Parental mental illness and the risk of offspring cancer in childhood: a pooled meta-analysis of English and Swedish national cohorts

Parental mental illness’ effects on risk of childhood cancers is largely unknown. This study determined the association between maternal or paternal mental illness and risk of childhood cancers. Retrospective cohort studies and meta-analysis using population-based registers from England and Sweden. 591,092 children born 1996–2017 (England) and 2,192,476 children born 1991–2011 (Sweden) were linked to their mothers (both countries) and fathers (Sweden), followed until latest December 2016 (Sweden) or July 2017 (England). Parental mental illness (depression/anxiety, psychosis, alcohol/substance use disorders, eating/personality disorders) were identified through primary (England) or secondary care (Sweden) databases as time-varying exposure, measured from one year before birth until the end of follow-up. Childhood cancer were identified from secondary care data. Hazard ratios (HRs) were estimated using Cox models separately in both countries, adjusted for potential confounders. Random-effects meta-analyses were used to estimate the association for maternal exposure. All estimates were characterised by uncertainty, with 95% confidence intervals including scenarios where there was no association. The point estimate represented an increased risk of childhood cancer associated with maternal alcohol or substance use disorders; however, the 95% confidence interval narrowly crosses the null, therefore, this cannot be inferred with certainty (pooled HR 1.30, 95% CI 0.97–1.75). Conversely, point estimates suggest a decreased risk of childhood cancer associated with maternal psychosis (HR 0.76, 95% CI 0.53–1.09) and paternal depression/anxiety (HR 0.85, 95% CI 0.71–1.02); however, these confidence intervals also cross one, so a null effect—or even a small increased risk—cannot be fully ruled out. There was some tentative evidence of an association between parental mental illness and childhood cancer; however, the uncertainty in the estimates precludes strong conclusions. Further data may clarify these associations.