Factors associated with participation over time in the Avon Longitudinal Study of Parents and Children: a study using linked education and primary care data

Abstract

Background Dropout from studies can lead to biased exposure-outcome estimates if the outcome is associated with continued participation, but this cannot be investigated using incomplete data. Linkage to external datasets provides a means of obtaining outcome - or proxy outcome - data on non-responders. Methods We examined the association between baseline socio-demographic factors and participation in the Avon Longitudinal Study of Parents and Children. We investigated whether child and adolescent outcomes measured in linked education and primary care data were associated with participation after accounting for baseline factors. To demonstrate the potential for bias, we examined whether the association between maternal smoking and these outcomes differed in the subsample who completed the 19-year questionnaire. Results Lower levels of school attainment, lower GP consultation and prescription rates, higher BMI, special educational needs (SEN) status, not having an asthma diagnosis, depression and being a smoker were all associated with lower participation after adjustment for baseline factors. For example, adjusted odds ratio (OR) for participation comparing ever smokers (by 18 years) to non-smokers: 0.65, 95% CI (0.56, 0.75). The association of maternal smoking differed between the subsample of participants at 19 years and the entire sample, although differences were small and confidence intervals overlapped. For example: for SEN status OR=1.19 (1.06, 1.33) (all participants); OR=1.03 (0.79, 1.45) (subsample). Conclusions Linkage to routine data provides a unique opportunity to compare continuing participators to those who drop out, and the impact this self-selection can have on results. Cohort studies should use linkage to routine data to explore participation and conduct sensitivity analyses.

Bias from questionnaire invitation and response in COVID-19 research: an example using ALSPAC

“…Hereafter we will refer to these as "candidate predictors of selection", acknowledging that we are not studying all possible sources of selection bias in ALSPAC and its COVID-19 data. We selected those variables a priori based on our knowledge of ALSPAC, COVID-19 and factors that are hypothesised to, or are known to, shape patterns of invitation and response 14,16,21,30 . For analyses involving continuous variables, these were transformed into standard deviation (SD) units in order for all odds ratios to be interpretable on the same scale.…”

“…This can result in selection bias [7][8][9][10][11] , which may be exacerbated by the ways in which data collection has changed during the pandemic, for example, with the increased use of online questionnaires 12,13 . Importantly, having an available and rich collection of pre-pandemic data in existing longitudinal cohorts can be useful in efforts to explore potential selection pressures that lead to bias [14][15][16][17] .…”