Comparison between Self-Completed and Interviewer-Administered 24-Hour Dietary Recalls in Cancer Survivors: Sampling Bias and Differential Reporting

Abstract

Self-completed 24 h dietary recalls (24-HRs) are increasingly used for research and national dietary surveillance. It is unclear how difficulties with self-completion affect response rates and sample characteristics. This study identified factors associated with being unable to self-complete an online 24-HR but willing to do so with an interviewer. Baseline 24-HRs from the ASCOT Trial were analysed (n = 1224). Adults who had been diagnosed with cancer in the past seven years and completed treatment, were invited to self-complete 24-HRs online using myfood24®. Non-completers were offered an interviewer-administered 24-HR. One third of participants willing to provide dietary data, were unable to self-complete a 24-HR. This was associated with being older, non-white and not educated to degree level. Compared to interviewer-administered 24-HRs, self-completed 24-HRs included 25% fewer items and reported lower intakes of energy, fat, saturated fat and sugar. This study highlights how collection of dietary data via online self-completed 24-HRs, without the provision of an alternative method, contributes to sampling bias. As dietary surveys are used for service and policy planning it is essential to widen inclusion. Optimisation of 24-HR tools might increase usability but interviewer-administered 24-HRs may be the only suitable option for some individuals.