Validation of the second version of a quantitative food-frequency questionnaire for use in Western Mali.

Objective: To assess the relative validity of the second version of a quantitative food-frequency questionnaire (QFFQ), designed to measure the habitual food and nutrient intake in one season in rural populations in Western Mali, West Africa.

Design: The dietary intake during the previous week was assessed with the 164-item QFFQ administered by interview. This was compared with the intake from a 2-day weighed record (WR) with weighed recipes.

Setting: The village of Ouassala in the Kayes region, Western Mali.

Subjects: Thirty-four women and 36 men aged 15-45 years, from 29 households.

Results: The QFFQ gave a lower intake of lunch and dinner and a higher intake of snacks than the WR. The discrepancies were larger for women than for men. The median proportion of subjects classified in the same quartile of intake was 29% for food groups and 36% for energy and nutrients. For classification into extreme opposite quartiles, the median proportion was 6% for food groups and 7% for energy and nutrients. Spearman's rank correlation for energy and nutrients ranged from 0.16 (% energy from protein) to 0.62 (retinol equivalents).

Conclusions: The second version of the QFFQ tends to underestimate total food weight. The methods used for estimating food portion size should therefore be applied with caution. The changes made from the first version had little effect. The ability to rank subjects according to dietary intake is similar with both versions. The improved layout of the new QFFQ makes it a more user-friendly tool for comparing dietary intake between population groups and for measuring changes over time.