P1-409 Differential loss of participants does not necessarily cause selection bias
Journal of Epidemiology and Community Health
Objective To examine the effects of unplanned pregnancy on cognitive, behavioural and health outcomes of children at 5 years. Design Population-based cohort (Sweep 1 and 3 of the UK Millennium Cohort Study). Participant 13 529 singletons, born 2000e2001. Exposure measure Mothers reported whether their pregnancy was planned, and their feelings when first pregnant. The population was divided into "unplanned" pregnancies (unplanned and unhappy), "mistimed" pregnancies (unplanned but happy) and
... nned". Outcome measures At 5 years mothers answered questions about their child's health and behaviour (Strength and Difficulties Questionnaire). Children completed cognitive tests (British Abilities Scales II), including a component to measure verbal abilities. Results 17% of pregnancies were unplanned, 31% mistimed, and 52% planned. On average, children from unplanned pregnancies had cognitive scores indicating a delay of 5 months; were more likely to have behavioural problems (OR 2.9 (95% CI 2.3 to 3.3)); and more likely to be in poor health than the planned group (OR 2.3 (95% CI 1.8 to 3.0)). After adjustment the effects were reduced, though remained significant for behavioural and health outcomes: OR 1.3 (95% CI 1.1 to 1.6) and OR 1.6 (95% CI 1.2 to 2.1), respectively. The effects were smaller in the mistimed group, and after adjustment for confounding by socioeconomic and demographic factors, these effects disappeared. Conclusions While the adverse outcomes associated with mistimed pregnancies can be explained by confounding of socioeconomic circumstances, there remains a significant effect on the health and behaviour of children born after unplanned pregnancies. Potential confounding by other contextual or family factors will be examined.