Email administered web‐based dietary assessment from the UK Biobank captures chronotype‐specific reporting behaviors: implications for population‐based dietary sampling

The 24‐hr diet recall is traditionally administered in‐person and more recently electronically. Chronotype, a continuum from morningness to eveningness preference, impacts temporality of cognitive performance and timing of lifestyle behaviors. The purpose of this study was to compare responses to diet recalls administered in‐person (constrained method) and electronically (unconstrained method) of participants with different chronotypes in order to evaluate the need for respondent‐determined recall conditions. In the UK Biobank, dietary data were collected from 189,158 individuals aged 37–74 using the Oxford WebQ, a web‐based 24‐hr diet recall, administered (1) in‐person at assessment centers ( n =70,755), and (2) electronically by email ( n =157,737). From the recall data, we captured three response performance parameters: time‐of‐day (clock time of response) and day‐of‐week the response was submitted, response duration (in minutes), and delay in response to 24‐hr diet recall request (in days; for email administered recalls only). Chronotype was self‐reported using the question, “Do you consider yourself to be?” (definite‐morning person, more morning than evening, more evening than morning, and definite‐evening person). We observed a stronger association between chronotype and response times to the 24‐hr diet recalls administered by email. In multivariable linear regression models, we observed that compared to definite‐morning chronotypes, the more evening chronotype groups had a 45.4, 67.8, and 93.0 minutes later clock times to completing email administered recall, respectively ( p <2e‐16). This response pattern was greater on weekends compared to weekdays. In addition, response duration was longer for all chronotypes when assessment was administered by email compared to assessment center [M(SD) 19.3(32) vs. 14.2(6.7) mins; p =0.0001], particularly for more evening chronotypes and on weekdays. Lastly, delay in response to recall was also chronotype‐specific: same‐day responses to 24‐hr diet recall request was highest among definite‐morning chronotypes (62.8%). In multivariable logistic regression models, we observed that compared to definite‐morning chronotypes, the more evening chronotype groups had lower odds of same‐day responses [OR (95% CI) =0.92(0.89, 0.94), 0.87(0.84, 0.89), and 0.87(0.84, 0.91), respectively]. Interestingly, same‐day responses were also lower on weekends (53.2%) compared to weekdays (69.3%) for all chronotypes. Our results suggest that email administered web‐based dietary assessments allow for chronotype‐dependent optimal response conditions, including chronotype‐appropriate time‐of‐day responses and durations, which may be constrained by in‐person assessments. Targeting these conditions with respect to the administration of online recalls may be necessary to maximize response rates and reduce respondent‐burden. Yet it remains unclear whether dietary assessment at chronotype ‘inappropriate’ times‐of‐day may influence the reliability of self‐reported dietary data.