IN REVIEW


Prenatal Factors and Adult Mental and
Physical Health

Ezra B Susser, MD, DrPH1, Alan Brown, MD2, Thomas D Matte, MD, MPH3


Objective: To review research on prenatal influences on adult mental and physical health and draw implications for future directions in psychiatric research.

Method: Schizophrenia is selected as an example from mental health and cardiovascular disease as an example from physical health. For each of these disorders, empirical findings on prenatal influences are reviewed, and the methods used to demonstrate them are critiqued.

Results: Research on prenatal antecedents of these conditions has proceeded in parallel: intriguing findings have related fetal growth restriction or fetal insult to adult health; similar types of causal pathways have been proposed to explain the relationships; and research has been plagued by similar limitations, including lack of precise prenatal exposure data and difficulty of controlling confounding. The prevailing view of disease causation, which is not well-suited to investigation of prenatal antecedents, impedes research in both fields. Yet, there has been little interchange between researchers in the 2 fields.

Conclusions: We propose a causal paradigm that could serve as a guide for future investigations on the prenatal antecedents of adult health and promote interchange between research on mental and physical health. The paradigm reflects current thinking in epidemiology by encompassing not only risk factors as traditionally conceived but also causal chains over time and causal influences at multiple levels of organization. Implications for the design of new research are illustrated with reference to an ongoing study.

(Can J Psychiatry 1999;44:326–334)

Key Words: prenatal exposure, delayed effects, pregnancy, schizophrenia, cardiovascular disease

The hypothesis that early life experiences can have lasting effects on adult mental health has a long history in the field of psychiatry. Similarly, connections between early-life environment and adult physical health have been proposed since early in this century (1). The combination of an old but powerful paradigm—that early-life exposures influence adult health—with the tools of modern epidemiology and biomedical science now has the potential to greatly advance our understanding of the pathways to adult mental and physical health.

For those interested in the factors operating early in life that may exert long-term latent effects on adult mental health, some useful lessons can be drawn from research into early antecedents of adult physical health. For example, criticisms of existing research (2,3) and their implications for future directions are, for the most part, equally relevant to both domains of outcome. Despite these parallels, there previously has been only limited interchange between researchers in the 2 domains. This paper discusses the evolution of research concerning the early origins of adult health, considering both psychiatric and physical disease.

Early Developmental Insults and Schizophrenia

The Neurodevelopmental Hypothesis of Schizophrenia

Converging evidence from several areas of research supports a neurodevelopmental hypothesis of schizophrenia, which posits that a disruption of brain development plays an etiologic role in a substantial proportion of schizophrenia cases. First, numerous neuroimaging studies have demonstrated that first-episode patients with schizophrenia, on average, have structural brain abnormalities, including enlarged cerebral ventricles (4–6) and reduced temporolimbic (7) and other cortical volumes (8). Second, neurocognitive manifestations, including diminished IQ (9), attention (10), and neuromotor performance on average tend to be present in children destined to develop schizophrenia many years later. Third, an excess of minor physical anomalies, an indicator of in utero malformation, has also been found in patients with schizophrenia (11).

While genetic factors clearly play an important role in the etiology of schizophrenia and can be responsible for disruptions of fetal brain development, the potential contribution of environmental factors should not be underestimated. For example, in a landmark magnetic resonance imaging (MRI) study of monozygotic (MZ) twins, nearly every affected twin, compared with his or her unaffected cotwin, demonstrated structural brain abnormalities, including ventriculomegaly and diminished volume of the amygdala–hippocampus complex (12). Given that MZ twins share 100% of their genes, these findings suggest that the differences in brain morphology may be attributable to environmental factors.

The Search for Prenatal Insults

Motivated by these investigations, epidemiologic studies have examined prenatal insults known to be teratogenic to the brain as potential risk factors for schizophrenia. On the whole, these investigations have supported the view that several prenatal insults are associated with significant increases in risk for schizophrenia in the offspring, although the findings are not yet definitive and there are discrepancies between studies. There is compelling evidence for prenatal factors in schizophrenia, from crude ecologic studies to the birth cohort investigations that form the basis of a new era of epidemiology. Though not reviewed here, evidence also supports an effect of perinatal and infant exposures. The progress in schizophrenia research parallels the evolution of research methodology in epidemiologic studies in CVD, discussed later.

Prenatal Influenza. Many ecologic studies have examined whether individuals in utero during influenza epidemics evidenced an increased occurrence of schizophrenia. The first of these investigations, in Finland, demonstrated an approximately twofold association between second-trimester influenza exposure to the 1957 influenza A epidemic and schizophrenia (13). Despite the methodological limitations, its findings were replicated in subsequent ecologic studies in Great Britain (14,15), Japan (16), and Australia (17); most of these associations were also specific to the second trimester. Additional studies have demonstrated associations between influenza epidemics over long intervals and the occurrence of schizophrenia; again, the findings appeared to be specific to the second trimester. These include studies in Denmark (18), England (19,20), Wales (19), and the Netherlands (21).

Notwithstanding these intriguing findings, there have been some negative ecologic studies, including those in Holland (22), Croatia (23), and the United States (US) (24). In addition, the only 2 studies that used data on influenza exposure among individuals reported negative results (25,26).

Prenatal Rubella. Emerging evidence from our group suggests that prenatal rubella may increase the risk for nonaffective psychoses, including schizophrenia. The plausibility of this virus as a cause of nonaffective psychosis is supported by over 50 years of evidence that rubella is a known cause of neurodevelopmental anomalies, including deafness, mental retardation, and cerebral palsy (27,28). In 1971, Chess and colleagues reported substantially higher risks for several childhood psychiatric disorders, including autism and separation anxiety, in a birth cohort prospectively exposed during gestation to rubella (29), in whom the majority of exposures were serologically documented in early gestation. These findings led us to hypothesize that this birth cohort would be at higher risk for psychotic disorders in adulthood. We therefore examined the risk for nonaffective psychosis among those in this birth cohort who were of normal intelligence and were followed up by our group in young adulthood with a comprehensive psychiatric diagnostic interview. Our findings suggest a markedly increased rate of nonaffective psychosis, including schizophrenia, in this sample (Brown and others, submitted for publication).

