Original Communication

European Journal of Clinical Nutrition (2004) 58, 1580–1586. doi:10.1038/sj.ejcn.1601989 Published online 19 May 2004

Healthy Eating Index and obesity

X Guo1, B A Warden1, S Paeratakul2 and G A Bray2

  1. 1Constella Health Sciences, Constella Group, Inc., Durham, NC, USA
  2. 2Pennington Biomedical Research Center, Baton Rouge, LA, USA

Correspondence: X Guo, Constella Health Sciences, Constella Group, Inc., 2605 Meridian Parkway, Durham, NC 27713, USA. E-mail: xguo@constellagroup.com

Received 13 March 2003; Revised 11 February 2004; Accepted 13 February 2004; Published online 19 May 2004.

Top

Abstract

Background: There is a continuing need to examine the relationship between diet quality and health in the population. The Healthy Eating Index (HEI) has been developed as a composite measure of diet quality by the US Department of Agriculture.

Objectives: The first objective was to use the HEI to assess the diet quality of a representative sample of the US population and population groups. The second objective was to examine the association between HEI and obesity.

Design: Cross-sectional analysis of data from 10 930 adults who participated in the Third National Health and Nutrition Examination Survey. Sociodemographic, physical activity, dietary, and health data were used in the analysis. Diet quality was assessed with the HEI score, ranging from 0 to 100, based on 10 dietary criteria. A low HEI score indicates poor diet.

Results: A majority of survey participants had a low HEI score. The percentage of individuals classified as having a poor diet varied by age, gender, race/ethnicity, income, and education. A low HEI score was associated with overweight and obesity. There was a graded increase in the odds ratio of obesity across the HEI category after adjusting for age, gender, race/ethnicity, physical activity, smoking, alcohol use, income, and education.

Conclusions: An index of diet quality, such as HEI, may provide a comprehensive assessment of diet in the population. Since the HEI is based on the US Dietary Guidelines, the use of these guidelines as a way to improve health should be emphasized. However, the overall effectiveness of these guidelines in disease prevention needs to be studied further.

Sponsorship: None.

Keywords:

diet, diet quality, healthy eating index, obesity

Top

Introduction

One of the US national health objectives for the year 2010 is to reduce the prevalence of obesity among adults to less than 15%. However, research indicates that the situation is worsening rather than improving (Flegal et al, 2002). Not surprisingly, eating too much and exercising too little exert a great influence on the development of overweight and obesity. Recent epidemiologic studies of diet and health outcomes including obesity have changed the focus to the overall diet quality and dietary pattern instead of single nutrients, such as dietary fat (Hu et al, 2000; Fung et al, 2001a, 2001b). This concept was emphasized in the Dietary Approach to Stop Hypertension (DASH) trial, where a diet rich in fruits, vegetables, and whole grains with only small amount of fat and meat has been shown to be effective in reducing blood pressure (Appel et al, 1997).

Two methods are commonly used in the study of diet quality and health. First, statistical techniques such as factor analysis and cluster analysis may be used to identify the common dietary patterns and then relate these patterns to the health outcome. Using this approach, a diet with a high intake of red meat, processed meat, and refined products has been associated with higher cardiovascular risk compared to a diet with high intake of fruits, vegetables, and whole grains (Hu et al, 2000; Fung et al, 2001a). In addition, a higher intake of fruits, vegetables, and whole grains was recently confirmed to be associated with smaller gains in body mass index (BMI) and waist circumference (Newby et al, 2003), and dietary 'meat' pattern was positively associated with BMI in a multiethnic group of women (Maskarine et al, 2000).

The second approach to assessing an individual's diet is to score it with a set of criteria to produce a composite index of diet quality. Examples include the Diet Quality Index (Patterson et al, 1994; Haines et al, 1999), the Diet Diversity Score (Kant et al, 1995), and the Healthy Eating Index (HEI) (Kennedy et al, 1995; McCullough et al, 2000a). Some studies have examined dietary intake patterns, which were then used to describe the effects of dietary quality on health outcomes (Kant et al, 1995; McCullough et al, 2000a, 2000b). However, based on these studies, the nutritional etiology of obesity remains unclear and controversial. A review of eating patterns and BMI found that dietary patterns defined by using indexes, factors, or cluster analysis were inconsistently related to BMI (Togo et al, 2001). Inconsistencies among these earlier studies have been attributed to variations in the: age of the populations studied, study protocols, dietary assessment methods, composition and range of dietary index, and genetic predisposition for obesity, as well as limited control for social, environmental and physical activity influences, which may have introduced confounding through misclassification of dietary variables. Further, most of these studies did not examine associations of dietary patterns on obesity separately among different age groups.

