Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Eating patterns and dietary composition in relation to BMI in younger and older adults



To compare relative associations of eating patterns and dietary composition with body mass index (BMI) in younger (aged 20–59 years, n=1792) and older (aged 60–90 years, n=893) participants in the Continuing Survey of Food Intakes by Individuals, collected 1994–1996.


Data from two 24-h dietary recalls from individuals reporting physiologically plausible energy intake (within ±22% of predicted energy requirements, based on previously published methods) were used.


Mean reported energy intake was 96 and 95% of predicted energy requirements in younger and older subjects, respectively. Older subjects were less likely than younger subjects to skip a meal, but snacking was common in both age groups. Fiber density was significantly higher in the older group. A higher BMI in both age groups was associated with a higher total daily energy intake, and higher energy intakes at all eating occasions. In both age groups, eating frequency was positively associated with energy intake, and eating more than three times a day was associated with being overweight or obese. In the younger group but not the older group, a lower fiber density coupled with higher percentage of energy from fat was independently associated with having a higher BMI.


While no one eating occasion contributes more than any other to excess adiposity, eating more often than three times a day may play a role in overweight and obesity in both younger and older persons. A reduced satiety response to dietary fiber in addition to lower energy expenditure may potentially further contribute to weight gain in older persons.


Sixty-three percent of 20- to 59-year-old and 71% of 60- to 74-year-old US adults are overweight or obese.1 Typically, body fat doubles from young adulthood through middle age, and decreases after age 60–65 years.2, 3, 4 Dietary factors related to overeating and obesity, particularly at different life stages, are poorly understood. Dietary fat has long been presumed to be a leading cause of weight gain.5, 6 More recently, other dietary composition factors such as fiber and energy density,7, 8 and eating patterns such as eating frequency, snacking or skipping meals have been suggested to have a potentially important influence on energy regulation.9 However, the relative importance of these dietary factors is not known as few studies have examined eating patterns and dietary composition simultaneously in relation to body fatness.

In most studies in free-living adults of all age groups, a lower eating frequency has been associated with a higher body mass index (BMI).10, 11 Breakfast consumption has been suggested to be important for energy regulation;12, 13 however, while some studies have shown a lower percentage of total daily energy intake at breakfast associated with a higher BMI in younger12 and older14 persons, others have found no relationship.13, 15, 16, 17 Although snacking, usually defined as eating at occasions other than meals (e.g., breakfast, lunch or dinner), is commonly regarded as contributing to excess weight,18, 19 studies on snacking associated with BMI have also yielded mixed results.15, 20, 21 One potential explanation for the equivocal findings among studies is dietary reporting bias. As suggested previously,9, 10, 22 many of these studies were likely confounded by under-reporting of energy intake.23, 24 This under-reporting, primarily by overweight and obese persons, may be specific to energy-dense snack and dessert-type foods.25, 26, 27, 28 We recently reported that implausible energy intakes affect apparent eating patterns as well, being associated with lower reported meal and snack frequencies, energy densities and portions consumed.9, 10, 22

With aging comes a decline in the ability to regulate food intake. For example, older subjects failed to compensate at meals for energy preloads,29 and did not return to starting body weight after long-term overfeeding or underfeeding as did younger subjects.30, 31 These age-related changes in energy regulation are thought to be due to altered hormonal signals, such as cholecystokinin (CCK), glucagon-like peptide 1 (GLP-1) and peptide YY (PYY), and impaired senses of taste and smell.32, 33, 34 Potentially, changes in energy-regulating hormones during aging may result in changes in eating patterns that are associated with long-term body weight and fat changes.21, 35, 36, 37 Only a few studies, however, have specifically examined age-related changes in eating patterns.

We therefore conducted an analysis of eating patterns, dietary composition and their relative associations with BMI in younger and older adult participants in a US national survey. As it has been well established that overweight and obese individuals under-report energy intake by 30–50%,23, 24 and physiologically implausible reporters may contribute to inaccuracy in relationships between dietary factors and BMI,22, 27, 38, 39, 40, 41 we utilized our recently validated method for identifying implausible energy intake reports (i.e., those incompatible with long-term weight stability)22 and limited our analysis to only those subjects reporting physiologically plausible energy intakes. We hypothesized that older subjects would have weaker associations of dietary factors with BMI compared to younger subjects.

Subjects and methods


Data from the USDA Continuing Survey of Food Intake by Individuals (CSFII) collected in 1994–199642 were used. This survey of 16 103 non-institutionalized individuals aged 2–90 years residing in the US contains information on dietary intake (by one or two non-consecutive, multiple-pass 24-h recalls (day 1 in person; day 2 in person or by telephone)); socioeconomic, demographic and health parameters; and self-reported height and weight. From 8219 respondents aged 20 years, we excluded pregnant or lactating women, and individuals who were self-reported as food insecure (on welfare or ‘Meals on Wheels’, or did not have sufficient or suitable food), were on medically related diets, only completed one 24-h dietary recall, or did not provide height, weight or time of consumption for all eating occasions. From the remaining 6499 individuals, 41% (n=2685) were determined to have reported energy intakes (rEIs) that were physiologically plausible, as described below. We divided our final sample of plausible respondents into younger (20–59 years, n=1792) and older (60–90 years, n=893) groups, based on evidence that on average, BMI increases with age up to about 60 years and declines thereafter.4

Standardization of eating occasions

Subjects self-reported the type of eating occasion at which each food was consumed: breakfast, brunch, lunch, dinner, supper or snack. However, this resulted in non-standardized coding of eating occasions and multiple same meals consumed on the same day. Therefore, as in our previous studies,22, 39 meal coding was standardized so that not more than one each of breakfast, lunch and dinner, but multiple snacks were allowed, as follows.

