Agreement between a brief food frequency questionnaire and diet records using two statistical methods

Our plain-English summary of Agreement between a brief food frequency questionnaire and diet records using two statistical methods, by G.L. Ambrosini, N.H. de Klerk, A.W. Musk, et al (Public Health Nutrition, 2001).

The study

• Authors: G.L. Ambrosini, N.H. de Klerk, A.W. Musk, D. Mackerras
• Journal: Public Health Nutrition
• Published: 2001;4(2):255-64
• Type of article: Randomized trial
• Cited by: 25 articles, as counted by NIH iCite
• Read it: PubMed · Publisher

Abstract

Objective: To compare intra- and inter-method reliability of a semi-quantitative food frequency questionnaire (FFQ) designed specifically to measure beta carotene (BC) and retinol intake, using two methods - the limits of agreement (LOA) and the correlation coefficient.

Design: A cross-sectional study of dietary intake.

Setting: A randomized trial of vitamin A supplements in 2769 subjects with past asbestos exposure.

Subjects: Data from 57 men and 26 women, aged 28-72 years, living in Western Australia.

Methods: The FFQ was administered at baseline (FFQ1) and repeated 1 year later (FFQ2). Four 1-week diet records (DRs) were completed during the year.

Results: Mean agreement between FFQ2 and FFQ1 was 120% for BC and 98% for retinol. LOA were 47-306% and 21-459%, respectively. Mean agreement between FFQ2 and the DR was 149% for BC and 63% for retinol; LOA were 50-447% and 11-349%, respectively. Mean agreement and LOA varied across energy intakes. Between the DR and FFQ2, correlation coefficients were 0.36 for BC and 0.51 for retinol. These varied considerably across age, gender and energy intakes and were not in accordance with limits of agreement findings.

Conclusion: Although correlation coefficients were positive and significant, there was less than ideal intra-method and inter-method reliability shown by the limits of agreement method. Bias was uneven across the range of intakes, the LOA were wide and, compared with the DR, the FFQ significantly over-estimated BC and under-estimated retinol. This shows the limitations of calculating correlation coefficients alone, for assessing reliability and validity.

In plain English

This study tested how well a short food questionnaire measured intake of beta carotene and retinol (pre-formed vitamin A). It was run inside the Vitamin A Program, a supplement trial of 2769 people from two cohorts of former workers and residents of an asbestos mining and milling town in Western Australia. The study did not examine asbestos-related disease. Of 570 people joining the trial during one period, 118 agreed to take part. Of these, 83 (57 men, 26 women) completed four 1-week diet records, and 76 also completed the questionnaire twice, one year apart. On average, the repeat questionnaire gave 120% of the first beta carotene estimate and 98% of the first retinol estimate. For retinol, though, the limits expected to contain 95% of individual differences ran from 21% to 459% of the first estimate. Compared with the diet records, the questionnaire on average gave 149% of beta carotene intake and 63% of retinol intake. Correlations with the diet records were 0.36 for beta carotene and 0.51 for retinol. The authors concluded that agreement was less than ideal and that correlation coefficients alone have limitations for judging reliability and validity.

We wrote this plain-English summary from the full article with the help of AI, and a second AI model checked each sentence against the article. It is not the authors' wording; the abstract and the article are their own account.

What the study did and found

• Design: Cross-sectional reliability study of dietary intake methods, nested within a randomized trial of vitamin A supplements
• Who was studied: The study drew on people entering the Vitamin A Program trial (n = 2769), who came from two cohorts of former workers and residents of an asbestos mining and milling town in Western Australia. Of 570 people enrolling during the sixth-month recruitment period, 118 (21%) agreed to take part. 83 (57 men, 26 women) completed four diet records, and 76 (55 men, 21 women) completed both questionnaires and the diet records.
• Exposure: The quantity measured was dietary beta carotene and retinol intake. It was assessed by a brief semi-quantitative food frequency questionnaire given at baseline and 1 year later, and compared with four 1-week diet records. Past asbestos exposure was the trial's background; no fiber type is named in the article.
• Main results:
  • Repeat questionnaire vs first questionnaire (76 subjects): mean agreement was 120% for beta carotene and 98% for retinol. The limits of agreement were 47 to 306% for beta carotene and 21 to 459% for retinol. The repeat questionnaire significantly over-estimated beta carotene compared with the first.
  • Repeat questionnaire vs the four diet records (76 subjects): mean agreement was 149% for beta carotene and 63% for retinol. The limits of agreement were 50 to 447% for beta carotene and 11 to 349% for retinol. The questionnaire gave significantly higher beta carotene and significantly lower retinol than the diet records.
  • Individual differences between the questionnaire and the diet records varied up to almost seven-fold for beta carotene and six-fold for retinol.
  • The differences between the questionnaire and the diet records depended on intake level. The dependence was pronounced for retinol and weaker for beta carotene.
  • Pearson correlations between the diet records and the repeat questionnaire were 0.36 for beta carotene and 0.51 for retinol. Across subgroups, the correlations did not track the limits-of-agreement findings.
• The authors' conclusion: The authors concluded that, although the correlation coefficients were positive and significant, the limits of agreement method showed less than ideal reliability, both for the repeated questionnaire and against the diet records. They said this shows the limitations of relying on correlation coefficients alone.
• Limitations the authors state:
  • The participation rate was low (21%).
  • Keeping diet records between the two questionnaires may have affected the repeat questionnaire's reliability.
  • Entering the trial, with smoking and dietary advice, may have led participants to change their diets.
  • The questionnaire did not measure total energy intake, so intakes could not be adjusted for energy.
  • The questionnaire contained few retinol food sources, which may have led to under-reporting of retinol.
  • Supplement intake reliability was not examined, because participants were told to stop non-trial vitamin A supplements.

Topics

Medical terms in this study, explained in our Lexicon: Asbestos Exposure.

Related studies

• Comparison of an Australian food-frequency questionnaire with diet records: implications for nutrition surveillance — G.L. Ambrosini, D. Mackerras, et al. (2003)
• No dose-dependent increase in fracture risk after long-term exposure to high doses of retinol or beta-carotene — G.L. Ambrosini, A.P. Bremner, et al. (2013)
• Vitamin A and cancer prevention II: comparison of the effects of retinol and beta-carotene — N.H. de Klerk, A.W. Musk, et al. (1998)
• The protective effect of beta-carotene and retinol on ventilatory function in an asbestos-exposed cohort — P. Chuwers, S. Barnhart, et al. (1997)
• Risk factors for lung cancer and for intervention effects in CARET, the Beta-Carotene and Retinol Efficacy Trial — G.S. Omenn, G.E. Goodman, et al. (1996)
• Nonenzymatic β-Carotene Degradation in Provitamin A-Biofortified Crop Plants — P. Schaub, F. Wüst, et al. (2017)

Browse return to medical research.