Self-reported rate of eating correlates with body mass index in 18-y-old Japanese women


OBJECTIVE: To examine associations between rate of eating and macronutrient and dietary fiber intake, and body mass index (BMI).

DESIGN: Cross-sectional study.

SUBJECTS: A total of 1695 18-y-old female Japanese dietetic students.

MEASUREMENTS: Macronutrient intake (protein, carbohydrate, and fat) and dietary fiber intake were assessed over a 1-month period with a validated, self-administered, diet history questionnaire. Body height and weight and rate of eating (according to five categories) were self-reported.

RESULTS: Among the nutrients examined, only dietary fiber intake weakly, but significantly, and negatively correlated with BMI in a multiple regression analysis. The rate of eating showed a significant and positive correlation with BMI. The mean BMI was higher by 2.2, 1.5, 1.0, and 0.5 kg/m2 in the ‘very fast’, ‘relatively fast’, ‘medium’, and ‘relatively slow’ groups, respectively, compared with the ‘very slow’ rate of eating group. This correlation remained evident after adjustment for nutrient intake.

CONCLUSIONS: Rate of eating showed a significant and positive correlation with BMI, whereas only dietary fiber intake showed a weak correlation with BMI.


Both rate of eating and macronutrient intake balance have long been of interest as factors that may contribute to obesity. Although several previous studies have explored potential associations, the results have been inconsistent.1,2,3,4,5,6,7,8,9,10,11,12,13,14,15 Most studies of rate of eating have examined the association using test meals and have compared the rate of eating between obese and nonobese subjects, or have observed weight change by decreasing the rate of eating in obese subjects.1,2,3 Few epidemiological studies have examined this topic to date.4,5 In contrast, several epidemiological studies have examined the role of differing macronutrients in the development of obesity. These results have varied across the studies and populations examined.6,7,8,9,10,11 Other studies have reported the importance of dietary fiber.12,13 As palatability differs between foods with different compositions of fat and dietary fiber,14,15 rate of eating may correlate with intake of these nutrients. Therefore, when dietary factors associated with obesity are examined, both nutrient intake and rate of eating should be considered. To our knowledge, no studies to date have considered both rate of eating and nutrient intake with respect to obesity. We, therefore, investigated associations between rate of eating and macronutrient and dietary fiber intake, and body mass index (BMI).

Subjects and methods

Survey method

Subjects were students who entered dietetic courses at 22 colleges and technical schools in Japan in April 1997 (n=2069).16 The survey was conducted during April 1997. A total of 2063 students (2017 women and 46 men) responded to the survey (response rate = 99.7%). Staff at each school checked the submitted questionnaires according to the survey protocol. When missing values and/or logical errors were identified, the staff asked the subjects to complete the questions again. The questionnaires were checked at least once by staff at each school and by staff at the survey center. Most surveys were completed by the end of May.


We used two questionnaires—a self-administered diet history questionnaire (DHQ)17,18 and a questionnaire on general lifestyle. The DHQ is a validated 16-page questionnaire that recalls dietary habits over a 1-month period. The general lifestyle questionnaire used was a four-page questionnaire designed for this survey. It addressed general lifestyle factors, including participation of sports club activities over the previous month, current dietary counseling, the period of intentional change in dietary habits, experience of dieting, and smoking status. Body height and weight were self-reported as part of the DHQ. The period of intentional change in dietary habits was asked with four categories, from ‘none’, ‘within 1 y’, ‘within 3 y’, to ‘more than 3 y ago’. When asking about experience of dieting, dieting was defined as at least 2 kg intentional reduction of body weight within 1 month. Smoking status was asked with three categories, that is, ‘nonsmoker’, ‘past smoker’, and ‘current smoker’.