Our goal in undertaking a comprehensive analysis of dietary patterns in relation to overweight and obesity was to clarify some of the inconsistencies in the earlier studies. Therefore, in this study, we focused on the HEI, which has been developed by the US Department of Agriculture as a measure of diet quality. The HEI is constructed for monitoring dietary intake and nutrition promotion activities for the US population (Kennedy et al, 1995), and employs 10 criteria for evaluating diet quality with possible HEI scores ranging from 0 to 100 (see Table 1).


The HEI criteria are based on the Dietary Guidelines for Americans and the Food Guide Pyramid (USDA, 1992; USDHHS & USDA, 1995). Although these guidelines are aimed at improving health and reducing the risk of chronic diseases, the data supporting the overall effectiveness of these guidelines are sparse (McCullough et al, 2000a, 2000b). The background, rationale, and significance of focusing on the HEI as a measure of dietary quality has been documented in detail elsewhere (Bowman et al, 1998). The relationship between HEI and nutrient intake has been validated in a sample of 340 women, where a higher HEI score was associated with a higher plasma concentration of alpha-carotene, beta-carotene, beta-cryptoxanthin, lutein, and vitamin C (Hann et al, 2001). However, the HEI profile of the US population and population subgroups has not been adequately described. In addition, whether the HEI can be related to obesity has not been studied. In this paper, we examined the HEI profile of the US population and its subgroups. We hypothesized that a low HEI score was associated with overweight status, because the HEI included the most prominent components from the Dietary Guidelines for Americans and the Food Guide Pyramid, which were regarded as the promoting a healthy American population (USDA, 1992; USDHHS & USDA, 1995).

Top

Subjects and methods

The Third National Health and Nutrition Examination Survey (NHANES III) is a nationally representative survey of the US civilian population using a stratified, multistage, probability sample design. The details of survey design and methods are available elsewhere (CDC & NCHS, 1994). Briefly, the data were collected by using questionnaire and personal interview administered by health interviewers, and the mobile examination centers were used to obtain the anthropometric, clinical, and laboratory measurements. For NHANES III, 39 695 individuals were selected over 6 years. Of these, 33 994 (86%) were interviewed and 30 818 (78%) participated in the clinical examination. For this study, we focused on 16 046 individuals 20 to 75 y of age from the total of 30 818 individuals aged 2 months to greater than 80 y old. We excluded 3133 participants whose interviews were coded as unreliable. These included mostly individuals whose response to the interview was 'Don't Know' and to a lesser extent, obvious outliers identified from the schematic plots of energy intake, body weight, and height. We also excluded 1525 individuals with missing or incomplete data and 230 pregnant women. Our study sample consisted of 10 930 individuals.

Data for analysis included anthropometric and dietary intake data. Additional data included sociodemographic data (age, sex, race/ethnicity, household income, and education), physical activity, smoking status, and alcohol use. Body weight was measured using a beam balance scale and height was measured using a stadiometer. From these, BMI was calculated as weight in kilograms divided by height in meters squared (kg/m2). We defined three weight categories using the following recommended criteria (National Institutes of Health (NIH), 1998): normal weight (BMI of 18.5–24.9 kg/m2), overweight (BMI of 25–29.9 kg/m2), and obesity (BMI of 30 kg/m2 or greater). We excluded underweight individuals with BMI of 18.4 kg/m2 or lower (n=228) from the analysis.

Dietary intake data were obtained from 24-h dietary recall using an automated, interactive interview. From these data, the food and nutrient intake of each individual was calculated. Accordingly, the HEI score for each individual was computed from 10 equally weighted components that represent different dietary guidelines. Each component has a minimum score of 0 and a maximum score of 10. Individuals with intake between the minimum and maximum ranges were assigned scores proportionately (Bowman et al, 1998). The total HEI score ranges from 0 to 100.

