![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
The Journal of Nutrition Vol. 128 No. 10 October 1998,
pp. 1665-1671
, and
* Division of Applied Human Nutrition and Department of Mathematics and Statistics, University of Guelph, Guelph, Ontario, N1G 2W1 Canada and ** Pediatric Nutrition Research and Development, Ross Products Division, Abbott Laboratories, Columbus, OH 43215
 |
ABSTRACT |
|---|
Abstract
Introduction
Methods
Results
Discussion
References
|
|---|
The purpose of this study was to validate a food-frequency questionnaire (FFQ) and a 3-d weighed food record (3d-WFR) by comparing nutrient intakes estimated using these methods with serum folate, RBC folate and serum vitamin B-12 concentrations in 105 females aged 16-19 y. During an early morning clinic visit, subjects completed a self-administered, 116-item FFQ, blood was collected and they were trained to complete a 3d-WFR. Folate intakes as determined by the 3d-WFR (r = 0.65, P < 0.01) exhibited a stronger association with serum folate than did intakes from the FFQ (r = 0.48, P < 0.01) (P = 0.017). The correlations between folate intakes and RBC folate as determined by the FFQ (r = 0.42, P < 0.01) and 3d-WFR (r = 0.50, P < 0.01) methods did not differ. Vitamin B-12 intakes showed only a modest association with serum vitamin B-12 when supplement users were included in the analyses (FFQ, r = 0.25, P < 0.05; 3d-WFR, r = 0.32, P < 0.05). After excluding supplement users from the analyses, the relationship between vitamin B-12 intakes as determined by FFQ and serum vitamin B-12 was no longer significant. Median daily folate intakes (346 vs. 212 µg) and vitamin B-12 (4.9 vs. 1.9 µg) estimated from the FFQ were higher than those obtained from the 3d-WFR. In sum, these data suggest that both the FFQ and 3d-WFR are valid measures of assessing the folate intake of young women, and both appear to be useful in determining vitamin B-12 intake when supplemental users are included. The markedly different conclusions about absolute folate and vitamin B-12 intakes obtained using these two dietary methodologies should be taken into consideration when making recommendations about optimal folate intakes in relation to disease prevention.
KEY WORDS: young women · folate · vitamin B-12 · dietary assessment
Two recent trials, one coordinated by the British Medical Research Council (MRC Vitamin Study Research Group 1991) and the other conducted in conjunction with the Hungarian Family Planning Committee (Czeizel and Dudas 1992 The weak association between dietary folate intake and the incidence of NTD may in fact reflect the difficulty in estimating individual folate intakes rather than a lack of a protective effect of dietary folate per se. Researchers examining the relationship between maternal dietary folate intake and the incidence of NTD have relied almost exclusively on food-frequency questionnaires (FFQ) to obtain estimates of usual folate intake during the periconceptional period (Bower and Stanley 1989 Another approach commonly employed to determine nutrient intake is the use of weighed food records. In contrast to FFQ, weighed food records allow more precise determination of portion sizes, do not rely on memory and are not limited to selection from a predetermined list of foods. Weighed food records are not very practical in large epidemiologic studies, however, because they require extensive participant training, have a high respondent burden and require lengthy data entry by trained personnel (Willett 1990 Food-frequency questionnaires have been validated for determining folate intake in a variety of populations in the United States including the following: ranchers and volunteers from South Dakota and Wyoming (Longnecker et al. 1993 The purpose of this study was to estimate intakes of folate and vitamin B-12 in a group of young women (16-19 y of age) by using a semiquantitative FFQ and a 3-d weighed food record (3d-WFR). In addition, the validity of a FFQ and 3d-WFR in estimating usual intakes of folate and vitamin B-12 was assessed by correlating folate and vitamin B-12 intakes derived using these two dietary measures with biochemical indices. Further, the ability of the FFQ and the 3d-WFR to classify participants' nutrient intakes correctly into quartiles of blood vitamin levels was determined.
Study population.