If two or more meals of the same eating occasion were reported within 59 min of each other, they were considered one meal and combined, using the average of the consumption times. Otherwise, the occasion with the largest energy content was coded as a meal and the others as snacks. Brunch was coded as breakfast if it occurred before 1100 hours and lunch if it occurred between 1100 and 1600 hours. In some cases, lunch was missing, but dinner and supper were both reported (some individuals named the current conventional concept of lunch in the US, that is, the mid-day meal, dinner). If any dinner and any supper occurred more than an hour apart, and dinner had lower energy than supper, dinner was coded as lunch and supper coded as dinner. If more than one dinner was reported, a second dinner could not be recoded to lunch unless the first dinner was already recoded to lunch, to preserve the temporal order of meals. After recoding, if breakfast, lunch or dinner were missing, the meal was considered to have been skipped. Approximately 20% of eating occasions were recoded.

Data analysis

Determination of physiologically plausible reports

Physiologically plausible reports of dietary energy intake were determined by comparing rEI with predicted energy requirements (pERs) for a low active individual (physical activity level=1.4–1.6)43 using previously detailed procedures.22, 41 Briefly, we calculated age group-, sex- and weight class-specific cutoffs for rEI as a percentage of pER (rEI/pER × 100%), taking into account measurement and biological intra-individual variation in rEI, and total energy expenditure (TEE),41, 43, 44, 45 as measured by doubly labeled water and error in equations for pER. When different cutoff levels were tested (±2 s.d., ±1.5 s.d., ±1 s.d. and 0.05 increments of ±1 s.d. between ±1 and ±1.5 s.d. of rEI/pER × 100%), only cutoff values between ±1 and ±1.4 s.d. resulted in a relationship between rEI and body weight that did not differ significantly from the relationship between TEE and body weight.22 For the present analysis, we used the ±1 s.d. cutoff (±22% for these data) in order to maximize the validity of rEI. This required that an individual's rEI be within 78–122% of pER in order to be considered physiologically plausible.


BMI (kg/m2) was calculated using self-reported weight and height. Average dietary intakes over 2 days were used for all analyses. Energy intake and dietary composition variables, including percentage energy from fat, protein, and carbohydrate, fiber (g), fiber density (g/MJ) and energy density (kJ/g), were calculated as average daily values and by eating occasion (meals and snacks). Total daily energy density was calculated in two ways: (1) as total energy/total weight of all foods and beverages, and (2) as total energy/weight of all foods and beverages minus that of low-energy beverages, coffee and tea (pre-defined food groups in CSFII). For most analyses, nutrient intake at snacks was expressed as the average of all snacks consumed over a day, as snack frequency varied among individuals from 0 to 14/day. Eating pattern variables included total daily eating frequency and separate meal and snack frequencies, all as continuous variables. Eating frequency was also examined categorically as 3, 3.5–6 and >6/day. Meal skipping (yes/no) and snacking (yes/no) were also calculated according to whether they occurred on one or both days of recorded intake.


Descriptive demographic and dietary variables (mean±s.e.m.) were calculated, and independent t-tests and χ2-tests were used to compare characteristics between younger and older subjects. Energy and dietary composition variables across eating occasions were compared within- and between-age groups by analysis of variance (ANOVA), with Bonferroni adjustments for multiple comparisons. ANOVA was also used to compare energy intake distribution across eating occasions among normal weight (BMI <25 kg/m2), overweight (BMI 25–29.9 kg/m2) and obese (BMI 30 kg/m2) subjects within each age group.

Multiple regression analysis was conducted to determine associations between energy intake and BMI in each age group. Age group differences in these associations were tested for significance as explained above. Multiple regression analysis was also used to determine the relative associations of dietary composition and eating pattern variables with BMI in each age group. Macronutrients, fiber density, energy density, eating frequency (continuous and categorical), meal and snack frequency, meal skipping and snacking were considered as independent variables. Selected interactions between variables were also examined, including fiber density by the percentage of energy from fat (based on our previous analysis38), and eating frequency by dietary composition variables. To determine whether dietary composition at specific eating occasions was associated with BMI, additional models with dietary composition at each eating occasion entered simultaneously were also tested. To determine whether meal skipping was associated with increased snack frequency, models with meal skipping (yes/no) were tested with all three meals considered simultaneously. Finally, the association of eating frequency with energy intake was also tested.

All regression models were controlled for age, sex, education level (high school or less vs beyond high school), current smoking (yes/no), self-reported chronic disease (yes/no for at least one of the following: diabetes, hypertension, heart disease, cancer, hypercholesterolemia and/or stroke), ethnicity (white vs non-white), annual household income (0–130, 131–350 and >350% of poverty threshold), urbanicity (urban, suburban, rural), geographic region (northeast, midwest, south, west) and television viewing (h/day) (as a proxy for inactivity). In separate analyses, models were additionally controlled for low-energy containing beverages, coffee and tea. This was carried out because eating occasions could conceivably consist of only non-energy or low-energy containing beverages, and because our preliminary analyses indicated positive associations between these variables and BMI, thereby potentially confounding the association between eating frequency and BMI. Regression analyses were also conducted on the subset of individuals without self-reported chronic disease (n=1417 younger and 312 older subjects), with qualitatively similar results (data not shown). Regression analysis of eating frequency associated with energy intake was additionally controlled for height because there is a strong association between height and energy requirements independent of weight or BMI.43

The unweighted mean variation of rEI/pER × 100% (sample-specific) and weighted partial R2's from regression analyses were calculated using SAS v.8.2 (Cary, NC, USA). T-tests, R2 and χ2-tests, and linear regression analyses were performed using SUDAAN v.8 (Research Triangle Park, NC, USA) and weighted for sampling design, with alpha set at 0.05. SUDAAN was used to incorporate the CSFII sample design (stratified, multistage area probability sample) for variance estimation. Failure to account for sampling design is known to underestimate standard errors of parameters, hence increasing the risk of rejecting true null hypotheses. A P-value of 0.05 was considered significant.


