Académique Documents
Professionnel Documents
Culture Documents
ABSTRACT INTRODUCTION
Background: There is strong epidemiologic evidence that dietary Overweight and obesity are substantial public health challenges. A
fiber intake is protective against overweight and obesity; however,
1514 Am J Clin Nutr 2017;106:1514–28. Printed in USA. Ó 2017 American Society for Nutrition
assessed the between-study heterogeneity of soluble fiber–in- from the Cochrane Library) (Figure 1). After duplicate abstracts
tervention effects using the I2 statistic (44). The following I2 were removed, a total of 7013 abstracts remained, in which 6824
interpretive categories were used: ,30% was considered low were excluded on the basis of the article title and abstract
heterogeneity, 30–75% was considered moderate heterogeneity, screening. We included 189 articles in the full-text review dur-
and .75% was deemed considerable heterogeneity (38, 44). A ing which 174 articles were excluded for not meeting study-
random-effects model was used as the default method to esti- selection eligibility criteria. In the excluded articles, 30 studies
mate pooled effects and 95% CIs. This model was used instead were ,1 wk in duration (47–76); 3 studies included a duplicate
of a fixed-effects model because of the observed variation in study population as had been evaluated in other articles (77–79);
study treatments and protocols (38). We conducted meta- 14 studies provided ineligible study treatments (78, 80–90); 23
regressions to assess outcomes in relation to the following 3 studies provided dietary fiber to both treatment and placebo
prespecified factors: intervention duration, daily dose of soluble groups (91–113); 2 studies only included normal-weight adults
fiber, and soluble fiber category. Factors were selected on the (114, 115); 11 studies reported central-tendency measures for
basis of the likelihood of influence on outcomes of interest. The participants who were not solely in the overweight or obese
intervention duration was defined as the time period (weeks) categories (115–125); 11 studies were published in languages
when participants received the treatment or placebo; the soluble other than English (126–136); 8 studies were conference pro-
fiber dose was defined as daily grams of fiber treatment as ceedings (137–144); 9 studies reported nonrelevant outcomes
provided during the intervention; and the categorization of sol- (145–152); 5 studies involved participants with previously di-
uble fibers was based on their physicochemical properties agnosed and currently treated metabolic disease (153–157); 48
[e.g., 1) soluble, viscous, and fermentable or 2) soluble, non- studies were review papers (18, 26, 29, 159–203); and 10 studies
viscous, and fermentable]. Studies that provided .1 dose of included energy restriction or weight-loss programs in addition
tolerance (220), and 3 studies included measures of energy in- provided a varied mix of viscous and nonviscous fermentable fi-
take or appetite as primary outcomes (43, 223, 224). Fiber- bers. Baseline total dietary fiber intakes were reported and were
intake compliance was high (85–100%) when reported and shown to be similar between treatment and placebo groups in 3 of
was evaluated in 8 studies through in-person observations (222), the 12 included studies (43, 223, 225). One study presented
packet returns (43, 224, 225), or self-reported consumption re- baseline intake values that were greater than the mean US intake
cords (42, 43, 219, 220, 225, 226). (31.6 g/d for both treatment and control groups) with the use of
The types of soluble fiber treatments and daily dosages varied data from one 24-h dietary recall (223); a second study reported
in studies. One study (220) administered individualized dosages intakes that were similar to mean US consumption values
of soluble fiber on the basis of body weight, and the remaining (i.e., w16–17 g/d) at baseline but did not include the methodology
studies provided daily soluble fiber treatments between 3 and 34 g. for how these data were collected (43); and a third study reported
Six of the included studies included soluble fiber treatments #10 g/d mean baseline dietary fiber intakes of 9.6 g/d for both treatment
(42, 217, 219, 221, 226, 227), and 5 studies included treatments and placebo groups with the use of data from 7-d diet records
between 15 and 34 g soluble fiber/d (43, 222–225). Two studies (225). Measures of energy intake were reported in 8 studies (67%)
provided 2 doses of soluble fiber (42, 43). Five (42%) of the 12 (43, 220–225, 227), 3 of which observed significant reductions in
included studies provided nonviscous but fermentable fiber intake in the treatment group compared with in the placebo group
supplements in the form of manno-oligosaccharides, galacto- (222–224). Four studies (33%) used additional screening pro-
oligosaccharides, and fructo-oligosaccharides (220, 223, 224, cedures to include individuals without a previous diagnosis of
226, 227). Five studies (42%) used fiber treatments that were re- metabolic disease but with elevated glucose (42, 221), lipids
ported to be soluble, viscous, and fermentable (42, 217, 219, 222, (220), or metabolic syndrome risk factors (227) at screening.
