Loss and representativeness in a biomedical survey at age 45 years: 1958 British birth cohort

Abstract

Objective: A biomedical survey of the 1958 British birth cohort at age 45 years provides a baseline for future studies of chronic disease. The extent and nature of bias in this sample was examined. Methods: Follow-up of all births in Great Britain in one week in March 1958. At 45 years the sample was compared with the surviving cohort on characteristics recorded at birth and seven years, and in adulthood (42 years). Results: Sample attrition to age 45 years was chiefly through avoidable (35.8%) than unavoidable loss through death or emigration (13.7%). 11 971 individuals were invited to participate at 45 years. Of 9377 participants (78.3%), most consented to, and had valid values for, physical and mental measurements, survey questionnaires, and blood and saliva sampling; 8302 (88.5%) provided a blood sample. Groups moderately underrepresented in the 45-year sample included those with externalising or internalising behaviours, poor reading or maths scores, and shorter stature. For example, 8.8% of the 45-year sample had been poor readers at age seven years compared with 11.1% of the total surviving cohort; for shorter stature the figures were 7.2% versus 8.4%, respectively. There was also underrepresentation of some minority groups (non-whites, births in households with no male head and children in social care). Most bias was present before the 45-year survey. Conclusion: The 45-year sample remains broadly representative of the surviving cohort, but specific biases may need to be taken into account in future research. Renewed efforts to re-engage all cohort members will improve the representativeness and value of the study.