Demographic and dietary characteristics of younger and older persons

Demographic characteristics are shown in Table 1. Mean BMI and prevalence of underweight or overweight and obesity did not differ between age groups. However, older subjects were less educated and less likely to smoke, and were more likely to be Caucasian, have a major disease and spend more time watching TV.

Table 1 Characteristics of younger and older adult participants in the Continuing Survey of Food Intakes by Individuals 1994–1996 reporting physiologically plausible energy intakesa

Dietary energy and composition are shown in Table 2. In both younger and older subjects, energy intake reporting plausibility (i.e., energy intake as a percentage of energy requirements) was similar and very high on average. Older subjects, however, reported a lower absolute energy intake than younger subjects, reflecting the lower energy requirements of older persons.43 Dietary composition differences between age groups were significant but tended to be very small on average. The exception was fiber density, which was much higher in older subjects compared to younger subjects.

Table 2 Eating patterns and dietary composition in younger and older adult participants in the CSFII 1994–1996 reporting physiologically plausible energy intakesa

Table 2 also shows the eating patterns of the two age groups. Older subjects consumed more meals and fewer snacks per day than did younger subjects (Table 2), so that no significant difference in total eating frequency existed between the two age groups. Meal skipping was common, with 44% of all subjects having skipped at least one meal on at least one of the two dietary recall days. Older subjects tended to skip fewer meals than younger subjects, particularly breakfast. In addition, both age groups skipped lunch more often than they skipped breakfast. Snacking was also very common in both groups, with the majority of subjects (>92%) having snacked on at least one recall day.

Meal skipping was associated with a higher snacking frequency in both age groups. In the younger group, skipping breakfast (β=0.20±0.09, P=0.03) and dinner (β=0.32±0.11, P<0.02), and in the older group, skipping breakfast (β=0.73±0.19, P<0.001) and lunch (β=0.33±0.11, P<0.01) were independently associated with increased snack frequency after controlling for socioeconomic and lifestyle factors. Because these associations implied that snacks were replacing meals (or vice versa), and because the designation of meals and snacks can be somewhat arbitrary,18, 46 we also calculated the distribution of eating frequency categories among younger and older subjects. The percentage of subjects who ate 3, 3.5–6 and >6 times a day, respectively, did not differ significantly between age groups, being 15, 75 and 10% in the younger group, and 14, 77 and 9% in the older group.

Energy and macronutrient intake by eating occasion are shown in Table 3. Energy intake at breakfast did not differ significantly between age groups, but the older group consumed less energy than the younger group during lunch, dinner and snacks. Both age groups consumed the highest proportion of their total energy intake at dinner and, while the younger group's lowest proportion of their total energy intake was consumed at breakfast, the older group's lowest proportion of their total energy intake was consumed in snacks. The macronutrient and fiber composition of meals and snacks also differed. In general, in both age groups, lunch and dinner were highest in percent energy from protein and fat and energy density, and dinner was highest in total fiber (g). However, while dinner was also highest in fiber density (g/MJ) for younger subjects, breakfast was highest in fiber density for older subjects. Snacks were the least fiber-dense eating occasions in both age groups.

Table 3 Energy intake and macronutrient composition by eating occasion in younger and older adult participants in the CSFII 1994–1996 reporting physiologically plausible energy intakes

Relationships of total energy, meal and snack energy and composition with BMI

Figure 1 shows energy intake at each eating occasion by weight class. In the younger age group, energy intake at all eating occasions was lower in normal weight compared to overweight and obese subjects. In older subjects, energy intake at breakfast was significantly lower in normal weight than overweight (but not obese) subjects. Also, energy intake in snacks was significantly lower in older normal weight and overweight subjects compared to older obese subjects. In both age groups, differences between weight classes no longer existed when energy intake was expressed as the percentage of total daily energy intake (data not shown).

Figure 1

Adjusted mean energy intakes at meals and aggregate snacks in normal weight (BMI<25 kg/m2), overweight (BMI 25–29.9 kg/m2) and obese (BMI30 kg/m2) younger (20–59 years) and older (60–90 years) subjects participating in the CSFII and reporting physiologically plausible energy intakes (see text for explanation). Bars with the same letter within an age group are not significantly different. n=1792 younger, 893 older.

Total energy intake, controlled for socioeconomic and lifestyle parameters, was positively associated with BMI in both age groups (younger: β=1.23±0.07, P<0.001; older: β=1.21±0.11, P<0.001), and when data from both age groups were pooled for analysis, these associations did not differ significantly. Energy intake accounted for 18% of the total variance in BMI in the younger group and 12% in the older group. When energy intakes at all meals and snacks were entered simultaneously in a model predicting BMI, all were independently and positively associated with BMI in both the younger and older groups (P<0.0001 for all). When associations of dietary composition (fiber density, percent energy from fat, their interaction and energy density) at breakfast, lunch, dinner and snacks were entered in a model simultaneously, only percent energy from fat at lunch was positively associated with BMI in the younger group. In the older group, none of the variables were associated.

