Relative validity and reproducibility of a food frequency questionnaire used in pregnant women from a rural area of China

ObjectiveFood frequency questionnaires are relatively inexpensive, easy and quick to administer, but the construction of a food frequency questionnaire that can capture Chinese food habits is challenging given the diverse lifestyle and eating habits in different parts of the country. The aim of this study was to assess the validity and reproducibility of a self-administered food frequency questionnaire against a 3-day dietary recall in a rural region of western China. DesignProspective cohort study. SettingChinese maternal and child healthcare hospital. PopulationA total of 168 healthy pregnant women. MethodsPregnant women completed a food frequency questionnaire at 16-24weeks gestation, and again at 29-31weeks; during weeks 26-27 they completed a 3-day dietary recall. ResultsIn general, mean intake was higher when assessed with food frequency questionnaires compared with dietary recall. Spearman and intra-class correlation coefficients between the two food frequency questionnaires ranged from 0.31 to 0.69 and from 0.27 to 0.79, respectively. For the second food frequency questionnaire and the dietary recall, the crude and de-attenuated Spearman correlations ranged from 0.12 to 0.55 and 0.14 to 0.58, respectively. The correlation both between the two food frequency questionnaires and between the second food frequency questionnaire and the dietary recall decreased after adjustment for energy. Ranking women, 31-57% and 1-8% were classified into the same and the opposite quartile, respectively, by both food frequency questionnaires; 30-45% and 1-11% were classified into the same and the opposite quartile respectively for the second food frequency questionnaire and dietary recall. ConclusionThe food frequency questionnaire showed good reproducibility and correlations with dietary recall; it is useful for ranking study participants according to dietary intake, which is of great importance to future etiological studies in this cohort.