*Geek Box: Dietary Assessment in the Nurses’ Health Study
When it comes to dietary assessment in nutritional epidemiology, we know that validation process is integral to the generation of reliable data. The validation process in the Nurses Health Study [NHS] is an example of particular rigour in this regard.
Starting in 1980, an initial food-frequency questionnaire was administered, and over the course of the following year a total of four separate 1-week diet records, with foods and drinks measured, were completed by a representative subgroup in the cohort. Doing 4, spread across seasons, allowed for any potential seasonal variations in dietary intake to be captured.
The FFQ was administered then for a second time in 1981, providing the investigators with a second measure to compare the performance of the FFQ in measure diet against the data from the 4-weeks of dietary records. In 1984, the FFQ was revised, and in 1986-1987, a further two 1-week dietary records were completed to compare the reproducibility, i.e., the consistency of the FFQ across repeated measures in the same person. The FFQ was further revised based on this updated data.
This process resulted in strengthened correlation coefficients for major nutrients of interest, highlighting that rigorous validation processes improve the reliability of dietary data in epidemiology. The NHS validation process, due to its rigour, has been repeated in other cohorts, notably the EPIC-Oxford cohort in the UK.