Fiber intake and laxation in people with normal bowel function: a systematic review

Abstract

Background

Optimal evidence-based dietary recommendations for dietary fiber have not been established.

Objectives

This study aims to conduct a systematic review of fiber and gut motility (laxation) in noninfants with normal bowel function.

Methods

We searched PubMed, Embase, and CINAHL from inception through 4 March, 2025, and existing systematic review reference lists. We included randomized controlled trials (RCTs) in the general population without bowel dysfunction that compared fiber amounts or types and assessed stool consistency, fecal weight, fecal frequency, or gut transit time. We categorized fibers according to their solubility, viscosity, and fermentability and performed hierarchical Bayesian regression models and spline analyses across reported total and added fiber intake doses, with subanalyses by fiber type. We assessed study risk of bias and strength of evidence (SoE).

Results

We identified 113 eligible trials, with low or moderate risk of bias. Models provided moderate SoE that increasing fiber intake (per g/d total fiber) yields softening stool consistency [0.013 units, 95% credible interval (CrI): 0.009, 0.016 on the Bristol Stool Scale], increasing total fecal weight (1.76 g/d, 95% CrI: 1.58, 1.94) and dry fecal weight (0.47 g/d, 95% CrI: 0.42, 0.51), increasing fecal frequency (0.053 additional bowel movements/wk, 95% CrI: 0.042, 0.065), and shortening gut transit times (−0.24 h, 95% CrI: −0.36, −0.12). Associations between fiber intake and outcomes varied across fiber doses and type. Low-solubility and low-fermentability fibers had the strongest associations with laxation outcomes, with low to medium SoE. However, findings are indirect, based on models, and there is insufficient evidence in specific subpopulations.

Conclusions

RCTs support small to modest effects of dietary fiber on laxation in generally healthy people with normal bowel function. Low-solubility and low-fermentability fibers yielded the strongest effects. Future studies should better characterize fiber properties, assess wider ranges of fiber doses, and comprehensively report participants' total and background fiber intake.

This study was registered at PROSPERO as CRD42024522380.

Introduction

Dietary fiber is an important factor in the management of gastrointestinal function, particularly constipation [1,2]. Persistent constipation has a negative effect on quality of life, healthcare expenditures and is associated with increased mortality and poorer health outcomes [3,4]. Currently, the majority of United States and Canadian populations consume less than the dietary reference intake (DRI) adequate intake recommendations of 14 g/1000 kcal [5,6]. Up to 20% of United States adults report experiencing constipation of unknown cause, with nearly half using medication for laxation [7].

Dietary fibers are carbohydrate polymers that resist hydrolysis by endogenous enzymes in the small intestine [8]. Current DRI criteria distinguish between dietary fiber (nondigestible carbohydrates and lignin naturally occurring in plants) and functional fiber (isolated, nondigestible carbohydrates with demonstrated physiological benefits), with total fiber representing their sum [9,10]. Contemporary considerations of dietary fiber emphasize functional properties—particularly viscosity and fermentability—as key determinants of gastrointestinal utility, moving beyond physicochemical classifications such as solubility [[11], [12], [13]].

Solubility characterizes the degree to which fibers attract, retain, and dissolve in water. Low-solubility fibers hasten gastrointestinal transit time [12,14]. Fiber viscosity has dual counterbalancing roles. More viscous fibers increase stool hydration, which adds bulk, thus potentially hastening transit time, whereas resistance to flow from the gel-forming properties of more viscous fibers may slow transit time [12,15]. Fermentability, the extent to which gut microbiota digest a fiber, affects how much of ingested fibers are excreted intact. Less fermentable fibers persist and continue to exert effects throughout the large bowel to elimination [12,15]. High-fermentability fibers yield by-product gases that may delay transit time. Conversely, the bulking effect of increased microbial biomass could potentially enhance laxation. Given that each fiber source has a unique solubility, fermentability and viscosity profile, the complex mechanisms and interplay between these characteristics drive the laxation effects of dietary fibers.

The current DRI values for the United States and Canada for dietary fiber, published in 2005, were based on narrative literature review. In preparation for a planned rereview of the DRI macronutrient values, this systematic review aims to summarize interventional studies examining associations between fiber amount, type, and characterization with laxation outcomes related to gut motility in the general population with normal bowel function. Treatment of conditions, such as constipation or colitis, is outside the remit of the committee that will update DRIs.