Prenatal Famine. A series of investigations, also from our group, using psychiatric registry data in birth cohorts exposed and unexposed to the Dutch “hunger winter” of 1944/1945, has provided important evidence that prenatal famine increases the risk for schizophrenia (30). The initial finding to emerge from these studies was that birth cohorts exposed to the hunger winter in early but not later gestation had a twofold increase in the risk for schizophrenia (31,32). A subsequent study using military induction data demonstrated that prenatal famine during the same early period of gestation was associated with a twofold elevation in risk for schizoid or schizotypal personality disorders (33).

Maternal Stress. Two studies have drawn attention to the potential role of maternal stress in schizophrenia. In one study, children of pregnant mothers in Finland whose husbands died during the pregnancy (mostly during World War II) had a significantly increased risk for schizophrenia and other psychiatric disorders (34). In the other study, the May 1940 invasion of the Netherlands was associated with an increased risk for schizophrenia among those in utero at that time (35). One potential explanation for these findings is excessive release of maternal cortisol, which may be toxic to the developing fetal hippocampus.

Rhesus Incompatibility. Rhesus (Rh) incompatibility, characterized by an Rh-negative mother pregnant with an Rh-positive fetus, has been associated with an elevated risk for schizophrenia. Hollister and others compared the risk for being hospitalized with schizophrenia, ascertained from the Danish Psychiatric Hospital Register, with Rh-compatible and -incompatible pregnancies, using data from the Danish Perinatal Cohort (36). Rh incompatibility can give rise to Rhesus hemolytic disease of the newborn (Rh HDN), a hemolytic reaction that results in, among other consequences, neuropsychiatric disturbances with some interesting parallels to schizophrenia. These include childhood neuromotor abnormalities, including choreoathetosis from basal ganglia damage, and childhood behavioural disturbances, such as emotional instability, that could be related to hippocampal dysfunction (37). Rh HDN most commonly occurs in second and later children born to an Rh-negative mother who has previously delivered an Rh-positive fetus thus triggering the production of maternal antibody against the Rh(D) antigen. As hypothesized based on the above findings for Rh HDN, the risk for schizophrenia among males in the Danish Perinatal Cohort Study was increased over threefold in second- and later-born offspring from Rh-incompatible pregnancies, but there was no increased risk for first-born Rh-incompatible children (36).

Low Birthweight and Diminished Head Circumference. Low birthweight, sometimes considered a crude indicator of a prenatal developmental delay or disruption, has been associated with schizophrenia in several studies (38–40). However, there is disagreement as to whether the relationship of low birthweight to schizophrenia risk is accounted for by prematurity, intrauterine growth retardation, or some combination. Jones and others showed that the proportions of individuals born below the 10th percentile for weight by gestational age were virtually identical between the schizophrenia and control groups, suggesting that prematurity may be the more important of the 2 factors in relation to schizophrenia (40). The relationship of low birthweight to schizophrenia is also not conclusive: contrary to expectations, Hultman and others found that increased birthweight by body length was more common in schizophrenia (41).

A second proxy of a prenatal developmental disruption that has been examined in studies of schizophrenia is diminished head circumference, which has been demonstrated in some studies of patients who develop schizophrenia (42). These findings suggest that a general impairment of brain growth in utero may predispose individuals to schizophrenia.

Early Origins of CVD

The identified adult CVD risk factors could not account for much of the social and geographic gradient in CVD risk; this has led to research of CVD causes early in life. Forsdahl, for example, noted a correlation between infant mortality rates and adult mortality rates from arteriosclerotic heart disease (ASHD) decades later in the same geographic areas of Norway (43). This suggested to him a connection between undernutrition early in life and an increased susceptibility to elevated cholesterol levels when exposed to a more plentiful diet in adult life. Subsequently, investigators have replicated some of the ecologic associations between infant mortality (or other measures of deprivation in early life at the community level) and adult mortality (44,45) noted by Forsdahl. These ecologic studies, however, did not permit the examination of specific pre- and postnatal factors that might cause or mediate an increase in subsequent CVD.

So, investigators have sought out populations for which at least limited individual-level data on early-life factors are available that attained sufficient age to permit examination of CVD endpoints or risk factors. Most of these studies have been carried out by investigators at the Environmental Epidemiology Unit at the University of Southampton, England (46). The early antecedents most closely examined by this group can be loosely classified as fetal and infant growth and nutrition. Birth weight is the measure of fetal growth most commonly used, and it has been linked to CVD mortality and morbidity. For example, in 6 districts of Hertfordshire, England, data from records kept by midwives and local health workers (“visitors”) between 1911 and 1930 have been used to construct a historical cohort, which has been followed up to great advantage. Barker and colleagues found that, among 5654 men for whom early life and outcome data were available, birthweight and weight at 1 year were inversely correlated with death rates from coronary heart disease (47).