Relationships of eating patterns and total dietary composition with BMI

Regression models for the relative associations of eating patterns and dietary composition with BMI are shown in Table 4. In the younger group, the best-fit model (Younger Model 1) consisted of fiber density, a fiber density by percent energy from fat interaction and eating frequency. Fiber density was inversely associated with BMI, and examination of the fiber density by percent energy from fat interaction indicated that a low fiber density coupled with a high percent energy from fat was positively associated with BMI. The main effect of percent energy from fat was not significant. In this model, dietary composition accounted for 2.9% of the between-subject variation in BMI, whereas eating frequency, which was positively associated with BMI, accounted for 0.4% of the between-subject variation in BMI. We found similar associations when meal and snack frequencies were entered (Younger Model 2) instead of eating frequency, but in this case only snack frequency was significantly associated with BMI (although meal frequency approached significance). A third model (Younger Model 3), with eating frequency as a categorical variable, is shown for comparison with the model obtained in older group (Older Model 1). In the older group, only when eating frequency was coded categorically was it significantly associated with BMI. Furthermore, none of the dietary composition variables that were significantly associated with BMI in the younger group were significant in the older group. In the older group, eating frequency accounted for approximately 1.3% of the between-subject variance in BMI, whereas in the younger group it accounted for approximately 0.3% (as a categorical variable). Other dietary composition and eating pattern variables, including snacking (yes/no) and meal skipping (yes/no), did not predict BMI in either group. In each of the models in the younger and older groups, all of the variables that were significant remained so even after controlling for intake (g) of low-energy beverages, coffee and tea. Furthermore, low-energy beverages themselves were independently associated with BMI in the younger group, and both low-energy beverages and coffee and tea were independently associated with BMI in the older group (β=0.0009–0.0025; P<0.0001–0.03).

Table 4 Models predicting BMI from dietary factors and eating patterns in younger and older adults reporting plausible energy intakes1

Mean BMI and energy intake in relation to eating frequency categories are shown in Figure 2, adjusted for socioeconomic and lifestyle parameters as well as the other dietary variables shown in Table 4. In general, both BMI and energy intake increased with increasing eating frequency in both age groups. In the younger group, BMI was significantly higher among subjects eating >6 times a day compared to those eating 6 times a day. In the older group, BMI was significantly higher among subjects eating >3 times a day compared to those eating 3 times a day. In both age groups, energy intake increased significantly among subjects in each increasing category of eating frequency.

Figure 2

Adjusted mean BMI and energy intake in relation to daily eating frequency (number per day) in younger (20–59 years) and older (60–90 years) subjects participating in the CSFII and reporting physiologically plausible energy intakes (see text for explanation). Eating frequency was categorized as 3, 3.5–6 and >6/day. The 0.5 increments in eating occasions is due to the averaging of 2 days of intake. Values are adjusted for sex, age, education level, current smoking, self-reported chronic disease, ethnicity, household income, urbanicity, geographic region and television viewing. Bars with the same letter within an age group are not significantly different. n=1792 younger, 893 older.


This is the first study to simultaneously examine eating patterns and dietary composition in relation to BMI in younger and older adults. We found that a higher total energy intake, and energy intake at all meals and snacks, rather than any particular eating occasion, was associated with a higher BMI in both age groups. In addition, a higher BMI in both age groups was associated with a higher eating frequency. Also, despite their more fiber-rich diets than younger subjects, BMI in older subjects was not associated with fiber density or the fiber density by percent energy from fat interaction as it was in younger subjects. Along with previous studies, our findings provide further evidence for increased energy dysregulation with aging, and suggest that older adults may have a somewhat reduced sensitivity to dietary factors that typically help regulate appetite in younger adults.

Age-related differences in physiology may help explain the small variations between younger and older adults in energy compensation and the associations of dietary factors with BMI observed in this study. Gastric emptying is delayed after meals in older subjects, increasing satiation.47 Older individuals also have a decreased olfactory sense and increased taste threshold48, 49 and may experience less intense hunger signals;29, 31, 47 therefore, they may be less sensitive to stimulation by the presence of food than younger persons. However, differences with age in hormonal mechanisms that regulate sensitivity to dietary factors are uncertain. Stimulation of satiety hormones such as CCK, GLP-1 and PYY in response to dietary factors may be altered as people age.49, 50 Some studies have reported that CCK increases more markedly in response to fat in older persons,51 but a heightened sensitivity to CCK with advanced age has not been clearly demonstrated.50 Likewise, basal concentrations of GLP-1 and PYY are comparable and increase similarly after fat infusion in younger and older subjects.51 Behavioral factors may also contribute to different eating patterns and an apparent reduced sensitivity to satiety cues with age. For example, older people may be less attentive to hunger or satiety cues than younger persons and instead follow certain eating patterns out of habit, concern for health or according to a schedule.52, 53 The lower activity of older persons may also contribute to a reduced appetite.54 Thus, the extent to which hormonal, physical and behavioral changes influence appetite in older people is still unclear.

We found that in younger adults, both dietary composition and eating frequency were independently associated with BMI. A higher fiber density was associated with a lower BMI. In addition, while percent energy from fat was not by itself associated with BMI, a high percent energy from fat coupled with a diet low in fiber density was positively associated with BMI. These findings are consistent with intervention studies showing an average of −1.9 kg weight loss over 3.8 months with the addition of 1–25 g/day of fiber to ad libitum diets,55 and greater weight loss with both a decrease in fat intake and an increase in fiber intake compared to either one alone.56 We also found that dietary composition may be more strongly related to BMI than eating frequency in younger adults, as dietary composition accounted for approximately six times the variance in BMI than did eating frequency (2.9 vs 0.4%). In older adults, however, eating frequency (categorically) was the only dietary factor that we examined that was associated with BMI. However, the relatively smaller sample size of the older group may have precluded finding dietary associations. Our eating frequency categorical models showed that in older adults, eating frequency accounted for about 1.3% of the variation in BMI, approximately four times that of younger adults. Both younger and older adults consumed more energy and had correspondingly higher BMIs with each increasing category of eating frequency. However, eating frequency may be a more critical factor in weight control in older adults, perhaps due to their lower energy requirements,43 whereas it may be only one of several factors in younger adults. These differences need to be substantiated in future studies.

