Eck LH, Klesges LM, Klesges RC. Precision and estimated accuracy of two short-term food frequency questionnaires compared with recalls and records. J Clin Epidemiol 49, 1195-1200
ABSTRACT Two widely used food frequency questionnaires (Block FFQ, Willett FFQ) were modified to reflect intake over the past 7 days and compared to intake information gathered from diet records and 24-hr recalls covering the same 7-day period. The Block FFQ and the Willett FFQ were also gathered at the beginning of the assessment period to reflect the 7-day period of time before records and recalls were gathered. Fifty-six subjects were assigned to either recording diet for 3 days, recording diet for 6 days, or providing three 24-hr recalls. Results indicate similar levels of within-method test-retest reliabilities for 3-day RECORDS and 6-day RECORDS, and within the two Block FFQs and within the two Willett FFQs from each subject, while lower reliabilities were seen in 24-hr RECALLS. When the FFQs were compared to the 6-day RECORDS with between-method agreement coefficients, there was a moderate level of agreement, with most values between 0.5 and 0.8 for both FFQs. Significant differences between mean levels of nutrients estimated by the three methods indicated differences only in the estimates of carbohydrate and vitamin A. The use of FFQs to gather short-term intake information is discussed.
- SourceAvailable from: jssm.org
[Show abstract] [Hide abstract]
- "ICCs were chosen in addition to correlations to allow for evaluation of how interchangeable the LMU RAM is with 3DRs. ICCs, whose use is supported by others, allow for assessment of equivalency, unlike the Pearson correlations which simply evaluate proportionality (Eck et al.,1996; Ward et al., 2004). "
ABSTRACT: When conducting research in the area of bone health, accurate measurement of calcium intake is crucial. The rapid assessment method (RAM) is one technique that has frequently been used for its measurement of calcium intake. However, the RAM and other currently established questionnaires lack the assessment of dietary supplement use, which is common for athletes. Our objective was to evaluate the validity of a RAM questionnaire designed to assess daily calcium consumption which was further modified to meet the needs of athletes who frequently consume dietary supplements. Usefulness of the modified RAM for athletes and non-athletes was evaluated as well as utility among those who do and do not use supplements. The 47 volunteers (n = 31 women, 16 men) were between the ages of 18 and 25 including, 33 athletes and 14 controls. The population also contained 23 supplement users and 24 non-supplement users. Participants completed the modified RAM and were instructed to complete a three-day diet record (3DR), logging food intake for 2 weekdays and 1 weekend day. The data collected via the modified RAM was compared with the 3DR. Mean calcium intake was 935mg ± 420mg and 1085mg ± 573mg, for the modified RAM and 3DR respectively. A strong positive correlation (r) was found between calcium intake measured with the modified RAM and 3DRs (r(45) = 0.854, p < 0.01). Intraclass correlation coefficients (ICC) revealed that agreement between the two instruments was good (ICC = 0.76, df = 45, p < 0.01) and much improved when compared to agreements without consideration of supplements (ICC = 0.05, df = 21, p > 0.05). We have found the modified RAM to be a valid tool which can be used to estimate calcium intake in the athletes and controls we strive to study. The accuracy of this instrument improved by including assessment of dietary supplement sources of calcium. Key pointsWhen conducting research on bone health, accurate measurement of calcium intake is crucial. The rapid assessment method (RAM) is one technique that has frequently been used for its measurement; however, currently established questionnaires lack assessment of dietary supplement use, which is common for athletes.We report that estimated calcium intake from the LMU RAM modified to evaluate supplement use has good agreement with three-day diet records (3DRs). There was a strong correlation between the two methods with about 69% (r = 0.83, r(2) = 0.69) of the variability in calcium intake quantified via the LMU RAM being accounted for by the 3DR.Calculated intraclass correlation coefficients between 0.63 and 0.77 reveal that the LMU RAM appears to be a valid tool of measuring daily calcium intake in athletes and non-athletes and among those who do and do not use supplements.When evaluating calcium intake without considering supplements, agreement (ICC) and correlation (r) values decreased considerably.We found the LMU RAM to be a valid measurement of calcium intake in athletes and controls. Without the addition of a section on supplement use, estimated calcium intake would have decreased an average of 32%.Journal of sports science & medicine 01/2009; 8(2):225-9. · 0.90 Impact Factor
[Show abstract] [Hide abstract]
- "There was little difference in all measures of agreement for the two FFQs, except perhaps for the slightly better correlations observed with the ®rst FFQ. Other investigators (Willett et al, 1985; Pietinen et al, 1988; Salvini et al, 1989; Eck et al, 1996) have found better agreement with the second rather than ®rst FFQ administration. In the present study, differences between correlations for the ®rst and second FFQ were small, suggesting any bias was probably not great. "
ABSTRACT: To assess the ability of a 16-item food frequency questionnaire (FFQ) to measure consumption of cereal foods, fruit and vegetables in Chinese Singaporeans. Subjects completed the questionnaire twice, at the beginning and end of a six-week period during which they also provided three 24 h diet recalls. Estimates of intake from the questionnaire were compared with those from diet recalls. Subjects were recruited from a range of occupational groups through random sampling across divisions in the headquarters of the Singapore Ministry of Health. Of the 81 subjects initially recruited, three failed to complete the diet recalls, one was excluded due to changes in diet, and seven did not return the second questionnaire. Mean difference in food group consumption estimated by the two methods did not differ significantly from zero for fruit (0.00 serving, s.d.=0.54, 95% CI= -0.13, +0.12, P=0.95) or vegetables (-0.05, s.d.=0.29, 95% CI= -0.12, +0.02, P=0.13). For cereal foods, the mean difference was small, but significantly different from zero only in women (-0.32 servings, s.d. = 0.92, 95% CI = -0.59, -0.06, P=0.02). At an individual level, cereal food intake as measured by the FFQ may be 37% below or 59% above the diet recall values; and values for total fruit and vegetables may be half or double the recall values. Among subjects whose intake was classified into the lowest quartiles by diet recalls, 78% and 94% respectively, fell into the lowest two questionnaire quartiles for cereal foods, and total fruit and vegetables. The ability of the questionnaire to predict those having inadequate intake based on recall data was more than 90% for the three food groups. The short questionnaire cannot replace the three-day recalls in intake assessment for individuals, but it could be used to screen for low consumers in intervention programmes, to assess mean food group intake in population groups, and to rank individuals into broad categories of food group intake.European Journal of Clinical Nutrition 09/1998; 52(8):557-64. DOI:10.1038/sj.ejcn.1600605 · 2.95 Impact Factor
Chapter: Dietary Assessment Methodology01/2001: pages 3–39; Elsevier Academic Press.