|
1
| 2
| 3 | 4
| 5
Other Characteristics of Prevention Programs That Affect Intervention
Effectiveness
We have described how the dimensional aspect of broadly population-based
and targeted risk groups helps identify the target group for a prevention
program. Other criteria are also important, such as the timing and
duration of the intervention, the person or persons who will deliver
the intervention, and the social context within which it is delivered.
A complete mapping of the combined risk and protective factors over
the life course at the individual or contextual levels would provide
valuable knowledge for designing prevention programs. Even incomplete
mapping or mappings generated by combining findings across different
samples can be useful. This is the perspective of developmental
epidemiology (118), which guides our understanding of the development,
formation, and interactions of a defined population of individuals
within their environments.
A developmental perspective on schizophrenia suggests that we should
consider not the predictability of risk factors across age groups
but their predictability when measured within and across different
stages of development. It seems likely that schizophrenia genes
will express themselves differently at these different stages. For
example, current neurobiological theories of schizophrenia implicate
dysfunction of the frontal cortex and of circuits connecting to
that region. Because frontal cortex is developing throughout childhood,
it would be reasonable to suspect age-related changes in the expression
of premorbid predictors of schizophrenia.
Within a developmental epidemiology framework, numerous factors
demand consideration when models of disease causation are being
developed and ultimately tested in preventive intervention trials.
For example, the timing of the intervention and its position relative
to other interventions will dramatically influence the prevention
program’s success. To illustrate: a schizophrenia prevention
program can aim for improved host resistance (that is, prophylactic
neuroleptics), behavioural change (that is, improved attention and
processing), and environmental change (that is, less stressful peer
and intrafamily relationships), each of which must be implemented
at the appropriate stage in both the individual’s development
and in the target’s progression.
Determining the proper timing of an intervention is sometimes difficult,
because it is known that, for schizophrenia, risk factors exist
much earlier than the period within which incidence rises. It is
critical to intervene during the premorbid period, in the stage
prior to disease onset, or even prior to prodromal onset, when frank
signs and symptoms are not evident: targeting early risk and protective
factors can lead to long-term benefits. For example, the Israeli
High-Risk Study of offspring of parents with schizophrenia reported
a dramatic difference in risk, based on social adaptational status.
Only those who performed poorly in each important stage of life
from preschool, early school, and adolescence were at risk for schizophrenia;
none of those who made successful transitions through most of these
stages were diagnosed with the disorder (119). Thus, poor social
adaptation throughout the life course among offspring of parents
with schizophrenia constitutes a particularly high-risk developmental
process.
It should be noted that low neighborhood socioeconomic status,
as well as family poverty status, both have a direct impact on a
child’s aggressive behaviour ratings from elementary school
to early adolescence (120). Poverty is also a risk factor for schizophrenia
(121). Aggressiveness, particularly when accompanied by shy or withdrawn
behaviour, is strongly associated with poor peer and parent relationships
and low academic performance. When concentrated in a genetically
vulnerable group of children, this cluster of multiple problems
may also be an important developmental pathway toward later psychopathology.
Further, in 2 independent studies of children of schizophrenia patients,
Fish and others described a syndrome of motor abnormalities that
predicted subsequent schizophrenia or related disorders (92). Similarly,
in both the Copenhagen and New York high-risk projects, neuromotor
impairment predicted schizophrenia onset (93). These findings are
consistent with Walker and Lewine’s finding of poorer fine
and gross motor coordination in videotapes of children who subsequently
developed schizophrenia (95). In addition to neuromotor impairment,
attentional deficits have also been found to predict subsequent
schizophrenia and related disorders (122).
Variation in Intervention Impact
To test these developmental epidemiologic or early prevention approaches,
we will need to compare the growth trajectories of these risk behaviours
across time. Change in these proximal risk targets should theoretically
carry forward to changes on the more distal outcome. Thus, to examine
impact on proximal growth trajectories and distal diagnostic outcomes,
statistical growth-modelling techniques used to chart changes in
proximal risk behaviours can be combined with the analysis of a
distal binary outcome—whether diagnosis or another outcome.
Similarly, the techniques can also be combined with a survival analysis
involving time to diagnosis. Both theoretical and empirical work
indicate that an intervention effect is likely to differ by early
risk status, so recent evaluations examine the intervention’s
effect on different subgroups of the population that are often divided
based on their early risk status (123). In schizophrenia research,
one would logically attempt to examine (based on sample size limitations)
whether an intervention varies in effectiveness among those who
have a family risk (including a history of schizophrenia); or pregnancy
and delivery complications, or early attention problems.