Later work examined the relations of impaired fetal growth to presumed and potential CVD antecedents such as higher blood pressure, adverse serum lipid profiles, elevated plasma fibrinogen, impaired glucose tolerance, and decreased arterial compliance. In the same Hertfordshire cohort, 297 women underwent examinations to assess blood pressure, glucose tolerance, serum lipids, and central adiposity. Lower birthweights were associated with insulin resistance, higher blood pressure, higher ratio of waist to hip circumference (a measure of central adiposity associated with CVD risk), higher serum triglyceride levels, and lower serum high-density lipoprotein (HDL) levels (48). Higher body-mass index (BMI) at the time of the adult examination was independently associated with adverse CVD risk profiles. Together, these findings have been interpreted by some as indicating that an environment which impairs growth in utero or during early childhood can influence an individual’s metabolism and/or physiology in a way that increases the risk for CVD in adult life. The hypothesis as stated by David Barker is: “The fetus responds to undernutrition with permanent changes in its physiology, metabolism, and structure, and these lead to coronary heart disease and stroke in adult life” (46). Animal studies support at least some of these connections (49). The impact of adult BMI indicates the potential for environmental and behavioural influences on CVD risk to operate over the life course of an individual.

Postulated Mechanisms

Despite considerable efforts, the causal mechanisms that link prenatal insults with later health remain largely undetermined. There are 2 especially prominent views on causal mechanisms. One posits that specific prenatal insults, individually or in combination, cause structural and functional damage to certain organs; for example, postulated lesions affect specified brain regions and neurotransmitters, which increase vulnerability to schizophrenia. The second view is that early developmental factors result in a more generalized disruption of fetal or child health, which is reflected in impaired indices of growth, in turn predisposing the child to ill health in adult life through metabolic programming or other less specific mechanisms. In general, the first type of causal model has guided research into the early origins of schizophrenia, while the second has been dominant in interpreting evidence for early life factors and adult CVD.

Structural brain abnormalities correlated with schizophrenia partly support the possibility that a discrete prenatal insult and brain injury could lead to schizophrenia. Other observations indicate that such abnormalities might have their origins early in life. For example, an increased frequency of periventricular white-matter lesions and associated ventricular enlargement was observed in a large series of preterm infants (50,51). Potential causal mechanisms that may underlie this finding include ischemic injury to the subcortical white matter, which is especially sensitive to hypoperfusion (52), and metabolic insults (51) occurring before 32 weeks of gestation. In addition, animal models have been used to examine whether a neurodevelopmental disruption of prefrontal-hippocampal circuitry could play a role in the pathogenesis of schizophrenia (53). In these studies, neonatal excitotoxic lesions in the hippocampus resulted in several dopamine-mediated and prefrontal-cortical abnormalities, consistent with neurochemical and neurocognitive findings in schizophrenia (54).

Hypotheses are also being formulated concerning possible nutrient deficiencies predisposing individuals to brain lesions that may be associated with schizophrenia. For example, evidence has emerged to suggest a causal mechanism for periconceptional folate deficiency in neural tube defects, which involves a concurrent impairment in homocysteine metabolism (55). This metabolic defect in some cases may be due to a specific single base-pair mutation of the gene encoding methylenetetrahydrofolate reductase (MTHFR), which leads to high homocysteine levels. In this scenario, the effects of the genetic defect can be overcome by folic-acid intake during early gestation, since the derivatives of folic acid enhance homocysteine metabolism. In the original Dutch famine study (56), there was a single but salient finding among the neurodevelopmental outcomes examined: the birth cohort conceived during the peak of the famine—the same cohort that later demonstrated a peak in schizophrenia incidence—evidenced an increased prevalence of congenital anomalies of the central nervous system, mostly neural tube defects. The remarkable concordance in timing between neural tube defects and schizophrenia provides the basis for a compelling hypothesis: that periconceptional folate deficiency—which has been linked to neural tube defects—may also play an etiologic role in schizophrenia. Conceivably, some cases of schizophrenia may also involve a genetic defect in homocysteine metabolism.

Links between a discrete prenatal insult and adult physical health are also possible and fall under the broadest use of the term “programming,” describing a “process whereby a stimulus or insult, at a sensitive or ‘critical’ period of development, has lasting or lifelong significance” (46). In the search for early antecedents of CVD, however, most researchers have focused not on specific insults or lesions, but on more general indices of fetal growth, such as birthweight. The fetal programming hypothesis of CVD risk asserts that “malnutrition” (usually operationalized as lower birthweight)—including that due to placental insufficiency—induces adaptations in the developing fetus with consequent lifelong differences in key precursors to CVD, such as hypertension, obesity, insulin resistance, glucose intolerance, elevated levels of triglycerides and low-density lipoproteins (LDL), and reduced levels of HDL.

At the cellular level, 3 potential fetal programming mechanisms have been proposed that could lead to CVD and other impacts on adult physical health (57): 1) Permanent alteration of the expression of certain genes (for example, those controlling the production of hormones) and receptors involved in regulating glucose and lipid metabolism. 2) Permanent reduction in cell numbers in selected organs and tissues and perhaps changes in organ structure. Examples might include decreased numbers of nephrons leading to an increased risk hypertension and decreased pancreatic beta cells leading to an increased risk of glucose intolerance. 3) The selection of cellular clones. For example, the ratio of periportal to perivenous liver cells may be influenced by nutrition, with possible effects on cholesterol and fibrinogen synthesis (58).

Animal studies support at least some of the proposed causal relations between the early-life environment and CVD precursors. For example, in a guinea pig model, unilateral uterine artery ligation during pregnancy caused reduced birthweight and subsequent increased blood pressure in offspring (59). Offspring of pregnant rats treated with low-dose dexamethasone showed similar effects (60). In rats, dietary protein restriction in utero causes a permanently impaired insulin response to glucose (61).

A link between more general impairment of fetal growth and schizophrenia, rather than a specific brain lesion as suggested earlier, provides an alternative explanation for the findings of the Dutch famine study. Perhaps a general state of maternal malnutrition led to an impairment of overall fetal growth, of which impaired brain development was but 1 feature. In that study, early gestational famine was also associated with an increased incidence of very low birthweight, a possible risk factor for schizophrenia.