A sample of 105 adolescent female volunteers aged 16-19 y was recruited from September 1994 to March of 1995 from Southern Ontario, Canada, via publicity in local newspapers, universities, community colleges, shopping malls and other commercial areas frequented by adolescents. Exclusion criteria included the use of drugs known to interfere with folate metabolism and the presence of chronic disease. In addition, pregnant females or females who had a pregnancy lasting more than 20 wks in the year before the scheduled clinic visit were also excluded. Informed written consent was obtained from each participant after the purpose, significance and the protocol of the study were fully explained. The research protocol was approved by the University of Guelph Human Ethics Committee.
Biochemical assessment.
Participants attended an early morning clinic visit after an overnight fast. Blood was collected via venipuncture from participants into evacuated containers with and without heparin. Hematocrits were determined on freshly collected heparinized blood. For the RBC folate assay, two 100-µL aliquots of heparinized whole blood were diluted with 1.0 mL of 4 g/L ascorbate solution to prevent the oxidation of folate. Plasma from heparinized containers and serum from containers with no anticoagulant were separated from whole blood by centrifugation (620 × g for 20 min at 5°C). Blood samples were stored at Dietary assessment.
After providing a blood sample, participants completed a self-administered 116-item semiquantitative FFQ developed by Willett et al. (1985 and 1988). The questionnaire was designed to assess usual intake over the previous year. Where possible, the portion sizes used in the questionnaire were based on the typical or natural portion consumed (e.g., a slice of bread, one egg, one cup of coffee). When a typical or natural portion size was not obvious, a commonly used portion size was selected (e.g., one cup rice). For the purposes of this study, portion sizes provided with the FFQ were not adjusted for the age or gender of subjects. The reproducibility (precision) of various forms of this questionnaire has been determined in a variety of adult populations, and correlations between repeated FFQ for folate intake have ranged from 0.57 to 0.87 (Longnecker et al. 1993 Statistical analysis.
Statistical analyses were performed by using the General Linear Model (GLM) procedure of the Statistical Analysis System (SAS/STAT Version 6, SAS Institute, Cary, NC). A probability level of 5% was chosen as the level of significance. Biochemical indices and folate intakes (FFQ and 3d-WFR) were normalized using logarithmic transformations; vitamin B-12 intakes (FFQ and 3d-WFR) were normalized using a square root transformation. T tests were used to compare folate and vitamin B-12 intakes as estimated by the FFQ and the 3d-WFR, as well as biochemical indices of supplement users vs. supplement non-users. Pearson correlations (r) were generated between each of the nutrient intakes and their respective biochemical indices (Snedecor and Cochran 1989
The mean age of the participants was 18.5 ± 0.1 y (mean ± SEM). Fourteen, 19, 29 and 43 of the young women were 16, 17, 18 and 19 y of age, respectively. Ninety-five of the participants were Caucasian, one was Black, and nine were Asian. Of the 105 participants, 61 lived at home with one or both parents, 39 lived with friends or in a university dormitory, three lived with a boyfriend and two lived alone. At the time of the survey, 64 of the participants were in high school or had completed high school and 41 were in university or college.
Results from this study suggest that both the FFQ and 3d-WFR are valid measures for assessing the folate intakes of young women aged 16-19 y. Using serum folate concentration as the sole biochemical criterion, it appeared that the 3d-WFR was superior to the FFQ as a tool to predict the folate intakes of young women (P = 0.017). The correlation coefficient between folate intakes estimated using the 3d-WFR and serum folate was 0.65 (adjusted for the ratio of intra- to intersubject variation) vs. 0.48 for the FFQ. Garry et al. (1984)
INTRODUCTION
Abstract
Introduction
Methods
Results
Discussion
References
), provide compelling evidence that periconceptional folic acid supplementation reduces the number of pregnancies affected by neural tube defects (NTD).3 These studies confirmed a series of earlier case-control and cohort studies that showed a protective effect of supplemental folic acid on the incidence of NTD (Picciano et al. 1994). Despite the well-characterized relationship between periconceptional vitamin supplement use and NTD, the association between dietary folate intake and NTD has not been extensively studied. Data from the few available studies suggest, however, that this relationship is less remarkable (Bower and Stanley 1989, Brown et al. 1997, Cuskelly et al. 1996, Daly et al. 1995, Milunsky et al. 1989, Werler et al. 1993).
