![[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}
|
|
|
|
|
||
Department of Mathematics, Statistics and Computer Science, Bar Ilan University, Ramat Gan, Israel, and Gertner Institute for Epidemiology and Health Policy Research, Tel Hashomer, Israel;
* Biometry Research Group, Division of Cancer Prevention, National Cancer Institute, Bethesda, MD
Department of Statistics, Texas A&M University, College Station, TX;
** Applied Research Program, Division of Cancer Control and Population Sciences, National Cancer Institute, Bethesda, MD;
Nutritional Epidemiology Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, Bethesda, MD; and
International Agency for Research into Cancer, Lyon, France
2To whom correspondence should be addressed. E-mail: victor_kipnis{at}nih.gov.
 |
ABSTRACT |
|---|
 TOP ABSTRACT MATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
KEY WORDS: • biological markers • energy intake • nutrition assessment • questionnaires • reference values • underreporting • surveillance
A major purpose of dietary surveillance or monitoring is to evaluate dietary intake relative to some standard. Standards may be averages, around which the population’s intake should be distributed, or thresholds, above or below which the population’s intake should fall, but they are all established with regard to usual intake, generally defined as the long-run average daily intake. This is important because diets vary considerably from day to day. Nonetheless, the primary assessment method used in dietary surveillance is the 24-h dietary recall (24HR)3 (1), an instrument that inherently captures intake only a day at a time and thus yields an excessive amount of within-person variation. What is needed instead is an estimate of the distribution of usual food intake.
FFQs (2–6) are designed to capture usual intake, but there is no general agreement regarding their use, as evidenced by the spirited exchanges between scientists over the relative merits of FFQs vs. 24HRs in surveillance of dietary intake (7–15). For example, Liu (14) concluded that FFQs "may not be appropriate for comparisons of mean intakes among different populations and for estimation of nutrient intake distributions." Our paper sheds further light on this issue, and evaluates, in depth, the use of 24HRs.
In previous work on estimating the distribution of dietary intakes, investigators either presented the distribution of the reported values (16) or, in some studies where repeat 24HRs were used, adjusted the variance of the distribution to exclude within-person variation (17–19). Both approaches implicitly assume that the instrument is unbiased at the individual level, i.e., that the mean of a sufficient number of repeat observations on an individual will yield a person’s true usual intake value. The first approach also assumes that within-person variation is either negligibly small or can be ignored for other reasons. The second approach assumes that, in repeat administrations of the instrument, within-person random errors are independent of each other and of true intake levels, and these assumptions form the basis of the variance adjustment. We regard these conditions, unbiasedness and independence of within-person errors, as the requirements for a valid reference instrument for surveillance.
Recent evidence strongly suggests that common self-report instruments are unlikely to satisfy these requirements. Studies with the few biomarkers of dietary intake that do represent valid reference measurements ("reference" biomarkers), such as doubly labeled water (DLW) for energy and urinary nitrogen (UN) for protein, demonstrate serious biases in dietary self-report instruments (20–29). Such biases are likely to affect the usual intake distributions derived from these instruments.
In this paper, we use data from the Observing Protein and Energy Nutrition (OPEN) Study (28) to examine the performance of an FFQ and of various methods applied to 24HRs for estimating usual intake distributions of energy, protein, and protein density (30) by comparing their results with those derived from reference biomarkers for these nutrients. We also compare reported intakes of potassium with the urinary potassium biomarker.
Finally, we suggest and evaluate methods for adjusting the reported values from 24HRs when reference biomarker measurements for other nutrients are available, in particular, adjusting 24HR reported total energy intake by using the ratio of biomarker protein intake to 24HR reported protein intake.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
3 mo later, within the 2 wk prior to Visit 3. The 24HR, administered by an interviewer, was completed at Visit 1 and 3 mo later at Visit 3. The adopted FFQ was the Diet History Questionnaire developed at NCI (2–6). The 24HR was a highly standardized version that used the 5-pass method developed by the USDA for national dietary surveillance (1).
Participants received their dose of DLW at Visit 1 and returned 2 wk later (Visit 2) to complete the DLW assessment. Repeat DLW measurements were collected in 25 volunteers (14 men, 11 women) who received their second DLW dose at the end of Visit 2 and returned 2 wk later to complete the assessment. It was assumed that total energy expenditure, which is assessed by DLW, equals energy intake under conditions of stable body weight.
