C ROSS -S ECTIONAL S TUDIES
9.4.2 C ROSS -S ECTIONAL S TUDIES
Cross-sectional studies that examined the relationship between carbohydrate intake and adiposity are both much more numerous and consistent in their results than longitudinal studies. We identified 47 cross-sectional studies published since 1991 (Table 9.2) that evaluated the relationship between adiposity and total carbohydrate intake (38 studies), fiber intake (13) studies), or sugars and simple vs. complex carbohydrates (15 studies).
Significant negative associations were reported between total carbohydrate intake and BMI or some other measure of adiposity for 71% of 17 studies conducted
among adults. 23,31,47,59–68 Of the remaining five studies, two reported no significant association, 19,69 two reported mixed results, 31,61 and only one study reported a positive
association with BMI. 70 Of the 21 studies to examine the relationship between carbohydrate intake and adiposity in children, 7 found a significant negative
association 54,71–76 and 2 found a significant negative association among boys, but no significant association among girls. 77,78 The remainder of the studies found no sig-
nificant association between these two variables. 52,79–89 Two of the studies 60,64 that reported significant negative associations between adult BMI and total carbohydrate intake used nationally representative data sets (National Health and Nutrition Exam- ination Survey (NHANES) and CSFII). Twenty-nine of the studies examined pop- ulations outside the U.S. (in Europe, Japan, and Canada). All seven of the studies conducted among adults in the U.S. 31,47,60,61,64,66,67 found a significant negative asso-
ciation between total carbohydrate intake and adiposity, although one study 31 found no association when the dependent variable was continuous (as opposed to categor- ical). Among children, four of the seven studies conducted in the U.S. 54,74–76 found significant negative associations between BMI and carbohydrate intake. The 39 cross-sectional studies varied widely in the extent to which they controlled for potentially confounding variables, yet even after controlling for an array of potential confounders, significant and negative associations between total carbohydrate intake and adiposity in adults were still observed in several studies. Taken as a whole, these results establish that intakes of a relatively larger percent of calories from carbohy- drates are consistently associated with lower BMI, at least among adults. However,
Ludwig et al. 47 reported that the association between total carbohydrate intake and adiposity was weak and substantially attenuated when adjusted for fiber, suggesting
that fiber may explain most of the association among the cohort studied. Fewer studies have examined different types of carbohydrates and their associ-
ation with some measure of adiposity. We identified 13 cross-sectional studies that reported on the association between dietary fiber and adiposity, of which 8 reported
a significant negative association 47,65–67,72,74,84,90,91 and 1 reported a significant negative association among females but not males. 77 Only one study (conducted in Sweden)
reported a significant positive association. 70 These studies were conducted in Europe, the U.S., Canada, and Japan. Six were conducted among children, over half of which
found a significant negative association. 72,74,77,84 The studies among children, how- ever, were less consistent in their associations and the study designs tended to be
weaker. Outcome measures were not the same for all studies; they included both measured and self-reported BMI, as well as height for weight based on underwater
336 TABLE 9.2
Summary of Cross-Sectional Studies That Examined the Relationship between Carbohydrate Intake and Adiposity (in Descending Order by Sample Size, Adults First)
Association with Adiposity a Study
Study Characteristics
Subjects
Total CHO b Fiber
Sugars
Gonzalez (2000), 59 EPIC Study, Spain
14,374 men, 23,289 women, ages 29–69, diverse SES
Lewis (1992), 15 Nationwide Food Consumption Survey (NFCS), 30,770 adults, children over 4, nationally representative 1977–78
Added sugars Trichopulou (2002), 69 EPIC study, Greece
27,862 adults, ages 25–82, healthy volunteers
Kromhout (2001), 90 Seven Countries Study, Europe, Japan, and 12,763 men, middle aged, diverse SES
U.S., 1958 and 1964 Bolton-Smith (1994), 63 Scottish Health Heart Study,
5768 men, 5858 women, ages 25–65
(p value not
Total sugar
reported)
– Extrinsic sugar
Intrinsic sugar, lactose, starch Stam-Moraga (1999), 62 Belgian Interuniversity Research on
Functional F –
5837 men, 5243 women, ages 25–74, nationally
Nutrition and Health, 1979–1984
Total sugar, d strongest association Kennedy (2001), 64 Continuous Survey of Intake of Individuals
representative
9786 adults, nationally representative
(CSFII), 1994–96
ood Carboh
Yang (2003), 60 NHANES III
3754 men, 4074 women, ages 25–64, nationally
representative
Appleby (1998), 91 Oxford Vegetarian Study, 1980–84
1914 men, 3378 women, nonsmokers, about half nonmeat
eaters
Slattery (1992), 61 CARDIA, 5 U.S. cities
5115 adults, white and black, ages 18–30, diverse SES
ydrates
study baseline
Men
Carboh
Ludwig (1999), 47 CARDIA, 5 U.S. cities
2902 adults, ages 18–30, black and white, study endpoint
– Adults,
Gibson (1996), 92 Dietary and Nutritional Survey of British
1087 men, 1110 women, ages 16–64
(weak) 1986–87
ydrates and Obesity
High-sugar/high-fat
– High-sugar/low-fat consuming men c – Sugar among high-fat consumers Macdiarmid (1998), 29 1986–87, Britain
–/0 Total sugar males/females –/+ High-fat/sweet males/females Ruidavets (2002), 68 France
1239–1853 healthy, nondieting adults, ages 16–64
330 men, ages 45–64
Wamala (1997), 70 Stockholm Female Coronary Risk Study
300 healthy women, ages 30–65
+ Sucrose
Tucker (1992), 31 Western U.S.
