Birth weight, gestational age and late adolescent liver function using twin status as instrumental variable in a Hong Kong Chinese birth cohort: "Children of 1997"

Abstract

Birth weight (BW) is inversely associated with diabetes and liver function in Mendelian Randomization studies. Observationally, lower BW is usually also associated with poorer liver function. However, these studies could be confounded by socioeconomic position. Here we assessed if BW is associated with liver function in a unique population with little socio-economic patterning of BW, using both instrumental variable and an observational analysis. We used instrumental variable analysis (IVA) to assess the association of BW with liver function (alanine transaminase (ALT), alkaline phosphatase (ALP), bilirubin, and albumin) at ~17 years with twin status as an instrumental variable in the prospective population-representative "Children of 1997" birth cohort (n = 8327). We also conducted an observational analysis adjusted for sex, maternal age, maternal migrant status, smoking and parental socio-economic position. A generalized linear model with gamma family was used for ALT, ALP, and bilirubin because they are not normally distributed. Using IVA, BW was not associated with ALT, ALP or bilirubin, but was possibly negatively associated with albumin (-1.12 g/L, 95% confidence interval (CI) -2.08 to -0.16). Observationally, BW was negatively associated with ALT (-1.23 IU/L, 95% CI -2.16 to -0.30), ALP (-1.72 IU/L, 95% CI -3.43 to -0.01) and higher albumin (-0.23 g/L, 95% CI -0.40 to -0.06). Poor liver function may be a pathway by which the risks of lower BW are actuated. This insight might help identify post-natal targets of intervention to mitigate the adverse health effects of lower birth weight.

Keywords: Birth weight; Epidemiology; Instrumental variable analysis; Liver function; Twin status.