JACN
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Hudson, T. S.
Right arrow Articles by Lanza, E.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Hudson, T. S.
Right arrow Articles by Lanza, E.
Journal of the American College of Nutrition, Vol. 25, No. 5, 370-381 (2006)
Published by the American College of Nutrition

Dietary Fiber Intake: Assessing the Degree of Agreement between Food Frequency Questionnaires and 4-Day Food Records

Tamaro S. Hudson, PhD, Michele R. Forman, PhD, Marie M. Cantwell, PhD, Arthur Schatzkin, MD, Paul S. Albert, PhD and Elaine Lanza, PhD

Division of Cancer Prevention (T.S.H)
Laboratory of Biosystems and Cancer, Center for Cancer Research (M.R.F., M.M.C.)
Divisions of Cancer Epidemiology and Genetics (A.S.)
Biometric Research Branch, Division of Cancer Treatment and Diagnosis (P.S.A.)
Laboratory of Cancer Prevention, Center for Cancer Research (E.L.), National Cancer Institute, Bethesda, Maryland

Address reprint requests to: Elaine Lanza, PhD, Laboratory of Cancer Prevention, National Cancer Institute, National Institute of Health, 6116 Executive Boulevard, Bethesda, MD 20892-8329. E-mail: el33t{at}nih.gov


   ABSTRACT
TOP
 FOOTNOTES
 ABSTRACT
INTRODUCTION
 METHODS
 RESULTS
 ACKNOWLEDGMENTS
 REFERENCES
 
Objective: To assess the degree of agreement (comparability) between dietary fiber intakes reported on a food frequency questionnaire (FFQ) with 4-day food records (4DFR) and determine whether demographic, behavioral and biological factors influence comparability.

Methods: At baseline and year one, all participants in the Polyp Prevention Trial (PPT), a multi-center randomized, clinical trial of a low-fat, high fiber, high fruit/vegetable eating plan and recurrence of large bowel adenomatous polyps were instructed in dietary assessment and completed a 106-item FFQ and 4DFR that trained nutritionists reviewed. A random sub-cohort of participants (n = 399) was selected from the intervention and control arms of the PPT for analysis of both FFQ and 4DFR.

Results: Baseline crude and energy-adjusted fiber intakes were significantly higher in the 4DFR than the FFQ (P = 0.001). Using Bland-Altman statistics, the mean difference (FFQ-4DFR) was –0.11 g/MJ; while the limits of agreement were –1.45, 1.23 g/MJ. The mean fiber difference increased with increasing average intake (FFQ + 4DFR)/2, (P = 0.004) for men, but not women (P = 0.10), suggesting that fiber intake was under-estimated in the FFQ, relative to the 4-DFR, for men with low fiber intakes and over-estimated for men with high intakes. Smoking and gender significantly influenced the average intake at baseline, whereas other demographic and behavioral factors did not. Education was significantly associated with average difference in fiber intake at baseline, but not at year 1.

Conclusions: This study of clinical trial volunteers revealed differences in the ability to comparably report fiber intake across tools by gender, smoking, and education, however participants’ repeated training in dietary assessment improved comparability in reporting over time.

Key words: Dietary fiber, dietary assessment, food frequency questionnaire, food records


   INTRODUCTION
TOP
 FOOTNOTES
 ABSTRACT
 INTRODUCTION
METHODS
 RESULTS
 ACKNOWLEDGMENTS
 REFERENCES
 
In 1971, Burkitt proposed that a high fiber diet might reduce the risk for colon cancer, based on the very low rates of this malignancy in Africa, where high quantities of fiber-rich foods were consumed [1]. Since that time, there has been a great deal of investigation into the potential protective effect of fiber on colorectal neoplasia [2]. Most case-control studies have demonstrated a protective effect of fiber intake on colorectal neoplasia [3–9]. Whereas earlier prospective studies reported no association [10–14], two recent large prospective investigations reported a significant inverse association between dietary fiber intake and risk of colorectal adenomas [15] and cancer [16]. Furthermore, two large, randomized clinical trials failed to show protection from recurrent adenomas with increased fiber and/or decreased fat in the diet [17,18]. In the Wheat Bran Fiber (WBF) trial, rates of recurrent adenomatous polyps were similar in those randomized to a high-fiber or to a low-fiber supplement [17]. In the Polyp Prevention Trial (PPT), there was no reduction in risk of colorectal adenoma recurrence after four years in participants who consumed a low-fat, high-fiber diet compared to those on their usual diet [18]. Thus, the relationship between fiber intake and colorectal neoplasia in observational studies and clinical trials remains controversial.

Inconsistent epidemiologic associations between dietary fiber intake and colorectal cancer may be due to inaccurate measurement of dietary fiber intake, which could attenuate a true diet-cancer relationship. The food frequency questionnaire (FFQ) is a dietary tool commonly used in epidemiological studies, because it is relatively inexpensive, and easy to analyze in large populations. Recently researchers have suggested the discrepancies between retrospective and prospective diet-cancer studies could be due to the use of food frequency questionnaires for dietary assessment [19–21]. Furthermore, several biomarker studies have suggested that the FFQ has insufficient precision to allow detection of moderate but important diet—disease associations [22–24]. Day [22] validated 7-day diet-diaries and a FFQ against urinary biomarkers of nitrogen, potassium, and sodium, and found the records to be more accurate than the FFQ. The Observing Protein and Energy Nutrition (OPEN) study compared the performance of a FFQ with that of a 24-hour recall (24HR) using biomarkers for energy and protein, and concluded the FFQ could be recommended as an instrument for evaluating relations between absolute intake of energy or protein and disease [24].

Because a gold standard for fiber intake is not available, the comparability of the FFQ is usually based on reference data from 24-hour recalls or food records. The range in the correlations between dietary fiber estimates reported in the FFQ and food records is from 0.35 for women [25] to 0.72 for men [26]. In this paper, we analyzed dietary fiber intake at baseline and year one in a random sub-cohort of participants in the Polyp Prevention Trial for the following purposes: 1) to estimate dietary fiber intake from a food-frequency questionnaire (FFQ) and 4-day food records (4DFR) at baseline, by demographic and behavioral characteristics of the participants, a family history of colon cancer, or histopathologic features of adenomatous polyps; 2) to assess the degree of concordance in estimates of total dietary fiber intake from a FFQ and 4DFR at baseline by the Bland-Altman method [27]; and 3) to examine the joint classification of and correlations in dietary fiber intake by dietary tool at baseline and year one.