Methods

We followed established systematic review methodologies per the Agency for Healthcare Research and Quality (AHRQ) Methods Guide [16]. The protocol was finalized in discussion with the review’s sponsors (United States Department of Health and Human Services Office of the Assistant Secretary for Health and United States Department of Agriculture Food and Nutrition Service) and a technical expert panel comprised of invited nutritionists and gastroenterologists. The protocol was prospectively registered with PROSPERO (CRD42024522380). We reviewed the association between fiber intake and laxation/gut motility in the general population. This review is based on a broader report commissioned by AHRQ, which also includes studies in infants and comparisons of specific fibers [17].

We conducted literature searches from inception through 4 March, 2025 in PubMed, Embase, and CINAHL. We also reviewed reference lists from existing systematic reviews published since 2015 and assessed studies suggested by outside experts, peers, and invited and public reviews. We searched index terms and free-text words for concepts related to dietary fiber and laxation (details in Supplemental Material A-1).

All citations were independently screened in duplicate in the Systematic Review Data Repository Plus (SRDR+), which is currently defunct. It incorporated machine learning (details in Supplemental Material A-2). Conflicts were adjudicated by the whole team during pilot screening rounds, then by a senior researcher.

We included randomized controlled trials (RCTs), both parallel and crossover design, conducted in the general population, and excluded people with acute or chronic conditions that may affect gut motility (e.g., Crohn’s disease, chronic constipation). The current analyses excluded infants. Interventions of interest included any assessment of total daily fiber intake or added fiber consumed as either food or supplement. The full AHRQ report includes analyses by fiber format (foodstuff or supplement), which are not included here [17]. Comparators included no added fiber, different dosages of the same fiber, or different fiber types. Fibers had to be consumed for ≥2 wk to allow for the effect of fiber on gut motility to stabilize. Given the focus on people with normal bowel function, prioritized outcomes pertained to measures of gut motility, including stool consistency, fecal weight (total and dry), fecal frequency, and gut transit time. Per the sponsors’ requirements, studies had to be published in English. Detailed eligibility criteria are listed in Supplemental Table B-1.

We extracted data and conducted risk of bias assessments using a customized form in SRDR+. Data extraction and risk of bias assessment were conducted by 1 reviewer and confirmed by a second reviewer. Two experienced fiber researchers (postdoctoral researchers in diet and gastroenterology) confirmed the amounts of total and added fiber, and characterized the investigated fibers by their solubility, viscosity, and fermentability, primarily based on a preestablished categorization of specific fibers determined using peer-reviewed literature (Supplemental Table C-1) [12,[18], [19], [20], [21], [22], [23], [24]]. Their determinations were then reviewed by senior systematic reviewers. We standardized stool consistency scales to the Bristol Stool Scale (BSS) [scale ranges from 1 (hard) to 7 (watery)] [25]. All outcome units were standardized to g/d for fiber, stools (defecations)/wk, g/d for total and dry fecal weights, and h/wk for gut transit time. Study risk of bias was assessed with items from the Cochrane Risk of Bias tool, with additional questions pertaining to crossover trials (carry-over effects, equal numbers in each group, and period effects) [26]. Risk of bias was further reviewed by a senior researcher. Both the daily amount of fiber added to participants’ diets and their total dietary fiber intake were extracted. A posteriori, we conducted a sensitivity analysis excluding high risk of bias studies.

For each outcome, we conducted analyses for: 1) total dietary fiber, 2) added fiber, and 3) the fiber characteristics of added fiber (solubility, viscosity, and fermentability). Studies with missing data precluding analysis were excluded. The primary analyses were based on hierarchical Bayesian regression models that accounted for both fixed and random effects [27]. All analyses used minimally informative prior distributions that matched ranges reported in the literature. Convergence was assessed by visual inspection of trace plots and with the German Rubin diagnostic for between-study means and variances [28]. We also compared model predictions with the smoothed empirical data. Metaregression was conducted to allow analyses across the full range of evaluated fiber doses and analyses by fiber types, which could not be evaluated by individual studies. For total and added fiber analyses, we conducted spline analyses applying knots at various fiber amounts, to determine if plateau effects or other substantive changes in associations were evident at higher or lower fiber amounts. All Bayesian analyses were conducted using Stan software (version 2.32.6) for Bayesian analysis in R (version 2024.09.0 + 375) [29,30]. For analyses of similar doses of fibers with different characteristics, we conducted standard restricted likelihood meta-analyses.

