The effect of birth weight discordance on twin neonatal mortality

OBJECTIVE: To estimate the association between birth weight discordance and neonatal mortality controlling for the effects of fetal growth, and to understand the differences in the incidence of mortality between larger and smaller infants. METHODS: This analysis is based on the National Center for Health Statistics matched multiple birth data set file containing all twin births in the United States from 1995 through 1997. Birth weight discordance was grouped into four levels (15-19%, 20-24%, 25-29%, and 30% or more). Generalized estimating equations were used to obtain adjusted odds ratios and 95% confidence intervals to estimate the mortality risk associated with discordance after adjusting for fetal growth. RESULTS: Mortality was 11 times higher among highly discordant smaller twins (30% or more) compared with nondiscordant smaller twins (43.4 and 3.8 per 1000, respectively). Risk estimates ranged from 1.08 (95% confidence interval 0.85, 1.38) among 15-19% discordant twins to 2.05 (95% confidence interval 1.66, 2.51) among 30% or more discordant twins. Larger twins had similar risk estimates. After accounting for the association between fetal growth and discordance, mortality risk was substantially higher among smaller and larger twins who were highly discordant (30% or more). In addition, there was little difference in the magnitude of risk estimates between highly discordant smaller and larger twins. CONCLUSION: After controlling for fetal growth, smaller and larger twins affected by higher levels of birth weight discordance (25% or more) remain at disproportionate risk for neonatal mortality when compared with other smaller or Larger twins. Additionally, smaller twins do not have an elevated risk compared with Larger twins after adjusting for their different fetal growth distributions.