Vanishing twins, selection in utero, and infant mortality in the United States.

Background and objectives: Research to identify fetal predictors of infant mortality among singletons born in the United States (US) concludes that poorly understood and unmeasured "confounders" produce a spurious association between fetal size and infant death. We argue that these confounders include Vanishing Twin Syndrome (VTS)-the clinical manifestation of selection against frail male twins in utero. We test our argument in 276 monthly conception cohorts conceived in the US from January 1995 through December 2017.

Methodology: We use Box-Jenkins transfer function modeling to test the hypothesis that among infants born from 276 monthly conception cohorts conceived in the US from January 1995 through December 2017, the sex ratio of twins born in the 37th week of gestation will correlate inversely with infant mortality among singleton males born at the 40th week of gestation.

Results: We find support for our hypothesis and infer that the contribution of survivors of VTS to temporal variation in infant mortality among the hardiest of singleton male infants, those born at 40 weeks gestation, ranged from a decrease of about 7% to an increase of about 5% over our 276 monthly conception cohorts.

Conclusions and implications: We conclude that an evolutionary perspective on fetal loss makes a heretofore "unmeasured confounder" of the relationship between fetal size and infant mortality both explicable and measurable. This finding may help clinicians better anticipate changes over time in the incidence of infant mortality.