Critiques of Early Antecedents Research

Reports indicating prenatal origins of adult mental and physical health have been highly provocative, challenging some well-established paradigms about illness causation. In addition, such work is subject to the usual challenges to validity extant in any observational epidemiologic research. So the validity of causal inferences drawn from relations of various prenatal factors to the risk of serious mental and physical health problems years later can be questioned. Prior research in both domains has limitations and therefore important implications for the design of future studies. Readers are referred elsewhere for further reviews (2,3,62,63).

Crude Measures of Prenatal and Early Postnatal Exposures

The frequent application of ecologic data to this field has limited the precision of exposure measures. For example, in all but 2 studies cited concerning prenatal exposure to influenza and schizophrenia risk, it was known that the person was in midgestation at the time of an influenza epidemic but not whether that person’s mother actually was infected with influenza. Ecologic studies of CVD risk have also used surrogate markers of fetal and infant health, such as infant and maternal mortality (45,47). Such measures are likely correlated with community-level social adversity at the time of birth and potentially with adult socioeconomic status as well.

Further limitations in exposure metrics are related to the long latent period between early-life exposures and proposed health outcomes. Thus, investigators have often relied on data collected years earlier, usually for nonresearch purposes such as documenting routine prenatal care by midwives (47).

Fetal programming studies using birthweight as the exposure measure often assume that birthweight is influenced by maternal nutrition. In the Dutch famine cohort, severe caloric restriction did reduce birthweight (64). However, since birthweight in humans is relatively insensitive to dietary restriction until it becomes severe, birthweight variation in nonfamine conditions in industrialized countries is probably due more to variation in placental perfusion and function than to variation in maternal diet (65,66).

Poor Quality of Outcome Measures

Some research into early origins of schizophrenia used unrefined diagnoses, including those obtained for clinical rather than research purposes, diagnoses based on systems with relatively nonspecific criteria such as those in the second Diagnostic and Statistical Manual of Mental Disorders (DSM-II) or the ninth International Classification of Diseases (ICD-9), and diagnoses based on review of hospital records without any direct assessment by the research team. Cardiovascular studies based on mortality data have similar limitations. While more recent measures of CVD risk are easier to standardize, the link of some measures to actual CVD risk is still to be proven.

Confounding Bias

Research of early antecedents in both cardiovascular and psychiatric disease may not adequately control for socioeconomic factors. An adverse social environment early in life has been associated with early developmental insult, including low birthweight, and in some reports with schizophrenia. Early social disadvantage can also influence CVD risk through an association with health risk behaviours. In addition, social adversity and chronic strain might directly affect CVD risk through proposed neuroendocrinologic and other physiologic mechanisms (67). Early-life social adversity, then, clearly is a potential confounder of the relation of birthweight to schizophrenia or to CVD risk. Even when simple indices of social position, such as income or occupation, are used to adjust for social adversity, adjustment may be incomplete if the indices used are imprecise measures of relative social and economic deprivation (2). In some studies where exposure metrics determined at birth, such as birthweight and obstetric complications, are associated with the outcome of interest, an unmeasured confounding prenatal insult might in fact be to blame.

Finally, it has been argued that the relation of low birthweight to CVD risk is confounded by a common, perhaps genetic, antecedent. In support of such a hypothesis, the mothers who gave birth to lower-birthweight infants were at increased risk for CVD years later, even when adult lifestyle factors were accounted for (68). It could be that inheriting a trait predisposing to both CVD and low birthweight, rather than having a low birthweight per se, places offspring at increased risk. For example, a genetic predisposition to insulin resistance could lead both to impaired fetal growth and to glucose intolerance and CVD later in life (69). A similar argument can be made regarding schizophrenia.

Selection Bias

In the few cohort studies of schizophrenia (40), persons lost to follow-up were an important source of potential bias, because there may be differential loss between individuals who would develop schizophrenia and those who would not. Similarly, in the Hertfordshire cohort cardiovascular study, outcome data were available for only slightly more than one-third of eligible subjects (62). While it is likely that those examined differed on health status from those not examined, it is less clear if selective follow-up would bias the observed associations between health status and birthweight. Such bias would require not only differential loss to follow-up but also an interaction between exposure status and schizophrenia or CVD risk in effecting loss to follow-up.

Causal Pathways—Lack of Assessment or Ambiguous Hypotheses

Most of the schizophrenia studies cited above have not had sufficient data to control for confounding or to examine the mediators that may connect prenatal exposures to schizophrenia. For instance, linking Rh incompatibility to schizophrenia, Hollister and others noted the possibility that the effects may be mediated by perinatal complications (36), but their data did not enable them to test this hypothesis. The practice of adjusting associations between birthweight and adult blood pressure for adult BMI fails to consider that BMI may mediate rather than confound the relation of birthweight to adult blood pressure  (2). In regard to mediation of CVD pathways, without adjustment for BMI the relation of birthweight to adult blood pressure has been nonsignificant or positive, while the opposite pattern (a significant inverse relation of birthweight to adult blood pressure) often emerges with adjustment for BMI. To the extent that higher birthweights contribute to higher adult BMI, which is in turn associated with higher blood pressure, such “overadjustment” can suggest that heavy babies are protected from hypertension, when the opposite may be true.

Hypotheses about relations between early growth and later health have not always been clearly specified, perhaps because of varying exposure metrics or outcomes. For example, results concerning early origins of CVD have not been entirely consistent (3): some fairly large studies fail to find the inverse relation of birthweight to blood pressure, and others find a positive relation. Paneth and others (62) noted that at least 43 different cardiovascular or metabolic outcomes were examined by Barker (70). Given that 12 perinatal markers of prenatal and early postnatal “nourishment” were also examined, they argue that many associations would be expected to emerge by chance alone. A similar difficulty arises when considering studies of “obstetric complications” and psychiatric outcomes, in which diverse clinical events comprise the exposures considered.