The positive association of eating frequency with BMI that we observed in this study is in contrast to previous reports of either no relationship or an inverse relationship between eating frequency and BMI or percent body fat in free-living persons.10, 11, 57, 58, 59, 60, 61, 62, 63 Also contrary to previous reports showing breakfast skipping, reduced energy intake at breakfast and/or increased energy intake later in the day being associated with a higher BMI or energy intake,11, 12, 14, 64 we found no such relationships. Instead, total energy intake, or energy at all meals and snacks rather than any particular eating occasion, was associated with a higher BMI in both age groups. Some reasons that our findings conflict with those of previous studies are that we took into account all eating occasions at one time in our statistical models, and we excluded subjects with physiologically implausible reports of energy intake from our analysis.22

It has been demonstrated unequivocally that overweight and obese subjects are more likely to provide physiologically implausible reports of energy intake than normal weight subjects,23, 24, 41, 65, 66 typically under-reporting energy intake by 30–50%. Older persons may also under-report energy intake.67, 68, 69 In our previous analyses, we showed that including only plausible reports resulted in stronger relationships of both eating patterns and dietary composition with BMI in subjects across all age groups.22, 38, 39 Our studies also provided strong evidence that when implausible reporting occurs, whole eating occasions are omitted, and reported portion sizes and energy density of both meals and snacks are lower in the total sample compared to the sample limited to plausible reporters.22, 39 Thus, it is not surprising that we observed different relationships of both eating patterns and dietary composition with BMI in our study than have been previously reported, and we recommend that future epidemiological studies on dietary associations with BMI also identify and screen out implausible energy intake reports.

We observed that, in both younger and older persons, a higher BMI was associated with higher total daily energy intakes, and in a separate model, energy intake at all meals and snacks. We also found that increased eating frequency was associated with increases in energy intake and BMI. Skipping meals was not associated with BMI. Skipping meals was, however, associated with an increase in snacking frequency, which suggests that either snacks were replacing meals, or that, as suggested by others,18, 46 the designation of eating occasions as a meal or a snack is somewhat arbitrary. We considered the potential different influences of meal vs snack frequency on energy regulation, and found that a higher snack frequency was associated with a higher BMI in younger persons, whereas meal frequency was only marginally associated. These differential associations with BMI may be explained by the lower range of meal frequencies (0–3/day) vs snack frequencies (0–14/day) in our sample. Taken together, these data suggest that an individuals' energy intake may be achieved by several avenues, either a combination of a low eating frequency and higher portion consumed, or a higher eating frequency with a lower portion consumed, in any combination of meals and snacks. Our study provides strong evidence, however, that free-living subjects who eat more often, regardless of whether the eating occasions are designated as meals or snacks, have higher energy intakes. Therefore, higher eating frequencies may be associated with higher total daily energy intakes and hence over time, weight gain, unless cognitive efforts to reduce portion size are practiced consistently.

Interestingly, older subjects skipped meals less often than younger subjects, and meal frequency was slightly higher and snacking frequency slightly lower. This supports the notion that older persons may eat somewhat more regularly or have a tendency to eat according to a schedule, whereas younger persons may eat more haphazardly and perhaps schedule eating around other activities. In particular, breakfast skipping was much less common in the older group: only 5% of older subjects vs 22% of younger subjects skipped breakfast on either one or both recall days. The most frequently skipped meal by both age groups was lunch, with 27% of subjects skipping lunch on one or both days. Snacking was quite common among both older and younger persons, with over 92% of subjects snacking on at least one of the recall days, as has been observed by others.70, 71 However, as noted previously, neither meal skipping nor snacking per se, nor energy intake at any one eating occasion, was associated with BMI; rather, total eating frequency was the eating pattern that had the strongest positive association with energy intake and BMI in both age groups.

The major strength of this study is that when we excluded implausible reporters of energy intake, we obtained 95% reporting accuracy in a large national data set. However, there were some limitations. Our previous analysis in younger subjects suggests that there may be differences between men and women in dietary composition relationships with BMI.38 Nonetheless, in our preliminary analyses in older subjects, only minor differences between men and women were present. Therefore to maximize statistical power, particularly in the older group, which had a smaller sample size than the younger group, we elected to present our analysis with men and women combined, controlling for sex. Weight and height were self-reported. Our method of recoding meals may have misclassified meals as snacks or vice versa; however, as discussed in the methods section, it was necessary to implement a coding scheme that would allow us to interpret the eating pattern data, as self-reported eating occasions were originally not standardized across individuals. Regarding physical activity, we did not have a good measure but relied on TV viewing hours per day as a proxy of inactivity level. Finally, while the CSFII 1994–1996 may be considered somewhat ‘out of date’, it is a very rich data set with 2 days of dietary intake per participant, compared with more recent surveys that contain only 1 day of dietary intake per person or rely upon the less precise food frequency questionnaires to determine intake. It would be very useful, however, to analyze additional data sets with the same questions in mind as the present study had, so that consistency among findings from different surveys can be assessed.


Older people may be less sensitive to satiety cues than younger people, despite having somewhat more regular eating patterns and choosing a more fiber-dense diet. Our data suggest that a reduction in eating frequency may be one way to reduce excess energy intake in overweight and obese persons of all ages, while in younger persons additional advice on increasing dietary fiber and reducing dietary fat is warranted.

Accession codes




  1. 1

    Flegal KM, Carroll MD, Ogden CL, Johnson CL . Prevalence and trends in obesity among US adults, 1999–2000. JAMA 2002; 288: 1723–1727.

    Article  Google Scholar 

  2. 2

    Steen B . Body composition and aging. N Engl J Med 1988; 46: 45–51.

    CAS  Google Scholar 

  3. 3

    Shimokata H, Tobin JD, Muller DC, Elahi D, Coon PJ, Andres R . Studies in the distribution of body fat: I. Effects of age, sex, and obesity. J Gerontol 1989; 44: M66–M73.

