Validity and reproducibility of a food frequency questionnaire designed for residents in north China.

Xiaoqin Wang, Rina Sa, Hong YanPubMed 2008

TL;DR

Researchers have developed and validated a new food frequency questionnaire (FFQ) specifically for residents of Xi'an, China, to better understand dietary links to colorectal cancer. The FFQ demonstrates reasonable validity and reproducibility, suggesting it's a useful tool for dietary assessment in this population, with potential improvements from including supplements and seasonings.

What Problem Does This Paper Address?

Investigating the complex relationship between diet and chronic diseases like colorectal cancer requires accurate and reliable methods for assessing dietary intake. Traditional methods, like 24-hour dietary recalls, can be resource-intensive and may not capture long-term dietary habits. Food frequency questionnaires (FFQs) offer a more practical approach for large-scale studies by assessing habitual food consumption over a longer period. However, existing FFQs are often designed for Western populations and may not accurately reflect the unique dietary patterns of other regions, such as northern China. This study addresses this gap by developing and validating a semi-quantitative FFQ tailored for residents of Xi'an, China, to be used in an international case-control study on colorectal cancer.

Methods

The study employed a two-pronged approach to evaluate the newly developed Xi'an FFQ:

  • Validation: To assess how well the FFQ measures actual nutrient intake, it was compared against a "gold standard" method – 24-hour dietary recalls (24-HDRs). Participants completed the FFQ twice over a one-year period, and four 24-HDRs were conducted at different seasons within that year. Nutrient intakes from the second FFQ (FFQ-2) were compared with the average of the four 24-HDRs.
  • Reproducibility: To determine if the FFQ consistently yields similar results over time, the nutrient intakes reported in the first FFQ (FFQ-1) were compared with those from the second FFQ (FFQ-2), administered one year apart.

Statistical analyses included calculating Pearson correlation coefficients to measure the association between nutrient intakes from different methods. To account for potential overestimation due to within-person variation in the 24-HDRs, a deattenuation formula was applied. Nutrient intakes were also categorized into quartiles to assess agreement between methods, providing a measure of how well the FFQ ranks individuals by their intake levels. Energy adjustment was also performed to assess the impact of overall energy consumption on nutrient correlations.

Datasets

The study utilized data collected from 125 participants residing in Xi'an City, Shaanxi province, China. These participants were aged 50-79 years and were recruited from the Municipal Resident Registry Office. The data collected included:

  • Two administrations of the Xi'an Food Frequency Questionnaire (FFQ-1 and FFQ-2) over a one-year interval.
  • Four 24-hour dietary recalls (24-HDRs) conducted seasonally.

No new benchmarks were introduced; the study focused on validating an existing FFQ design against established dietary assessment methods.

Key Insights and Ideas

  • Cultural Adaptation is Crucial: The study highlights that simply adapting existing FFQs isn't enough. The Xi'an FFQ was specifically modified from the Shanghai FFQ by removing certain food items and adding local ones to better reflect the distinct dietary habits of northwest China (e.g., flour-based staples vs. rice). This underscores the importance of culturally relevant dietary assessment tools.
  • Supplements and Seasonings Matter: Including questions about nutritional supplements and common seasonings (like cooking oil, vinegar, soy sauce) significantly improved the correlation coefficients between the FFQ and 24-HDRs. This suggests that neglecting these components can lead to an incomplete picture of dietary intake and potentially weaken the validity of the FFQ.
  • Ranking is Key for Epidemiological Studies: While absolute nutrient values might differ between methods, the ability of an FFQ to correctly rank individuals into intake categories (like quartiles) is critical for epidemiological studies investigating diet-disease associations. The Xi'an FFQ showed reasonable agreement in quartile classification, suggesting it can effectively identify individuals with different levels of nutrient intake.

Key Results

  • Validation: The mean deattenuated correlation coefficients between the 24-HDRs and FFQ-2 ranged from 0.35 to 0.84 for males and were slightly lower for females. For quartile agreement, approximately 79% of males and 72% of females were classified into the same or adjacent quartiles.
  • Reproducibility: The mean crude correlation coefficients between FFQ-1 and FFQ-2 were between 0.41-0.68 for males and 0.36-0.66 for females. The mean agreement in the same quartile between the two FFQs was 36% for both males and females.
  • Nutrient Intakes: Nutrient intakes assessed by FFQ-2 were generally lower than those assessed by FFQ-1.
  • Impact of Supplements/Seasonings: Including seasonings and supplements increased the mean crude correlation coefficient by 14% in males and 16% in females when comparing FFQ and 24-HDR estimates.

Limitations and Future Work

  • Dietary Fiber Correlation: The correlation for dietary fiber was notably lower (less than 0.4) for both males and females. The authors suggest this might be due to difficulties in estimating fiber intake using FFQs or other reference methods, as well as high within-person variation in fiber intake. This could potentially attenuate risk estimations in epidemiological studies.
  • Measurement Bias: The paper acknowledges that the actual source of measurement or subject bias in the FFQ is not yet known.
  • Future Work: The authors suggest that the Xi'an FFQ would benefit from further refinement, particularly in the inclusion of nutritional supplements and seasonings, to enhance its accuracy for assessing dietary intake in older persons in Xi'an. They also propose that this tool could be validated and applied to other Northern Chinese populations.

Summary updated

Abstract

The objective of this study was to evaluate the validity of a Xi'an food frequency questionnaire (FFQ) designed for application in an international case-control colorectal cancer study. The FFQ was administered to 125 participants twice over 1-year interval. Four 24-hour dietary recalls (24-HDRs) were conducted with the same participants in each season of the year. Comparative validation was assessed by comparing the nutrient intakes derived from the 24-HDRs and the FFQ-2, and reproducibility was estimated by comparing the nutrient intakes from two FFQs. In the validation study, the mean deattenuated correlation coefficients for nutrients between the 24-HDRs and the FFQ-2 ranged from 0.35 to 0.84 in the males. The female results were slightly lower than the male's. The mean percentage of classification into the same quartile was 38% for the males and 35% for the females. In the reproducibility study, the mean crude correlation coefficients between the two FFQs were from 0.41 to 0.68 in the males and from 0.36 to 0.66 in the females. The newly developed Xi'an FFQ appears to be reasonably valid and reliable for most nutrients but would benefit from the addition of nutritional supplements and seasonings for assessing dietary intake in older persons in Xi'an, China.
BibTeX
@inproceedings{wang2008validity,
  author = {Xiaoqin Wang and Rina Sa and Hong Yan},
  title = {{Validity and reproducibility of a food frequency questionnaire designed for residents in north China.}},
  booktitle = {PubMed},
  year = {2008},
}
Figure