For each set of analyses, we applied standard AHRQ Evidence-based Practice Center Program methods to determine the strength of evidence (SoE) for each conclusion [16].

Results

The literature searches yielded 9669 citations, from which we identified 113 eligible RCTs (Figure 1). Existing systematic reviews and experts did not yield additional studies. We included 40 parallel design RCTs (N = 2896 total) and 73 crossover trials (N = 2369 total). Approximately one-third of trials were conducted in the United States (27 studies) or Canada (9 studies); about half in Western Europe, Australia, New Zealand, or Israel; about 10% in East or Southeast Asia; and 4 studies were from Brazil. Studies were published between 1976 and 2024. Most studies did not fully describe the fiber interventions and/or sources of total daily fiber. The trials were highly heterogeneous, evaluating 64 different fiber types or sources, across a wide range of fiber amounts: total daily fiber 2.8–75 g/d (median 24.5 g/d); added fiber 0.6–60 g/d (median 12.0 g/d). The median duration of the fiber interventions across studies was 4 wk (range 2 wk to 6 mo).

FIGURE 1.

Literature flow diagram.

About half of trials were at low risk of bias; one-third were at moderate risk of bias; and 15% at high risk of bias (Supplemental Table D-1). Detailed descriptions of the included studies and studies results are provided in Supplemental Tables E-1 to E-3, F-1, and F-2.

Briefly, no single study effectively addressed the research question of this review, namely, the overall effect of dietary fiber on laxation across a wide range of fiber intake amounts and in different types of fibers. Overall, models found linear associations between increasing fiber intake (per g/d) and softening stool consistency, increasing total and dry fecal weight, increasing fecal frequency, and shortening gut transit times, with moderate SoE (Table 1). However, the effects of fiber on laxation outcome in generally healthy people were typically small. The doses at which increasing fiber was associated with the largest laxation effects varied by outcome and fiber type. In general, low-solubility and low-fermentability fibers had stronger effects on laxation outcomes than other fibers. Relative effects of different viscosity fibers were inconsistent.

TABLE 1.

Overall summary of findings

Fiber/analysis	Stool consistency	Total fecal weight	Dry fecal weight	Fecal frequency	Transit time
Total fiber	▲▲ 0.013 units per g/d Strongest >30 g/d	▲▲ 1.76 g/d per g/d Stable across doses	▲▲ 0.47 g/d per g/d Stronger at lower doses	▲▲ 0.053/wk per g/d Stable across doses	▼▼ −0.24 h per g/d Stronger at lower doses
Added fiber	▲▲ 0.010 units per g/d Strongest >15 g/d	▲▲ 1.41 g/d per g/d Stronger at lower doses	▲▲ 0.39 g/d per g/d Stronger at lower doses	▲▲ 0.030/wk per g/d Stronger at higher doses	▼▼ −0.25 h per g/d Strongest <5 g/d
Solubility	▲ Low solubility	●	▲ Low solubility	▲ Low solubility	▲▲ Low, medium sol
Viscosity	●	●	●	▲▲ Medium, high viscosity	▲ Medium viscosity
Fermentability	▲ Low fermentability	▲▲ Medium, low fermentability	▲▲ Low fermentability	▲▲ Low fermentability	●

▲▲ = moderate SoE of a positive association/effect, ▲ = low SoE of a positive association/effect, ▼▼ = moderate SoE of a negative association/effect, ● = low SoE of no association/effect.

Abbreviations: g/d, grams per day of fiber; SoE, strength of evidence.

Exemplar results tables and figures are presented here. Results tables for all other analyses are reported in Supplemental Material G.

Stool consistency

Fifty trials reported analyzable data for stool consistency: 27 reported total fiber intake [[31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42], [43], [44], [45], [46], [47], [48], [49], [50], [51], [52], [53], [54], [55]], all 50 reported added fiber [[31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42], [43], [44], [45], [46], [47], [48], [49], [50], [51], [52], [53], [54], [55], [56], [57], [58], [59], [60], [61], [62], [63], [64], [65], [66], [67], [68], [69], [70], [71], [72], [73], [74], [75], [76], [77]]. Tables, figures, and additional details are provided in Supplemental Material G.