205 adult females, mostly white, mean age = 34.6
– (categorical)
(continuous)
Nelson (1996), 66 Utah
203 men, mostly white, mean age = 41
Simple CHO – Complex CHO Alfieri (1995), 65 Ontario
150 adults, mostly women, ages 18–65
– R2 = 17%
Miller (1994), 67 Indiana
46 men, 32 women
Total sugars + Added sugars
TABLE 9.2 (continued)
Summary of Cross-Sectional Studies That Examined the Relationship between Carbohydrate Intake and Adiposity (in Descending Order by Sample Size, Adults First)
Association with Adiposity a Study
Study Characteristics
Subjects
Total CHO b Fiber
Sugars
Westerterp (1996), 23 The Netherlands
34 obese women, 34 nonobese women, ages 20–50, case-
control
Cox (1999), 19 England
41 lean and 35 obese
nondieting healthy adults
Lluch (2000), 142 Stanislas Family Study, France
1320 members of 387 families, ages 11–65
– Extrinsic sugar (both age groups for boys; 1 group for girls) Lewis (1992), 15 USDA NFCS
Gibson (1993), as reported by Hill (1995), 14 Britain
897 boys, 747 girls, ages 10–11 and 14–15
– Added sugars (g/kg of body weight) Davies (1997), 79 NDNS, U.K.
4682 children, ages 4–10
0 Functional F
1444 British children, ages 1.5–4.5
Bao (1996), 83 Bogalusa Heart Study, Louisiana
1419 children, age 10, 35% black, 65% white
Rodriguez-Artalejo (2002), 87 Spain
557 boys, 555 girls, ages 6–7
Guillaume (1998), 73 Belgium
955 children, ages 6–12
ood Carboh
Maffeis (2000), 78 Italy
278 boys, 252 girls, ages 7–11
– Males
Females
Stewart (1999), 80 FRESH, Maryland
468 children, grades 2–5 (mean age = 8.9)
0 ydrates
Carboh
Garaulet (2000), 77 Spain
192 girls, 139 boys, ages 14–18, Spanish
– Males
Males
ydrates and Obesity
Females
Females
Tucker (1997), 74 Utah
253 children, ages 9–10
– Association found only when no variables controlled for
Hanley (2000), 84 Canada
242 children, ages 2–19, native Canadians
0 – Fiber
Simple sugars
Starch Lee (2001), 54 Pennsylvania
192 white girls, age 5
(cross-sectional results from longitudinal study)
(g)
Gills (2002), 93 Canada
181 children, ages 4–16
+ Total sugar (g)
Dennison (1997), 89 New York
168 predominantly white children, 94 age 2, 74 age 5
(g)
Scaglioni (2000), 52 Italy
80 white boys, 67 white girls, age 5 (cross-sectional results from longitudinal study)
0 total sugar Maffeis (1998), 56 Italy
McGloin (2002), 94 Northern Ireland
66 boys, 48 girls, ages 6–8, predominantly white
112 white children, mean age = 8.7, Italian (cross-sectional results from longitudinal study)
Maffeis (1996), 81 Italy
82 white children, 30 obese, 52 nonobese
Atkin (2000), 86 Great Britain
39 boys, 38 girls, ages 1.5–4.5
Ortega (1995), 71 Spain
37 boys, 27 girls, ages 15–17, Spanish
Gazzaniga (1993), 76 Iowa
25 girls, 23 boys, ages 9–11
Bandini (1999), 75 Massachusetts
11 obese girls, 10 obese boys, 12 nonobese girls, 10
nonobese boys, ages 12–18 Case-control
Koivisto (1994), 82 Sweden
15 overweight children, 24 normal-weight children, ages
(g)
Sucrose
Case-control
TABLE 9.2 (continued)
Summary of Cross-Sectional Studies That Examined the Relationship between Carbohydrate Intake and Adiposity (in Descending Order by Sample Size, Adults First)
Association with Adiposity a Study
Study Characteristics
Subjects
Total CHO b Fiber
Sugars
Rocandio (2001), 72 Spain
16 boys, 16 girls, age 11
– Mean fiber
intake (g)
Francis (1999), 88 Texas
12 pairs white nonobese children (12 with obese mother,
12 with nonobese mother)
Note : CHO = carbohydrate; SES = socioeconomic status.