We defined five racial/ethnic groups: non-Hispanic white (n=4284), non-Hispanic black (n=3073), Mexican Americans (n=3114), other Hispanics (n=297) and other non-Hispanics (n=162). We defined three income groups using the percentage of poverty, based on total household income adjusted for household size (CDC & NCHS, 1994; USDHHS, 1998): lower income (less than 185%), middle income (185–350%), and higher income (more than 350% of poverty). Educational attainment was reported as the actual number of years of schooling completed. This was categorized into two groups: lower (high school education or less) and higher (college education or more).

Physical activity was assessed by the self-reported frequency of eight specific leisure time activities during the previous month. These were summed together and used as a continuous variable in the multivariate analysis. Similarly, the frequency of alcohol consumption was used in the analysis as a continuous variable. Smoking was classified into two categories: smokers and non-smokers/ex-smokers.

Top

Statistical analysis

Appropriate sampling weights were used in the analysis to account for complex survey design, where sampling weight assigned to each individual was commonly defined as the inverse of the probability of being selected into the sample (CDC & NCHS, 1997). The mean HEI score was calculated, stratified by age group (20–40, 41–60, and 61–74 y), gender, race/ethnicity, income, and education level. We used the recommended HEI criteria (Bowman et al, 1998) to define diet quality as 'good' (a score of 81 or more), 'needs improvement' (a score between 51 and 80), and 'poor' (a score of 50 or less). The percentage of individuals in each of these HEI categories was calculated and compared across the sociodemographic groups.

The mean HEI score and the proportion of subjects in each HEI category among the normal weight, overweight, and obese subjects were calculated and compared. Comparisons of HEI categories across sociodemographic groups were performed by the linearization (Taylor series) variance estimation method (Wald's F tests) for continuous variables. Cochran–Mantel–Haenszel x2 tests were used for categorical variables. Age, gender, race/ethnicity, income, education, physical activity, smoking, and alcohol use were considered to be the confounders and were included in all models. Other potential confounding factors were marital status, residential area, and time of migration. To assess the effect of these factors, the odds ratio (OR) for overweight and obesity was calculated by adding these variables, one at a time, to the original model. Any variable that changed the OR by 10% or more compared to the original model was considered to be a confounder. A similar approach was used in the multivariate regression model. Marital status, residential area, and time of migration did not meet this criterion.

In addition, we fitted regression models by using quadratic terms for age, because age may not be linearly related to obesity. We checked for effect modification by creating interaction terms for age and HEI, sex and HEI, and age and sex. We then tested the interaction terms in the final model to see if the model fit improved. The final multivariate regression models were used to estimate the least-square mean (LSM) for HEI score, and OR for weight status, controlling for age, age squared, sex, race/ethnicity, income, education, physical activity, smoking, and alcohol use. In order to identify the effect of sex hormones on obesity, these same analyses were employed for men and women separately. All analyses were performed with SUDAAN, version 7.5 (Research Triangle Institute, Research Triangle Park, NC, USA) and SAS for WINDOWS software, version 8.2 (SAS Institute, Cary, NC, USA).

Top

Results

The mean HEI score and the proportion of subjects in each HEI category (good, needs improvement, and poor), stratified by sociodemographic groups, are presented in Table 2. The mean HEI score for the total sample was 63.2. As shown, the mean HEI score varied across sociodemographic groups, being lower in the younger age group, men, non-Hispanic blacks, and among individuals with lower income and education. About 18% of the total sample was classified as having a poor diet. The proportion of individuals in each HEI category also varied, with a higher percentage of individuals consuming a poor diet were found among younger age groups, men, non-Hispanic blacks, smokers, those with a lower income, and the less educated compared to other groups.


Table 3 compares the LSM for the HEI score and the percentage of individuals in each HEI category among the normal weight, overweight, and obese subjects after adjusted for the sociodemographic characteristics described above. The LSM for HEI score was significantly lower among obese individuals compared to the normal weight individuals in the total sample and in women. The proportion of individuals classified as having a good diet was also lower among the obese as compared to normal weight subjects. The proportion of individuals classified as having a poor diet was higher in the obese as compared to normal weight subjects, but the difference was not statistically significant.


