Overcoming attenuation bias in regressions using polygenic indices

Separating Scarring Effect and Selection of Early-Life Exposures With Genetic Data

Causal life course research examining consequences of early-life exposures has largely relied on associations between early-life environments and later-life outcomes using exogenous environmental shocks. Nonetheless, even with (quasi-)randomized early-life exposures, these associations may reflect not only causation ("scarring") but also selection (i.e., which members are included in data assessing later life). Investigating this selection and its impacts on estimated effects of early-life conditions has, however, often been ignored because of a lack of pre-exposure data. This study proposes an approach for assessing and correcting selection, separately from scarring, using genetic measurements. Because genetic measurements are determined at the time of conception, any associations with early-life exposures should be interpreted as selection. Using data from the UK Biobank, we find that in utero exposure to a higher area-level infant mortality rate is associated with genetic predispositions correlated with better educational attainment and health. These findings point to the direction and magnitude of selection from this exposure. Corrections for this selection in examinations of effects of exposure on later educational attainment suggest underestimates of 26-74%; effects on other life course outcomes also vary across selection correction methods.

An analysis of the accuracy of retrospective birth location recall using sibling data

Many surveys ask participants to retrospectively record their location of birth. This paper examines the accuracy of such data in the UK Biobank using a sample of full siblings. Comparison of reported birth locations for siblings with different age gaps allows us to estimate the probabilities of household moves and of misreported birth locations. Our first contribution is to show that there are inaccuracies in retrospective birth location data, showing a sizeable probability of misreporting, with 28% of birth coordinates, 16% of local districts and 6% of counties of birth being incorrectly reported. Our second contribution is to show that such error can lead to substantial attenuation bias when investigating the impacts of location-based exposures, especially when there is little spatial correlation and limited time variation in the exposure variable. Sibling fixed effect models are shown to be particularly vulnerable to the attenuation bias. Our third contribution is to highlight possible solutions to the attenuation bias and sensitivity analyses to the reporting error.

Genetics, epigenetics, and neurobiology of childhood-onset depression: an umbrella review

Depression is a serious and persistent psychiatric disorder that commonly first manifests during childhood. Depression that starts in childhood is increasing in frequency, likely due both to evolutionary trends and to increased recognition of the disorder. In this umbrella review, we systematically searched the extant literature for genetic, epigenetic, and neurobiological factors that contribute to a childhood onset of depression. We searched PubMed, EMBASE, OVID/PsychInfo, and Google Scholar with the following inclusion criteria: (1) systematic review or meta-analysis from a peer-reviewed journal; (2) inclusion of a measure assessing early age of onset of depression; and (3) assessment of neurobiological, genetic, environmental, and epigenetic predictors of early onset depression. Findings from 89 systematic reviews of moderate to high quality suggest that childhood-onset depressive disorders have neurobiological, genetic, environmental, and epigenetic roots consistent with a diathesis-stress theory of depression. This review identified key putative markers that may be targeted for personalized clinical decision-making and provide important insights concerning candidate mechanisms that might underpin the early onset of depression.

A Bayesian Approach to Correcting the Attenuation Bias of Regression Using Polygenic Risk Score

Polygenic risk score (PRS) has become increasingly popular for predicting the value of complex traits. In many settings, PRS is used as a covariate in regression analysis to study the association between different phenotypes. However, measurement error in PRS causes attenuation bias in the estimation of regression coefficients. In this paper, we employ a Bayesian approach to accounting for the measurement error of PRS and correcting the attenuation bias in linear and logistic regression. Through simulation, we show that our approach is able to obtain approximately unbiased estimation of coefficients and credible intervals with correct coverage probability. We also empirically compare our Bayesian measurement error model to the conventional regression model by analyzing real traits in the UK Biobank. The results demonstrate the effectiveness of our approach as it significantly reduces the error in coefficient estimates.