31–33 Birth weight was classified as ≥2.5 kg (‘normal’) or <2.5 kg (‘low’). Gestation was recorded in weeks and classified as <28, 28–32, 33–36 or ≥37 weeks. The mode of Gemcitabine DNA Synthesis delivery was categorised as ‘normal’, ‘instrumental’ or ‘caesarean’. Sociodemographic factors Ethnicity was analysed as ‘white’ or ‘other ethnic group’. Parity was the number of children the mother

had (including the cohort member) and was coded as 1, 2, 3 or >3 children. Family status was categorised as ‘lone parent’, ‘cohabiting’ or ‘married’. Household socioeconomic class was measured by taking the occupation of the parent with the highest socioeconomic position according to the four UK National Statistics socioeconomic categories. Household income was calculated from the self-reported data on the questionnaire. Mother’s education was determined by the highest attainment of a National Vocational Qualification or equivalent group. These qualifications were grouped as follows: ‘higher’ (bachelor’s degree or equivalent),

‘medium’ (end of schooling at age 18, A Level or equivalent), ‘lower’ (end of compulsory schooling at age 16, General Certificate of Secondary Education (GCSE) or equivalent) or others. Inclusion and exclusion criteria The Millennium Cohort Study did not recruit families if the child had been taken into care at the time of initial assessment. One study participant, who withdrew consent after the study began, was excluded. For this analysis, mothers were included if they were the birth mother of the Millennium Cohort Study participant. Mothers who did not answer the question of whether they lived away from home were excluded, as were mothers who answered the question as “I don’t know”. Statistical analysis First, we compared the following characteristics of the ‘exposed’ with ‘unexposed’ groups using the χ2 statistic: age at delivery, ethnic group, social class, household income, education, family status, parity, smoking during pregnancy, symptoms of depression, mode of delivery, gestational age, birth weight and duration of feeding. We then used logistic

regression to estimate ORs for a history of time spent in care and the outcomes of smoking during pregnancy, birth weight, breastfeeding initiation and symptoms of depression, with adjustment for potential confounding Cilengitide factors. A plausible model was developed based on background literature and included the following potential confounders: age at delivery, ethnic group, social class, household income and education. Previous evidence suggests that these factors are associated with poor perinatal outcomes,34 although there has been very little previous evidence on how factors relating to time in care manifest in maternal and neonatal outcomes. All of these potential confounders were significantly associated with the outcome (indicated by Wald, p<0.