Future Directions

A causal paradigm could serve future research on early antecedents of adult health problems in virtually any domain (for a more complete descripton of this paradigm, see 71). The paradigm has specific implications for study designs for schizophrenia and CVD.

The causal paradigm entails broadening—but not discarding—the notion of multiple causes or “risk factors,” which was developed in epidemiology and other health sciences in the post-WW II period. This paradigm, sometimes referred to as “ecoepidemiology,” extends the notion of causation in 2 directions. First, the element of development over time is introduced; causation is considered in terms of a pathway over the life course rather than in terms of combined factors at some point in time. For most health outcomes, it is not biologically plausible that early influences are absent or that they cannot be altered by adult environment. We must think in terms of dynamic processes and developmental pathways toward a health outcome. In epidemiologic research, the life course model has emerged, incorporating such elements as cumulative insults over the life course, critical periods of susceptibility throughout life, a chain of causation from child to adult lifestyle, and interaction between early and late factors (72). Waddington and colleagues have adapted this view as a model for schizophrenia (73), arguing that the developmental path of the disorder needs to be traced over the life course as a crucial foundation for biologic, genetic, and epidemiologic research to discover the “cascade” of events that culminate in schizophrenia.

Second, the paradigm allows for causes to be considered at multiple levels of organization (74). This means, for instance, that, in an investigation (single or series) of schizophrenia, we may need to consider characteristics that may influence schizophrenia, such as those of societies (level of development [75]), family units, individuals (schizotypal personality traits), and specific genes within individuals that may combine with many other genes to determine individual characteristics or otherwise influence the development of the disease. The presence of distinct levels of organization is often overlooked in the conception and presentation of current research, as when we purport to seek “a gene” for personality traits predisposing to schizophrenia rather than a set of genes that may only produce a trait when acting in combination.

The paradigm of ecoepidemiology has many specific implications for the design of research. It was noted earlier that confounding by social environment has been a critical limitation of research in this field. If causal chains can operate across developmental periods, static measures of social standing may be inadequate: we need to control for the social environment over specified periods of the life cycle. Regarding CVD, for example, childhood and adult social environments could play very different roles. The former might influence growth, development, and the acquisition of risk behaviours; the latter might influence stress-induced physiologic changes. Given the important limitations of prior research in addressing potential confounding by socioeconomic status or genetic factors, sibling-pair studies are an important design strategy for refuting or confirming potential confounding. Matching on family unit can achieve tighter control of the childhood social environment than could any measure of it. In addition, the presence of the maternal genome, which seems to be the predominant genetic influence on birthweight, is partially controlled for in the sibling-pair design. Sibling-pair designs are feasible within existing large US cohorts that were enrolled prenatally and have now reached early middle age (76,77).

Example From Schizophrenia

As an example of future directions, we are applying these principles to the study of causes of schizophrenia in the Prenatal Determinants of Schizophrenia (PDS) study, one of a series of investigations undertaken by our group. The PDS builds on an Oakland birth cohort of 19 044 live births during 1959–1966 that have been followed since the mothers’ first prenatal visits in a broad investigation of human development and health (77). The cohort for the PDS study comprises a subgroup of approximately 12 000 individuals who were included in the medical care databases used to identify cases of schizophrenia.

The PDS is designed primarily to incorporate the time dimension of the causal paradigm and thereby to apply a life course approach to investigating schizophrenia. Among its strengths, perhaps the most striking is the rich array of high-quality, prospectively collected exposure data pertaining to the early prenatal period. These data include, for example, detailed measures of maternal health, smoking, and use of prescribed and nonprescribed drugs, of pregnancy complications, and of prenatal care. Serum samples drawn during the pregnancies of enrolled mothers were frozen and stored and are available for precise studies of a wide range of prenatal exposures including maternal infections (rubella, influenza), nutritional states (low folate, fatty acids), hormones (neuroendocrine), and neurotoxins (lead, polychlorinated biphenyls). In addition, being nested within an ongoing and long-term birth-cohort follow-up study, the PDS is suited to examine causal pathways that may underlie observed associations between early developmental events and schizophrenia. These analyses will capitalize on the full array of data in addition to prenatal, perinatal, and neonatal exposures, including the social and occupational histories of the parents, early childhood development, and in some cases adolescent development. Moreover, future research will extend the analysis of causal pathways to include neuroanatomic, neurochemical, and neuropsychological disturbances in relation to early developmental insult. Molecular genetic studies will elucidate, among other things, interactions between early brain insults and genetic mutations associated with schizophrenia.

Conclusion

Much can be gained by cross-fertilization between psychiatric and other chronic disease epidemiologists who are pursuing questions about the early origins of adult health. The same cohorts, methods, and even specific exposure measurements can often be useful to researchers in diferent domains. In addition, mechanisms proposed and studied for long-term physical effects of early environment have direct relevance to psychiatric outcomes. Both fields have undergone similar controversies and have evolved in similar ways. In each area, polarization between those emphasizing early (including genetic) factors and those emphasizing later environmental and behavioural factors seems to be resolving in favour of a life-course approach (72), in which both early and late factors and their interactions are examined. More significantly, it is increasingly evident that mental and physical health outcomes are themselves linked in diverse ways throughout the life course. The interplay of mental health and lifestyle on the one hand (78) and of physical health and cognitive functioning (79) on the other suggests to us that it should be the rule, rather than the exception, for research concerning the lifelong consequences of the early environment to consider outcomes in both domains.

Clinical Implications

Limitations

References

1. Kuh D, Davey Smith D. When is mortality risk determined? Historical insights in to a current debate. Society for Social Histroy of Medicine 1993;6:101–23.

2. Paneth N, Susser M. Early origin of coronary heart disease (the “Barker hypothesis”). BMJ 1995;310:411–2.

3. Joseph KS, Kramer MS. Review of the evidence on fetal and early childhood antecedents of adult chronic disease. Epidemiol Rev 1996;18:158–74.