   METHODS
TOP
 FOOTNOTES
 ABSTRACT
 INTRODUCTION
 METHODS
RESULTS
 ACKNOWLEDGMENTS
 REFERENCES
 
Design
The PPT was a multi-center randomized, clinical trial to determine the effect of a low-fat, high fiber, high-fruit/vegetable eating plan on the recurrence of large bowel adenomatous polyps [18,28–30]. In this study, we analyzed PPT dietary data at baseline and at one year after the intervention began. A health and lifestyle questionnaire (HLQ) was administered by an interviewer at baseline and year one to collect information on general health and lifestyle, prescription medications, over-the-counter preparations, and vitamin and mineral supplements taken on a regular basis. Participants completed a FFQ and 4DFR at baseline and year one to estimate dietary intake in the intervention and control groups, including estimates of total dietary fiber intake.

Sample
The overall design, rationale, dietary intervention, endpoint ascertainment, and trial results for the PPT have been reported previously [18,28–30]. The study enrolled 2079 men and women ≥35 years of age between 1991 and 1994 with at least one histologically confirmed large bowel adenomatous polyp, that was removed during a colonoscopy procedure within the previous six months, at one of eight clinical centers in the U.S. (listed in the Acknowledgements Section). Of the 2079 randomized to the dietary intervention arm or their usual diet, 1905 subjects completed the study after four years. At baseline and each annual follow-up visit, all participants were trained in and completed a FFQ and 4DFR. The nutrient intake from FFQs was analyzed for all participants. A 40% random sub-cohort of PPT participants was selected for blood biomarker analysis to monitor dietary compliance. Neither participants nor trial investigators knew the membership of the blood sub-cohort, since blood was drawn on all 1905 subjects. A 50% random sample of this blood-monitoring cohort (n = 415) was selected for nutrient analysis of 4DFR. In order to reduce bias from editing only members of the 4DFR sub-cohort, all PPT food records were reviewed and edited by nutritionists before identification of the blood-monitoring/4DFR sub-cohort. The 4DFR sample of 415 was representative of the clinical center, gender and intervention status of the entire cohort. Eligibility criteria included: ≥35 years old; no history of colorectal cancer, surgical resection of adenomas, bowel resection or disease; <150% of the recommended weight for height; satisfactory completion of a pretest of the FFQ and 4DFR; and no medical conditions or dietary restrictions that would substantially limit their ability to complete the study. The Institutional Review Boards of the National Cancer Institute and each participating center approved the study. All participants provided written and informed consent. The statistical analysis was completed on 399 of the 415 participants with complete 4DFRs and FFQs at baseline and at year one.

Dietary Assessment
Participants in the PPT completed a modified, 106-item, Block-National Cancer Institute (NCI) FFQ, and 4DFR. At baseline and at each annual follow-up, participants viewed instructional videos demonstrating food portion size estimation and proper completion of dietary assessment forms, and were instructed to always complete the FFQ prior to the 4DFR. Participants estimated their food intake over the past year in a self-administered FFQ at home. The trained nutritionists, who reviewed all FFQs and 4DFRs, were not involved in the dietary counseling. Additionally, all computer edits, such as out-of-range FFQ values and missing responses were reviewed with the participants and revised if incorrect. It is important to note that the same food composition database was used to estimate total dietary fiber intake in the FFQ and 4DFR. The Block-NCI FFQ was validated in clinical trial participants [31] and in the general population of adults aged 25–50 years [32].

Statistical Analysis
The statistical analysis was completed in several phases. In the first phase, the mean (± standard deviation) of the crude and energy-adjusted dietary fiber intakes as reported in the FFQ and the 4DFRs were computed by demographic characteristics (age, gender, ethnicity, education, and marital status), behavioral or lifestyle characteristics (smoking status, alcohol intake, use of non-steroidal anti-inflammatory drugs, and use of multivitamins), anthropometric status (body-mass index), and biologic characteristics (advanced adenoma, and family history of colorectal cancer) of participants at baseline. An advanced adenoma had either a diameter of ≥1 cm or ≥25% villous elements or high-grade dysplasia, including carcinoma. Histograms and Q-Q plots were examined for normality of all relevant parameters. The Student’s T-test was used to test whether differences in the mean total (crude or energy-adjusted) dietary fiber intakes as reported in the FFQ and 4DFR were significantly different by demographic, behavioral, or biologic subgroup.

In the second phase, the Bland-Altman [27] technique was used to assess the degree of agreement in estimates of the energy-adjusted dietary fiber intake between the FFQ and 4DFR. The Bland-Altman procedure was implemented as follows: 1) The average of the two instruments (FAI), i.e. that is the sum of total dietary fiber intake from the two instruments (FFQ and 4DFR) divided by two, after adjusting for the average energy intake in each, forms the X-axis of the plot 2) The mean difference of the two instruments (FDI) is based on the mean difference in fiber intake between the FFQ and 4DFR after adjusting for energy intake. The FDI forms the Y-axis of the plot. In the Bland Altman method, the FDI is plotted against the FAI, with the 95% limits of agreement for individuals at ± 2 standard deviation (SD) units from the mean difference. These limits provide an assessment of the difference between the two measures. We also examined whether there was a positive linear relationship between the FDI and FAI using simple linear regression. When there is a linear relationship, the standardized difference between the two instruments (FRI i.e. fiber ratio of intake), which is calculated as the mean difference in fiber intake (g/MJ) from the two instruments relative to the average fiber intake (g/MJ) of the same instruments (FRI = FAI/FDI), provides a standardized metric for examining discrepancy between the two instruments. T-tests and analysis of variance (ANOVA) were used to examine whether any differences between the two instruments, as measured by the FAI, FDI, and FRI, were influenced by demographic, behavioral and biologic variables.

