; DSM–5; American Psychiatric Association) or International Classification of Diseases. Second, some method is needed to monitor the effect of a prevention program at the population level. These have been set up in some countries, through organizations such as the Centers for Disease Control and Prevention in the United States, or using national case registers such as those in force in the Scandinavian countries. Evaluation is often the weak point of universal prevention programs; for example, the vast system of health visitors, operating for more than a century in the United Kingdom with the goal of educating mothers and protecting young children, has hardly been evaluated at all (Laming, 2009). Another example is the provision of specialty mental health and counseling services to school-age children. There is good evidence that these services reach more children than any other form of mental health care (Brener, Martindale, & Weist, 2001; Burns et al., 1995), but we know very little about their effectiveness.

In other areas of health it is very clear that universal prevention has been extraordinarily effective, for example, in lowering infant and child mortality and morbidity and improving healthy physical development. There is also indirect evidence that mean intelligence levels have increased dramatically in the past 100 years, at a far faster speed than could be attributable to evolutionary selection (Flynn, 1984). No such data are available for mental health. The second step for instituting universal systems of mental health care is clearly a national (and eventually international) surveillance system that can monitor need for services and change in demand.

It is likely that when and if such systems were to be set up, we should find that some of the universal health care programs already in place have a marked effect on children’s mental health. Prenatal care and nutrition is an obvious example, given the evidence for the damage that can be caused by low birth weight (Costello, Worthman, et al., 2007; Sommerfelt, Troland, Ellertsen, & Markestad, 1996; Szatmari, Saigal, Rosenbaum, Campbell, & King, 1990). Mandated screening for hearing and visual deficits is another example. Programs that affect whole communities, such as the removal of lead from the environment, have effects on both behavioral and cognitive development (Needleman & Bellinger, 1991).

It is nevertheless the case that, despite what universal prevention has done to protect them, far too many children have psychiatric disorders. For example, our 20-year prospective study of a community sample found that by age 21 more than 50% of the participants had experienced at least one DSM-IV (American Psychiatric Association, 1994) psychiatric disorder, and a further 10%–15% had significant functional impairment associated with psychiatric symptoms (Copeland, Shanahan, Costello, & Angold, 2011). Other longitudinal studies show similar results(Jaffee, Harrington, Cohen, & Moffitt, 2005). Barely one in four received any specialty mental health care, and those who did waited 2 or 3 years for services (Burns et al., 1995; Costello, He, Sampson, Kessler, & Merikangas, 2013).

CONCLUSIONS

Returning to the title, “Early Detection and Prevention of Mental Health Problems: The Role of Developmental Epidemiology in Planning Systems of Support,” the role of developmental epidemiology in the prevention of metal disorder has both optimistic and pessimistic sides. In the case of targeted prevention we can be optimistic because, as other articles in this issue show, preventive interventions can be effective, and the systems needed to bring them to scale—screening, training of field staff, destigmatization of mental illness—have been worked out at least in principle (of interest, a lot of the work has been done in the context of global mental health; Collins et al., 2011; Patel, Flisher, Hetrick, & McGorry, 2007). Pessimism is induced by the high cost of instrument and program development and by the likely service gap that will yawn when screening reveals the true numbers of children in need of clinical care.

In the case of universal prevention, we can be more optimistic that programs already in the field, and well accepted by almost everyone, will be shown to yield mental health benefits that have not yet been evaluated. Pessimistically, we note the high rate of (largely untreated) mental disorder in the population and the lack of surveillance systems that can track the effects of existing or new programs.