4. DeGreef G, Ashtari M, Bogerts B. Volumes of ventricular system subdivisions measured from magnetic resonance images in first-episode schizophrenic patients. Arch Gen Psychiatry 1992;49:531–7.

5. Nopoulos P, Torres I, Flaum M, Andreasen NC, Ehrhardt JC, Yuh WT. Brain morphology in first-episode schizophrenia. Am J Psychiatry 1995;152:1721–3.

6. Lim KO, Tew W, Kushner M, Chow K, Matsumoto B, DeLisi LE. Cortical gray matter volume deficit in patients with first-episode schizophrenia. Am J Psychiatry 1996;153:1548–53.

7. Bogerts B, Ashtari M, DeGreef G, Alvir JMJ, Bilder RM, Lieberman JA. Reduced temporal limbic structure volumes on magnetic resonance images in first episode schizophrenia. Psychiatry Res 1990;35:1–13.

8. Gur RE, Cowell P, Turetsky BI, Gallagher F, Cannon T, Bilker W, and others. A follow-up magnetic resonance imaging study of schizophrenia. Relationship of neuroanatomical changes to clinical and neurobehavioral measures. Arch Gen Psychiatry 1998;2:145–52.

9. Jones P, Rogers B, Murray R, Marmot M. Child developmental risk factors for adult schizophrenia in the British 1946 birth cohort. Lancet 1994;344:1398–1402.

10. Cornblatt BA, Erlenmeyer-Kimling L. Global attentional deviance as a marker of risk for schizophrenia: specificity and predictive validity. J Abnorm Psychol 1985;94:470–86.

11. Green MF, Satz P, Gaier DJ, Gancell S, Kharabi F. Minor physical anomalies in schizophrenia. Schizophr Bull 1989;15:91–9.

12. Suddath RL, Christianson GW, Torrey EF, Casanova MF, Weinberger DR. Anatomical abnormalities in the brains of monozygotic twins discordant for schizophrenia. N Engl J Med 1990;322:789–94.

13. Mednick SA, Machon RA, Huttenen MO, Bonett D. Adult schizophrenia following prenatal exposure to an influenza epidemic. Arch Gen Psychiatry 1988;45:189–92.

14. O’Callaghan E, Sham P, Takei N, Glover G, Murray RM. Schizophrenia after prenatal exposure to 1957 A2 influenza epidemic. Lancet 1991;337:1248–50.

15. Takei N, Sham PC, O’Callaghan E, Murray GK, Glover G, Murray RM. Prenatal exposure to influenza and the development of schizophrenia: is the effect confined to females? Am J Psychiatry 1994;151:117–9.

16. Kunugi H, Nanko S, Takei N, Saito K, Hayashi N, Kazamatsuri H. Schizophrenia following in utero exposure to the 1957 influenza epidemics in Japan. Am J Psychiatry 1995;152:450–2.

17. McGrath J, Pemberton MR, Welham JL, Murray RM. Schizophrenia and the influenza epidemics of 1954, 1957 and 1959: a southern hemisphere study. Schizophr Res 1994;14:1–8.

18. Barr CE, Mednick SA, Munk-Jorgensen P. Exposure to influenza epidemics during gestation and adult schizophrenia. Arch Gen Psychiatry 1990;47:869–74.

19. Sham PC, O’Callaghan E, Takei N, Murray GK, Hare EH, Murray RM. Schizophrenia following pre-natal exposure to influenza epidemics between 1939 and 1960. Br J Psychiatry 1992;160:461–6.

20. Adams W, Kendell RE, Hare EH, Munk-Jorgensen P. Epidemiological evidence that maternal influenza contributes to the aetiology of schizophrenia. An analysis of Scottish, English, and Danish data. Br J Psychiatry 1993;163:522–34.

21. Takei N, van Os J, Murray RM. Maternal exposure to influenza and risk of schizophrenia: a 22-year study from The Netherlands. J Psychiatr Res 1995;29:435–45.

22. Susser ES, Lin SP, Brown AS, Lumey LH, Erlenmeyer-Kimling L. No relation between risk of schizophrenia and prenatal exposure to influenza in Holland. Am J Psychiatry 1994;151:922–4.

23. Erlenmeyer-Kimling L, Folnegovic Z, Hrabak-Zerjavic Boracic B, Folnegovic-Smalc V, Susser E. Schizophrenia and prenatal exposure to the 1957 A2 influenza epidemic in Croatia. Am J Psychiatry 1994;151:1496–8.

24. Torrey EF, Bowler AE, Rawlings R. An influenza epidemic and the seasonality of schizophrenic births. In: Kurstak E, editor. Second World Congress on viruses and mental health. New York: Plenum; 1991.

25. Crow TJ, Done DJ, Johnstone EC. Schizophrenia and influenza. Lancet 1991;38:116–7.

26. Cannon M, Cotter D, Coffey VP, Sham PC, Takei N, Larkin C, and others. Prenatal exposure to the 1957 influenza epidemic and adult schizophrenia: a follow-up study. Br J Psychiatry 1996;168:368–71.

27. Gregg NM. Congenital cataract following German measles in the mother. Trans Ophthalmology Society of Australia 1941;3:35–45.

28. South MA, Sever JL. Teratogen update: the congenital rubella syndrome. Teratology 1985;31:297–307.

29. Chess S, Korn S, Fernandez P. Psychiatric disorders of children with congenital rubella. New York: Brunner/Mazel; 1971.

30. Susser E, Hoek HW, Brown A. Neurodevelopmental disorders after prenatal famine. Am J Epidemiol 1998;147:213–6.

31. Susser E, Lin SP. Schizophrenia after prenatal exposure to the Dutch Hunger Winter of 1944–45. Arch Gen Psychiatry 1992;49:983–6.

