The Intergenerational Transmission of Low Birth Weight: A Large Multigenerational Cohort Study in Taiwan

We study the origins of health disparities by identifying the intergenerational fetal programming effect. We create two unique three-generational samples for matrilineage and patrilineage to ask whether the fetal programming effect is inheritable and how socioeconomic status (SES) can help to modify the transmission cycle. We contribute to the literature by differentiating the maternal fetal programming transmissions from paternal ones and identifying a gender specific inheritance pattern.

Fetal programming hypothesis states that fetus facing with compromised intrauterine environment may not only slow down its growth rate to reduce nutritional requirements but might also modify its physiology in a durable fashion, leading to increased risk of chronic disease in later life. Therefore, low birth weight (LBW) is used as a major proxy for fetal programming in our study. We first examine the strength of the intergenerational transmission of fetal programming by exploiting the LBW correlations between two adjacent generations. Grandparent fixed effects are controlled to account for family specific time invariant unobservables. We then attempt to identify possible modifier by including interactions between SES and parental LBW in the model. Three surrogates for SES are employed: town-level income, parental years of schooling, and changes in years of schoolings from the first to the second generation.

Two datasets across three consecutive generations are created based on confidential Taiwan annual Birth Certificate Records (BCR) from 1978 to 2006. The BCR provide us with information on infant health at birth, parental demographics as well as parents’ birth dates and personal identification (ID) numbers, making it possible for us to obtain information on three generations by liking the BCR of two generations. The mother (father) sample is constructed by matching infant’s BCR to mother’s (father’s) BCR according to mother’s (father’s) date of birth and personal ID number. We obtained town-level income from Ministry of Finance in Taiwan.

We find that the intergenerational fetal programming effect occurs matrilineally, which is consistent with animal and human studies in epidemiology (Drake and Walker, 2004). The inheritance is resistant to income change but moderately responsive to educational improvement. And females show more responsiveness to the modifier. Similar results are found using indicators for intrauterine growth restrictions (IUGR) such as LBW at term, small for gestational age at 5^th percentile, and less than two standard deviation below mean of gestation.

The findings suggest that improving maternal health will generate positive spillovers to future generation. And education policy is relatively more effective than income policy for population at the lower end of the birth weight distribution, as it is possible to modify the transmission of the fetal programming effect when at least one of the parents has high school equivalent education level or the family educational background has been improved fast over the past years.