The OR for occurrence of overweight and obesity according to the HEI category is shown in Table 4. These estimates were adjusted for age, race/ethnicity, physical activity, smoking, alcohol use, income, and education using the high HEI as the referent category (OR=1.0). In addition, the distribution of HEI score by weight groups is shown in Figure 1. With HEI score decrease, there was a significant increase in likelihood of being overweight in the total sample and in men, but this was not observed in women. The trend for increasing occurrence of overweight was not statistically significant either in the total sample or men. Similarly, there was a significant and graded increase in likelihood of obesity with descending HEI score in the total sample and in both men and women. A significant trend of toward in increased likelihood of obesity was observed in all groups. For example, the OR of obesity among men classified as having a poor diet was about twice that of men classified as having a good diet.

Figure 1.
Figure 1 - Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact help@nature.com or the author

Prevalence of overweight and obesity by HEI category and age.

Full figure and legend (26K)


Top

Discussion

Our results show that the diet quality as assessed by HEI varied according to sociodemographic factors such as age, gender, race/ethnicity, income, and education as well as lifestyle factors, such as smoking, alcohol use, and physical activity. The mean HEI score was higher among the older individuals, women, other Hispanics, other non-Hispanics, and individuals with higher income and education. Conversely, more of the younger age groups, men, non-Hispanic blacks, and individuals with low income and education were classified as having poor diet. The HEI-based diet quality in our study sample is comparable to that reported in the 1994–1996 Continuing Survey of Food Intakes by Individuals (Bowman et al, 1998). For example, only about 11% of the total sample was classified as having a good diet while about 72% had a diet that needs improvement.

The analysis of the relationship between HEI and obesity showed that the mean HEI score was significantly lower among the obese subjects compared to normal weight subjects in the total sample and in women. The percentage of overweight and obese individuals classified by HEI category (poor diet, diet that needs improvement, and good diet) was generally higher among individuals classified as having a poor diet or a diet that needs improvement (HEI scores of less than 50 and 80, respectively) compared to those with a good diet (HEI score of 81 or higher), but none of the differences was statistically significant.

A diet index such as the HEI provides a comprehensive assessment of diet quality although the number of criteria used in computation of the index limits it and it does not provide information about specific foods. While a high HEI score obviously indicates a good diet, whether a high score is also favorably associated with health outcomes, in this case obesity is not known. Our results show that a poor diet, as defined by the HEI based on the Dietary Guidelines for Americans and the Food Guide Pyramid, was associated with overweight and obesity even though the HEI is not obesity-specific, that is, neither the dietary guidelines nor the food guide Pyramid were specifically designed to prevent or reduce obesity.

An important question raised by this study was whether the HEI adequately captures the overall diet quality. Likewise, the effectiveness of the Dietary Guidelines for Americans in chronic disease prevention has also been questioned. Two recent studies of HEI and health outcomes in large cohorts of health professional men and women found little or no association between HEI and the risk of major chronic diseases (McCullough et al, 2000a, 2000b). However, these studies used the modified form of HEI (HEI-f) calculated from the food frequency questionnaire (FFQ), which does not provide an absolute value of intake, and did not include obesity as an outcome. In our study, the HEI score computed from 24-h dietary recall was associated with a significantly higher OR of obesity after controlling for confounding factors such as age, sex, race/ethnicity, physical activity, education, income, smoking, and alcohol use. The OR of being obese was nearly twice as high among individuals with a poor diet compared to those with a good diet, and in general, there was a graded increase in occurrences of obesity as the HEI score decreased. Our results are consistent with other studies of dietary patterns using factor analysis or cluster analysis, which showed that the 'Western' dietary pattern with high intake of red meat, processed meat, refined grains, and high-fat dairy products was associated with higher risk of coronary heart disease compared to the 'prudent' dietary pattern rich in fruits, vegetables, fish, poultry, and whole grains in both men and women (Hu et al, 2000; Fung et al, 2001a; Newby et al, 2003). However, studies of diet quality and obesity using either a diet index, factor analysis, or cluster analysis produced contradictory results (Togo et al, 2001). These earlier studies concluded that the heterogeneity of food intake patterns and the lack of a gold standard for the application of these techniques impeded the consistent finding of a relationship between food intake patterns and health outcomes. Using factor analysis, however, a recent study in Brazil has shown that the 'traditional' dietary pattern that relies mainly on rice and beans was associated with lower risk of obesity, but an association between the 'Western' or 'mixed' dietary patterns and obesity was not clear (Sichieri, 2002).