32. Susser E, Neugebauer R, Hoek HW, Brown AS, Lin S, Labovitz D, and others. Schizophrenia after prenatal famine: further evidence. Arch Gen Psychiatry 1996;53:25–31.

33. Hoek HW, Susser E, Buck KA, Lumey LH, Lin SP, Gorman JM. Schizoid personality disorder after prenatal exposure to famine. Am J Psychiatry 1996;153:1637–9.

34. Huttenen MO, Niskanen P. Prenatal loss of father and psychiatric disorders. Arch Gen Psychiatry 1978;35:429–31.

35. van Os J, Selten JP. Prenatal exposure to maternal stress and subsequent schizophrenia: the May 1940 invasion of the Netherlands. Br J Psychiatry 1998;172:324–6.

36. Hollister JM, Laing P, Mednick SA. Rhesus incompatibility as a risk factor for schizophrenia in male adults. Arch Gen Psychiatry 1996;53:19–24.

37. Hollister JM, Brown A. Rhesus incompatibility as a risk factor for schizophrenia. In: Susser ES, Brown AS, Gorman JM, editors. Prenatal exposures in schizophrenia. Washington (DC): American Psychiatric Press. Forthcoming.

38. Lane EA, Albee GW. Comparative birth weights of schizophrenics and their siblings. J Psychol 1966;64:227–39.

39. Pollin W, Stabenau JR. Biological, psychological and historical differences in a series of monozygotic twins discordant for schizophrenia. In: Rosenthal D, Kety SS, editors. The transmission of schizophrenia. London: Pergamon Press; 1968.

40. Jones PB, Rantakallio P, Hartikainen AL, Isohanni M, Sipila P. Schizophrenia as a long-term outcome of pregnancy, delivery, and perinatal complications: a 28-year follow-up of the 1966 North Finland general population birth cohort. Am J Psychiatry 1998;155:355–64.

41. Hultman CM, Ohman A, Cnattingius S, Wieselgren IM, Lindstrom LH. Prenatal and neonatal risk factors for schizophrenia. Br J Psychiatry 1997;170:128–33.

42. McNeil TF, Cantor-Graae E, Nordstrom LG, Rosenlund T. Head circumference in ‘preschizophrenic’ and control neonates. Br J Psychiatry 1993;162:517–23.

43. Forsdahl A. Living conditions in childhood and adolescence - an important risk factor for arteriosclerotic heart disease? British Journal of Preventive and Social Medicine 1977;31:91–5.

44. Williams DRR, Roberts SJ, Davies TW. Deaths from ischemic heart disease and infant mortality in England and Wales. J Epidemiol Community Health 1979;33:199–202.

45. Barker DJP, Osmond C. Infant mortality, childhood nutrition, and ischemic heart disease in England and Wales. Lancet 1986:1077–81.

46. Barker DJP. Mothers, babies, and health in later life. Edinburgh: Churchill Livingstone; 1998.

47. Barker DJP, Winter PD, Osmond C, Margetts B, Simmonds SJ. Weight in infancy and death from ischemic heart disease. Lancet 1989;2:577–80.

48. Fall CHD, Osmond C, Barker DJP, Clark PMS, Hales CN, Stirling Y, and others. Fetal and infant growth and cardiovascular risk factors in women. BMJ 1995;310:428–32.

49. Langley SC, Jackson AA. Increased systolic blood pressure in adult rats induced by exposure to maternal low protein diets. Clin Sci 1994;86:217–22.

50. Paneth N, Rudelli R, Kazam E, Monte W. Brain damage in the preterm infant. London: MacKeith Press; 1994.

51. Leviton A, Paneth N. White matter damage in preterm newborns: an epidemiologic perspective. Early Hum Dev 1990;24:1–22.

52. Moody DM, Bell MA, Chalia VR. Features of the cerebral vascular pattern that predict vulnerability to perfusion or oxygenation deficiency: an anatomic study. American Journal of Neuroradiology 1991;11:431–9.

53. Weinberger DR. Implications of normal brain development for the pathogenesis of schizophrenia. Arch Gen Psychiatry 1987;44:660–9.

54. Lipska BK, DR Weinberger. Delayed effects of neonatal hippocampal damage on haloperidol-induced catalepsy and apomorphine-induced stereotypic behaviors in the rat. Brain Res Dev Brain Res 1993;75:13–22.

55. Whitehead AS, Gallagher P, Mills JL, Kirke PN, Burke H, Molloy AM, and others. A genetic defect in 5,10 methylenetetrahydrofolate reductase in neural tube defects. QJM 1995;11:763–6.

56. Stein Z, Susser M, Saenger G, Marolla F, editors. Famine and human development: the Dutch hunger winter of 1944–45. New York: Oxford University Press; 1975.

57. Lucas A. Programming by early nutrition in man. In: Bock GR, Whelen J, editors. The childhood environment and adult disease. Chichester: John Wiley; 1991. p 38–55.

58. Desai M, Crowther NJ, Ozanne SE, Lucas A, Hales CN. Adult glucose and lipid metabolism may be programmed during fetal life. Biochem Soc Trans 1995;23:331–5.

59. Persson E, Jansson T. Low birth weight is associated with elevated adult blood pressure in the chronically catheterized guinea-pig. Acta Physiol Scand 1992;145:195–6.

60. Edwards CRW, Bendiktsson R, Lindsay RS, Seckl JR. Dysfunction of placental glucocorticoid barrier: link between fetal environment and adult hypertension. Lancet 1993;341:355–7.

61. Dehri S, Snoeck A, Reusens-Billen B, Remacle C, Hoet JJ. Islet function in offspring of mthers on a low-protein diet during gestation. Diabetes 1991;40(Suppl 2):115–20.