Rate of eating was self-reported according to one of five qualitative categories—‘very slow’, ‘relatively slow’, ‘medium’, ‘relatively fast’, and ‘very fast’. We examined the validity of these rating categories in a subpopulation (n=222) using the rate of eating reported by a close friend as the standard. We asked subjects to categorize the rate of eating of three close friends and obtained 498 eligible responses. The percentage of exact and adjunct agreement was 46 and 47%, respectively, which indicated high levels of agreement between self- and friend-reported rates of eating.

Statistical analysis

For the purposes of statistical analysis, we selected female subjects aged 18 y (n=1744). Subjects considered likely to have severe under- or over-reported energy intake were excluded. This included subjects with a reported energy intake less than half the energy requirement for the lowest physical activity category (6276 kJ/day) according to the recommended dietary allowance for Japanese, sixth revised edition or a reported energy intake more than or equal to 1.5 times the energy requirement of the highest physical activity category (9623 kJ/day), respectively.19 We also excluded subjects with missing values in nondietary variables. The 1695 women met the criteria and were included in the present analyses.

Intakes of energy, macronutrients, and dietary fiber were computed using an ad hoc algorithm for DHQ. As alcohol intake was extremely low, that is, the mean was 0.6 g/day, alcohol was excluded from the present analyses. Validity was higher in energy-adjusted values than in crude values in most of the nutrients in DHQ17 as was observed in other several dietary assessment questionnaires with similar structures.20,21 Owing to the lower validity of crude values, energy-adjusted values were recommended to be used in analyses for nutrient–disease association.22 We therefore used both crude and energy-adjusted values in the present analyses. The energy density model, that is, percentage of energy intake (%E) for macronutrients and grams per 4184 kJ energy intake (g/4184 kJ) for dietary fiber, was used for energy adjustment.

The subjects with participation of sports club activities at least once per week were physically active. BMI (kg/m2) was computed as body weight (kg) divided by square of body height (m).

Before analyzing correlations between nutrient intake, rate of eating, and BMI, we computed correlation between nondietary variables, and rate of eating and BMI in order to find out confounders.

Then, we computed means and 95% confidence intervals of body height, body weight, BMI, intakes of energy, three macronutrients, and dietary fiber by rate of eating.

Next, we calculated partial regression coefficients and 95% confidence intervals for nutrient intake and rate of eating by multiple regression analysis with BMI as the dependent variable. We tested the association using three different models, that is, nutrient intakes only, rate of eating only, and both nutrient intakes and rate of eating as independent variables, respectively. In this analysis, nondietary variables with significant correlations with BMI were included in the models as covariates. Crude intake values were expressed as ‘kJ/day’ for macronutrients, and ‘10 g/day’ for dietary fiber. In the model with nutrient intake of energy density model, carbohydrate intake was not included in the model because of very high correlation with fat intake (Pearson's correlation coefficient=−0.94).

All analyses were carried out using SAS statistical software, version 6.12 (SAS Institute Inc., Cary, NC, USA). A P-value of less than 0.05 was considered statistically significant.


Table 1 shows the characteristics of the study population. The distribution of rate of eating differed significantly between two categories on experience of dieting (P<0.001). Mean BMI was significantly different both between two physical activity levels (P<0.01) and between two categories of experience of dieting (P<0.001). Neither did intentional dietary change, current dietary counseling, nor smoking status affect BMI.

Table 1 Characteristics of the study population, rate of eating, and BMI (n=1695)

Table 2 shows body height, body weight, BMI, and intakes of energy, macronutrients, and dietary fiber by rate of eating. Body weight and BMI increased steadily with the increase in the rate of eating. In contrast, dietary fiber intake adjusted for energy decreased steadily with the increase in the rate of eating.

