UK Biobank, big data, and the consequences of non-representativeness

• Published in: The Lancet (British edition) 2019-03, Vol.393 (10178), p.1297-1297
• Main Authors: Keyes, Katherine M, Westreich, Daniel
• Format: Article
• Language: English
• Subjects: Big Data - supply & distribution; Biological Specimen Banks - organization & administration; Biological Specimen Banks - statistics & numerical data; Data management; Environmental Exposure - statistics & numerical data; Health Information Systems - supply & distribution; Health Resources - supply & distribution; Humans; Population studies; United Kingdom - epidemiology; Validity; Volunteers - statistics & numerical data
• ISSN: 0140-6736; 1474-547X
• DOI: 10.1016/S0140-6736(18)33067-8

The sampling population is volunteer-based and is not representative of the UK population.1 Investigators state that although the estimates of prevalence and incidence should be interpreted with caution, valid measures of association and estimates of causal effect can be more readily interpreted as they do “not require participants to be representative of the population at large”.2 This statement is a puzzling claim: sample selection can indeed influence measures of association. Specifically, whether or not an association observed in a study is similar in some other target population (ie, has external validity) depends on a number of factors, including the distribution of effect measure modifiers of the exposure–outcome relationship in the study sample and target population.3 Critically, a study can have restricted external validity even when it has internal validity, which might occur in a randomised trial.4 Thus, researchers should not be quick to set aside issues of representativeness in interpreting UK Biobank results. [...]larger sample size in a skewed sample only leads to confidence in answers that might not apply to the target population. [...]it is paramount that external validity be taken more seriously in the UK Biobank and other large data resources.