Validity and reliability of a semi-quantitative food frequency questionnaire in groups at high risk for cardiovascular diseases

Abstract

Abstract Background Diet is a modifiable risk factor for cardiovascular diseases (CVD), but there is still a lack of tools to assess dietary intakes of this high-risk population in Ningxia, China. Objective We aim to evaluate the validity and reliability of the semi-quantitative food frequency questionnaire (SFFQ) in the groups in Ningxia using a 24-hour dietary recall method. Method Two hundred five participants were included in the analysis. The two FFQs were 6 months apart, and during this time two 24-hour dietary recalls (24HDRs) were completed. Statistical methods were compared using inter-class correlation coefficient, unadjusted, energy-adjusted, de-attenuated correlation coefficient, quartile classification, weighted K values, and 95% limits of agreement (LOA). Results The inter-class correlation coefficients between FFQ1 and FFQ2 ranged from 0.25 to 0.73. The number of subjects classified as identical or adjacent was 72.2 to 85.9%. The crude correlation coefficient between FFQs and 24HDRs was 0.30 ~ 0.81, the energy-adjusted correlation coefficient was 0.16 ~ 0.83, and the de-attenuated correlation coefficient was 0.19 ~ 0.98. Weighted k statistics and Bland-Altman plots showed acceptable agreement between FFQs and 24HDRs. Conclusion The FFQ developed for the population at high risk of cardiovascular and cerebrovascular diseases in areas of Ningxia can be used to measure the dietary intake of nutrients and food groups reliably and validly.