225). Four of these studies used individual fiber types including
b-glucan (42), flaxseed mucilage (219), mannans (225), and a
proprietary formulation consisting of wheat and maize-derived Study quality and biases
dextrin (222). One study (43) administered a 50:50 mixture of All but one of the included studies [i.e., Walsh et al. (217)]
pectin (a soluble, viscous, fermentable fiber) and oligofructose (a could be considered high quality according to the MQS (Table 1,
soluble, nonviscous, fermentable fiber), and another study (221) Supplemental Table 2). Two studies reported the use of
TABLE 1
Details of RCTs (n = 12) included in the meta-analysis1
1518
Bays et al., 2011 (42) DB, parallel arm Barley b-glucan in flavored Flavored water beverage HD: 6 g/d Not reported 12 9 30–70 y old 44 4 Glucose
water beverage without fiber LD: 3 g/d BMI (in kg/m2): 25–40 HD treatment: 15 Y Insulin, HD
LD treatment: 15 Y HOMA-IR, HD
Placebo: 14
Brahe et al., 2015 (219) SB, parallel arm Flaxseed mucilage in Breakfast buns with 10 g/d Not reported 6 8 Women aged 40–70 y 39 4 Glucose
breakfast buns maltodextrin BMI: 30–45 Treatment: 19 4 Insulin
Placebo: 20 4 HOMA-IR
Genta et al., 2009 (220) DB, parallel arm Fructo-oligosaccharide in syrup Sugar-free syrup without 0.14 g/kg Not reported 17 8 Women aged 31–49 y 35 Y Weight
fiber BMI $30 Treatment: 20 Y BMI
Placebo: 15 Y Body fat percentage
Y WC
4 Glucose
Y Insulin
Y HOMA-IR
Kobayakawa et al., DB, parallel arm Galactomannan, glucomannan, Powdered starch without 7.5 g/d Not reported 12 9 Asian men aged 20–65 y 30 Y Weight
2013 (221) b-glucan, and alginic acid fiber in water BMI ,35 with visceral obesity Treatment: 15 Y BMI
powder in water Placebo: 15 Y Glucose
Li et al., 2010 (222) DB, parallel arm Soluble dextrin from maize and Fruit juice with 34 g/d Not reported 12 10 Asian men aged 20–35 y 120 Y Weight
wheat starch in fruit juice maltodextrin BMI .24 Treatment: 60 Y BMI
Placebo: 60 Y Body fat percentage
4 Glucose
4 Insulin
Y HOMA-IR
Morel et al., 2015 (223) DB, parallel arm Galacto-oligosaccharide in tea Tea and glucose syrup 18 g/d Treatment: 31.6 2 11 Asian adults aged 18–45 y 44 Y Weight
without fiber Placebo: 31.6 BMI: 25–28 Treatment: 22 Y BMI
Placebo: 22 4 Body fat percentage
4 WC
Parnell and Reimer DB, parallel arm Oligofructose powder in Maltodextrin powder in 21 g/d Not reported 12 9 20–70 y old 39 Y Weight
2009 (224) participant-selected participant-selected BMI .25 Treatment: 21
THOMPSON ET AL.
Bays et al., 2011 (42) Unclear Unclear Low Low Low Low Low
Brahe et al., 2015 (219) Low Low High Unclear Low Low Low
Genta et al., 2009 (220) Unclear Unclear Low Low High Low Low
Kobayakawa et al., 2013 (221) Low Low Low Low Low Low Low
Li et al., 2010 (222) Low Low Low Low Low Low Low
Morel et al., 2015 (223) Low Unclear Low Low Low Low Low
Parnell and Reimer, 2009 (224) Low Unclear Low Unclear Low Low Low
Reimer et al., 2013 (225) Unclear Unclear Low Low Low Low Low
Salinardi et al., 2010 (226) Low Low Low Low Low Low Low
Savastano et al., 2014 (43) Low Low Low Low Low Low Low
Vulevic et al., 2013 (227) Unclear Unclear Low Low Low Low Low
Walsh et al., 1984 (217) High Unclear Low Low Unclear Unclear Low
1
Bias designations by study criteria are indicated by 7 domains with categories including low risk if negative aspects of the study design were not likely
to influence the study findings, high risk if the study design was likely to influence the study findings, or unclear risk if high or low risk could not be assigned
because of a lack of evidence. Ref, reference.