The major limitation of our study is the use of cross-sectional study design, which cannot provide evidence of a causal relationship between HEI and obesity. However, it is generally accepted that poor diet is associated with the subsequent development of several chronic diseases. In addition, we could not control for the genetic predisposition for obesity in our analyses. This is because we failed to find an optimal variable with enough sample size to represent participant's parent(s) weight status, especially among elderly. Another possible limitation of this study is the under-reporting of food intake by some overweight and obese individuals, a well-known difficulty in the population study of diet and obesity. Since the HEI score is a specific algorithm that represents a summary measure of dietary quality, and combines information on the amount and variety of foods as well as recommendations for consumption of specific food components (Kennedy et al, 1995), like multiple-day dietary recalls, the systematic error, such as that caused by diet under-reporting, could be alleviated when more dietary components were used. In our study, this effect is not a significant problem, because under-reporting of intake would only serve to underestimate the association between HEI and obesity.

In conclusion, our results suggest that HEI is associated with obesity in the US population. Since the HEI is based on the Dietary Guidelines for American, the use of these guidelines as a way to improve health should be emphasized. However, the overall effectiveness of these guidelines in disease prevention needs to be investigated further in prospective studies and among different populations, especially populations outside the US.

Top

References

  1. Appel LJ, Moore TJ, Obarzanek E, Vollmer WM, Svetkey LP, Sacks FM, Bray GA, Vogt TM, Cutler JA, Windhauser MM, Lin P-H, Karanja N, Simons-Morton D, McCullough M, Swain J, Steele P, Evans MA, Miller ER & Harsha DW (1997): The DASH Collaborative Research Group. A clinical trial of the effects of dietary pattern on blood pressure. N. Engl. J. Med. 336, 1117–1124. | Article | PubMed | ISI | ChemPort |
  2. Bowman SA, Lino M, Gerrior SA & Basiotis PP (1998): The Healthy Eating Index: 1994–96. US Department of Agriculture, Center for Nutrition Policy and Promotion. CNPP-5.
  3. Centers for Disease Control Prevention (CDC) & National Center for Health Statistic (NCHS) (1994): Plan and operation of the Third National Health and Nutrition Examination Survey, 1988–94. Vital and Health Statistics, Series 1, No. 32.
  4. Centers for Disease Control Prevention (CDC) & National Center for Health Statistics (NCHS) (1997): National Health and Nutrition Examination Survey. NHANES III Data CD-ROM series 11, No. 1. http://www.cdc.gov/nchs/about/major/nhanes/nh3data.htm).
  5. Flegal KM, Carroll MD, Ogden CL & Johnson CL (2002): Prevalence and trends in obesity among US adults, 1999–2000. JAMA 288, 1723–1727. | Article | PubMed | ISI |
  6. Fung TT, Willett WC, Stampfer MJ, Manson JE & Hu FB (2001a): Dietary patterns and the risk of coronary heart disease in women. Arch. Intern. Med. 161, 1857–1862. | Article | PubMed | ISI | ChemPort |
  7. Fung TT, Rimm EB, Spiegelman D, Rifai N, Tofler GH, Willett WC & Hu FB (2001b): Association between dietary patterns and plasma biomarkers of obesity and cardiovascular risk. Am. J. Clin. Nutr. 73, 61–67. | PubMed | ISI | ChemPort |
  8. Haines PS, Siega-Riz AM & Popkin BM (1999): The Diet Quality Index revised: a measurement instrument for populations. J. Am. Diet Assoc. 99, 697–704. | Article | PubMed | ISI | ChemPort |