In the third phase, Analysis of Variance (ANOVA) models were used to: examine the main effects and interactions of categorical variables that were significantly associated with the energy-adjusted FAI, FDI and FRI (as described above) at baseline; and by randomization group and other covariates after one year of the PPT intervention.

In the fourth phase, the joint classification (%) of the participants in 4DFR and FFQ was calculated. Specifically, how many participants in the lowest and highest quartiles of fiber intake in the 4DFRs, (as the reference method), were classified in the same quartile, in the same and adjacent quartiles, and in the extreme opposite quartile of the FFQ at baseline. Finally, the Pearson and Spearman Correlation coefficients for dietary fiber intakes from the 4DFR and FFQ were estimated by gender and by smoking status at baseline and after one year of the trial. All statistically significant differences were based on a P value of ≤0.05.


   RESULTS
TOP
 FOOTNOTES
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
ACKNOWLEDGMENTS
 REFERENCES
 
Demographic characteristics of the 399 participants who completed the 4DFRs and FFQ at baseline and at year one are summarized in Table 1. The majority were white men; the mean age was 61 years (SD ± 9.8; range = 36–86 y). Approximately 70% had more than a high school education, 81% were married, and 14% were smokers at baseline. The mean body-mass index (BMI) of the participants was 27 (SD ± 3.8), while their alcohol consumption was 7.4 g/day (SD ± 12.5) based on the FFQ. Thirty-one percent reported use of non-steroidal anti-inflammatory drugs (NSAIDs), 35% took multivitamins, 26% had a family history of colorectal cancer, and 37% had advanced adenomas at baseline. The distributions of the baseline data in the random sub-sample of 399 participants in Table 1 were comparable to the distributions in the 1506 non-participants in this analysis of the PPT and were also comparable in the same sub-sample stratified by randomization group at baseline.


View this table:
[in this window]
[in a new window]
  		Table 1. Baseline Characteristics for the 399 Participants1 and Non-Participants: PPT

 
The mean (±SD) total crude dietary fiber intake was significantly higher in the 4DFR than the FFQ (i.e. 19.5 ± 7.0 g/day and 17.6 ± 7.5 g/day, respectively, P = 0.001), and likewise, the mean (±SD) energy adjusted fiber intake was significantly higher in the 4DFR than the FFQ (i.e. 2.4 ± 0.8 g/MJ and 2.3 ± 0.8 g/MJ, respectively, P = 0.0007). The mean difference in fiber intake (FFQ-4DFRs) for the entire cohort was –0.11 g/MJ; the 95% limits of agreement were –1.45 and 1.23 g/MJ.

In Table 2, the mean dietary fiber intake as reported in each tool at baseline differed by gender, smoking status, alcohol consumption, and marital status. Specifically, men reported higher total dietary fiber intake (g/d) from 4DFRs than women, however, after adjustment for caloric intake, women reported higher fiber intake in both the FFQ and 4DFR than men. Compared to smokers, non-smokers reported significantly higher (crude and energy-adjusted) dietary fiber intakes. We compared fiber intake in PPT participants based on their reported alcohol consumption from the FFQ and the 4DFR. Non-drinkers, compared to those who drank >20 g/d, reported significantly higher dietary fiber (g/MJ) intake on the FFQ. Additionally fiber intake from 4DFR (g/d) was significantly higher in married participants compared to those not married; however, the effect did not remain after adjustment for caloric intake. No significant differences in fiber intake were observed by level of education, BMI, family history of colorectal cancer, NSAID or multivitamin supplement use, or in participants with an advanced adenoma at trial entry.


View this table:
[in this window]
[in a new window]
  		Table 2. Mean ± SD of Dietary Fiber Intake (based on the FFQ and 4DFR) in 399 Participants by their Baseline Characteristics

 
To examine agreement in estimates of dietary intake in the FFQ and 4DFR, we used the Bland-Altman method of plotting the mean difference between the FFQ and 4DFR (FDI) against the average of the two (FAI). Fig. 1a, 1b, and 1c, show the plots for the entire cohort as well as the cohort stratified by gender at baseline. In Fig. 1a, for men and women combined, there is a tendency for the mean FDI to increase with the FAI; a linear regression of the FAI on the FDI was highly significant (P = 0.009). The difference between the FFQ and the 4DFR increases across the range of fiber intake i.e. as the FAI increases. The resulting slope of the regression line of the FDI on the FAI was positive, suggesting that there is at least a sizable fraction of the population who over-estimate fiber intake on the FFQ relative to the mean 4DFR. When we stratified by gender, this linear relationship between FAI and FDI appeared in men (Fig. 1b) not women (Fig. 1c); and the linear regression analysis of the FAI on the FDI was highly significant in men (P = 0.004), but not women (P = 0.10). This observation led us to further characterize the factors, including gender and educational level, which might potentially contribute to the FDI, FAI, and FRI (Table 3). Women reported a significantly higher energy-adjusted FAI compared to men. Similarly, non-smokers reported a significantly higher FAI than smokers. Those with ≤ high school education reported a higher FDI and a higher FRI compared to those who have > high school education. Moreover, the FRI is significantly lower in non-drinkers than in those who drink ≥20 grams per day, using alcohol consumption from the FFQ or 4DFR. There were no differences in FAI, FDI, or FRI by intervention group at baseline, when participants were randomized to intervention or control group in the PPT. There were also no significant differences in the FAI, FDI, and FRI by marital status, NSAID use, vitamin supplement use, family history of colorectal cancer, BMI, family history of colorectal cancer, or advanced adenomatous polyp in the colonoscopy before enrollment.


Figure 1
View larger version (11K):
[in this window]
[in a new window]
  		Fig. 1a. Bland-Altman plot for all subjects at baseline.

x-axis Average Fiber intake at baseline (FFQ-4 day food record)/2, g/MJ.

y-axis Difference in Fiber intake at baseline (FFQ-4 day food record), g/MJ.

Fig. 1b.

Bland-Altman plot for Males at baseline.

x-axis Average Fiber intake at baseline (FFQ-4 day food record)/2, g/MJ.

y-axis Difference in Fiber intake at baseline (FFQ-4 day food record), g/MJ.