TABLE 3
Meta-analysis and publication bias results for included studies1
P
Outcome Included studies, n (references) Pooled difference2 P Egger’s test Begg’s test
BMI, kg/m 2
7 (219–225) 20.84 (21.35, 20.32) 0.001 0.62 0.65
Body weight, kg 10 (43, 217, 219–226) 22.52 (24.25, 20.79) 0.004 0.28 0.42
Body fat, % 4 (219, 222–224) 20.41 (20.58, 20.24) ,0.00001 0.95 1.00
Waist circumference, cm 6 (219–223, 225) 22.01 (25.36, 1.35) 0.24 0.69 0.85
Glucose, mmol/L 8 (42, 43, 219–221, 224, 225, 227) 20.17 (20.28, 20.06) 0.002 0.87 1.00
Insulin, pmol/L 7 (42, 219–221 224, 225, 227) 215.88 (229.05, 22.71) 0.02 0.11 0.88
HOMA-IR 5 (42, 219–221, 225) 21.33 (22.98, 0.32) 0.12 0.54 1.00
1
Meta-analysis was conducted with the use of random-effects models. Publication bias was assessed with the use of Egger’s and Begg’s tests.
2
All values are between-group means (95% CIs) between soluble fiber–treatment and placebo groups.
1520 THOMPSON ET AL.
FIGURE 2 Forest plots for randomized controlled trials of soluble fiber–intervention studies included in a BMI (in kg/m2) subgroup meta-analysis (n = 358) (A),
a body weight subgroup meta-analysis (expressed as kg) (n = 520), body fat subgroup meta-analysis (expressed as %) (n = 237) (C), and waist-circumference subgroup
meta-analysis (expressed as cm) (n = 320) (D). Random-effects models were used to calculate mean differences (gray squares), 95% CIs (horizontal lines through gray
squares), and pooled-effect sizes (black diamonds). The study weight (expressed as %) indicates the relative contribution of individual studies to the overall pooled-
effect size. Between-study heterogeneity was calculated via the I2 statistic.
SUPPLEMENTAL FIBER IMPROVES WEIGHT AND GLYCEMIA 1521
Table 3 and Figure 3. Soluble fiber supplementation reduced Publication bias assessment
fasting glucose concentrations by 0.17 mmol/L (95% CI: 20.28, No publication bias was indicated for any outcomes as de-
20.06 mmol/L; I2 = 56%, P = 0.002) and fasting insulin con- termined via a funnel plot inspection and Begg’s test and Egger’s
centrations by 15.88 pmol/L (95% CI: 229.05, 22.71 pmol/L; test P values .0.05 (Table 2).
I2 = 88%, P = 0.02) compared with the effects of the placebo.
The numeric reduction in HOMA-IR of 1.33 U with soluble fiber
supplementation was NS (95% CI: 22.98, 0.32 U; I2 = 97%, P = Meta-regressions for intervention duration, fiber dose, and
0.12). The omission of Genta et al. (220) from the meta-analysis fiber type
reduced I2 measures of between-study heterogeneity to ,50% The intervention duration (range: 2–17 wk) and fiber dose
for fasting glucose, fasting insulin, and HOMA-IR (Supplemental (range: 3–34 g/d) were not significantly related to study out-
Figure 3) while maintaining the significant differences between comes. Meta-regressions that were conducted by fiber type
soluble fiber treatment and the placebo. The percentage of mean indicated a more pronounced reduction in HOMA-IR (re-
difference change was.10% in 3 of 8 studies for fasting glucose gression coefficient: 24.11; P = 0.009; 95% CI: 25.77, 22.46)
(219, 221, 224), in 2 of 7 studies for fasting insulin (220, 224), and by nonviscous, fermentable, soluble fiber supplements than by
in 4 of 5 studies for HOMA-IR (219–221, 225). viscous, fermentable fiber types.
1522 THOMPSON ET AL.