, Daly et al. 1995, Milunsky et al. 1989, Werler et al. 1993). Food-frequency questionnaires are an attractive method in epidemiologic studies because of their low respondent burden and ease of administration (Willett 1990). The use of FFQ in estimating nutrient intake is based on the frequency with which a fixed list of foods of predetermined portion sizes is consumed over an extended period of time. This method relies heavily on memory, and the questions posed to respondents are open to interpretation (Willett 1990).
). A larger concern is that food records may cause participants to alter their food intake and may not be representative of the respondents' usual dietary intake (Gibson 1990).
), male health professionals (Rimm et al. 1992) and men and women aged 40 y and older (Jacques et al. 1993, Munger et al. 1992). However, to our knowledge, no studies have validated the use of a FFQ in a population of young women. Young women, in particular, are at risk of low folate intakes and suboptimal folate status as a result of poor dietary habits along with self-imposed restriction of energy secondary to concerns about weight gain (Bailey 1990a). Further, pregnancy, which places heavy demands on folate supplies, is a possibility in this population.
MATERIALS AND METHODS
Abstract
Introduction
Methods
Results
Discussion
References
75°C until analyzed for folate and vitamin B-12 concentrations.
, Sauberlich 1995). As demonstrated by Gunter et al. (1996), we found in our laboratory that analysis of serum samples with the use of either the microbiological method or radioassay with the newer Quanta Phase Folate/B-12 kit yielded virtually identical folate concentrations. In contrast, the two methods yielded markedly different results for RBC folate concentrations. Hence, whole blood folate and heparinized plasma folate concentrations (used for determination of RBC folate) were determined using the microtiter technique described by Tamura (1990) with Lactobacillus casei (ATCC 7469) as the test microorganism. The interassay coefficient of variation for the microtiter technique was 6.7% based upon repeated measurements of a pooled serum sample.
, Munger et al. 1992, Rimm et al. 1992). Similarly, vitamin B-12 correlations have ranged from 0.37 to 0.73 (Longnecker et al. 1993, Munger et al. 1992, Rimm et al. 1992). The nutrient database used to calculate nutrient intakes from this FFQ was derived primarily from information from the Consumer and Food Economic Institute (1976-1991) compiled by the USDA with additional published data and information from food manufacturers (Willett et al. 1985). Because Canadian cold breakfast cereals are fortified with considerably lower levels of folate and vitamin B-12 than those in the U.S., the intakes of these vitamins derived from the FFQ were modified to reflect these differences. Nutrient values for cold breakfast cereals were obtained from Health and Welfare Canada (1988).
). Most of the food composition values in the database were from the Consumer and Food Economic Institute (1976-1991). The folate and vitamin values for cereals and grains were modified by information from Health and Welfare Canada (1988) to incorporate the differences in Canadian fortification laws for folate and vitamin B-12. Additional values for certain food items were obtained from Holland et al. (1991). There were no missing values for any folate or vitamin B-12-containing foods in the database.
). Partial correlations between nutrient intakes and biochemical measures were determined after adjusting for energy and age using multivariate ANOVA.
![<RAD><RCD>1 + [(σ<SUP>2</SUP><SUB><IT>intra</IT></SUB>/σ<SUP>2</SUP><SUB><IT>inter</IT></SUB>)/<IT>n</IT>]</RCD></RAD>](/content/vol128/issue10/fulltext/1665/img001.gif)
where ro is the unadjusted correlation coefficient between the nutrient intake as determined from 3d-WFR, 
2intra is the intrasubject variation, 2inter is the intersubject component of variance and n is the number of days of records (Anderson 1986, Beaton et al. 1979).