a Association with adiposity: – indicates an inverse and significant association with adiposity; + indicates a direct and significant association with adiposity; 0 indicates the association was not
significant; blank space indicates no association was reported. Findings were the same for all age, gender, and ethnic groups unless otherwise indicated. b All studies defined carbohydrate intake in terms of percent of total calories unless otherwise indicated by g (grams). c Characteristics of subgroup to which the results apply are indicated when they do not apply to all groups studied. d Definition of independent variable is indicated if different from that indicated in the column heading.
Functional F
ood Carboh
ydrates
Carbohydrates and Obesity
weighing and on skinfold measurements. Several studies controlled for numerous potential confounders. No study included a nationally representative sample from
the U.S., but the CARDIA study 47 did include blacks and whites from five U.S. cities. One study in Canada 65 found that fiber accounted for 17% of the variance in BMI, and another 47 found that dietary fiber accounted for a larger proportion of the variation in BMI than total carbohydrates, fats, or proteins. These studies provide strong evidence that a higher fiber intake is consistently associated with lower adiposity.
Fourteen cross-sectional studies were identified that examined the relationship between sugar or simple carbohydrates and adiposity. 14,15,29,62,63,66–68,70,82,84,92–94 Of
those that looked at total sugar or simple carbohydrates, two found significant negative associations with adiposity, 62,63 one found a negative association for males
but none for females, 29 and five found no significant association. 66–68,84,94 One found
a positive association, but only when intake was measured as an absolute value. 93 Of the seven that looked at added sugars or sucrose, three found a negative associ- ation, 14,15,63 two found a positive association, 67,70 and two found no significant asso- ciation with adiposity. 15,82 Three studies analyzed the relationship between starch 63,84
or complex carbohydrates 66 and adiposity. The former two found no significant association, whereas the latter found a negative association. Only one study looked at intrinsic sugar (sugar naturally present in foods as opposed to being added during
processing or preparation) and found no significant relationship with adiposity. 63 All of the studies were conducted in the U.S., Canada, or Europe. The only study of
children in the U.S. 15 examined a nationally representative data set (NFCS) and reported a negative association between added sugars and BMI, as determined by self-reported height and weight. Using the same data set, but including adults and
children, Lewis 15 found no association between added sugars and adiposity. The only study of native Canadian children found no association between total sugar intake and adiposity.
An additional study looked at intake of high-sweet and -fat foods in Britain and found that those with a high-sugar, low-fat intake had the lowest BMI of all groups. 92 Because sugar was also negatively associated with adiposity among high-fat con-
sumers, Gibson 92 concludes that sugar and fat have antagonistic rather than syner- gistic influences on adiposity, with sugar having a protective effect. Macdiarmid et
al. 29 found that high-fat/high-sweet foods were negatively associated with adiposity in males and positively associated with adiposity in females. These two studies suggest that the relationship between sugar intake and adiposity is a complex one, is not independent of the relative intake of other nutrients, and may have distinct impacts on adiposity among males and females.
When interpreting these results, it must be kept in mind that underreporting of intake can be particularly problematic with foods that are high in sugar (or fat), given that these foods are considered less desirable from a nutritional standpoint. 32
Macdiarmid et al., 29 for example, found that an observed negative association between the intake of high-fat and -sweet products was reversed (and became significant) when low-energy reporters were excluded from the analysis. Overweight
individuals may be the most likely to underreport these “forbidden” foods. 95 This
Functional Food Carbohydrates
may explain in part the conflicting results when examining the association between sugar intake and adiposity in studies using approaches that rely on self-reporting.
In summary, findings from cross-sectional studies suggest that both greater intakes of total sugar/simple carbohydrates (three negative associations and six
nonsignificant results,) and complex carbohydrates (one negative association, two nonsignificant) may be protective and no conclusion can be made regarding added sugars (three negative, two positive, and two nonsignificant associations). But, given the inconsistent definitions of sugar, wide variations in the number of potential confounders controlled for, and lack of data from a more recent nationally representative sample, as well as the mixed results, it is not possible to draw a firm conclusion regarding the relationship between intake of simple carbohydrates and obesity.