Objective To examine the effects of pregnancy planning, time to conception and assisted reproductive treatment on expressive language ability at 3 years.

Design Population-based cohort (Sweeps 1 and 2 of the UK Millennium Cohort Study).

Setting UK.

Participants 11 790 English-speaking singletons born 2000–2001.

Exposure measure Mothers reported whether the pregnancy was planned, and their feelings when first pregnant. Women who planned their pregnancy provided time to conception (TTC) and details of any assisted reproductive treatment (ART). The population was divided into “unintended” pregnancies (unplanned and unhappy about pregnancy), “unplanned” pregnancies (unplanned but happy), “planned” (planned and TTC≤12 months), “subfertile” (planned and TTC>12 months), “ovulation stimulated (OS)” (received clomiphene citrate) and “ART” (in vitro fertilisation or intracytoplasmic sperm injection).

Outcome measure Expressive language ability was assessed using the Naming Vocabulary component of the British Ability Scale. There is evidence that expressive language skills predict cognitive ability.

Results 15% of pregnancies were unintended and a further 26% were unplanned. 53% were planned; 4% subfertile; 2% used OS and 1% ART to help them conceive. The mean BAS naming vocabulary score was 76.5 in the planned group and was not significantly different in the OS or subfertile groups. In the unadjusted analysis the mean score was lower among the unintended and unplanned groups (difference in mean score was −4.8 (95% CI −6.0 to 3.7) and −2.8 (95% CI −3.4 to −1.51) respectively) and higher in the ART group (3.8 (95% CI −0.2 to 7.85)). This equates to a 3.8 month cognitive delay in the unintended group, and to a 3.0 month advantage in the ART group, compared to the planned group. These effects almost entirely disappear once the models are adjusted for sociodemographic factors (parent's income, qualifications, mother's age, whether the baby is her first child and alcohol in pregnancy). Compared to the fertile group, the difference in means becomes −0.9 (95% CI −1.9 to 0.2) in the unintended group, −0.4 (95% CI −1.2 to 0.4) in the planned group and 1.2 (95% CI −2.0 to 4.4) in the ART group. Further adjustment for lifecourse mediating factors (birthweight, parent-child relationship, and parental involvement) makes little difference to the final results.

Conclusions We find no evidence that pregnancy planning, subfertility or ART adversely affects children's expressive language ability at 3 years of age. The effects observed in the crude analyses are almost entirely explained by confounding by socio-economic factors, indicating that maternal education and poverty are strong influences on children's cognitive tests in early childhood.