Systematic review and meta-analysis on the association of prepregnancy underweight and miscarriage.

BACKGROUND: Maternal underweight, overweight and obesity have been associated with a higher risk of miscarriage. Most individual reports and all meta-analyses have addressed high body mass index.

OBJECTIVES: To review the literature and summarize the risk of miscarriage in underweight women vs those with normal weight.

METHODS: A Medline Search (1st January 1990-20th November 2015, human, in English, French, Italian, Spanish or Portuguese) was conducted. Both spontaneous pregnancies and pregnancies after assisted reproduction techniques were considered. Cohort and case control studies were included if they reported data on the outcome of interest (clinical miscarriage), in underweight and normal weight women. Information on clinical miscarriage in other body mass index categories was collected when available. Two investigators reviewed the abstracts, full text papers and extracted data. Review Manager 5.1 software was used to summarize the results.

RESULTS: 32 studies (30 cohort, 2 case control) and a total of 265,760 women were included. In cohort studies, the relative risk (RR) of clinical miscarriage in underweight women was 1.08, 95% CI 1.05-1.11; p<0.0001). The corresponding figures were RR 1.09, 95% CI 1.04-1.13; p<0.0001 for overweight women and RR 1.21, 95% CI 1.15-1.27; p<0.00001 for obese women. In case control studies, the odds ratio (OR) of clinical miscarriage in underweight women was 1.02, 95% CI 0.46-2.30; p=0.95). The corresponding figures were OR 1.01, 95% CI 0.88-1.16; p=0.89 for overweight women and OR 1.26, 95% CI 1.01-1.57; p=0.04 for obese women. The limitations of this study are that it is restricted to studies with information on underweight women and that I(2) ranges from 0 to 91% in different subgroups.

CONCLUSION: We conclude that maternal underweight is associated with a slightly increased risk of clinical miscarriage, similar to that of overweight women and lower than the risk observed in obesity. The heterogeneity displayed in some subgroups limits the strength of the conclusion.