    CAS  Article  Google Scholar 

  4. 4

    Roberts SB, Williamson DF . Causes of adult weight gain. J Nutr 2002; 132: 3824S–3825S.

    CAS  Article  Google Scholar 

  5. 5

    Roberts SB, Pi-Sunyer FX, Dreher M, Hahn R, Hill JO, Kleinman RE et al. Physiology of fat replacement and fat reduction: effects of dietary fat and fat substitutes on energy regulation. Nutr Rev 1998; 56: S29–S49.

    CAS  Article  Google Scholar 

  6. 6

    Willett WC . Dietary fat plays a major role in obesity: no. Obes Rev 2002; 3: 59–68.

    CAS  Article  Google Scholar 

  7. 7

    Roberts SB . High-glycemic index foods, hunger, and obesity: is there a connection? Nutr Rev 2000; 58: 163–170.

    CAS  Article  Google Scholar 

  8. 8

    Roberts SB, McCrory MA, Saltzman E . The influence of dietary composition on energy intake and body weight. J Am Coll Nutr 2002; 21: 140S–145S.

    Article  Google Scholar 

  9. 9

    Kirk RR . 2nd plenary session of ‘body weight balance and regulation’. Role of dietary carbohydrate and frequent eating in body-weight control. Proc Nutr Soc 2000; 59: 349–358.

    CAS  Article  Google Scholar 

  10. 10

    Bellisle F, McDevitt R, Prentice AM . Meal frequency and energy balance. Br J Nutr 1997; 77: S57–S70.

    CAS  Article  Google Scholar 

  11. 11

    Ma Y, Bertone ER, Stanek EJI, Reed GW, Hebert JR, Cohen NL et al. Association between eating patterns and obesity in a free-living US adult population. Am J Epidemiol 2003; 158: 85–92.

    Article  Google Scholar 

  12. 12

    Fricker J, Giroux S, Fumeron F, Apfelbaum M . Circadian rhythm of energy intake and corpulence status in adults. Int J Obes 1990; 14: 387–393.

    CAS  PubMed  Google Scholar 

  13. 13

    Cho S, Dietrich M, Brown CJP, Clark CA, Block G . The effect of breakfast type on total daily energy intake and body mass index: results from the Third national Health and Nutrition Examination Survey (NHANES III). J Am Coll Nutr 2003; 22: 296–302.

    Article  Google Scholar 

  14. 14

    Ortega RM, Redondo MR, Lopez-Sobaler AM, Quintas ME, Zamora MJ, Andres P et al. 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.

    CAS  Article  Google Scholar 

  15. 15

    Andersson I, Rossner S . Meal patterns in obese and normal weight men: the ‘Gustaf’ study. Eur J Clin Nutr 1996; 50: 639–646.

    CAS  Google Scholar 

  16. 16

    Summerbell CD, Moody RC, Shanks J, Stock MJ, Geissler C . Relationship between feeding pattern and body mass index in 220 free-living people in four age groups. Eur J Clin Nutr 1996; 50: 513–519.

    CAS  PubMed  Google Scholar 

  17. 17

    Song Y, Buring JE, Manson JE, Liu S . A prospective study of red meat consumption and type 2 diabetes in middle-aged and elderly women. Diabetes Care 2004; 27: 2108–2115.

    CAS  Article  Google Scholar 

  18. 18

    Drummond S, Crombie N, Kirk T . A critique of the effects of snacking on body weight status. Eur J Clin Nutr 1996; 50: 779–783.

    CAS  Google Scholar 

  19. 19

    Forslund HB, Torgerson J, Sjostrom L, Lindroos AK . Snacking frequency in relation to energy intake and food choices in obese men and women compared to a reference population. Int J Obes 2005; 29: 711–719.

    Article  Google Scholar 

  20. 20

    Basdevant A, Craplet C, Guy-Grand B . Snacking patterns in obese French women. Appetite 1993; 21: 17–23.

    CAS  Article  Google Scholar 

  21. 21

    Summerbell CD, Moody RC, Shanks J, Stock MJ, Geissler C . Sources of energy from meals versus snacks in 220 people in four age groups. Eur J Clin Nutr 1995; 49: 33–41.

    CAS  Google Scholar 

  22. 22

    Huang TT-K, Roberts SB, Howarth NC, McCrory MA . Effect of screening out implausible energy intake reports on relationships between diet and BMI. Obes Res 2005; 13: 1205–1217.

    Article  Google Scholar 

  23. 23

    Schoeller DA, Bandini LG, Dietz WH . Inaccuracies in self-reported intake identified by comparison with the doubly labeled water method. Can J Physiol Pharmacol 1990; 68: 941–949.

    CAS  Article  Google Scholar 

  24. 24

    Schoeller DA . Limitations in the assessment of dietary energy intake by self-report. Metabolism 1995; 44: 18–22.

    CAS  Article  Google Scholar 

  25. 25

    Bingham SA, Cassidy A, Cole TJ, Welch A, Runswick SA, Black AE et al. Validation of weighed records and other methods of dietary assessment using the 24 h urine nitrogen technique and other biological markers. Br J Nutr 1995; 73: 531–550.

    CAS  Article  Google Scholar 

  26. 26

    Hirvonen T, Mannisto S, Roos E, Pietinen P . Increasing prevalence of underreporting does not necessarily distort dietary surveys. Eur J Clin Nutr 1997; 51: 297–301.

    CAS  Article  Google Scholar 

  27. 27

    Lafay L, Basdevant A, Charles MA, Balkau B, Borys JM, Eschwege E et al. Determinants and nature of dietary underreporting in a free-living population: the Fleurbaix Laventie Ville Sante (FLVS) Study. Int J Obes Relat Metab Disord 1997; 21: 567–573.

