It is widely acknowledged that selection bias will distort prevalence estimates. This can be clearly seen in differences between participants at baseline and at subsequent assessments in cohort studies, such as when we compare the original ALSPAC sample with those who attended later clinics (see Box 1). It can also be seen in differences between an actual study sample and the source population from which it is drawn (i.e., the intended study population); for example, the UK Biobank study differs relative to the general population in the UK (see Box 2 UK BiobankThe UK Biobank is a cross-sectional study which recruited over 500 000 individuals aged between 40 and 69 years between 2006 and 2010 [http://www.ukbiobank.ac.uk/]. Individuals in this age group living within a 25-mile radius of any of the 22 assessment centres across the UK were identified from NHS patient registers.37 In total, around 9 million individuals were invited to participate. However, UK Biobank was only able to achieve a 5% response rate (∼500 000 participants recruited from ∼9 000 000 invited, personal communication, UK Biobank, 8 July 2016),
