There is a strong positive association between paternal age and schizophrenia

2004;329:1070 (6 November), doi:10.1136/bmj.38243.672396.55 (published 22 October 2004)

Paper
Paternal age and schizophrenia: a population based cohort study
Attila Sipos, honorary senior clinical lecturer in psychiatry1, Finn Rasmussen, senior clinical lecturer and associate professor of epidemiology2, Glynn Harrison, professor of mental health1, Per Tynelius, senior statistician2, Glyn Lewis, professor of psychiatric epidemiology1, David A Leon, professor of epidemiology3, David Gunnell, professor of epidemiology4

1 Academic Unit of Psychiatry, Cotham House, University of Bristol BS6 6JL, 2 Department of Public Health Sciences, Karolinska Institute, Norrbacka, SE-17176 Stockholm, Sweden, 3 Department of Epidemiology and Population Health, London School of Hygiene and Tropical Medicine, London WC1E 7HT, 4 Department of Social Medicine, Canynge Hall, Bristol BS8 2PR

Correspondence to: F Rasmussen finn.rasmussen@phs.ki.se


Abstract
Abstract
Introduction
Method
Results
Discussion
References

Objective To investigate the association of paternal age at conception with the risk of offspring developing schizophrenia.
Design A population based cohort study.

Setting Sweden.

Subjects 754 330 people born in Sweden between 1973 and 1980 and still alive and resident in Sweden at age 16 years.

Main outcome measures Hospital admission with schizophrenia or non-schizophrenic, non-affective psychosis.

Results After adjustment for birth related exposures, socioeconomic factors, family history of psychosis, and early parental death the overall hazard ratio for each 10 year increase in paternal age was 1.47 (95% confidence interval 1.23 to 1.76) for schizophrenia and 1.12 (0.98 to 1.29) for non-schizophrenic non-affective psychosis. This association between paternal age and schizophrenia was present in those with no family history of the disorder (hazard ratio for each 10 year increase in paternal age 1.60, 1.32 to 1.92), but not in those with a family history (0.91, 0.44 to 1.89) (P = 0.04 for interaction).

Conclusions Advancing paternal age is an important independent risk factor for schizophrenia. The stronger association between paternal age and schizophrenia in people without a family history provides further evidence that accumulation of de novo mutations in paternal sperm contributes to the overall risk of schizophrenia.


........



Our findings confirm an association between increased paternal age and schizophrenia in offspring, which remained even after we controlled for a wide range of potential confounding factors. The association seems to be relatively specific to schizophrenia compared with non-schizophrenic non-affective psychosis and was stronger in those with no family history of the disorder and those with normal Apgar scores at birth.
Strengths and weaknesses of study
We used routinely recorded data on variables related to birth, parents, and adulthood collected before the onset of disease. Furthermore, as cases were ascertained from a national inpatient register the possibility of selection bias was reduced. The large number of cases gives us statistical power to control for a wide range of important confounding factors and investigate whether the associations differ in relation to family history of psychosis or environmental risk factors.

The main limitation of our analysis is that case ascertainment was based on people admitted to hospital only with diagnoses recorded on an administrative database. Though we will have missed people who were not admitted to hospital, studies in the United Kingdom indicate that in the first three years after presentation over 80% of patients are admitted, even in areas with community oriented services.18 Furthermore, analyses of diagnoses recorded on the Swedish inpatient discharge register indicate that schizophrenia is diagnosed with reasonable accuracy.19-20 Another limitation is that a family history of admission with schizophrenia is only a marker for and not the equivalent of genetic vulnerability.

Comparison of findings with earlier research
In comparing our findings with those of earlier studies1-3 5 6 it is important to bear in mind that we were able to distinguish between narrowly defined schizophrenia (ICD-10 F20) and other non-schizophrenic non-affective psychosis (ICD-10 F21-29). One previous study1 also reported a stronger association with paternal age in relation to narrowly defined schizophrenia.

We found a pattern of association across the categories of paternal age that suggested a J shaped rather than a linear association, in line with the initial unadjusted findings of Byrne et al.2 In our study this pattern remained even after we controlled for possible confounders. Two other studies also investigated the possibility of a stronger association of increased paternal age in people with schizophrenia without a family history as opposed to those with a family history, as this would provide further evidence to support the hypothesis that this association is due to accumulating de novo mutations in the germ cells of older fathers. Malaspina et al showed that paternal age was significantly higher for people with schizophrenia without a family history in a small study that compared 35 familial cases with 68 sporadic cases.12 Zammit et al found no evidence for such effects, but their study lacked statistical power.6

The finding that the association between increased paternal age and schizophrenia was evident only in people with normal Apgar scores was based on a relatively small number of cases in people with low scores (n = 31) and should be viewed as preliminary until replicated in other studies.

Conclusions
Our findings confirm advancing paternal age as a strong independent risk factor for schizophrenia and indicate that 15.5% of cases of schizophrenia in our cohort could be due to the patient having a father who was aged > 30 years at birth. We found a stronger association in subjects without a family history of schizophrenia, providing further evidence to support the theory that accumulating de novo mutations in the germ cells of older fathers might contribute to an increased risk of schizophrenia in their offspring.




What is already known on this topic
Increased paternal age is associated with several diseases, possibly due to the age associated increase in sporadic de novo mutations in male germ cells

Several studies have reported an association between paternal age at conception and their offspring's risk of schizophrenia

If this association was due to de novo mutations one would expect to find a stronger association between paternal age and schizophrenia in cases with no family history of the disorder

What this study adds

There is a strong positive association between paternal age and schizophrenia that is not due to sociodemographic, birth related, or socioeconomic factors or family history or early parental death

Paternal age is only weakly associated with other non-affective, non-schizophrenic psychosis

This association is stronger in those with no family history of schizophrenia, supporting the hypothesis that accumulating de novo mutations in the germ lines of older fathers could play an important part in the aetiology of schizophrenia






In England and Wales the average paternal age has increased from 29.2 years in 1980 to 32.1 in 2002.21 Using our data we can estimate the hazard ratio associated with an average increase of paternal age by 2.9 years as 1.12. Assuming a conservative background annual incidence rate for schizophrenia of 10/100 00022 for the whole of the UK population (59.2 million in 2002) and that the observed associations are truly causal, this increase in paternal age since 1980 could have led to an additional 710 cases of schizophrenia in 2002 over and above the 5923 cases due to the background incidence rate.




--------------------------------------------------------------------------------
Two extra tables of data can be found on bmj.com
We thank Christina Dalman, Peter Allebeck, and Susanne Wicks for their helpful comments, Stan Zammit for his help with accessing all the relevant literature, and Geoff Adams for statistical/database support.

Contributors: DG, FR, and GH developed the core idea, and FR, AS, DG, GH, GL, DAL, and PT were involved in the design of the study. DG, FR, GH, GL, and DAL raised the research funds. PT prepared the cohort data and did the database linkages. AS carried out the statistical analysis and literature search and wrote the first draft of the paper. DG, DAL, and GL gave advice on statistical methodology. All authors critically reviewed and contributed to the final draft of the paper. FR and AS are guarantors.

Funding: Stanley Medical Research Institute.

Competing interests: None declared.

Ethical approval: The ethics committee at the Karolinska Institute, Stockholm, Sweden approved this study.