Events causing stress responses during sensitive periods of rapid neurological development in childhood may be early determinants of all-cause premature mortality. Using a British birth cohort study of individuals born in 1958, the relationship between adverse childhood experiences (ACE) and mortality ≤50 year was examined for men (n = 7,816) and women (n = 7,405) separately. ACE were measured using prospectively collected reports from parents and the school: no adversities (70 %); one adversity (22 %), two or more adversities (8 %). A Cox regression model was carried out controlling for early life variables and for characteristics at 23 years. In men the risk of death was 57 % higher among those who had experienced 2+ ACE compared to those with none (HR 1.57, 95 % CI 1.13, 2.18, p = 0.007). In women, a graded relationship was observed between ACE and mortality, the risk increasing as ACE accumulated. Women with one ACE had a 66 % increased risk of death (HR 1.66, 95 % CI 1.19, 2.33, p = 0.003) and those with ≥2 ACE had an 80 % increased risk (HR 1.80, 95 % CI 1.10, 2.95, p = 0.020) versus those with no ACE. Given the small impact of adult life style factors on the association between ACE and premature mortality, biological embedding during sensitive periods in early development is a plausible explanatory mechanism.Electronic supplementary materialThe online version of this article (doi:10.1007/s10654-013-9832-9) contains supplementary material, which is available to authorized users.

BackgroundTo analyse whether Adverse Childhood Experiences (ACE) are associated with an increased risk of cancer.MethodsThe National child development study (NCDS) is a prospective birth cohort study with data collected over 50 years. The NCDS included all live births during one week in 1958 (n = 18558) in Great Britain. Self-reported cancer incidence was based on 444 participants reporting having had cancer at some point and 5694 reporting never having cancer. ACE was measured using reports of: 1) child in care, 2) physical neglect, 3) child’s or family’s contact with the prison service, 4) parental separation due to divorce, death or other, 5) family experience of mental illness & 6) family experience of substance abuse. The resulting variable had three categories, no ACEs/ one ACE/ 2 + ACEs and was used to test for a relationship with cancer. Information on socioeconomic characteristics, pregnancy and birth were extracted as potential confounders. Information on adult health behaviours, socioeconomic environment, psychological state and age at first pregnancy were added to the models. Multivariate models were run using multiply-imputed data to account for missing data in the cohort.ResultsThe odds of having a cancer before 50 y among women increased twofold for those who had 2+ ACEs versus those with no ACEs, after adjusting for adult factors and early life confounders (OR: 2.1, 95% CI: 1.42-3.21, p < 0.001).ConclusionThese findings suggest that cancer risk may be influenced by exposure to stressful conditions and events early on in life. This is potentially important in furthering our understanding of cancer aetiology, and consequently in redirecting scientific research and developing appropriate prevention policies.

When estimating the causal effect of an exposure of interest on change in an outcome from baseline, the choice between a linear regression of change adjusted or unadjusted for the baseline outcome level is regularly debated. This choice mainly depends on the design of the study and the regression-to-the-mean phenomena. Moreover, it might be necessary to consider additional variables in the models (such as factors influencing both the baseline value of the outcome and change from baseline). The possible combinations of these elements make the choice of an appropriate statistical analysis difficult. We used directed acyclic graphs (DAGs) to represent these elements and to guide the choice of an appropriate linear model for the analysis of change. Combined with DAGs, we applied path analysis principles to show that, under some functional assumptions, estimations from the appropriate model could be unbiased. In the situation of randomized studies, DAG interpretation and path analysis indicate that unbiased results could be expected with both models. In the case of confounding, additional (and sometimes untestable) assumptions, such as the presence of unmeasured confounders, or effect modification over time should be considered. When the observed baseline value influences the exposure ("cutoff designs"), linear regressions adjusted for baseline level should be preferred to unadjusted linear regression analyses. If the exposure starts before the beginning of the study, linear regression unadjusted for baseline level may be more appropriate than adjusted analyses.