Table 2 Body height and weight, and energy-providing nutrient and dietary fiber intakes (means and 95% confidence intervals) by self-reported rate of eating

Table 3 shows partial regression coefficients by multiple regression analysis with BMI as the dependent variable. As physical activity category and experience of dieting showed a significant association with BMI in Table 1, they were included in the models as covariates. In the model with nutrient intakes as independent variables, only dietary fiber significantly and negatively correlated with BMI (P<0.05). In the model with rates of eating as independent variables, the mean BMI steadily increased with the increase in the rate of eating. In the models with both nutrient intakes and rates of eating as independent variables, BMI steadily increased with the increase in the rate of eating, that is, 0.5, 1.0, 1.5, and 2.2 kg increase in ‘relatively slow’, ‘medium’, ‘relatively fast’, and ‘very fast’ groups compared to the ‘very slow’ group, respectively. Among the nutrients, only dietary fiber adjusted for energy attained the significance level (P<0.05). Additionally, energy intake did not significantly correlate with BMI, that is, Pearson's correlation coefficient was −0.01 and −0.04 without and with adjustment for rate of eating, respectively.

Table 3 Partial regression coefficients (95% confidence intervals) by multiple regression analysis expressing changes in BMI (kg/m2) for change in rate of eating and nutrient intakes with adjustment for physical activity category and experience of dieting


Although cross-sectional studies on nutrient intake and obesity have reported inconsistent results, the majority of studies conducted in Western populations have indicated a significant and positive association between fat intake and BMI.6,7,8 In contrast, two studies conducted in Korea and China in which fat intake was relatively low (mean was 24.7 and 24.8%E, respectively) did not report a significant association between fat intake and BMI.9,10 The mean fat intake, 29.4%E, in this study was higher than in those studies, but still lower than those in Western studies. There is no current explanation for the discrepancy in results seen between high and low fat intake populations. In view of the possibility of increased reporting bias for fat intake with greater degrees of obesity,7,8 the findings observed in Asian countries may be more accurate. However, in this population, energy intake did not significantly correlate with BMI category, and the highest energy intake was observed in the leanest BMI group (data not shown). As this was difficult to understand from physiology, under-reporting of intake in obese subjects apparently existed in this population as was seen in several Western populations.23,24,25 Considering this problem, the results of analysis using energy-adjusted values may be more reliable than those using crude values.

Dietary fiber showed a negative correlation with BMI, regardless of adjustment of rate of eating (P<0.05) (Table 3). A similar finding with respect to dietary fiber intake and BMI has been reported in previous studies.12,13 In this study, dietary fiber showed a strongly negative correlation with rate of eating (P<0.001) (Table 2). Considering the slight decrease in partial regression coefficient when rate of eating was adjusted for (Table 3), the association between dietary fiber intake and BMI may have two pathways, that is, a direct one and an indirect one through rate of eating. Further studies are needed in this issue.

Some experimental studies have compared rate of eating of test meals by obese and nonobese subjects.1,2,3 Results were inconsistent, and the number of subjects were generally small. In the present study, there were few very slow and very fast eaters—only 5% for both groups—and the rate of eating explained only 7% of the observed variation in BMI. This relatively narrow range of distribution for rate of eating and the small contribution to obesity suggests that it would be difficult to design experimental studies to determine an effect. To our knowledge, only two epidemiological studies have previously examined the association between rate of eating and obesity, with both reporting a positive association.4,5 The current study similarly observed a positive association between self-reported rate of eating and BMI.

If fast eating was recognized as an important risk factor for obesity, obese subjects would tend to eat foods more slowly and reported rate of eating would be confounded by BMI status. However, ‘eating slower’ was not ranked among 20 major methods of dieting by Japanese female college students.26 This indicates that reduction in eating rate as a weight-control measure is unlikely to be a serious source of bias in this population. Moreover, we included intentional dietary change into a regression analysis as the covariate in order to avoid this possible confounding.

We used a self-administered semiquantitative dietary assessment questionnaire for data collection.17,18 Since actual dietary habits were not observed, the results should be interpreted cautiously. In order to minimize data inaccuracy, we used a previously validated questionnaire. The Pearson's correlation coefficients for the 3-day diet records were 0.48–0.55 for the three macronutrients used in the study and 0.48 for energy. As previously mentioned, under-reporting of energy intake assessed with this questionnaire among obese subjects was obvious because of no significant correlation between energy intake and BMI. Therefore, the analysis for correlation between nutrient intake, at least expressed as a crude value, and BMI may be inappropriate.