Participants provided two 24-h urine collections during the 2-wk period between Visit 1 and Visit 2, verified for completeness by the 3 para-amino benzoic acid check method (31). Since 81% of nitrogen intake is excreted through the urine (27,31) and nitrogen constitutes 16% of protein, the UN values were divided by 0.81 and multiplied by 6.25 to estimate individual protein intake. On the basis of several published studies comparing potassium intake with urinary excretion (see Discussion), the values for urinary potassium were divided by 80% to estimate the individual potassium intake. The study was approved by the NCI Special Studies Institutional Review Board for Human Subjects Research.
Estimation of the distribution of usual nutrient intake. The simplest approach to estimating the usual nutrient intake distribution is to use the reported intakes as if they represent true usual intake. When using an FFQ, the usual intake distribution is estimated by the histogram of the reported values. Using a 24HR, when there is only 1 administration of the instrument, investigators often use the mean of the sample to estimate the population mean usual intake and to track changes over time (16,32,33). When there are 2 or more administrations of a 24HR available, some investigators report the mean of the first assessment as the population mean but calculate the SDs and percentiles of the distribution from the mean reported intakes of each individual (16). We investigated the performance of this approach and referred to it as "traditional." We note that the approach is somewhat inconsistent because it adjusts the mean but not the percentiles for the often-observed tendency for repeat 24HR reports to yield lower intakes than the first report. The method does not remove within-person variation in the 24HR.
We also investigated the performance of 2 methods of estimating the distribution of usual intake that do adjust for within-person variation. The first, which we refer to as the National Research Council (NRC) method, was broadly outlined in 1986 (17).4 The other, known as the Iowa State University (ISU) method, is described in Nusser et al. (18). The 2 methods are conceptually similar but differ in details. Both methods require that a sizable random subgroup of the survey participants complete repeated assessments and are based on the assumption that the dietary instrument is unbiased for usual intake but contains random measurement error (34–37). Both methods assume that the reported nutrient intake is normally distributed in the population, after a suitable transformation. The main difference between the 2 methods is in the scale on which the reported intake is assumed unbiased, and the type of transformation used. The NRC method assumes that reported intake is unbiased on the transformed scale, whereas the ISU method assumes it is unbiased on the original untransformed scale. The NRC method uses, if necessary, a logarithmic, square root, or other power transformation as appears suitable for the data, while the ISU method uses a more complex, nonparametric method of transformation and is more suitable for nutrients with highly skewed distributions (18). Both methods assume that in a series of repeated administrations of the adopted instrument, the true intake is measured with the following: a) zero bias at the first administration and a bias thereafter that depends only on the timing of the administration but not on the individual, and b) random error that is independent of the true intake and independent of the error in the other repeats. The allowance of bias after the first administration is designed to accommodate for the often-observed decrease in mean dietary report at repeat administrations.
Based on these assumptions, both methods, after applying the transformation, i) calculate an adjusted intake for each individual that is "shrunk" toward the sample mean; ii) transform the shrunken value back to the original scale; and iii) then use the resulting values for plotting the histogram of the adjusted distribution or for calculating means, SDs, and percentiles.
The currently recommended approach for estimating usual intake distributions is to apply the NRC method or the more complex ISU method to 24HR measurements (17,38). As mentioned above, however, studies with reference biomarkers have demonstrated serious biases in 24HRs as well as in FFQs (20–29). Thus, it is unlikely that 24HR measurements satisfy the assumptions a and b, listed above, casting into question both current practice and recommendations for estimating usual intake distributions.
To examine these issues, we compared several methods for estimating the usual intake distributions of energy, protein, and protein density for which we had reference biomarker measurements. Because both DLW and adjusted UN are proper reference instruments for energy and protein intake, respectively, on the logarithmic scale, application of the NRC method to these biomarkers should produce unbiased estimates of the distributions of usual intake, and we considered these estimates to be our gold standard. We also studied potassium intake [and potassium density (g/MJ)], for which urinary potassium excretion may be a valid reference biomarker. The methods we compared were as follows:
For each method, we estimated the mean, SEM, SD, and the major percentiles of the usual intake distribution and the percentage of the population consuming less than the 5th or 10th percentile of the gold standard distribution or more than its 90th or 95th percentile. Distributions for men and women were estimated separately.
Adjustment of reported intake. As detailed in the Results section, for both the FFQ and 24HR, we found evidence of reporting bias in energy and protein intake but little bias for protein density. This suggested to us a simple adjustment of reported values of 1 nutrient when using the reported values and the biomarker of the other nutrient. For example, multiplying reported energy intake by the ratio of biomarker protein intake to reported protein intake should give an improved estimate of usual energy intake. Formally, we calculated for each individual as follows: adjusted energy = reported energy x (biomarker protein/reported protein). We then applied the NRC method4 to these adjusted energy values.
By using the same idea, we also adjusted protein by using energy misreporting; that is, by multiplying by the ratio of biomarker energy to reported energy (if we had a biomarker value of energy intake). We also adjusted potassium by using either energy or protein misreporting. Note, however, that all such adjustments perform well only if the relative error in reporting the 2 nutrients is approximately equal. Thus, we added to the 5 comparative methods described above:
It is understood that method 6 cannot be used for estimating the usual protein intake distribution, nor can method 7 be used for estimating the energy intake distribution. Note also that to apply method 6 requires that each survey participant provide 24-h urinary samples that are assessed for nitrogen. To apply method 7 requires that each participant is assessed for energy intake by using DLW, which currently would be prohibitively expensive.
Statistical analysis. When dealing with nutrient intakes adjusted for energy intake, we used nutrient densities (30) (rather than energy-adjusted residuals) calculated for protein as the percentage of total energy coming from protein and for potassium as intake divided by total energy intake (mg/MJ). Protein and potassium densities were calculated for each instrument by using the corresponding nutrient and energy intakes as measured by this instrument.5
For all dietary variables, we excluded extreme outlying values that fell outside the interval given by 25th percentile minus twice the interquartile range to 75th percentile plus twice the interquartile range on the logarithmic scale. The logarithmic transformation produced a reasonable approximation to normality for all the nutrients analyzed here and was used both to exclude outliers and with the NRC method. For each variable and each instrument, no more than 6 outlying values for men and 4 for women were excluded from the analyses.
All analyses were performed by using in-house programs written in SAS (version 8.2, 2001; SAS Institute).
|
RESULTS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
|
Underreporting was less severe with the 24HR. By using the traditional method (i.e., the first report), the mean reported energy intake was 8% and 12% lower than for the energy biomarker in men and women, respectively. The corresponding figures were 8% and 4% for protein intake. The different methods (traditional, NRC, and ISU) applied to the 24HR reports did not affect the mean values very much (e.g., for energy intake in males, the range for the mean across the 3 methods was 10,775–11,073, a 3% difference).
Regarding the variance, the FFQ grossly overestimated the SD for protein and energy. Using the mean of two 24HR reports also overestimated the SD, but application of the NRC or ISU method improved matters considerably.
We next estimated and compared the percentiles of the distributions on their original scale by using the different methods (energy and protein entries of Tables 2 and 3). There was gross disagreement between FFQ-based percentiles and biomarker-based percentiles for both protein and energy. All of the 24HR-based methods severely underestimated the energy percentiles, especially in the lower tail of the distribution. They also underestimated the percentiles in the lower tail of the protein distribution, although, with the ISU method, the underestimation was less severe. Graphical display of the distributions of usual energy intake estimated by different methods for men and women, respectively, confirmed this conclusion.6
|
|
As before, we found that these results played out in the estimation of the percentiles of the distribution (protein density entry of Tables 2, and 3). The protein density percentiles were estimated reasonably well either by the FFQ or by the 24HR-based NRC or ISU methods, although the 24HR methods tended to overestimate the lower percentiles, whereas the FFQ tended to overestimate the upper percentiles.
Biomarker adjustments to the 24HR. The estimated mean and SD obtained from the biomarker-adjusted 24HR agreed reasonably well with the biomarker-based values for energy and protein absolute intakes (energy and protein entries sections of Table 1). In particular, results were much improved after adjusting the 24HR absolute energy intake by the protein biomarker. The bias was almost entirely eliminated, and the SDs were close to those of the biomarker-based distribution. When we adjusted absolute protein intake by the energy biomarker, agreement with the protein biomarker-based results also appeared quite good (energy and protein entries of Tables 1, 2, 3).6
Percentages of individuals exceeding or failing to exceed certain cutoffs. For each method, we also compared the percentages of individuals falling below or above the 5th, 10th, 90th, and 95th percentiles of the biomarker distribution (Table 4). For a method that works well, values close to 5, 10, 10, and 5, respectively, would be expected. The results mirrored those of Tables 1, 2, 3. There were gross discrepancies in many cases, and the only methods that yielded reasonable agreement with biomarker-based results were as follows: for energy, the protein biomarker-adjusted 24HR; for protein, the energy biomarker-adjusted 24HR, with the ISU 24HR-based method as a possible second best; and for protein density, the FFQ or any 24HR-based method.
|
Comparison with the mean of the biomarker-based distribution of potassium density showed that both FFQ and 24HR overestimated the mean, with the FFQ performing worse (potassium density entry of Table 1). This overestimation was expected because absolute potassium intake was well estimated by the self-report instruments, whereas energy intake was underreported when using these instruments. Accordingly, none of the methods came close in estimating the potassium density percentiles (potassium density entry of Tables 2, 3, 4). The agreement improved by using an adjusted potassium density equal to the ISU 24HR potassium intake divided by the protein-corrected 24HR energy, although we do not present the results here.
|
DISCUSSION |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
For absolute protein and energy intakes, we found that the FFQ severely underestimated the mean and severely overestimated the SD. It therefore appears of little use in estimating usual intake distributions of absolute nutrients. This finding reinforced our previous conclusion [made in relation to the attenuation of relative risks in epidemiologic studies (27,29)] about the inadvisability of using an FFQ in studies of absolute intakes. The FFQ appeared to be better at measuring protein density than absolute protein, as previously claimed (30). Further, it is likely that other widely used FFQs would perform similarly, given results of comparisons of this instrument to the Block and Willett FFQs (5).
Our results indicated that using the first 24HR for estimating mean absolute intakes was more successful than using an FFQ but that some underestimation is still evident for energy (8–12%) and protein intake (4–8%). [These levels of underreporting are slightly lower than previous reports from the OPEN Study (28), which used the mean of two 24HRs rather than the first 24HR.] The 24HR instrument appeared to do better at estimating the SD, after the NRC or ISU method was applied; however, because of the underestimation problem, these methods did not reliably estimate percentiles of the distribution of usual intake of energy and also were not very satisfactory for protein. For example, the 10th percentiles of the distribution of actual energy intake for men and women were estimated as the 32nd and 35th percentiles, respectively, by the 24HR-based ISU method. These problems may have major public health implications, especially in the context of understanding what percentage of the population meets dietary recommended intakes. The question is: can anything better be achieved?
We suggested in this paper an adjustment method that may prove useful, especially for energy intake. Our adjustment to 24HR reported energy intake, based on assessments of UN on each individual, markedly improved the accuracy of the 24HR for estimating usual energy intake distributions. In the example of the previous paragraph, the 10th percentile of the distribution of absolute energy intake would now be estimated as the 8th percentile for men and the 17th percentile for women, using this protein-adjusted 24HR energy. The option of adjusting energy intake by using UN measurements thus deserves serious consideration, especially in view of the importance of energy intake to the current problems of obesity in the United States.
Before the adoption of such an approach, several questions need to be considered carefully. Is it necessary to take two 24-h urinary samples from all participants? Could the method of estimating the usual energy intake distribution be implemented if by design some participants were to submit only 1 sample, and others none? What compliance problems would be encountered, and what proportion of participants would submit satisfactory urine samples? What would be the cost of adding the extra UN assessments in terms of time, money, field staff, laboratory equipment, and personnel? Would these costs be justified by the increased quality of the information obtained? We do not attempt to answer these questions in this paper, but we argue that the results that we presented for estimating usual energy intake are good enough to require their serious consideration. We therefore advocate further research into these questions.
In contrast to the results for energy and protein intake, there appeared to be no systematic underreporting of absolute potassium intake when using the FFQ, nor when using the 24HR. There are 2 possible explanations for this. It may be that there is much less underreporting of potassium-containing foods compared with other energy-providing foods, or it may be that our urinary potassium biomarker did not capture the full intake of potassium. We will consider each of these possibilities in turn.
The possibility that some foods are underreported more than others has been proposed before (39–41) and seems intuitively reasonable, given that underreporting is a sociological/psychological phenomenon (22). Moreover, our finding that, among women, underreporting of energy (12%) appears to be about triple the level of underreporting of protein (4%) supports this possibility. Among the major sources of potassium in the U.S. diet, only beef and milk—accounting for <17%—are also major sources of protein and energy (42). Thus, there is not much overlap in the food sources of these constituents, and differential underreporting is entirely possible. Although urinary potassium excretion has not been studied as extensively as UN, there are, nevertheless, several published studies comparing potassium intake with urinary excretion (43–49). In a well-designed study, Mickelson et al. (43) report a stable proportion of (of 0.83) potassium intake excreted intourine in 20 normal males, although this proportion has shown considerable between-person variation (0.61–1.00) and between-study variation (0.72–0.87) in other reports. It is still not entirely clear whether this variation is due to methodological problems in measuring true potassium intake and collecting complete urinary output or because the proportion really does depend on individual characteristics or environmental conditions. Thus, although the evidence for the potassium marker points in a generally positive direction, there is still some doubt as to whether the proportion correction applied in our study (0.80) is in fact valid. If it was too large, then true potassium intake would be underestimated by the biomarker, which could create the false impression that potassium intake was not underreported on the FFQ or 24HR.
In summary, the potassium results may be interpreted in 1 of 2 ways. The simpler interpretation is to defer drawing conclusions from them until firmer evidence regarding the potassium biomarker is obtained. In that case, it seems possible that our results for energy might extend to other nutrients that are sufficiently correlated with protein intake and that a protein-biomarker adjustment of 24HR reported intakes of other nutrients might yield improved estimates of the percentiles of usual intake distributions for which there are currently no reliable biomarkers.
The more complex interpretation is to accept the potassium results and infer that there are substantial differences in the misreporting of different foods and, therefore, also nutrients. If this were the case, we would have to admit at least temporary ignorance of which nutrients would be estimated well by 24HRs and which would not. Our present knowledge would simply extend to placing protein and energy in the "bad" list and potassium in the "good" list. A strong interest in energy intake would still suggest including a protein biomarker as part of the survey.
It therefore is clear that we need to know more about the potassium biomarker before advocating the use of a protein-marker adjustment for estimating nutrient intakes other than energy. Larger controlled feeding studies linking potassium intake to potassium excretion are needed to establish more clearly the extent of individual variation in the intake/excretion ratio and its dependence, if any, on the level of potassium intake and other factors.
Finally, our data allowed cross-sectional estimates of the distribution of usual intakes but not distributions of longitudinal change. It is entirely possible that 24HR or FFQ will do a reasonable job of tracking relative changes in mean intake over time, even though poorly estimating the actual mean. Longitudinal biomarker studies will be necessary to check whether self-report instruments are adequate for this task.
|
FOOTNOTES |
|---|
3 Abbreviations used: DLW, doubly labeled water; ISU, Iowa State University; NCI, National Cancer Institute; NRC, National Research Council; OPEN, Observed Protein and Energy Nutrition; UN, urinary nitrogen; 24HR, 24-h recall.
4 Our implementation of the NRC method is included as a supplement to the online posting of this article at www.nutrition.org.
5 The convention we used for dealing with biomarker-based densities is included as a supplement to the online posting of this article at www.nutrition.org.
6 Graphical displays (Figs. 1–2) are available as a supplement to the online posting of this article at www. nutrition.org.
Manuscript received 30 June 2003. Initial review completed 21 October 2003. Revision accepted 13 April 2004.
|
LITERATURE CITED |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
1. Moshfegh, A. J., Raper, N., Ingwersen, I., Cleveland, L., Anand, J., Goldman, J. & LaComb, R. (2001) An improved approach to 24-hour dietary recall methodology. Ann. Nutr. Metab. 45(Suppl. 1):156 (abs.).
2. Subar, A. F., Thompson, F. E., Smith, A. F., Jobe, J. B., Ziegler, R. G., Potischman, N., Schatzkin, A., Hartman, A., Swanson, S. & Kruse, L. (1995) Improving food frequency questionnaires: a qualitative approach using cognitive interviewing. J. Am. Diet. Assoc. 95:781-788.[Medline]
3. Subar, A. F., Midthune, D., Kulldorff, M., Brown, C. C., Thompson, F. E., Kipnis, V. & Schatzkin, A. (2000) An evaluation of alternative approaches to assigning nutrient values to food groups in food frequency questionnaires. Am. J. Epidemiol. 152:279-286.
4. Subar, A. F., Ziegler, R. G., Thompson, F. E., Johnson, C. C., Weissfeld, J. L., Reding, D., Kavounis, K. H. & Hayes, R. B. (2001) Is shorter always better? Relative importance of dietary questionnaire length and cognitive ease on response rates and data quality. Am. J. Epidemiol. 153:404-409.
5. Subar, A. F., Thompson, F. E., Kipnis, V., Midthune, D., Hurwitz, P., McNutt, S., McIntosh, A. & Rosenfeld, S. (2001) Comparative validation of the Block, Willett, and National Cancer Institute food frequency questionnaires: The Eating at America’s Table Study (EATS). Am. J. Epidemiol. 154:1089-1099.
6. Thompson, F. E., Subar, A. F., Brown, C. C., Smith, A. F., Sharbaugh, C. O., Jobe, J. B., Mittl, B., Gibson, J. T. & Ziegler, R. G. (2002) Cognitive research enhances accuracy of food frequency questionnaire reports: results of an experimental validation study. J. Am. Diet. Assoc. 102:212-225.[Medline]
7. Sempos, C. T., Briefel, R. R. & Flegal, K. M. (1992) Factors involved in selecting a dietary survey methodology for national nutrition surveys. Austr. J. Nutr. Diet. 49:96-101.
8. Baghurst, K. I. (1992) The food frequency technique and its relevance to population surveys in Australia: a commentary. Austr. J. Nutr. Diet. 49:101-104.
9. Block, G. & Subar, A. F. (1992) Estimates of nutrient intake from a food frequency questionnaire: the 1987 National Health Interview Study. J. Am. Diet. Assoc. 92:969-977.[Medline]
10. Briefel, R. R., Flegal, K. M., Winn, D. M., Loria, C. M., Johnson, C. L. & Sempos, C. T. (1992) Assessing the nation’s diet: limitations of the food frequency questionnaire. J. Am. Diet. Assoc. 92:959-962.[Medline]
11. Hartman, A. M. & Block, G. S. (1992) Dietary assessment methods for macronutrients. Micazzi, M. S. Moon, T. E. eds. Macronutrients: Investigating Their Role in Cancer 1992:87-124 Marcel Dekker New York, NY. .
12. Rimm, E. B., Giovannucci, E. L., Stampfer, M. J., Colditz, G. A., Litin, L. B. & Willett, W. C. (1992) Reproducibility and validity of an expanded self-administered semiquantitative food frequency questionnaire among health professionals. Am. J. Epidemiol. 135:1114-1126.
13. Sempos, C. T. (1992) Some limitations of semiquantitative food frequency questionnaires. Am. J. Epidemiol. 135:1127-1132.
14. Liu, K. (1994) Statistical issues related to semiquantitative food-frequency questionnaires. Am. J. Clin. Nutr. 59(Suppl.):262S-265S.
15. Carroll, R. J., Freedman, L. & Hartman, A. (1996) The use of semiquantitative food frequency questionnaires to estimate the distribution of usual intake. Am. J. Epidemiol. 143:392-404.
16. US Department of Agriculture, Agricultural Research Service (1998) Food and Nutrient Intakes by Individuals in the United States, by Sex and Age, 1994–96 () Nationwide Food Surveys Report No. 96–2 Washington, DC.
17. National Research Council, Subcommittee on Criteria for Dietary Evaluation, Coordinating Committee on Evaluation of Food Consumption Surveys, Food and Nutrition Board (1986) Nutrient Adequacy: Assessment Using Food Consumption Surveys 1986 National Academy Press Washington, DC.
18. Nusser, S. M., Carriquiry, A. L., Dodd, K. W. & Fuller, W. A. (1996) A semiparametric transformation approach to estimating usual daily intake distributions. J. Am. Stat. Assoc. 91:1440-1449.
19. Guenther, P. M., Kott, P. K. & Carriquiry, A. L. (1997) Development of an approach for estimating usual nutrient intake distributions at the population level. J. Nutr. 127:1106-1112.
20. Bandini, L. G., Schoeller, D. A., Cyr, H. N. & Dietz, W. H. () Validity of reported energy intake in obese and nonobese adolescents. Am. J. Clin. Nutr. 52:421-425.
21. Livingstone, M.B.E., Prentice, A. M., Strain, J. J., Coward, W. A., Black, A. E., Barker, M. E., McKenna, P. G. & Whitehead, R. G. (1990) Accuracy of weighed dietary records in studies of diet and health. Br. Med. J. 300:708-712.
22. Heitmann, B. L. (1993) The influence of fatness, weight change, slimming history and other lifestyle variables on diet reporting in Danish men and women aged 35–65 years. Int. J. Obes. 17:329-336.[Medline]
23. Heitmann, B. L. & Lissner, L. (1995) Dietary underreporting by obese individuals: is it specific or non-specific?. Br. Med. J. 311:986-989.
24. Martin, L. J., Su, W., Jones, P. J., Lockwood, G. A., Tritchler, D. L. & Boyd, N. F. (1996) Comparison of energy intakes determined by food records and doubly labeled water in women participating in a dietary-intervention trial. Am. J. Clin. Nutr. 63:483-490.
25. Sawaya, A. L., Tucker, K., Tsay, R., Willett, W., Saltzman, E., Dallal, G. E. & Roberts, S. B. (1996) Evaluation of 4 methods for determining energy intake in young and older women: comparison with doubly labeled water measurements of total energy expenditure. Am. J. Clin. Nutr. 63:491-499.
26. Black, A. E., Bingham, S. A., Johansson, G. & Coward, W. A. (1997) Validation of dietary intakes of protein and energy against 24 hour urinary N and DLW energy expenditure in middle-aged women, retired men and post-obese subjects: comparisons with validation against presumed energy requirements. Eur. J. Clin. Nutr. 51:405-413.[Medline]
27. Kipnis, V., Midthune, D., Freedman, L. S., Bingham, S. A., Schatzkin, A., Subar, A. F. & Carroll, R. J. (2001) Empirical evidence of correlated biases in dietary assessment instruments and its implications. Am. J. Epidemiol. 153:394-403.
28. Subar, A. F., Kipnis, V., Troiano, R. P., Midthune, D., Schoeller, D. A., Bingham, S., Sharbaugh, C. O., Trabulsi, J., Runswick, S., Ballard-Barbash, R., Sunshine, J. & Schatzkin, A. (2003) Using intake biomarkers to evaluate the extent of dietary misreporting in a large sample of adults: The Observing Protein and Energy Nutrition (OPEN) Study. Am. J. Epidemiol. 158:1-13.
29. Kipnis, V., Subar, A. F., Midthune, D., Freedman, L. S., Ballard-Barbash, R., Troiano, R. P., Bingham, S., Schoeller, D. A., Schatzkin, A. & Carroll, R. J. (2003) Structure of dietary measurement error: results of the OPEN biomarker study. Am. J. Epidemiol. 158:14-21.
30. Willett, W. C. (1998) Implications of total energy intake for epidemiologic analysis. Nutritional Epidemiology 2nd ed. 1998:273-301 Oxford University Press New York, NY.
31. Bingham, S. A. & Cummings, J. H. (1985) Urine nitrogen as an independent validatory measure of dietary intake: a study of nitrogen balance in individuals consuming their normal diet. Am. J. Clin. Nutr. 42:1276-1289.
32. McDowell, M. A., Briefel, R. R. & Alaimo, D. (1994) Energy and macronutrient intakes of persons ages 2 months and over in the United States: Third National Health and Nutrition Examination Survey, Phase I, 1988–91. Advance Data from Vital and Health Statistics. Number 255 1994 National Center for Health Statistics Hyattsville, MD.
33. Alaimo, D., McDowell, M. A. & Briefel, R. R. (1994) Dietary intake of vitamins, minerals and fiber of persons ages 2 months and over in the United States: Third National Health and Nutrition Examination Survey, Phase I, 1988–91. Advance Data from Vital and Health Statistics. Number 258 1994 National Center for Health Statistics Hyattsville, MD.
34. Beaton, G. H., Milner, J., Corey, P., McGuire, V., Cousins, M., Stewart, E., de Ramos, M., Hewitt, D. & Grambsch, P. V. et al. (1979) Sources of variance in 24-hour dietary recall data: implications for nutrition study design and interpretation. Am. J. Clin. Nutr. 32:2546-2559.
35. Freudenheim, J. L. & Marshall, J. R. (1988) The problem of profound mismeasurement and the power of epidemiologic studies of diet and cancer. Nutr. Cancer 11:243-250.[Medline]
36. Freedman, L. S., Schatzkin, A. & Wax, J. (1990) The impact of dietary measurement error on planning a sample size required in a cohort study. Am. J. Epidemiol. 132:1185-1195.
37. Kaaks, R. & Riboli, E. (1997) Validation and calibration of dietary intake measurements in the EPIC project: methodological considerations. Int. J. Epidemiol. 26(Suppl.):S15-S25.
38. Institute of Medicine (2000) Dietary Reference Intakes: Applications in Dietary Assessment 2000 National Academy Press Washington, DC.
39. Cook, A., Pryer, J. & Shetty, P. (2000) The problem of accuracy in dietary surveys. Epidemiol. Community Health 55:611-616.
40. Krebs-Smith, S. M., Graubard, B. I., Kahle, L. L., Subar, A. F., Cleveland, L. E. & Ballard-Barbash, R. (2000) Low energy reporters vs others: a comparison of reported food intakes. Eur. J. Clin. Nutr. 54:281-287.[Medline]
41. Pryer, J. A., Vrijheid, M., Nichols, R., Kiggins, M. & Elliott, P. (1997) Who are ‘low energy reporters’ in the dietary and nutritional survey of British adults. Intl. J. Epidemiol. 26:146-154.
42. Subar, A. F., Krebs-Smith, S. M., Cook, A. & Kahle, L. L. (1998) Dietary sources of nutrients among US adults, 1989 to 1991. J. Am. Diet. Assoc. 98:537-547.[Medline]
43. Mickelson, O., Makdani, D., Gill, J. L. & Frank, R. L. (1997) Sodium and potassium intakes and excretions of normal men consuming sodium chloride or a 1:1 mixture of sodium and potassium chlorides. Am. J. Clin. Nutr. 30:2033-2040.
44. McDonald, J. T. & Margen, S. (1979) Wine versus ethanol in human nutrition II. Fluid, sodium and potassium balance. Am. J. Clin. Nutr. 32:817-822.
45. Smith, S. J., Markandu, N. D., Sagnella, G. A., Poston, L., Hilton, P. J. & MacGregor, G. A. (1983) Does potassium lower blood pressure by increasing sodium excretion? A metabolic study in patients with mild to moderate essential hypertension. J. Hypertens. 1(Suppl. 2):27-30.[Medline]
46. Holbrook, J. T., Patterson, K. Y., Bodner, J. E., Douglas, L. W., Veillon, C., Kelsay, J. L., Mertz, W. & Smith, J. C., Jr (1984) Sodium and potassium intake and balance in adults consuming self-selected diets. Am. J. Clin. Nutr. 40:786-793.
47. Clark, A. J. & Mossholder, S. (1986) Sodium and potassium intake measurements: dietary methodology problems. Am. J. Clin. Nutr. 43:470-476.
48. Deriaz, O., Theriault, G., Lavallee, N., Fournier, G., Nadeau, A. & Bouchard, C. (1991) Human resting energy expenditure in relation to dietary potassium. Am. J. Clin. Nutr. 54:628-634.
49. McCullough, M. L., Swain, J. F., Malarick, C. & Moore, T. J. (1991) Feasibility of outpatient electrolyte balance studies. J. Am. Coll. Nutr. 10:140-148.[Abstract]
This article has been cited by other articles:
|
|
 |
|
  N. Tasevska, S. A. Runswick, and S. A. Bingham Urinary Potassium Is as Reliable as Urinary Nitrogen for Use as a Recovery Biomarker in Dietary Studies of Free Living Individuals J. Nutr., May 1, 2006; 136(5): 1334 - 1340. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. L. Carriquiry and G. Camano-Garcia Evaluation of Dietary Intake Data Using the Tolerable Upper Intake Levels J. Nutr., February 1, 2006; 136(2): 507S - 513S. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. D. Nobmann, R. Ponce, C. Mattil, R. Devereux, B. Dyke, S. O. E. Ebbesson, S. Laston, J. MacCluer, D. Robbins, T. Romenesko, et al. Dietary Intakes Vary with Age among Eskimo Adults of Northwest Alaska in the GOCADAN Study, 2000-2003 J. Nutr., April 1, 2005; 135(4): 856 - 862. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Jahns, A. Carriquiry, L. Arab, T. A. Mroz, and B. M. Popkin Within- and Between-Person Variation in Nutrient Intakes of Russian and U.S. Children Differs by Sex and Age J. Nutr., November 1, 2004; 134(11): 3114 - 3120. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
