Background Previous findings on the association between stress during pregnancy and pregnancy loss are inconsistent. We aimed to estimate this association using a large prospective cohort.

Methods This population-based study included all 1 303 660 clinically recognised pregnancies in Denmark between 1995 and 2008. We categorised women as exposed to severe stress if they lost a child, sibling or parent during pregnancy. Cox Proportional Hazards models were used to study the association between exposure and rate of fetal death, starting with the follow-up on the day of completion of week 4 of pregnancy. In an attempt to control for unknown potential confounders, we also designed a pregnancy-matched analysis in which each woman had her own baseline risk of pregnancy loss and controls therefore for genetic and time-stable environmental factors.

Results A total of 146 031 pregnancies ended in clinically recognised fetal loss (11.2%) and a total of 10 808 (0.8%) women were categorised as exposed. The overall risk of pregnancy loss was similar in the exposed and unexposed (aHR=1.05, 95% CI 0.95 to 1.15). Results from the pregnancy-matched analysis (performed in 423 women) showed stronger and significant associations (aHR=1.83, 95% CI 1.49 to 2.25). All the analyses indicated a stronger effect of bereavement when the mother lost a child or when the death was unexpected.

Conclusions Our main results suggested no strong association between severe stress during pregnancy and risk of pregnancy loss. Results from the pregnancy-matched analyses considered information from a selected and small group of women for whom there may exist a stronger association between stress during pregnancy and pregnancy loss. The fact that an unexpected death or the loss of a child had a stronger effect in both analyses may indicate that severe stressful situations increase the risk of pregnancy loss.