We observed a high level of concordance between self-reported and friend-reported rate of eating. However, as their friends knew the subject's general body shape, this may have influenced the independence of each rating. The observed agreement might have therefore been higher than the true value.

We used BMI values calculated from self-reported body height and weight. Previous studies have shown that BMI calculated from self-reported body height and weight is highly correlated with measured BMI values, although underestimation is common.27,28 The studies suggest that BMI calculated from self-reported body height and weight is a reliable measure for use in correlation analysis.

In conclusion, we observed a statistically significant positive association between rate of eating and BMI in 18-y-old Japanese women. This finding was consistent regardless of adjustment for nutrient intake. Conversely, only dietary fiber intake, but not macronutrients, correlated significantly with BMI. Further follow-up of subjects to elucidate the impact of rate of eating on subsequent body mass would be of interest but is beyond the scope of this cross-sectional study.


  1. 1

    Hill SW, McCutcheon NB . Contributions of obesity, gender, hunger, food preference, and body size, bite speed, and rate of eating. Appetite 1984; 5: 73–83.

  2. 2

    Spiegel TA, Wadden TA, Foster GD . Objective measurement of eating rate during behavioral treatment of obesity. Behav Ther 1991; 22: 61–67.

  3. 3

    Barkeling B, Ekman S, Rossner S . Eating behaviour in obese and normal weight 11-year-old children. Int J Obes Relat Metab Disord 1992; 16: 355–360.

  4. 4

    Gerace TA, George VA . Predictors of weight increases over 7 years in fire fighters and paramedics. Prev Med 1996; 25: 593–600.

  5. 5

    Ortega RM, Redondo MR, Lopez-Sobaler AM, Quintas ME, Zamora MJ, Andres P, Encinas-Sotillos A . Associations between obesity, breakfast-time food habits and intake of energy and nutrients in a group of elderly Madrid residents. J Am Coll Nutr 1996; 15: 65–72.

  6. 6

    Lissner L, Heitmann BL . Dietary fat and obesity: evidence from epidemiology. Eur J Clin Nutr 1995; 49: 79–90.

  7. 7

    Seidell JC . Dietary fat and obesity: an epidemiologic perspective. Am J Clin Nutr 1998; 67(Suppl): 546S–550S.

  8. 8

    Willett WC . Is dietary fat a major determinant of body fat? Am J Clin Nutr 1998; 67(Suppl): 556S–562S.

  9. 9

    Na HB, Kim SR . A study of energy intake and body fat percent of college students in Seoul. J Commun Nutr 2000; 2: 105–109.

  10. 10

    Stookey JD . Energy density, energy intake and weight status in a large free-living sample of Chinese adults: exploring the underlying roles of fat, protein, carbohydrate, fiber and water intakes. Eur J Clin Nutr 2001; 55: 349–359.

  11. 11

    Trichopoulou A, Gnardellis C, Benetou P, Bamita C, Trichopoulos D . Lipid, protein and carbohydrate intake in relation to body mass index. Eur J Clin Nutr 2002; 56: 37–43.

  12. 12

    Stevens J . Does dietary fiber affect food intake and body weight? J Am Diet Assoc 1988; 88: 939–945.

  13. 13

    Ludwig DS, Pereira MA, Kroenke CH, Hilner JE, Van Horn L, Slattery ML, Jacobs Jr DR . Dietary fiber, weight gain, and cardiovascular disease risk factors in young adults. JAMA 1999; 282: 1539–1546.

  14. 14

    Duncan KH, Bacon JA, Weinsier RL . The effects of high and low energy diets on satiety, energy intake, and eating time of obese and nonobese subjects. Am J Clin Nutr 1993; 57: 763–767.