Fig. 1c.

Bland-Altman plot for Females at baseline.

x-axis Average Fiber intake at baseline (FFQ-4 day food record)/2, g/MJ.

y-axis Difference in Fiber intake at baseline (FFQ-4 day food record), g/MJ.

 

View this table:
[in this window]
[in a new window]
  		Table 3. Energy-Adjusted FAI (g/MJ), FDI (g/MJ), FRI at Baseline

 
After identifying the factors that were significantly associated with the FAI, FDI, and FRI in the univariate analysis, ANOVA models were used to examine both the main effects and interactions between identified factors on the FAI, FDI, and FRI summary measures. Smoking and gender both independently influenced the energy-adjusted FAI at baseline (Table 4A); however there was no significant interaction between smoking and gender. Education was significantly associated with the energy adjusted FDI at baseline (F statistic = 7.9; P = 0.005). Neither education nor alcohol intake from the FFQ was significantly associated with the energy-adjusted FRI (Table 4B). Moreover there was no effect of group alone, or of the 2-way interaction of group by each parameter in any of the above analysis.


View this table:
[in this window]
[in a new window]
  		Table 4a. Energy-Adjusted FAI 1 by Gender, and Smoking Status Including Two-Way Interactions and Main Effects in an ANOVA at Baseline

 

View this table:
[in this window]
[in a new window]
  		Table 4b. Energy-Adjusted FRI 1 by Education, Alcohol and All Two-Way Interactions: ANOVA PPT at Baseline

 
The intervention group participants attended 20 counseling sessions in the first year of the trial to facilitate dietary behavior change. Since this training could have affected reported dietary fiber intake and other values in the intervention group, we examined differences in dietary fiber reporting by study group at year one (Table 5). Two factors were significantly associated with energy adjusted fiber intake from the 4DFR and the FFQ. Specifically, those in the intervention group reported significantly higher fiber intake and FAI than those in the control group; and non-smokers reported significantly higher fiber intake on the FFQ than smokers. Likewise, intervention group participants who were non-smokers reported significantly higher fiber intake on the FFQ and FAI than those who were smokers. Finally control group participants who were non-smokers reported higher fiber intake on the 4DFR and a higher FAI than those who were smokers. There was however, no association between any of the variables and the FDI or FRI (data not shown). In an ANOVA model, smoking and study group were independently associated with the FAI, but not the FDI or FRI, after adjustment for the other variable, at the end of the first year of the PPT (P = 0.0042 and P < 0.0001, respectively).


View this table:
[in this window]
[in a new window]
  		Table 5. Mean ± SD of Energy Adjusted (g/MJ) Dietary Fiber Intake (based on the FFQ and 4DFR) and the FAI in 399 Participants by Gender and Education at year 1

 
To provide data for comparison with other studies, we describe the joint classification of energy-adjusted fiber intake in quartiles using the 4DFR and FFQ at baseline for the entire sub-sample. At baseline, 61% and 97% of those in the lowest quartile of 4DFR reported fiber intakes categorized in the lowest or the two lowest quartiles of the FFQ, respectively. Among those in the highest quartile of 4DFR, 56% and 86% reported fiber intakes categorized in either the highest or two highest quartiles of FFQ, respectively. A very low percentage, if any, were misclassified in the opposite extremes on the FFQ compared to the 4DFR, i.e. 1% of those in the lowest quartile of the 4DFR were in the highest quartile of the FFQ and likewise 1% in the highest quartile of the 4DFR were in the lowest quartile on the FFQ. There were no differences in these percentages by gender. As expected at year 1, there were no differences in the percent concordance in the control group compared to the total sub-cohort at baseline. Finally, we computed the Pearson and Spearman correlations between energy-adjusted total dietary fiber intake from the FFQ and the 4DFR at baseline and year 1 (Appendix). Energy adjusted fiber correlations were consistently higher than unadjusted ones, and both the Pearson and Spearman correlations were higher in men than women and higher at year 1 than baseline.


View this table:
[in this window]
[in a new window]
  		APPENDIX Pearson and Spearman correlation coefficients of crude (g/d) and energy adjusted (g/MJ) dietary fiber intakes in the FFQ and 4DFR at baseline and year 1 by gender:

 

   DISCUSSION
 
In the present study, baseline fiber intake and fiber intake in the first year were measured using a FFQ (modified Block/NCI HHQ) and 4 DFRs, in a random sub-cohort of participants from the PPT. Mean energy-adjusted fiber intake at baseline, as estimated from the 4DFRs, was significantly higher than the estimates from the FFQ. Women reported a higher energy-adjusted fiber intake than men. Non-smokers reported a higher intake than smokers. Drinkers of 20+ grams of alcohol per day reported a significantly lower fiber intake than non-drinkers. People who were married reported higher crude, but not energy-adjusted, fiber intakes (g/d) by 4DFR, compared to those who were not married. Comparable data from earlier studies have not been reported by alcohol intake or marital status.

FFQs are used in epidemiologic research as they are generally considered to be cheaper to administer in large populations compared with other dietary assessment measures such as 4DFRs. However, diet-cancer research has reported inconsistent findings due to measurement error associated with the use of FFQs as well as participants’ characteristics such as age [18,23,24]. It is essential therefore to measure the degree of agreement or comparability in dietary assessment tools and to identify factors that influence comparability in reporting dietary intake. A 4DFR was considered the reference in the present study in the absence of a gold standard and it differs from the FFQ in several ways. Unlike the FFQ, food records do not rely on recall of usual or habitual intake over a specified period of time, as all food and drink are recorded at each eating occasion, and respondents are instructed to measure and report actual food intake on the food record.

