Estimating sibling spillover effects with unobserved confounding using gain-scores

Area of Science:

• Epidemiology
• Family Health Studies
• Biostatistics

Background:

• Sibling spillover effects, where one sibling's health exposure influences another's outcome, are an emerging area of epidemiological research.
• Unobserved familial factors can confound the identification of these health spillovers.
• Accurate identification is challenging due to potential bidirectional influences.

Purpose of the Study:

• To demonstrate a gain-score (fixed effects) regression method for identifying exposure-to-outcome spillover effects within sibling pairs.
• To assess the method's ability to distinguish between one-sided and two-sided spillover effects.
• To highlight the limitations of the method, particularly concerning outcome-to-exposure or outcome-to-outcome spillovers.

Main Methods:

• Employed a gain-score (fixed effects) regression model within linear frameworks.
• Applied the method to analyze sibling pairs from a large, population-based birth cohort in Wisconsin.
• Utilized Monte Carlo simulations and analytic results to validate the method and its limitations.

Main Results:

• Estimated the spillover effect of a child's preterm birth on an older sibling's literacy skills.
• Analyzed 20,010 sibling pairs, finding that preterm birth modestly decreases older siblings' literacy scores without covariate adjustment.
• Demonstrated the method's capacity to identify specific types of spillover effects.

Conclusions:

• Gain-score regression is a valuable strategy for identifying exposure-to-outcome spillover effects in sibling pairs.
• This method effectively controls for sibling-invariant unobserved confounding factors.
• The approach offers a promising avenue for understanding intrafamilial health dynamics.