Total fiber intake

Across 27 trials (N = 2355), total fiber intake ranged from 11 to 75 g/d (median 18 g/d) with overall mean BSS of 3.8 (where normal is 3–4 [78]). Metaregression showed a statistically significant slope between total fiber intake and BSS [0.013 units per g/d, 95% credible interval (CrI): 0.009, 0.016], indicating increasing stool softness with increasing total fiber intake (Supplemental Table G-1.1.1). Linear spline analyses suggested somewhat larger associations at higher total fiber intake, particularly >30 g/d fiber (slope 0.022 units per g/d, 95% CrI: 0.014, 0.031), compared with lack of association at <30 g/d fiber (slope 0.003 units per g/d, 95% CrI: −0.006, 0.011) (Supplemental Figure G-1.1.1). Increasing fiber intake had a small effect on stool consistency, with BSS increasing from a mean of 3.7 with 11 g/d total fiber to 4.3 with 55 g/d total fiber.

Added fiber intake

Across 50 trials (N = 3884), added fiber in intervention arms ranged from 0.6 to 60 g/d. Mean BSS was 3.7 across no added fiber arms. BSS increased with increasing added fiber intake (Table 2). Spline analyses suggested stronger associations at higher added fiber intake (>15 g/d fiber: slope 0.020 units per g/d, 95% CrI: 0.013, 0.028) than lower added fiber intake (<15 g/d fiber: slope 0.003 units per g/d, 95% CrI: −0.003, 0.009) (Figure 2). Mean stool consistency softened from about 3.7 with no added fiber to 4.6 with 60 g/d added fiber.

TABLE 2.

Added fiber and BSS

Analysis	Trials [arms] (N1)	Knot	Slope, BSS per g/d <Knot	Slope, BSS per g/d Overall or ≥Knot
Overall	50 [118] (3884)	—	—	0.010 (0.007, 0.013)2
Overall/knot	50 [118] (3884)	5 g/d	–0.024 (–0.040, –0.007)	0.018 (0.013, 0.023)2
Overall/knot	50 [118] (3884)	10 g/d	–0.004 (–0.012, 0.004)	0.019 (0.013, 0.025)2
Overall/knot	50 [118] (3884)	15 g/d	0.003 (–0.003, 0.009)	0.020 (0.013, 0.028)2
Overall/knot	50 [118] (3884)	20 g/d	0.006 (0.002, 0.011)2	0.020 (0.010, 0.031)2
Overall/knot	50 [118] (3884)	25 g/d	0.009 (0.006, 0.013)2	0.016 (0.004, 0.029)2
Solubility, low	8 [16] (591)3	—	—	0.010 (0.004, 0.016)2
Solubility, medium	3 [6] (298)3	—	—	0.005 (–0.012, 0.022)
Solubility, high	20 [48] (1236)3	—	—	0.002 (–0.002, 0.007)
Viscosity, low	21 [49] (1468)3	—	—	0.005 (0.002, 0.009)2
Viscosity, medium	8 [17] (529)3	—	—	0.001 (–0.014, 0.017)
Viscosity, high	4 [7] (214)3	—	—	0.008 (–0.007, 0.024)
Fermentability, low	6 [10] (302)3	—	—	0.019 (0.006, 0.032)2
Fermentability, medium	4 [9] (232)3	—	—	0.005 (–0.004, 0.013)
Fermentability, high	22 [53] (1677)3	—	—	0.004 (0.000, 0.008)2

Abbreviations: BSS, Bristol Stool Scale; g/d, grams per day of fiber; < (≥) Knot, analysis below (above) the knot point in the spline analysis; N, participant-arms; — indicates no relevant data for that cell.

¹N represents the total “participant-arms” from both parallel and crossover trials. From crossover trials, the N includes the same individuals repeatedly for each study arm (cross) they participated in.

²Indicates statistically significant slopes.

³In the overall model of fiber characteristics, there were 36 trials, with 78 study arms, and 2715 participant-arms.

