A Mendelian Randomization Approach to Assessing a Causal Relation Between Malaria and Stunting in Children

• Activity Number: 304
• Type: Contributed
• Date/Time: Tuesday, July 31, 2012 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Hyunseung Kang*+
• Companies: The Wharton School
• Keywords: Mendelian Randomization ; Malaria ; Stunting ; Child Development ; Ghana ; Instrumental Variables

Abstract

While a strong association between repeated malaria episodes and stunting in children has been observed, there is considerable difficulty in elucidating a causal relationship. Randomized experiments are impractical, presence of confounders including malnutrition, breastfeeding habits, and socioeconomic status introduce bias, and there is reason to believe reverse causality may be in play. In this study, we utilize Mendelian randomization, a type of instrumental variables method, where the sickle cell trait is used as the instrument. The trait is an ideal choice because, by Mendel's law of inheritance, the trait is randomly assigned to children at birth. We present a potential outcomes model and reveal conditions for consistent estimation of the causal parameters under Mendelian randomization. We also demonstrate the strength of the method on a data set containing 1070 children from the Ashanti Region in Ghana who were recruited and followed up with genetic, anthropometric, and demographic measurements until two years of age.