    CAS  Article  Google Scholar 

  28. 28

    Poppitt SD, Swann D, Black AE, Prentice AM . Assessment of selective under-reporting of food intake by both obese and non-obese women in a metabolic facility. Int J Obes 1998; 22: 303–311.

    CAS  Article  Google Scholar 

  29. 29

    Rolls BJ, Dimeo KA, Shide DJ . Age-related impairments in the regulation of food intake. Am J Clin Nutr 1995; 62: 923–931.

    CAS  Article  Google Scholar 

  30. 30

    Roberts SB, Fuss P, Heyman MB, Evans WJ, Tsay R, Rasmussen H et al. Control of food intake in older men. JAMA 1994; 272: 1601–1606.

    CAS  Article  Google Scholar 

  31. 31

    Moriguti JC, Das SK, Saltzman E, Corrales A, McCrory MA, Greenberg AS et al. Effects of a 6-week hypocaloric diet on changes in body composition, hunger, and subsequent weight regain in healthy young and older adults. J Gerontol Biol Sci 2000; 55A: B580–B587.

    CAS  Article  Google Scholar 

  32. 32

    Morley JE . Anorexia of aging: physiologic and pathologic. Am J Clin Nutr 1997; 66: 760–773.

    CAS  Article  Google Scholar 

  33. 33

    MacIntosh CG, Morley JE, Chapman IM . The anorexia of aging. Nutrition 2000; 16: 983–995.

    CAS  Article  Google Scholar 

  34. 34

    Roberts SB, Hays NP . Regulation of Energy Intake in Old Age. Functional Neurobiology of Aging. Academic Press: New York, 2001, pp 829–837.

    Google Scholar 

  35. 35

    Drewnowski A, Henderson SA, Driscoll A, Rolls BJ . The dietary variety score: assessing diet quality in healthy young and older adults. J Am Diet Assoc 1997; 97: 266–271.

    CAS  Article  Google Scholar 

  36. 36

    Hjartaker A, Lund E . Relationship between dietary habits, age, lifestyle, and socio-economic status among adult Norwegian women. The Norwegian women and cancer study. Eur J Clin Nutr 1998; 52: 565–572.

    CAS  Article  Google Scholar 

  37. 37

    Roberts SB, Hajduk CL, Howarth NC, Russell R, McCrory MA . Dietary variety predicts low body mass index and inadequate macronutrient and micronutrient intakes in community-dwelling older adults. J Gerontol Biol Sci 2005; 60: 613–621.

    Article  Google Scholar 

  38. 38

    Howarth NC, Huang TT-K, Roberts SB, McCrory MA . Dietary fiber and fat associations with excess weight in young and middle-aged US adults. J Am Diet Assoc 2005; 105: 1365–1372.

    Article  Google Scholar 

  39. 39

    Huang TT-K, Howarth NC, Lin B-H, Roberts SB, McCrory MA . Energy intake and meal portions: associations with BMI percentile in US children. Obesity Res 2004; 12: 1875–1885.

    Article  Google Scholar 

  40. 40

    Macdiarmid JI . The sugar–fat relationship revisited: differences in consumption between men and women of varying BMI. Int J Obes 1998; 22: 1051–1061.

    Article  Google Scholar 

  41. 41

    McCrory MA, Hajduk CL, Roberts SB . Procedures for screening out inaccurate reports of dietary energy intake. Public Health Nutr 2002; 5: 873–882.

    Article  Google Scholar 

  42. 42

    US Dept of Agriculture ARS AFSRG. Data (CD-ROM) and documentation for the 1994–1996 Continuing Survey of Food Intakes by Individuals (CSFII)-Diet and Health Knowledge Survey. National Technical Information Service Accession No. PB98-500457, 1998.

  43. 43

    Institute of Medicine. Dietary Reference Intakes Energy, Carbohydrate, Fiber, Fat, Fatty Acids, Cholesterol, Protein, and Amino Acids, Part 1. Institute of Medicine: Washington, DC, 2002, pp 7-1–7-69.

  44. 44

    Goldberg GR, Black AE, Jebb SA, Cole TJ, Murgatroyd PR, Coward WA et al. Critical evaluation of energy intake data using fundamental principles of energy physiology: 1. Derivation of cut-off limits to identify under-recording. Eur J Clin Nutr 1991; 45: 569–581.

    CAS  PubMed  Google Scholar 

  45. 45

    Black AE . Critical evaluation of energy intake using the Goldberg cut-off for energy intake:basal metabolic rate. A practical guide to its calculation, use and limitations. Int J Obes 2000; 24: 1119–1130.

    CAS  Article  Google Scholar 

  46. 46

    Chamontin A, Pretzer G, Booth DA . Ambiguity of ‘snack’ in British usage. Appetite 2003; 41: 21–29.

    Article  Google Scholar 

  47. 47

    Clarkston WK, Pantano MM, Morley JE, Horowitz M, Littlefield JM, Burton FR . Evidence for the anorexia of aging: gastrointestinal transit and hunger in healthy elderly vs. young adults. Am J Physiol Regul Integr Comp Physiol 1997; 41: R243–R248.

    Article  Google Scholar 

  48. 48

    Schiffman S . Changes in Taste and Smell with Age space:Psychophysical Aspects. Raven Press: New York, 1979.

    Google Scholar 

  49. 49

    Morley JE . Anorexia, sarcopenia, and aging. Nutrition 2001; 17: 660–663.

    CAS  Article  Google Scholar 

  50. 50

    Horwitz BA, Blanton CA, McDonald RB . Physiologic determinants of the anorexia of aging: insights from animal studies. Annu Rev Nutr 2002; 22: 417–438.