A consistent finding across several broad, population-based, preventive
interventions is that their impact can be highest among the highest-risk
group, even though the target group varies considerably in risk
(124). Because high-risk groups tend also to have few protective
factors, there may be more opportunity for improved outcomes should
the intervention focus on building the right protective factors
(125). It is not necessarily true, however, that a more targeted
intervention, focused on just the highest-risk group, would be equally
successful. Risk status on measures such as attention processing
can vary across time (126); a one-time targeted intervention would
miss those who developed the risk shortly afterwards (127). This
is not the case for a universal intervention, since everyone would
be exposed to it.
In preventive interventions, implementation can refer to behavioural,
systemic, or prescription practices. Of course, preventive intervention
effectiveness can diminish considerably when the intervention is
not delivered at full strength. Currently there is no uniformly
accepted pharmacologic approach for high-risk adolescent populations
at risk for schizophrenia, such as the population being studied
in a natural setting by Cornblatt and others (128). About one-half
of such adolescents are now being given antidepressants, with higher
levels being given to those with more severe symptoms. While there
is no definitive evidence of benefit or harm from such medication,
the overall benefit or potential harm is attenuated by this lack
of implementation standard for the field. For broad-based preventive
interventions where intervention involves delivering multiple components,
there is ample evidence that preventive effectiveness varies with
implementation level (6), and for a multiyear intervention, preventive
effectiveness varies by exposure level, determined by entrances
and exits to a school system (1).
The findings on participation in preventive interventions are somewhat
less generalizable. Participation can vary from nearly 100% for
interventions within classroom settings (7) to less than 50% when
individuals are asked to attend group sessions (89). Usually, the
participation level is not related in a strong systematic way to
risk. Sometimes, higher participation levels occur in low-risk groups,
and sometimes they occur in higher-risk groups. Nevertheless, since
there can be no intervention impact for those subjects who choose
not to participate, the overall participation level strongly affects
the preventive effectiveness for the population. That is, if one-half
of the eligible subjects participate in a preventive intervention,
we can anticipate roughly one-half the effect of a preventive intervention
delivered to the entire population.
Summary
This paper presents several methodological concepts guiding the
development of prevention programs and suggests ways to apply these
methods to the prevention of schizophrenia. To date, several important
risk factors for schizophrenia have been identified. These present
options for developing prevention programs. A few prevention programs
that focus exclusively on high-risk groups (especially those showing
prodromal symptoms of schizophrenia) are now being tested in randomized
trials.
To date, there have not been any attempts to evaluate either the
broader population prevention strategies or the strategies that
apply population-screening methods to select high-risk groups. For
example, reducing pregnancy and delivery complications may have
some impact on reducing the prevalence of schizophrenia and of other
disorders for which such complications are risk factors. Alternatively,
it may be feasible to screen births for pregnancy and delivery complications
relevant to the subsequent development of schizophrenia. Within
that high-risk sample, one could screen for other risk factors,
such as family history of schizophrenia, early signs of inattention,
poor motor skills, or poor socialization.
We should work now to diminish gaps in our knowledge about the
development of schizophrenia in the population. Future work needs
to answer several key questions: 1) What are the best early predictors
of subsequent schizophrenia? 2) Can these indicators predict illness
with sufficient accuracy to justify preventive intervention programs?
3) What are the relative strengths of risk factors that predict
schizophrenia? 4) Are these large enough to justify prevention efforts?
5) What types of treatments would prevent schizophrenia in high-risk
individuals? 6) Under what conditions, and in combination with which
other risk and protective factors, are the effects of a risk factor
greatest, and when are they lowest?
As we have shown, the methodological and statistical technologies
needed to answer these questions are available. If these are combined
with the contemporary tools of neuroscience and treatment research,
answers to these questions and the eventual prevention of schizophrenia
should be achievable.
Acknowledgements
We thank Dr Jane Pearson, Dr Sheppard Kellam, and Dr George Patton
for many helpful comments and insights on this manuscript. Our colleagues
in the Prevention Science and Methodology Group (PSMG) reviewed
presentations of this work and provided many additional comments.
Funding and Support
Dr Brown’s work on this article was supported by National
Institutes of Mental Health (NIMH) and National Institute on Drug
Abuse (NIDA) grant number MH40859, and by National Insitute of Child
Health and Human Development (NICHD) grant number HD040051. Dr Faraone
and Tsuang’s work was supported by NIMH grants RO1MH43518,
R01MH59624, and R25MH60485. This work was completed when Dr Glatt
was a NIMH-funded trainee (R25MH60485).
1
| 2
| 3 | 4
| 5
|