The Bland-Altman method [27] was applied to data in the present study to summarize the average fiber intake (FAI), the difference between dietary measures across a range of intake (FDI), and provide a measure of the average relative to the difference (FRI). The Bland-Altman method has only recently been incorporated in comparability studies and even fewer have used the method to assess fiber intake [33,34]. The Bland-Altman plot revealed a non-constant bias from individuals who reported a higher fiber intake on both methods (i.e. FAI) and also had a larger difference in intake (FDI). Men, not women, were more likely to overestimate fiber intake using the FFQ compared to the 4DFRs. Of the 2 previous studies [33,34] which applied the Bland-Altman method for fiber intake, both studies reported significant Pearson correlation coefficients (0.70 and 0.73 respectively) of intake from two different tools similar to the results in the present study. However, the mean fiber intake estimated from the FFQs was significantly different from the reference method in one [34] but not in the other [33]. In addition, the limits of agreement in this study indicate that for a new participant, we expect the two methods to differ in estimation of fiber intake by approximately –1.45 to 1.23 g/MJ. Earlier studies [33,34] reported wide limits of agreement for intake, with the potential for under- or over-estimation of fiber intake by as much as 14g/day compared to the reference method [33]. Further research is needed in two areas: to examine the magnitude of the difference in fiber and other nutrient intakes when assessing the comparability of reporting across tools, and to identify innovative technology for improving dietary assessment.

ANOVA models were computed to identify potential factors that influence average intake (FAI). Gender and smoking status were independently associated with the FAI at baseline, after taking into account the other variable. Educational level was the only variable associated with the difference in baseline fiber intake estimated from the two dietary assessment tools. Specifically those with less than a high school education had a larger difference in fiber intake and a larger ratio between the FAI and FDI (i.e. FRI) compared to those with a higher education. Non-drinkers had a lower FRI compared with drinkers at baseline. However none of these variables was significantly associated with the ratio between average intake and the difference in intake, when the association was adjusted for the other variable (Table 4b).

By year 1, smoking status was the only factor associated with average fiber intake (FAI). Dyer et al. [35] similarly reported that current smokers had a significantly lower fiber intake compared to non-smokers in Japan, the U.K. and the US, while smokers in China had non-statistically significantly lower fiber intakes than non-smokers. Investigators in Canada [36], and Australia [37] also reported lower fiber intakes in smokers than non-smokers and in men than women. By year 1, gender and education no longer modified a participant’s ability to estimate fiber intake, (i.e. there was no association with the FDI or FRI), which suggests that repeated training in dietary assessment and dietary counseling can improve dietary fiber estimation. As expected, over the first year of the study, the intervention group reported significantly higher fiber intakes than those on their usual diet. This difference in intake was anticipated as the intervention group received intensive dietary advice (i.e. 20 counseling sessions in the first year) to facilitate change to higher total fiber, fruit and vegetable intake, while those in the control group received no dietary counseling.

Results of our study indicate differences in fiber intake amongst subgroups. For example men had a higher crude fiber intake from both the FFQ and the 4DFRs compared to women, a finding which has been previously reported in two different countries [38,39]. However, women had a higher energy-adjusted fiber intake than men as previously shown by Lanza et al. [40]. At year one, mean energy-adjusted fiber intakes varied by study group and by smoking status. Smoking remained a determinant of dietary fiber intake at year one and smoking status by study group also revealed significant differences in FAI. For all PPT dietary goals, intervention participants made the greatest changes in intake during their first year in trial. In fact, the 4DFR completed by intervention participants after their first 6 months in the trial indicated that most of the changes were achieved by that time [28].

Baseline cross categorization and correlation coefficients in the PPT sub-cohort revealed 61% and 97% of those in the lowest quartile of fiber intake on the 4DFRs were correctly classified in the lowest or two lowest quartiles on the FFQ and only 1% was grossly misclassified (classified into the opposite quartile). These results are comparable to a previous study using the Block FFQ and a diet history method [41]. The agreement between various FFQs with food recalls or food records has been demonstrated in a number of observational studies [38,39,42–46] and clinical trials using Pearson correlations primarily [26,47]. Energy adjusted dietary fiber intake ranged from 0.35 for women [25] to 0.72 for men [26] compared to 0.53 for women and 0.71 for men in the PPT.

Prior dietary assessment research using food records as the referent compared to fiber intake from the FFQs have shown either higher [38,48], or lower intakes assessed by the food records compared with the FFQ [26,42–44,49]. Several potential sources of variation, that could explain this discrepancy, include: characteristics of the study participants’, the timing of administering each dietary assessment method and use of different food composition databases. A difference in intake assessed by food records and the FFQ may be exaggerated if the tools are administered during different seasons of the year or calculation of energy-adjusted intakes does or does not occur.

This study had certain limitations and strengths. The statistical power in the present study may have been insufficient to detect other differences between the dietary methods or differences in a participant’s ability to estimate fiber intake by demographic characteristics. In addition, the present study population is predominantly a white, middle-aged population, and therefore the results may not be generalizable to other ethnicities and age-groups. The strengths of the study rest in the participants’ volunteer bias notably being more highly educated and motivated; therefore PPT patients were able to follow instruction in dietary assessment. Further, training in dietary measurement was conducted repeatedly throughout the trial, leading to improved comparability in dietary reporting between tools over time. Comparability in reporting of dietary intakes across tools is not, however, an indication of improved assessment of true fiber intake in the participants.

In summary, our analyses suggest that when comparing estimates of energy adjusted fiber intake from a FFQ and 4DFR in the context of observational epidemiologic and intervention studies, it is important to consider the influence of smoking status, alcohol intake, gender and education. Among the factors influencing comparability in reporting dietary fiber intake, only smoking status remained in year one of this trial, therefore repeated training attenuated the effects of other demographic and lifestyle factors. It should also be noted that smoking, which is typically correlated with alcohol intake, may be reflective of a particular lifestyle which reduces a participant’s actual fiber intake or the ability to report fiber intake. It would therefore be prudent in epidemiologic studies to consider the influence of smoking, alcohol intake, gender and education on fiber intake estimates from an FFQ or 4DFRs. In addition to comparing fiber intake, studies should examine whether 4DFR and FFQs are identifying similar food sources of fiber. Finally, research on biomarkers of fiber intake could ameliorate the dependence on reference methods in dietary assessment.


