Loss to Follow-up in Cohort Studies

• DOI: 10.1097/ede.0b013e31827623b1
• Publication Date: 2012-12-04T19:12:47Z
• Authors: Laura D. Howe; Kate Tilling; Bruna Galobardes; Debbie A. Lawlor
• Keywords: Adolescent; Lost to follow-up; 333; socioeconomic status; Cohort Studies; 03 medical and health sciences; 0302 clinical medicine; Pregnancy; Risk Factors; cohort study; Birth Weight; Humans; Obesity; Child; 10. No inequality; Selection Bias; Models, Statistical; Infant, Newborn; Health Status Disparities; 16. Peace & justice; Breast Feeding; England; Data Interpretation, Statistical; Multivariate Analysis; Educational Status; Premature Birth; Female; Lost to Follow-Up

Background: Although cohort members tend to be healthy and affluent compared with the whole population, some studies indicate this does not bias certain exposure-outcome associations. It is less clear whether holds when socioeconomic position (SEP) exposure of interest. Methods: As an illustrative example, we use data from Avon Longitudinal Study Parents Children. We calculate estimates maternal education inequalities in outcomes for which are available on almost (birth weight length, breastfeeding, preterm birth, obesity, smoking during pregnancy, educational attainment). These calculated full (n~12,000) restricted subsamples defined by continued participation at age 10 years (n∼7,000) 15 (n∼5,000). Results: Loss follow-up was related both SEP outcomes. For each outcome, loss associated underestimation inequality, increased as rates decreased (eg, mean birth-weight difference between highest lowest 116 g [95% confidence interval = 78 153] sample 93 [45 141] 62 [5 119] those attending ages years, respectively). Conclusions: Considerable attrition may result biased inequalities, degree worsen decrease. However, even considerable (>50%), qualitative conclusions about direction approximate magnitude did change among most our examples. The appropriate analysis approaches alleviate depend missingness mechanism.