Comparison of dietary assessment methods in a southern French population: use of weighed records, estimated-diet records and a food-frequency questionnaire.

OBJECTIVE

The main objective of the study was to develop appropriate dietary assessment instruments for the French Mediterranean region and to validate the measurements they provide.

SUBJECTS AND METHODS

Three different assessment methods were submitted to a sample of 150 male and female volunteers. 98 completed the protocol, which consisted of a 4 d weighed dietary record (PETRA) and a 7 d estimated-diet record (S7) based on a check list and a set of photographs, both these records being completed once in each season of the year, and a semi-quantitative (standard portion) food-frequency questionnaire (FFQ) including questions eliciting socio-demographic and anthropometric data, which was completed once only. The days when PETRA was used to evaluate food consumption coincided with the first 4 d of S7 (S4).

RESULTS

Validation was based on nutrients and foods. Energy-adjusted Pearson correlation coefficients between S4 and PETRA ranged from 0.32 for vitamin E to 0.81 for vitamin C (mean: 0.65 for 21 nutrients). There was practically no misclassification in opposite extreme quartiles. Spearman correlation coefficients ranged from 0.63 for fish and sea-food to 0.90 for wine (mean: 0.76 for 16 food groups). There was practically no misclassification in opposite extreme quartiles. De-attenuated energy-adjusted Pearson correlation coefficients between FFQ and S7 ranged from 0.22 for proteins and monounsaturated fatty acids to 0.80 for iron (mean: 0.45). 10% or less of subjects were misclassified in opposite extreme quartiles (except for vitamin C, 12%). Spearman correlation coefficients ranged from 0.25 for green-yellow-red raw vegetables to 0.76 for wine (mean: 0.42). 8% or less of subjects were misclassified in opposite extreme quartiles (except for citrus fruit, 11%).

CONCLUSIONS

Portion estimation using the set of photographs was validated by the correlation between S4 and PETRA for both nutrients and foods. The FFQ provides a reasonably reliable measure of macronutrient intake and a good measure of micronutrient intake when compared with the data in the literature. It performs less well for food intake. Better results can be achieved for FFQ: (i) by using the set of photographs instead of standard portions and (ii) by adding further questions on foods which are insufficiently covered.