   ACKNOWLEDGMENTS
TOP
 FOOTNOTES
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 ACKNOWLEDGMENTS
REFERENCES
 
The members of the Polyp Prevention Study Group participated in the conduct of the Polyp Prevention Trial. However, the data presented in this manuscript and the conclusions drawn from them are solely the responsibility of the above listed co-authors. National Cancer Institute: Schatzkin A, Lanza E, Corle D, Freedman LS, Clifford C, Tangrea J; Bowman Gray School of Medicine: Cooper MR, Paskett E, Quandt S, DeGraffinreid C, Bradham K, Kent L, Self M, Boyles D, West D, Martin L, Taylor N, Dickenson E, Kuhn P, Harmon J, Richardson I, Lee H, Marceau E; University of New York at Buffalo: Lance MP, Marshall JR (currently University of Arizona), Hayes D, Phillips J, Petrelli N, Shelton S, Randall E, Blake A, Wodarski L, Deinzer M, Melton R; Edwards Hines, Jr; Hospital, Veterans Administration Medical Center: Iber FL, Murphy P, Bote EC, Brandt-Whittington L, Haroon N, Kazi N, Moore MA, Orloff SB, Ottosen WJ, Patel M, Rothschild RL, Ryan M, Sullivan JM, Verma A; Kaiser Foundation Research Institute: Caan B, Selby JV, Friedman G, Lawson M, Taff G, Snow D, Belfay M, Schoenberger M, Sampel K, Giboney T, Randel M; Memorial Sloan-Kettering Cancer Center: Shike M, Winawer S, Bloch A, Mayer J, Morse R, Latkany L, D’Amato D, Schaffer A, Cohen L; University of Pittsburgh: Weissfeld J, Schoen R, Schade RR, Kuller L, Gahagan B, Caggiula A, Lucas C, Coyne T, Pappert S, Robinson R, Landis V, Misko S, Search L; University of Utah: Burt RW, Slattery M, Viscofsky N, Benson J, Neilson J, McDivitt R, Briley M, Heinrich K, Samowitz W; Walter Reed Army Medical Center: Kikendall JW, Mateski DJ, Wong R, Stoute E, Jones-Miskovsky V, Greaser A, Hancock S, Chandler S; Data and Nutrition Coordinating Center (Westat) Cahill J, Hasson M, Daston C, Brewer B, Zimmerman T, Sharbaugh C, O’Brien B, Cranston L, Odaka N, Umbel K, Pinsky J, Price H, Slonim A; Central Pathologists: Lewin K (University of California, Los Angeles), Appelman H (University of Michigan); Laboratories: Bachorik PS, Lovejoy K (Johns Hopkins University); Sowell A (Centers for Disease Control); Data and Safety Monitoring Committee: Greenberg ER (chair) (Dartmouth University); Feldman E (Augusta, Georgia); Garza C (Cornell University); Summers R (University of Iowa); Weiand S (through June 1995) (University of Minnesota) and DeMets D (beginning July 1995) (University of Wisconsin).


   FOOTNOTES
TOP
 FOOTNOTES
ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 ACKNOWLEDGMENTS
 REFERENCES
 
Dr. Forman is now at the Department of Epidemiology, MD Anderson Cancer Center, Houston, Texas.

Received October 12, 2005. Accepted March 1, 2006.


