Sibling Comparison Designs

Abstract

Twins, full siblings, and half-siblings are increasingly used as comparison groups in matched cohort and matched case-control studies. The "within-pair" estimates acquired through these comparisons are free from confounding from all factors that are shared by the siblings. This has made sibling comparisons popular in studying associations thought likely to suffer confounding from socioeconomic or genetic factors. Despite the wide application of these designs in epidemiology, they have received little scrutiny from a statistical or methodological standpoint. In this paper we show, analytically and through a series of simulations, that the standard interpretation of the models is subject to several limitations that are rarely acknowledged.Although within-pair estimates will not be confounded by factors shared by the siblings, such estimates are more severely biased by non-shared confounders than the unpaired estimate. If siblings are less similar with regard to confounders than to the exposure under study, the within-pair estimate will always be more biased than the ordinary unpaired estimate. Attenuation of associations due to random measurement error in exposure will also be higher in the within-pair estimate, leading within-pair associations to be weaker than corresponding unpaired associations, even in the absence of confounding. Implications for the interpretation of sibling comparison results are discussed.

Age at Menarche and Cardiometabolic Health: A Sibling Analysis in the Scottish Family Health Study

“…This interpretation requires a strong assumption: that there is little individual‐level confounding. If siblings differ to a greater extent with regard to distributions of potential confounders than to the exposure of interest, the within‐sibships analysis may be more biased than a standard analysis 21, 33. Consequently, a key underlying assumption is that childhood adiposity (a key potential confounder in this study), and other lifestyle characteristics, are more similar within sisters than between unrelated individuals, and that the concordance for these potential confounders is greater than the concordance for age at menarche.…”

“…Sibling studies controls for confounding (measured and unmeasured) by characteristics shared within families 21, 22. The underlying assumption of this approach is that siblings share identical or very similar early life environments in addition to, on average, half of their genetic architecture—often referred to as fixed family characteristics —and thus individual‐level confounding by characteristics that vary between siblings will be minimal 23.…”

Farmers' mental health: A longitudinal sibling comparison – the HUNT study, Norway

“…Sibling comparison designs allow adjustment for unmeasured confounders shared by the siblings, and even though such studies cannot prove causality, they can support causal reasoning when also taking already existing knowledge and complementary analyses into account (11). To our knowledge, this study is the first to use a sibling comparison design to investigate the mental health of farmers.…”

“…One way to overcome this limitation is to compare siblings in the same family. When comparing mental health symptom load in relation to different occupations, a family design will control for all confounding from shared factors between the siblings (11). Further, the available literature on the mental health of farmers is largely based on cross-sectional data (1), and there is a need for prospective studies.…”

Age at immigration and crime in Stockholm using sibling comparisons

“…First, the nature of analyzing discordant siblings promotes a spurious correlation between confounders and age at immigration (see Frisell et al, 2012). This, however, is a minimal problem due to the generally exogenous nature of the different ages at immigration between siblings as they relate to shared family factors.…”