    CAS  Article  Google Scholar 

  51. 51

    MacIntosh CG, Andrews JM, Jones KL, Wishart JM, Morris HA, Jansen JBMJ et al. Effects of age on concentrations of plasma cholecystokinin, glucagon-like peptide 1, and peptide YY and their relation to appetite and pyloric motility. Am J Clin Nutr 1999; 69: 999–1006.

    CAS  Article  Google Scholar 

  52. 52

    Rolls BJ . Aging and appetite. Nutr Rev 1992; 50: 422–426.

    CAS  Article  Google Scholar 

  53. 53

    De Castro J . Age-related changes in spontaneous food intake and hunger in humans. Appetite 1993; 21: 255–272.

    CAS  Article  Google Scholar 

  54. 54

    Evans WJ, Meredith CN . Exercise and nutrition in the elderly. In: Munro HN, Danforth DE (eds). Nutrition, Aging and the Elderly. Plenum Press: New York, 1989, pp 89–126.

    Google Scholar 

  55. 55

    Howarth NC, Saltzman E, Roberts SB . Dietary fiber and weight regulation. Nutr Rev 2001; 59: 129–139.

    CAS  Article  Google Scholar 

  56. 56

    Yao M, Roberts SB . Dietary energy density and weight regulation. Nutr Rev 2001; 59: 247–258.

    CAS  Article  Google Scholar 

  57. 57

    Fabry P, Fodor J, Hejl Z, Braun T, Zvolankova K . The frequency of meals its relation to overweight, hypercholesterolaemia, and decreased glucose-tolerance. Lancet 1964; 2: 614–615.

    CAS  Article  Google Scholar 

  58. 58

    Metzner HL, Lamphiear DE, Wheeler NC, Larkin FA . The relationship between frequency of eating and adiposity in adult men and women in the Tecumseh Community Health Study. Am J Clin Nutr 1977; 30: 712–715.

    CAS  Article  Google Scholar 

  59. 59

    Kant AK, Ballard-Barbash R, Schatzkin A . Evening eating and its relation to self-reported body weight and nutrient intake in women, CSFII 1985–86. J Am Coll Nutr 1995; 14: 358–363.

    CAS  Article  Google Scholar 

  60. 60

    Whybrow S, Kirk TR . Nutrient intakes and snacking frequency in female students. J Hum Nutr Diet 1997; 10: 237–244.

    Article  Google Scholar 

  61. 61

    Drummond SE, Crombie NE, Cursiter MC, Kirk TR . Evidence that eating frequency is inversely related to body weight status in male, but not female, non-obese adults reporting valid dietary intakes. Int J Obes 1998; 22: 105–112.

    CAS  Article  Google Scholar 

  62. 62

    Wahlqvist ML, Kouris-Blazos A, Wattanapenpaiboon N . The significance of eating patterns: an elderly Greek case study. Appetite 1999; 32: 23–32.

    CAS  Article  Google Scholar 

  63. 63

    Ruidavets JB, Bongard V, Bataille V, Gourdy P, Ferrieres J . Eating frequency and body fatness in middle-aged men. Int J Obes 2002; 26: 1476–1483.

    CAS  Article  Google Scholar 

  64. 64

    Keski-Rahkonen A, Kaprio J, Rissanen A, Rose RJ . Breakfast skipping and health-compromising behaviour in adolescents and adults. Eur J Clin Nutr 2003; 57: 842–853.

    CAS  Article  Google Scholar 

  65. 65

    Johnson RK, Soultanakis RP, Matthews DE . Literacy and body fatness are associated with underreporting of energy intake in US low-income women using the multiple-pass 24-h recall: a doubly labeled water study. J Am Diet Assoc 1998; 98: 1130–1140.

    Google Scholar 

  66. 66

    Livingstone MBE, Black AE . Markers of the validity of reported energy intake. J Nutr 2003; 133: 895S–920S.

    CAS  Article  Google Scholar 

  67. 67

    Luhrmann PM, Herbert BM, Neuhauser-Berthold M . Underreporting of energy intake in an elderly German population. Nutrition 2001; 17: 912–916.

    CAS  Article  Google Scholar 

  68. 68

    Tomoyasu NJ, Toth MJ, Poehlman ET . Misreporting of total energy intake in older men and women. J Am Geriatr Soc 1999; 47: 710–715.

    CAS  Article  Google Scholar 

  69. 69

    Tomoyasu NJ, Poehlman ET . Misreporting of total energy intake in older African Americans. Int J Obes Relat Metab Disord 2000; 24: 20–26.

    CAS  Article  Google Scholar 

  70. 70

    Zizza C, Siega-Riz AM, Popkin BM . Significant increase in young adults' snacking between 1977–1978 and 1994–1996 represents a cause for concern! Prev Med 2001; 32: 303–310.

    CAS  Article  Google Scholar 

  71. 71

    Cross AT, Babicz D, Cushman LF . Snacking habits of senior Americans. J Nutr Elderly 1995; 14: 27–38.

    CAS  Article  Google Scholar 

Download references


This study was supported in part by USDA/ERS/FANRP Grant No. 43-3AEM-2-80088. Contents of this publication do not necessarily reflect the views or policies of the US Department of Agriculture. We thank Vanessa Howarth for technical assistance, Johanna Dwyer and Lynne Ausman for their helpful comments on the manuscript and Gerard Dallal for statistical help.

Author information



Corresponding author

Correspondence to M A McCrory.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Howarth, N., Huang, TK., Roberts, S. et al. Eating patterns and dietary composition in relation to BMI in younger and older adults. Int J Obes 31, 675–684 (2007).

Download citation


  • energy intake
  • eating frequency
  • meal skipping
  • snacking
  • dietary fiber
  • dietary misreporting

Further reading


Quick links