   REFERENCES
TOP
 FOOTNOTES
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 ACKNOWLEDGMENTS
 REFERENCES
 

  1. Burkitt DP: Epidemiology of cancer of the colon and rectum.Dis Colon Rectum36 :1071 –1082,1993 .[Medline]
  2. Trock B, Lanza E, Greenwald P: Dietary fiber, vegetables, and colon cancer: critical review and meta-analyses of the epidemiologic evidence.J Natl Cancer Inst82 :650 –661,1990 .[Abstract/Free Full Text]
  3. Martinez ME, McPherson RS, Annegers JF, Levin B: Association of diet and colorectal adenomatous polyps: dietary fiber, calcium, and total fat.Epidemiology7 :264 –268,1996 .[Medline]
  4. Martinez ME, McPherson RS, Levin B, Glober GA: A case-control study of dietary intake and other lifestyle risk factors for hyperplastic polyps.Gastroenterology113 :423 –429,1997 .[Medline]
  5. Negri E, Franceschi S, Parpinel M, La Vecchia C: Fiber intake and risk of colorectal cancer.Cancer Epidemiology, Biomarkers & Prevention7 :667 –671,1998 .[Abstract]
  6. Le Marchand L, Hankin JH, Wilkens LR, Kolonel LN, Englyst HN, Lyu LC: Dietary fiber and colorectal cancer risk.Epidemiology8 :658 –665,1997 .[Medline]
  7. Freudenheim JL, Graham S, Horvath PJ, Marshall JR, Haughey BP, Wilkinson G: Risks associated with source of fiber and fiber components in cancer of the colon and rectum.Cancer Res50 :3295 –3300,1990 .[Abstract/Free Full Text]
  8. Benito E, Obrador A, Stiggelbout A, Bosch FX, Mulet M, Munoz N, Kaldor J: A population-based case-control study of colorectal cancer in Majorca.I. Dietary factors. Int J Cancer45 :69 –76,1990 .
  9. Bidoli E, Franceschi S, Talamini R, Barra S, La Vecchia C: Food consumption and cancer of the colon and rectum in north-eastern Italy.Int J Cancer50 :223 –229,1992 .[Medline]
  10. McCullough ML, Robertson AS, Chao A, Jacobs EJ, Stampfer MJ, Jacobs DR, Diver WR, Calle EE, Thun MJ: A prospective study of whole grains, fruits, vegetables and colon cancer risk.Cancer Causes Control14 :959 –970,2003 .[Medline]
  11. Gaard M, Tretli S, Loken EB: Dietary factors and risk of colon cancer: a prospective study of 50,535 young Norwegian men and women.Eur J Cancer Prev5 :445 –454,1996 .[Medline]
  12. Pietinen P, Malia N, Virtanen M, Hartman TJ, Tangrea JA, Albanes D, Virtamo J: Diet and the risk of colorectal cancer in a cohort of Finnish men.Cancer Causes Control5 :387 –396,1999 .
  13. Terry P, Giovannucci E, Michels KB, Bergkvist L, Hansen H, Holmberg L, Wolk A: Fruit, vegetables, dietary fiber, and risk of colorectal cancer.J Natl Cancer Inst93 :525 –533,2001 .[Abstract/Free Full Text]
  14. Fuchs CS, Giovannucci EL, Colditz GA, Hunter DJ, Stampfer MJ, Rosner B, Speizer FE, Willett WC: Dietary fiber and the risk of colorectal cancer and adenoma in women.N Engl J Med340 :169 –176,1999 .[Abstract/Free Full Text]
  15. Peters U, Sinha R, Chatterjee N, et al: Dietary fibre and colorectal adenoma in a colorectal cancer early detection programme.Lancet361 :1491 –1495,2003 .[Medline]
  16. Bingham SA, Day NE, Luben R, et al: Dietary fibre in food and protection against colorectal cancer in the European Prospective Investigation into Cancer and Nutrition (EPIC): an observational study.Lancet361 :1496 –149501,2003 .[Medline]
  17. Alberts DS, Martinez ME, Roe DJ, et al: Lack of effect of a high-fiber cereal supplement on the recurrence of colorectal adenomas. Phoenix colon cancer prevention physicians’ network.N Engl J Med342 :1156 –1162,2000 .[Abstract/Free Full Text]
  18. Schatzkin A, Lanza E, Corle D, et al: Lack of effect of a low-fat, high-fiber diet on the recurrence of colorectal adenomas. Polyp prevention trial study group.N Engl J Med342 :1149 –1155,2000 .[Abstract/Free Full Text]
  19. Kristal AR, Peters U, Potter JD. Is it time to abandon the food frequency questionnaire?Cancer Epidemiol Biomarkers Prev.14 :2826 –8,2005 .[Free Full Text]
  20. Kipnis V, Midthune D, Freedman L, Bingham S, Day NE, Riboli E, Ferrari P, Carroll RJ. Bias in dietary-report instruments and its implications for nutritional epidemiology.Public Health Nutr5 :915 –23.
  21. Byers T: Food frequency dietary assessment: how bad is good enough?Am J Epidemiol154 :1087 –1088,2001 .[Free Full Text]
  22. Day NE, McKeown N, Wong MY, Welch A, Bingham S: Epidemiological assessment of diet: a comparison of a 7-day diary with a food frequency questionnaire using urinary markers of nitrogen, potassium and sodium.Int J Epidemiol30 :309 –317,2001 .[Abstract/Free Full Text]
  23. Bingham SA: Dietary assessments in the European prospective study of diet and cancer (EPIC).Eur J Cancer Prev.6 :118 –24,2001 .
  24. Schatzkin A, Kipnis V, Carroll RJ, Midthune D, Subar AF, Bingham S, Schoeller DA, Troiano RP, Freedman LS. A comparison of a food frequency questionnaire with a 24-hour recall for use in an epidemiological cohort study: results from the biomarker-based Observing Protein and Energy Nutrition (OPEN) study.Int J Epidemiol.32 :1054 –62,2003 .[Abstract/Free Full Text]
  25. Jain MG, Rohan TE, Soskolne CL, Kreiger N: Calibration of the dietary questionnaire for the Canadian Study of Diet, Lifestyle and Health cohort.Public Health Nutr6 :79 –86,2003 .[Medline]
  26. Pietinen P, Hartman AM, Haapa E, Rasanen L, Haapakoski J, Palmgren J, Albanes D, Virtamo J, Huttunen JK: Reproducibility and validity of dietary assessment instruments. II. A qualitative food frequency questionnaire.Am J Epidemiol128 :667 –676,1988 .[Abstract/Free Full Text]
  27. Bland JM, Altman DG: Statistical methods for assessing agreement between two methods of clinical measurement.Lancet1 :307 –310,1986 .[Medline]
  28. Lanza E, Schatzkin A, Daston C, Corle D, Freedman L, Ballard-Barbash R, Caan B, Lance P, Marshall J, Iber F, Shike M, Weissfeld J, Slattery M, Paskett E, Mateski D, Albert P: Implementation of a 4-y, high-fiber, high-fruit-and-vegetable, low-fat dietary intervention: results of dietary changes in the Polyp Prevention Trial.Am J Clin Nutr74 :387 –401,2001 .[Abstract/Free Full Text]
  29. Schatzkin A, Lanza E, Freedman LS, Tangrea J, Cooper MR, Marshall JR, Murphy PA, Selby JV, Shike M, Schade RR, Burt RW, Kikendall JW, Cahill J: The polyp prevention trial I: rationale, design, recruitment, and baseline participant characteristics.Cancer Epidemiol Biomarkers Prev5 :375 –383,1996 .[Abstract/Free Full Text]
  30. Lanza E, Schatzkin A, Ballard-Barbash R, Corle D, Clifford C, Paskett E, Hayes D, Bote E, Caan B, Shike M, Weissfeld J, Slattery M, Mateski D, Daston C, Clifford DC: The polyp prevention trial II: dietary intervention program and participant baseline dietary characteristics.Cancer Epidemiol Biomarkers Prev5 :385 –392,1996 .[Abstract/Free Full Text]
  31. Block G, Woods M, Potosky A, Clifford C. Validation of a self-administered diet history questionnaire using multiple diet records.J Clin Epidemiol43 :1327 –35,1990 .[Medline]
  32. Block G, Thompson FE, Hartman AM, Larkin FA, Guire KE: Comparison of two dietary questionnaires validated against multiple dietary records collected during a 1-year period.J Am Diet Assoc92 :686 –693,1992 .[Medline]
  33. Grootenhuis PA, Westenbrink S, Sie CM, de Neeling NJ, Kok FJ, Bouter LM: A semiquantitative food frequency questionnaire for use in epidemiologic research among the elderly: Validation by comparison with dietary history.J Clin Epidemiol48 :859 –868,1995 .[Medline]
  34. Cantwell MM, Gibney MJ, Cronin D, Younger KM, O’Neill JP, Hogan L, Flynn MA: Development and validation of a food-frequency questionnaire for the determination of detailed fatty acid intakes.Public Health Nutr8 :97 –107,2005 .[Medline]
  35. Dyer AR, Elliott P, Stamler J, Chan Q, Ueshima H, Zhou BF for the INTERMAP Research Group: Dietary intake in male and female smokers, ex-smokers, and never smokers: The INTERMAP Study.J Hum Hypertens17 :641 –654,2003 .[Medline]
  36. Palaniappan U, Jacobs Starkey L, O’Loughlin JO, Gray-Donald K: Fruit and vegetable consumption is lower and saturated fat intake is higher among Canadians reporting smoking.J Nutr131 :1952 –1958,2001 .[Abstract/Free Full Text]
  37. English RM, Najman JM, Bennett SA: Dietary intake of Australian smokers and non-smokers.Aus NZ J Public Health21 :141 –146,1997 .
  38. Sasaki S, Matsumura Y, Ishihara J, Tsugane S: Validation of a self-administered food frequency questionnaire used in the 5-year follow-up survey of the JPHC study cohort I to assess dietary fiber intake: comparison with dietary records.J Epidemiol13 :S106 –14,2003 .[Medline]
  39. Pisani P, Faggiano F, Krogh V, Palli D, Vineis P, Berrino F: Relative validity and reproducibility of a food frequency dietary questionnaire for use in the Italian EPIC centres.Int J Epidemiol26 (Suppl 1) :S152 –160,1997 .
  40. Lanza E, Jones DY, Block G, Kessler L: Dietary fiber intake in the U.S. population.Am J Clin Nutr46 :790 –797,1987 .[Abstract/Free Full Text]
  41. Caan BJ, Slattery ML, Potter J, Quesenberry CP, Coates AO, Schaffer DM: Comparison of the Block and the Willett self-administered semi-quantitative food frequency questionnaires with an interviewer-administered dietary history.Am J Epidemiol148 :1137 –1147,1998 .[Abstract/Free Full Text]
  42. Rimm EB, Giovannucci EL, Stampfer MJ, Colditz GA, Litin LB, Willett WC: Reproducibility and validity of an expanded self-administered semiquantitative food frequency questionnaire among male health professionals.Am J Epidemiol135 :1114 –1126,1992 .[Abstract/Free Full Text]
  43. Block G, Woods M, Potosky A, Clifford: Validation of a self-administered diet history questionnaire using multiple diet records.J Clin Epidemiol43 :1327 –1335,1990 .[Medline]
  44. Willet WC, Sampson L, Stampfer MJ, Rosner B, Bain C, Witschi J, Hennekens CH, Speizer FE: Reproducibility of a semi-quantitative food frequency questionnaire.Am J Epidemiol122 :51 –65,1985 .[Abstract/Free Full Text]
  45. Martinez ME, Marshall JR, Graver EG, Whitacre RC, Woolf K, Ritenbaugh C, Alberts DS: Reliability and validity of a self-administered food frequency questionnaire in a chemoprevention trial of adenoma recurrence.Cancer Epidemiol Biomarkers Prev8 :941 –946,1999 .[Abstract/Free Full Text]
  46. Kroke A, Klipstein-Grobusch K, Voss S, Moseneder J, Thielecke F, Noack R, Boeing H: Validation of a self-administered food-frequency questionnaire administered in the European Prospective Investigation into Cancer and Nutrition (EPIC) Study: comparison of energy, protein, and macronutrient intakes estimated with the doubly labeled water, urinary nitrogen, and repeated 24-h dietary recall methods.Am J Clin Nutr70 :439 –447,1999 .[Abstract/Free Full Text]
  47. Thomson CA, Giuliano A, Rock CL, Ritenbaugh CK, Flatt SW, Faerber S, Newman V, Caan B, Graver E, Hartz V, Whitacre R, Parker F, Pierce JP, Marshall JR: Measuring dietary change in a diet intervention trial: comparing food frequency questionnaire and dietary recalls.Am J Epidemiol157 :754 –762,2003 .[Abstract/Free Full Text]
  48. Goldbohm RA, van den Brandt PA, Brants HAM, van’t Veer P, Al M, Sturman F, Hermus JJ: Validation of a dietary questionnaire used in a large-scale prospective cohort study on diet and cancer.Eur J Clin Nutr48 :253 –265,1994 .[Medline]
  49. Bingham SA, Gill C, Welch A, Day K, Cassidy A, Khaw KT, Sneyd MJ, Key TJA, Roe L, Day NE: Comparison of dietary assessment methods in nutritional epidemiology: weighed records v. 24h recalls, food-frequency questionnaires and estimated-diet records.British J Nutr72 :619 –643,1994 .