  9. Hann CS, Rock CL, King I & Drewnowski A (2001): Validation of Healthy Eating Index with use of plasma biomarkers in a clinical sample of women. Am. J. Clin. Nutr. 74, 479–486. | PubMed |
  10. Hu FB, Rimm EB, Stampfer MJ, Ascherio A, Spiegelman D & Willett WC (2000): Prospective study of major dietary patterns and risk of coronary heart disease in men. Am. J. Clin. Nutr. 72, 912–921. | PubMed | ISI | ChemPort |
  11. Kant AK, Schatzkin A & Ziegler RG (1995): Dietary diversity and subsequent cause-specific mortality in the NHANES I epidemiologic follow-up study. J. Am. Coll. Nutr. 14, 233–238. | PubMed | ISI | ChemPort |
  12. Kennedy ET, Ohls J, Carlson S & Fleming K (1995): The Healthy Eating Index: design and applications. J. Am. Diet. Assoc. 95, 1103–1108. | Article | PubMed | ISI | ChemPort |
  13. Maskarine G, Novotny R & Tasaki K (2000): Dietary patterns are associated with body mass index in multiethnic women. J. Nutr. 130, 3068–3072. | PubMed | ISI | ChemPort |
  14. McCullough ML, Feskanich D, Rimm EB, Giovannucci EL, Ascherio A, Variyam JN, Spiegelman D, Stampfer MJ & Willett WC (2000a): Adherence to the Dietary Guidelines for Americans and risk of major chronic disease in men. Am. J. Clin. Nutr. 72, 1223–1231. | PubMed | ISI | ChemPort |
  15. McCullough ML, Feskanich D, Stampfer MJ, Rosner BA, Hu FB, Hunter DJ, Variyam JN, Colditz GA & Willett WC (2000b): Adherence to the Dietary Guidelines for Americans and risk of major chronic disease in women. Am. J. Clin. Nutr. 72, 1214–1222. | PubMed | ISI | ChemPort |
  16. National Institutes of Health (NIH) (1998): Clinical guidelines on the identification, evaluation, and treatment of overweight and obesity in adults—The evidence report. Obes. Res. 6 (Suppl 2), 51S–210S. | PubMed | ISI |
  17. Newby PK, Muller D, Hallfrisch J, Qiao N, Andres R & Tucker KL (2003): Dietary patterns and changes in body mass index and waist circumference in adults. Am. J. Clin. Nutr. 77, 1417–1425. | PubMed | ChemPort |
  18. Patterson RE, Haines PS & Popkin BM (1994): Diet quality index: capturing a multidimensional behavior. J. Am. Diet. Assoc. 94, 57–64. | Article | PubMed | ChemPort |
  19. Sichieri R (2002): Dietary patterns and their associations with obesity in the Brazilian city of Rio de Janeiro. Obes. Res. 10, 42–48. | PubMed |
  20. Togo P, Osler M, Sorensen TI & Heitmann BL (2001): Food intake patterns and body mass index in observational studies. Int. J. Obes. Relat. Metab. Disord. 25, 1741–1751. | Article | PubMed | ChemPort |
  21. US Department of Agriculture (USDA) (1992): The Food Guide Pyramid. Hyattsville, MD: Human Nutrition Information Service Publication HG 252.
  22. US Department of Health and Human Service (USDHHS) & US Department of Agriculture (USDA) (1995): Dietary Guidelines for Americans 4th ed. Washington, DC: US Government Printing Office.
  23. US Department of Health and Human Services (USDHHS) (1998): 1998 poverty guidelines. Federal Register 63, 9235–9238.
Top

Acknowledgements

We acknowledge the contributions of Ming Yin and Jean Orelien, who provided statistical consulting. None of us had any personal or financial affiliation with any organization involved in the study.

Top

MORE ARTICLES LIKE THIS

These links to content published by NPG are automatically generated

REVIEWS

Intramyocellular Lipid Content in Human Skeletal Muscle *

Obesity Perspective

Drug Insight: appetite suppressants

Nature Clinical Practice Gastroenterology & Hepatology Review (01 Feb 2005)

Mechanisms of Disease: nicotine???a review of its actions in the context of gastrointestinal disease

Nature Clinical Practice Gastroenterology & Hepatology Review (01 Nov 2005)

Extra navigation

.

natureevents

ADVERTISEMENT