Association of Genetic Risk for Schizophrenia With Nonparticipation Over Time in a Population-Based Cohort Study

Abstract

Progress has recently been made in understanding the genetic basis of schizophrenia and other psychiatric disorders. Longitudinal studies are complicated by participant dropout, which could be related to the presence of psychiatric problems and associated genetic risk. We tested whether common genetic variants implicated in schizophrenia were associated with study nonparticipation among 7,867 children and 7,850 mothers from the Avon Longitudinal Study of Parents and Children (ALSPAC; 1991-2007), a longitudinal population cohort study. Higher polygenic risk scores for schizophrenia were consistently associated with noncompletion of questionnaires by study mothers and children and nonattendance at data collection throughout childhood and adolescence (ages 1-15 years). These associations persisted after adjustment for other potential correlates of nonparticipation. Results suggest that persons at higher genetic risk for schizophrenia are likely to be underrepresented in cohort studies, which will underestimate risk of this and related psychiatric, cognitive, and behavioral phenotypes in the population. Statistical power to detect associations with these phenotypes will be reduced, while analyses of schizophrenia-related phenotypes as outcomes may be biased by the nonrandom missingness of these phenotypes, even if multiple imputation is used. Similarly, in complete-case analyses, collider bias may affect associations between genetic risk and other factors associated with missingness.

Keywords: Avon Longitudinal Study of Parents and Children; attrition bias; cohort studies; genetic risk; longitudinal studies; schizophrenia; study nonparticipation.

Figure 1.

Data availability in a sample of mothers and children with genetic data from the Avon Longitudinal Study of Parents and Children, 1991–2007. Black indicates persons who participated in data collection at a given time point; gray indicates persons with missing data.

Figure 2.

Polygenic risk scores (mean z score) for schizophrenia among children from the Avon Longitudinal Study of Parents and Children (1991–2007), depending on data availability. White bars represent persons with data available; gray bars represent persons with missing data. Error bars show standard errors.

Figure 3.

Polygenic risk scores (mean z score) for schizophrenia among mothers from the Avon Longitudinal Study of Parents and Children (1991–2007), depending on data availability. White bars represent persons with data available; gray bars represent persons with missing data. Error bars show standard errors.

Figure 4.

Polygenic risk scores (mean z score) for schizophrenia among children from the Avon Longitudinal Study of Parents and Children (1991–2007), according to the number of missing data points. The x-axis displays categories of persons with no missing data, missing data for only 1 time point, missing data for 1–2 time points, missing data for 1–3 time points, and so on. Error bars show standard errors.

Figure 5.

Polygenic risk scores (mean z score) for schizophrenia among mothers from the Avon Longitudinal Study of Parents and Children (1991–2007), according to the number of missing data points. The x-axis displays categories of persons with no missing data, missing data for only 1 time point, missing data for 1–2 time points, missing data for 1–3 time points, and so on. Error bars show standard errors.