12) was different from the correlation of the biochemical index with nutrient intake as determined from the 3d-WFR (13). The biochemical index (Y1), nutrient intake from the 3d-WFR (Y2) and nutrient intake from the FFQ (Y3) were assumed to follow a trivariate normal distribution. Briefly, the ratio of the likelihood evaluated assuming 12 = 13 over the likelihood making no assumptions on the correlations was evaluated. Finally, under the hypothesis 12 = 13, 2 loge (likelihood ratio) has approximately a chi-squared distribution with 1 df.
View this table:
Table 1.
Quartiles of energy, folate, and vitamin B-12 intake as determined by a food-frequency questionnaire (FFQ) and 3-d weighed food record (3d-WFR) in a group of women aged 16-19 y1
RESULTS
Abstract
Introduction
Methods
Results
Discussion
References
). Only 2.0% (2 of 100) of participants had both serum and RBC folate levels indicative of folate deficiency. Of the participants, 0 and 3.9% (4 of 103) had serum vitamin B-12 concentrations in the indeterminate (118-148 pmol/L) and deficient range (<118 pmol/L), respectively. Supplement users had higher serum folate, RBC folate and vitamin B-12 concentrations than supplement nonusers.
View this table:
Table 2.
Quartiles of biochemical folate and vitamin B-12 indices of a group of young women aged 16-19 y
View this table:
Table 3.
Correlations between the folate and vitamin B-12 intakes in a group of women aged 16-19 y estimated from 3-d weighed food records (3d-WFR) and food-frequency questionnaires (FFQ) vs. select biochemical indices1,2
View this table:
Table 4.
Percentage of participants correctly, closely or misclassified into quartiles of folate and vitamin B-12 intake as determined
by a 3-d weighed food record or a food-frequency questionnaire compared with classification by biochemical indices
DISCUSSION
Abstract
Introduction
Methods
Results
Discussion
References
reported a correlation of 0.49 (unadjusted) between folate intake as determined by 3d-WFR and plasma folate concentrations in a group of elderly men and women (n = 270) participating in the New Mexico Aging Process Study. Using a FFQ, Jacques et al. (1993) reported a correlation of r = 0.60 (energy, sex and age adjusted) between folate intakes and plasma folate concentrations in a group of 139 adults aged 40-83 y.
). A similar correlation of 0.49 (unadjusted) was observed between folate intake as determined by a 3d-WFR and RBC folate concentrations (Garry et al. 1984). In a small group of elderly British subjects, a correlation of 0.51 (unadjusted) was observed between folate intakes as determined by 1 y of continuous diet records (n = 19) and RBC folate concentrations (Bates et al. 1982). In contrast, no correlation was observed between folate intakes as determined by food records and RBC folate in another British study (Bingham et al. 1995).
). In this study, the 3d-WFR were collected at the time of blood sampling and reflect nutrient intakes at that time. In contrast, the FFQ is designed to assess usual nutrient intake over the previous year which may or may not represent very recent dietary intake. Had the 3d-WFR been collected before blood sampling, the correlation between folate intake as estimated by 3d-WFR and serum folate might have improved. Given the relatively long lifespan of red blood cells, it is unlikely that the timing of the 3d-WFR relative to the blood draw influenced the strength of this correlation.
). Our results for the FFQ are consistent with those of Jacques et al. (1993), who observed a correlation (age, sex and energy adjusted) of 0.35 between vitamin B-12 intakes estimated from a FFQ and serum vitamin B-12 concentrations.
), smoking (Senti and Pilch 1985), chronic alcohol consumption (Herbert 1990) and certain prescription drugs (Roe 1990).
