Reproducibility of the food frequency questionnaire in women

Abstract

The aim of this study, was to evaluate the reproducibility of portions and frequencies of a food frequency questionnaire (FFQ), previously validated. The study was conducted in a primary care center in Rio de Janeiro from July 2002 to May 2003. Sixty nine women answered the questionnaire twice with an interval of 30 to 60 days. The 78 item self-administered questionnaire had as time frame the previous year. Portions in the questionnaire had three categories; small, medium and large. For frequencies the eight categories ranged from never eaten to three or more times per day. Agreement was evaluated through the weighted kappa coefficient. The agreement for frequencies of food consum was better than the agreement for portions consum. None of the frequencies or portions had kappa <0.20 (poor agreement). Frequencies for tomato (k=0.40), beef (k=0.29) and chicken (k=0.28) showed the lowest agreement and pear (k=0.82), eggs (k=0.85), milk (k=0.93) and sodas (k=0.95) showed the greatest agreements. For portions about half of the items showed moderate agreement. Conclusion The FFQ presented good reproducibility, with greater agreement for frequency than for portions.