62. Paneth N, Ahmed F, Stein AD. Early nutritional origins of hypertension: a hypothesis still lacking support. J Hyperten 1996;14(Suppl 5):S121–S129.

63. Susser E, Brown A, Gorman J, editors. Prenatal exposures in schizophrenia. Washington (DC): American Psychiatric Association Press; 1999.

64. Stein AD, Ravelli ACJ, Lumey LH. Famine, third trimester weight gain and intrauterine growth retardation: the Dutch famine birth cohort study. Hum Biol 1995;67:135–49.

65. Kramer MS. Effects of energy and protein intake on pregnancy outcome: an overview of the research evidence from controlled clinical trials. Am J Clin Nutr 1993;58:627–35.

66. Perry IJ. Fetal Growth and development: the role of nutrition and other factors. In: Kuh D, Ben-Shlomo Y, editors. A lifecourse approach to chronic disease epidemiology. New York: Oxford University Press; 1997.

67. Siegrist J. Epidemiologic and medical sociological aspects of hypertension. Herz 1995;20:302–8.

68. Davey Smith G, Hart C, Ferrell C, Upton M, Hole D, Hawthorne V, and others. Birth weight of offspring and mortality in the Renfrew and Paisley study: a prospective observational study. BMJ 1997;315:1189–93.

69. Leon DA, Koupilova I, Lithell HO, Berglund L, Mohsen R, Vagero D, and others. Failure to realize growth potential in utero and adult obesity in relation to blood pressure in 50 year old Swedish men. BMJ 1996;312:401–6.

70. Barker DJP. Mothers, babies, and disease in later life. London: BMJ Publishing Group; 1994.

71. Schwartz S, Susser E, Susser M. A future for epidemiology? Annu Rev Public Health 1999;20:1–19.

72. Kuh D, Ben-Shlomo Y, editors. A lifecourse approach to chronic disease epidemiology. New York: Oxford University Press; 1997.

73. Waddington JL, Callaghan E, Joussef H, Buckley P, Lane A, Copper D, and others. Schizophrenia: evidence for a cascade process with neurodevelopmental origin. In: Susser E, Brown A, Gorman J, editors. Prenatal exposures in schizophrenia. Washington (DC): American Psychiatric Association Press; 1999.

74. Diez-Roux AV. Bringing context back into epidemiology: variables and fallacies in multilevel analysis. Am J Public Health 1998 Feb;88:216–22.

75. Jablensky A, Sartorius N, Ernberg G, Ander M, Korten A, Cooper JE, and others. Schizophrenia: manifestations, incidence and course in different cultures. A World Health Organization ten-country study. Psychol Med Monogr 1992;20:1–97.

76. Broman S. The collaborative perinatal project: an overview. In: Mednick SA, Harway M, Finello KM, editors. Handbook of longitudinal research. City: Praeger Publishers; 1984.

77. Van den Berg BJ, Christianson RE, Oechsli FW. The California Child Health and Development Studies of the School of Public Health, University of California at Berkeley. Paediatr Perinat Epidemiol 1988;2:265–82.

78. Zuckerman B, Amaro H, Davies TW. Depressive symptoms during pregnancy: relationship to poor health behaviors. Am J Obstet Gynecol 1989;160:1107–11.

79. Chyou PH, White LR, Yano K, Sharp DS, Burchfiel CM, Chen R, and others. Pulmonary function measures as predictors and correlates of cognitive functioning in later life. Am J Epidemiol 1996;143:750–6.

Résumé

Objectif : Examiner la recherche à propos des influences prénatales sur la santé physique et mentale des adultes, et en déduire les conséquences sur l’orientation future de la recherche psychiatrique.

Méthode : La schizophrénie a été choisie comme exemple de santé mentale, et la maladie cardiovasculaire, comme exemple de santé physique. Pour chacun de ces troubles, on a examiné les résultats empiriques des influences prénatales et critiqué les méthodes utilisées pour en faire la démonstration.

Résultats : La recherche sur les antécédents prénataux de ces états s’est effectuée en parallèle : des résultats intrigants ont lié le retard de croissance attribuable à une agression foetale à la santé adulte ; des types semblables de trajectoire causale ont été proposés pour expliquer les relations ; et la recherche a été semée d’embûches semblables, dont l’absence de données précises sur l’exposition prénatale et la difficulté d’éviter la confusion. Les opinions dominantes sur les causes des maladies, qui ne conviennent pas bien à la recherche sur les antécédents prénataux, font obstacle à la recherche dans les deux champs. Pourtant, il y a eu peu d’échanges entre les chercheurs des deux disciplines.

Conclusions : Nous proposons un paradigme causal qui pourrait servir de guide à la recherche future sur les antécédents prénataux de la santé adulte, et promouvoir les échanges entre chercheurs en santé mentale et en santé physique. Le paradigme reflète l’opinion actuelle en épidémiologie en ne considérant pas que les facteurs de risque comme on les conçoit habituellement, mais aussi les chaînes causales au fil du temps et les influences causales aux divers niveaux d’organisation. Les répercussions sur la conception d’une nouvelle recherche sont illustrées par rapport à une étude permanente.

Manuscript received January 1999.

1Chair, Division of Epidemiology; Professor of Clinical Psychiatry and Clinical Public Health, Joseph L Mailman School of Public Health, Columbia University; Director, Epidemiology of Brain Disorders Department, New York State Psychiatric Institute, New York, New York.

2Assistant Professor of Clinical Psychiatry, New York State Psychiatric Institute, New York, New York.

3Senior Epidemiologist, Center for Urban Epidemiologic Studies, New York Academy of Medicine, New York, New York.

Address for correspondence: Dr E Susser, Chair, Epidemiology Division, The Joseph L Mailman School of Public Health, 622 West 168th Street, PH 18th floor, New York, NY  10032

Can J Psychiatry, Vol 44, May 1999