). Second, the list of foods in the FFQ used in this study was designed for an American adult population and, consequently, contains foods most discriminating for this population (Willett 1990). Snack foods are known to contribute a large proportion of the energy and nutrient content of the adolescent diet (Bigler-Doughton and Jenkins 1987) and hence FFQ designed for adults may not adequately measure the nutrient contribution from snack foods among young women. Third, the portion sizes used in the FFQ employed in this study are based on natural portion sizes (e.g., slices of bread) whenever possible and commonly used portion sizes (e.g., 1 cup rice) when a natural portion size was not obvious (Willett et al. 1985). Adolescent females may consume smaller portion sizes, leading to an overestimation of usual nutrient intakes. The use of portion sizes specific for the population being studied has been emphasized by some researchers (Block et al. 1986). In this study, however, energy intakes did not vary greatly between the two intake assessment methods (8.3 vs. 7.5 MJ for FFQ and 3d-WFR, respectively), suggesting that these three explanations do not satisfactorily account for the different estimates of folate intake as determined by the FFQ and the 3d-WFR. Perhaps the difference in predicted folate intake could be due to the fact that the FFQ may overestimate the consumption of folate-containing, but low energy dense foods such as fruits, vegetables and folate-enriched cereals. This could be due at least in part to the fact that a generic folate content is assigned to specific groups of fruits and vegetables in this FFQ. Because the folate content of foods, particularly within traditional food groupings, varies widely and depends on whether it is fresh, frozen, canned or cooked, generic assignment of a single folate value may lead to a systematic overestimation of the folate content of many individual foods.
). Similarly, for vitamin B-12, 8 d would be required under similar conditions. Participant compliance, however, declines with increasing days of recording. Further, completion of more than 3 d of records is not feasible in most epidemiologic studies.
reported a dose-response relationship with negative slope between RBC folate concentrations at 15 wk post-conception and NTD risk. In their study, RBC concentrations < 340 nmol/L were associated with a more than eightfold difference in NTD risk compared with RBC folate concentrations > 905 nmol/L. In this study, only 13% (15 of 82) of young women not consuming supplemental folic acid achieved a RBC folate concentration > 905 nmol/L, whereas 73% (16 of 22) of supplement users had a RBC folate content greater than this level. These data are consistent with those of Brown et al. (1997), which indicated that among 189 healthy women aged 22-35 y, RBC folate levels 
906 nmol/L were found primarily in women who took folic acid supplements.
). Although both methods appear to be useful in determining vitamin B-12 intake when supplement users are included, the poor correlation between vitamin B-12 intake determined from a FFQ and serum vitamin B-12 suggests that the FFQ is not valid for the determination of vitamin B-12 intake from food alone in this population. Finally, these findings indicate that markedly different conclusions about absolute folate and vitamin B-12 intakes may be obtained depending on the dietary methodology employed. Given that these estimates of intake in relation to blood indices or desirable health outcomes (e.g., reduced risk of a NTD-affected pregnancy or cardiovascular disease) are often used to develop public policy recommendations, one must be mindful that the dietary instrument used may over- or underestimate actual nutrient intake.
| |
FOOTNOTES |
|---|
Manuscript received 8 December 1997. Initial reviews completed 23 February 1998. Revision accepted 28 May 1998.
| |
LITERATURE CITED |
|---|
|
Abstract
Introduction
Methods
Results
Discussion
References
|
|---|
This article has been cited by other articles:
|
|
 |
|
  K. L. Sherwood, L. A. Houghton, V. Tarasuk, and D. L. O'Connor One-Third of Pregnant and Lactating Women May Not Be Meeting Their Folate Requirements from Diet Alone Based on Mandated Levels of Folic Acid Fortification J. Nutr., November 1, 2006; 136(11): 2820 - 2826. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. L. Sedjo, P. Inserra, M. Abrahamsen, R. B. Harris, D. J. Roe, S. Baldwin, and A. R. Giuliano Human Papillomavirus Persistence and Nutrients Involved in the Methylation Pathway among a Cohort of Young Women Cancer Epidemiol. Biomarkers Prev., April 1, 2002; 11(4): 353 - 359. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. C. McKinley, J.J. Strain, J. McPartlin, J. M. Scott, and H. McNulty Plasma Homocysteine Is Not Subject to Seasonal Variation Clin. Chem., August 1, 2001; 47(8): 1430 - 1436. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. B Rasmussen, L. Ovesen, I. Bulow, N. Knudsen, P. Laurberg, and H. Perrild Folate intake, lifestyle factors, and homocysteine concentrations in younger and older women Am. J. Clinical Nutrition, November 1, 2000; 72(5): 1156 - 1163. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
