Riordan et al (Reference Riordan, Selvaraj and Stark2006) present evidence that maternal circumstances and foetal experience may have an impact upon the subsequent mental health of the offspring. Many studies describe gestational insults, obstetric complications and perinatal environment as risk factors for mental illness in later life. Such evidence has often been based upon longitudinal cohort studies which have the advantage of large sample sizes and masked assessments at both exposure and outcome. Such strategies have generated evidence supporting the neuro-developmental hypothesis of schizophrenia (Reference Done, Johnstone and FrithDone et al, 1991; Reference Jones, Rodgers and MurrayJones et al, 1994). There can be little doubt that such epidemiological evidence can lead to hypotheses of the pathogenesis of psychiatric illness. As suggested by Riordan et al, foetal nutrition, intra-uterine stressors, hypothalamic–pituitary–adrenal axis dysfunction and attachment theory may all be putative mechanisms by which the foetal–maternal interaction contributes to future psychiatric illness.

However, there are several inherent limitations to this methodology. Longitudinal studies have traditionally concentrated on descriptions of the progeny. Data relating to many maternal factors in birth cohort studies are limited or unavailable. Important confounding factors cannot be, and have not been, eliminated in such work. Social class, alluded to by Riordan et al, cannot be ignored as an important confounder for all of the findings. Riordan et al concede that assessment of economic circumstances is based only on parental occupation. Is such a measure valid over several generations? How can social class (a factor relating both to parity, and to young maternal age and depression and suicide) not be considered an important confounding variable? The importance of housing and maternal diet, alcohol, smoking and drug use will certainly influence birth weight and depression and suicide in later life. Family history of psychiatric illness is probably the most important confounding factor that has not been, and unfortunately cannot be, assessed in this study. Maternal depression may have an impact upon birth weight as an environmental factor, but may exert a genetic effect on psychiatric illness and suicide of the offspring. Although such a study does suggest important epidemiological trends, hypothesising that biological mechanisms are involved in psychiatric illness of offspring is premature without controlling for these important confounders.