This article has been cited by other articles:


Home page
 Am J EpidemiolHome page
L. B. Sansbury, K. Wanke, P. S. Albert, L. Kahle, A. Schatzkin, E. Lanza, and the Polyp Prevention Trial Study Group
The Effect of Strict Adherence to a High-Fiber, High-Fruit and -Vegetable, and Low-Fat Eating Pattern on Adenoma Recurrence
Am. J. Epidemiol., September 1, 2009; 170(5): 576 - 584.
[Abstract] [Full Text] [PDF]

Home page
 Cancer Epidemiol. Biomarkers Prev.Home page
G. Bobe, L. B. Sansbury, P. S. Albert, A. J. Cross, L. Kahle, J. Ashby, M. L. Slattery, B. Caan, E. Paskett, F. Iber, et al.
Dietary Flavonoids and Colorectal Adenoma Recurrence in the Polyp Prevention Trial
Cancer Epidemiol. Biomarkers Prev., June 1, 2008; 17(6): 1344 - 1353.
[Abstract] [Full Text] [PDF]

Home page
 J. Nutr.Home page
K. L. Wanke, C. Daston, A. Slonim, P. S. Albert, K. Snyder, A. Schatzkin, and E. Lanza
Adherence to the Polyp Prevention Trial Dietary Intervention Is Associated with a Behavioral Pattern of Adherence to Nondietary Trial Requirements and General Health Recommendations
J. Nutr., February 1, 2007; 137(2): 391 - 398.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Hudson, T. S.
Right arrow Articles by Lanza, E.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Hudson, T. S.
Right arrow Articles by Lanza, E.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS