Correspondence

Correction

Correction and Revision of Conclusions — Dietary Trends in the United States

N Engl J Med 1997; 337:1846-1848December 18, 1997

Article

To the Editor:

The article by Popkin et al. (Sept. 5, 1996, issue)1 on dietary trends presents a potentially misleading picture of fruit and vegetable intake in the United States. The implied serving sizes (grams of dietary intake in Table 4 of the article divided by numbers of servings in Table 3) range from 33.7 to 56.9 g (1.2 to 2.0 oz), lower than weights corresponding to recommended serving sizes (2 to 6 oz). In addition, the reported numbers of servings in Table 3 (4.3 to 4.8 for whites and 3.9 to 5.4 for blacks) generally exceed and run counter to those previously reported for the same survey (4.4 for whites and 3.9 for blacks).2 These differences are not easily explained by the information provided. For national estimates to be valid, sample weights should be used to adjust for nonresponse and for the different rates of selection.3 It appears that such weights were used in only one table. Furthermore, it appears that the authors did not use specialized software that adjusts for the clustered nature of the sample in the calculation of standard errors.3 Therefore, the standard errors reported may be too small and result in tests of significance indicating real differences when, in fact, the differences might be due to sampling variation. Finally, the use of a single 24-hour dietary recall to represent the distribution of intake is problematic. Because persons with extreme intakes on the recall day would receive higher or lower scores than their usual intake would merit, the proportions of both persons meeting the recommendations and those with very poor diets are likely to be overestimated in Table 2.

The measurement of fruit and vegetable intake is critical for monitoring dietary trends. The numbers of servings given in the article may represent an overstatement of both actual consumption and the proportion of the general population with good diets.

Blossom H. Patterson, M.A.
Susan M. Krebs-Smith, Ph.D.
Amy F. Subar, Ph.D.
National Cancer Institute, Bethesda, MD 20892-7354

3 References

1. 1

   Popkin BM, Siega-Riz AM, Haines PS. A comparison of dietary trends among racial and socioeconomic groups in the United States. N Engl J Med 1996;335:716-720
   Full Text | Web of Science | Medline

2. 2

   Krebs-Smith SM, Cook A, Subar AF, Cleveland L, Friday J. US adults' fruit and vegetable intakes, 1989 to 1991: a revised baseline for the Healthy People 2000 objective. Am J Public Health 1995;85:1623-1629
   CrossRef | Web of Science | Medline

3. 3

   Interagency Board for Nutrition Monitoring and Related Research. Nutrition monitoring in the United States: chartbook I: selected findings from the National Nutrition Monitoring and Related Research Program. Washington, D.C.: Government Printing Office, 1993.
   

Author/Editor Response

The authors reply:

To the Editor: While preparing our response to Patterson and colleagues, we identified an error in the program that created the Diet Quality Index used in our paper. The calculation of the overall Diet Quality Index scores was affected by this error, as were the estimates of the numbers of servings of fruits and vegetables and grains and legumes.

The corrected fruit, vegetable, and grain estimates and Diet Quality Index scores substantially change our conclusions. We concluded that “in 1965, there were large differences among groups in dietary quality, with whites of high socioeconomic status eating the least healthful diet, as measured by the index, and blacks of low socioeconomic status the most healthful. By the 1989–1991 survey, the diets of all groups had improved and were relatively similar.” The corrected results (Table 1Table 1Mean Diet Quality Index Scores and the Distribution of Individual Scores of <4 or >10, According to Socioeconomic Status and Race for Each Survey.) indicate that in 1965 the diets of whites of all socioeconomic groups and blacks of low socioeconomic status were similar. In 1989–1991, the differences among racial and socioeconomic groups remained small, except for blacks of high socioeconomic status, who had the highest-quality diets.

The primary outcome was the score (range, 0 to 16) on the Diet Quality Index, a composite of eight food-and-nutrient–based recommendations from the National Academy of Sciences. A score of 4 or less was considered to indicate a relatively more healthful diet, and a value of 10 or more a relatively less healthful diet. Our corrected results indicate that in 1965, 2.7 percent of whites of low socioeconomic status, 4.2 percent of blacks of low socioeconomic status, and 2.9 percent of whites of high socioeconomic status had mean scores of 4 or less. We incorrectly reported these percentages as 9.3 percent, 16.4 percent, and 4.7 percent, respectively. For the 1989–1991 survey, the correct percentages are 7.4 for whites of low socioeconomic status, 6.0 for blacks of low socioeconomic status, 10.7 for whites of high socioeconomic status, and 15.5 percent for blacks of high socioeconomic status. We incorrectly reported these percentages as 19.9 percent, 23.5 percent, 20.0 percent, and 2.7 percent, respectively.

The percentage of respondents with less healthful diets (scores of 10 or more) increased substantially in all groups after we corrected our findings (Table 1). Nonetheless, overall dietary quality still improved in all groups, from a mean index score of 8.3 in 1965 to 7.5 in 1989–1991. We incorrectly reported these scores as 7.4 and 6.4, respectively.

Corrected values for the number of servings of fruits and vegetables and grains and legumes are given in Table 2Table 2Intake of Dietary Components According to Socioeconomic Status and Race for Each Survey.. We previously stated: “The consumption of fruits and vegetables varied little over time, except for an increase among blacks of medium and high socioeconomic status.” Our corrected results indicate that the consumption of fruits and vegetables decreased among whites of middle and high socioeconomic status, decreased slightly among blacks of low and middle socioeconomic status, and increased among blacks of high socioeconomic status.

These errors do not affect the results of other studies that used the Diet Quality Index.1,2 The previous work done using the 1987 U.S. Department of Agriculture survey is correct. In applying the original Diet Quality Index program from 1987 to the 1965, 1977–1978, and 1989–1991 Department of Agriculture survey data, a line of coding was overlooked that assigned subjects with no intake of fruits, vegetables, grains, or legumes values of zero. Data for such respondents were mistakenly coded as missing, which caused the errors in our calculations of the mean numbers of servings for these foods, as well as the Diet Quality Index scores. The error had less effect on the results for grains and legumes, since most people consumed some of these foods daily. Other components of the Diet Quality Index were not affected by this error.

Barry M. Popkin, Ph.D.
Anna Maria Siega-Riz, Ph.D.
Pamela S. Haines, Ph.D.
University of North Carolina, Chapel Hill, NC 27514

2 References

1. 1

   Patterson RE, Haines PS, Popkin BM. Diet Quality Index: capturing a multidimensional behavior. J Am Diet Assoc 1994;94:57-64
   CrossRef | Web of Science | Medline

2. 2

   Patterson RE, Haines PS, Popkin BM. Health lifestyle patterns of U.S. adults. Prev Med 1994;23:453-460
   CrossRef | Web of Science | Medline

Author/Editor Response

The revised analyses and conclusions of Popkin et al. warrant a follow-up on my editorial about the implications of their analysis of U.S. dietary trends.1 The error in their original estimate of fruit and vegetable intakes apparently distorted the understanding of dietary trends between 1965 and 1989–1991 in two ways. The new finding that is most troubling from a public health perspective is that the proportion of the population meeting the guidelines for healthful diets, which was less than 25 percent in the original report, was overestimated, because those with the worst diets (in terms of fruit and vegetable intake on the survey day) had been inadvertently excluded. In the revised calculations, even though the situation did improve somewhat over time, the percentages with favorable dietary pattern scores were extremely low in all three periods. For example, in the 1989–1991 survey, only 1 in 10 high-income whites met the recommended dietary pattern on the day recalled.

Popkin et al. incorrectly reported that as compared with middle- or high-income groups, the least-advantaged populations (people with low incomes and black people) had significantly better dietary patterns initially and showed relatively less improvement over time. The revised data suggest similar, slightly upward trends across all groups in the percentage with favorable dietary-pattern scores. This result is more encouraging than the prior indication that things were worsening for the poor relative to the rich or for blacks relative to whites, but it is surely no cause for celebration. Now all groups are reported to be worse off than was previously thought. Most important, the desirable pattern — one in which the most disadvantaged would be gaining ground faster — has yet to be achieved.

Shiriki Kumanyika, Ph.D., M.P.H.
University of Illinois at Chicago, Chicago, IL 60612-7256

1 References

1. 1

   Kumanyika S. Improving our diet -- still a long way to go. N Engl J Med 1996;335:738-740
   Full Text | Web of Science | Medline

Citing Articles (8)

Citing Articles

1. 1

   Jamy D. Ard, Celette Sugg Skinner, Chuhe Chen, Mikel Aickin, Laura P. Svetkey. (2005) Informing Cancer Prevention Strategies for African Americans: The Relationship of African American Acculturation to Fruit, Vegetable, and Fat Intake. Journal of Behavioral Medicine 28:3, 239-247
   CrossRef

2. 2

   Kim M Gans, Gary J Burkholder, Patricia M Risica, Thomas M Lasater. (2003) Baseline fat-related dietary behaviors of white, Hispanic, and black participants in a cholesterol screening and education project in New England. Journal of the American Dietetic Association 103:6, 699-706
   CrossRef

3. 3

   Patricia G Moorman, Mary F Ricciuti, Robert C Millikan, Beth Newman. (2001) Vitamin supplement use and breast cancer in a North Carolina population. Public Health Nutrition 4:03, 821
   CrossRef

4. 4

   Bruce N. Ames. (2001) DNA damage from micronutrient deficiencies is likely to be a major cause of cancer. Mutation Research/Fundamental and Molecular Mechanisms of Mutagenesis 475:1-2, 7-20
   CrossRef

5. 5

   Anna H. Wu. (2000) Diet and breast carcinoma in multiethnic populations. Cancer 88:S5, 1239-1244
   CrossRef

6. 6

   BRUCE N. AMES. (1999) Micronutrient Deficiencies: A Major Cause of DNA Damage. Annals of the New York Academy of Sciences 889:1 CANCER PREVEN, 87-106
   CrossRef

7. 7

   Richard L. Nelson, Victoria Persky, Mary Turyk. (1999) Determination of factors responsible for the declining incidence of colorectal cancer. Diseases of the Colon & Rectum 42:6, 741-752
   CrossRef

8. 8

   Bruce N Ames. (1998) Micronutrients prevent cancer and delay aging. Toxicology Letters 102-103, 5-18
   CrossRef