Image-assisted dietary assessment Evaluating the potential of wearable cameras to enhance self - report in the 24-hour dietary recall method

Abstract

Background Traditional methods of dietary assessment are prone to self-report bias. Images captured by wearable cameras may reduce self-report bias for foods and dietary energy intake (EI). Aims To investigate the use of wearable cameras to (1) reduce the reporting bias associated with traditional self-reported dietary assessment, and (2) passively record and assess contexts of dietary behaviours. Methods Five modules of research were undertaken: (1) a secondary analysis of the 2008/09 New Zealand Adult Nutrition Survey (ANS 08/9) estimated the prevalence of low energy reporters (LERs), (2) a systematic review examined evidence for image-assisted methods of dietary assessment, (3) a feasibility study explored the use of wearable cameras to enhance self-report in the 24-hour dietary recall, (4) a doubly labelled water (DLW) study validated a wearable camera imageassisted 24-hour dietary recall, and (5) secondary analysis of images collected during the validation study explored the utility of wearable cameras to objectively record, and reliably assess, environmental and social contexts of eating episodes in free-living settings. Results The primary findings were: (1) 21% of New Zealand men and 25% of women were classified as LERs in the ANS08/9, and a systematic bias was observed with LERs more prevalent amongst women, people aged >65 years, and Maori and Pacific peoples, (2) literature published up to 2013 suggests images can provide objective information to independently verify and assess selfreported dietary intake but the limited existing evidence highlighted the need for further research, (3) a small study (n=10) using wearable cameras revealed unreported or misreported errors in the 24-hour dietary recall, which increased self-reported dietary EI, (4) a DLW study (n=40) showed that wearable cameras reduce reporting bias for dietary EI in 24-hour dietary recalls by 9% in men (from 17% to 9%) and 6% in women (from 13% to 7%), and (5) wearable cameras images can be analysed to objectively and reliably assess important contexts of dietary behaviours such as eating location, physical position, social interaction, and media screens. Conclusion Wearable cameras significantly reduce the reporting bias for dietary EI in the 24-hour dietary recall. Used in nutrition research, wearable cameras provide a new tool to verify and enhance self-reported dietary intake, and compared to self-report alone, allow additional information on dietary behaviours to be objectively assessed.