  15. 15

    Drewnowski A . Energy, density, palatability, and satiety: implications for weight control. Nutr Rev 1998; 56: 247–253.

  16. 16

    Sasaki S, Tsuji T, Katagiri A, Shimoda T for the Diets of the Fresh Students in Dietetic Courses Study Group. Association between the numbers of food items bought in convenience stores and nutrient and food-group intakes—a survey of first-year female college students taking dietetic courses. J Jon Soc Nutr Sci 2000; 53: 215–226 (in Japanese with English abstract).

  17. 17

    Sasaki S, Yanagibori R, Amano K . Self-administered diet history questionnaire developed for health education: a relative validation of the test-version by comparison with 3-day diet record in women. J Epidemiol 1998; 8: 203–215.

  18. 18

    Sasaki S, Ushio F, Amano K, Morihara M, Todoriki T, Uehara Y, Toyooka T . Serum biomarker-based validation of a self-administered diet history questionnaire for Japanese subjects. J Nutr Sci Vitaminol 2000; 46: 285–296.

  19. 19

    Ministry of Health and welfare. Recommended dietary allowance for Japanese: dietary reference intakes, 6th revised edn. Ministry of Health and Welfare; 1999 (in Japanese).

  20. 20

    Block G, Woods M, Potosky A, Clifford C . Validation of a self-administered diet history questionnaire using multiple diet records. J Clin Epidemiol 1990; 43: 1327–1335.

  21. 21

    Rimm EB, Giovannucci EL, Stampfer MJ, Colditz GA, Litin LB, Willett WC . Reproducibility and validity of an expanded self-administered semiquantitative food frequency questionnaire among male health professionals. Am J Epidemiol 1992; 135: 1114–1126.

  22. 22

    Willett W, Stampfer MJ . Total energy intake: implications for epidemiologic analysis. Am J Epidemiol 1986; 124: 17–27.

  23. 23

    Braam LA, Ocke MC, Bueno-de-Mesquita HB, Seidell JC . Determinants of obesity-related underreporting of energy intake. Am J Epidemiol 1998; 147: 1081–1086.

  24. 24

    Trabulsi J, Schoeller DA . Evaluation of dietary assessment instru-ments against doubly labeled water, a biomarker of habitual energy intake. Am J Physiol Endocrinol Metab 2001; 281: E891–E899.

  25. 25

    Ferrari P, Slimani N, Ciampi A, Trichopoulou A, Naska A, Lauria C, Veglia F, Bueno-de-Mesquita HB, Ocke MC, Brustad M, Braaten T, Jose Tormo M, Amiano P, Mattisson I, Johansson G, Welch A, Davey G, Overvad K, Tjonneland A, Clavel-Chapelon F, Thiebaut A, Linseisen J, Boeing H, Hemon B, Riboli E . Evaluation of under- and overreporting of energy intake in the 24-hour diet recalls in the European Prospective Investigation into Cancer and Nutrition (EPIC). Public Health Nutr 2002; 5: 1329–1345.

  26. 26

    Miyagi S . Relationship between obesity level of female students and their dietary habits, health and body perception. Jpn J Nutr 1998; 56: 33–45 (in Japanese with English abstract).

  27. 27

    Goodman E, Hinden BR, Khandelwal S . Accuracy of teen and parental reports of obesity and body mass index. Pediatrics 2000; 106: 52–58.

  28. 28

    Kuczmarski MF, Kuczmarski RJ, Najjar M . Effects of age on validity of self-reported height, weight, and body mass index: findings from the third national health and nutrition examination survey, 1988–1994. J Am Diet Assoc 2001; 101: 28–34.

Download references

Author information

Correspondence to S Sasaki.

Rights and permissions

Reprints and Permissions

About this article


  • eating rate
  • macronutrient intake
  • dietary fiber intake
  • Japanese
  • women
  • epidemiology

Further reading