FIGURE 2.

Added fiber compared with Bristol Stool Scale, overall and spline analyses. Each panel presents a spline analysis with knots at the indicated added fiber doses ranging from 5 to 25 g/d. The first panel presents the same data with no knot (the overall analysis). These models account for the pairing of study arms within trials; they are not adjusted for fiber characteristics. The red lines indicate the slope of the association between fiber dose and outcome. In each panel with knots, indicated by the vertical blue dashed lines, the slope of the line is allowed to change at that point. The red ribbon around the slope indicates the 95% credible interval. The black curves and gray ribbons highlight the adjusted mean outcome at each fiber dose across studies; these curves do not account for within-trial pairing of data. Each circle represents a study arm; the size of the circles is scaled to the inverse SE of the outcome so that study arms providing greater weight are larger. The apparent differences in colors among the circles are due to multiple study arms overlapping. g/d, grams per day of fiber.

Analyses of associations between added fiber and BSS for different categories of fiber based on solubility, viscosity, and fermentability did not yield substantial differences in change in BSS scores with increasing total daily intake across the different types of fibers (Table 2; Supplemental Figure G-1.2.1.1). Five trials that compared similar doses of fibers with different characteristics were mostly consistent across trials (Supplemental Material G-1.2.2 for further details) [32,45,50,64,76].

Total fecal weight

Total fiber intake

Across 37 trials (N = 2606), total fiber intake ranged from 11 to 65 g/d (median 18 g/d), with overall mean total fecal weight of 166 g/d [32,33,36,[43], [44], [45], [46],50,52,56,[79], [80], [81], [82], [83], [84], [85], [86], [87], [88], [89], [90], [91], [92], [93], [94], [95], [96], [97], [98], [99], [100], [101], [102], [103], [104]]. Metaregression showed a statistically significant slope between total fiber intake and total fecal weight (1.76 g/d of total fecal weight per g/d fiber, 95% CrI: 1.58, 1.94), indicating increasing total fecal weight with increasing total fiber intake (Supplemental Table G-2.1.1). Spline analyses suggested consistent associations across fiber intake, with a possible plateauing effect at total fiber intake >35 g/d (Supplemental Figure G-2.1.1).

Added fiber intake

Across 38 trials (N = 2677), added fiber in intervention arms ranged from 0.6 to 50 g/d [32,33,36,[43], [44], [45], [46],50,52,56,64,74,[79], [80], [81], [82], [83], [84], [85],90,91,[93], [94], [95],[97], [98], [99], [100], [101], [102], [103], [104], [105], [106], [107], [108], [109]]. Mean total fecal weight was 140 g/d across no added fiber arms. Total fecal weight increased with increasing added fiber intake (Supplemental Table G-2.2.1). Spline analyses suggested diminishing returns at higher added fiber intake (>5 g/d fiber: slope 1.25 g/d of total fecal weight per g/d fiber, 95% CrI: 0.88, 1.62) compared with lower added fiber intake (<5 g/d fiber: 2.00 g/d of total fecal weight per g/d, 95% CrI: 0.89, 3.12). Diminishing effects continued at even higher added fiber intakes with slopes of 1.1 for >20 g/d and 0.8 for >25 g/d fiber (Supplemental Figure G-2.2.1).

Analyses of different categories of fiber based on solubility, viscosity, and fermentability did not yield substantial differences in change in total fecal weight with fiber intake based on the fiber’s solubility or viscosity (Supplemental Table G-2.2.1 and Figure G-2.3.1.1). Fibers with low or medium fermentability were associated with greater increases in fecal weight with increasing fiber dose compared with fibers with high fermentability.

Ten trials compared similar doses of fibers with different characteristics [32,45,50,64,80,[82], [83], [84],98,100]. Meta-analysis yielded no statistically significant differences in total fecal weight between low- and high-solubility fibers [pooled mean difference (MD) 4.3 g/d, 95% confidence interval (CI): −49.0, 57.7; 3 trials], low- and medium-viscosity fibers (pooled MD 9.2 g/d; 95% CI: −46, 64; 3 trials) or low- and medium-fermentability fibers (pooled MD 73.6 g/d; 95% CI: −6.7, 153.9; 2 trials) (Supplemental Material G-2.3.2 for further details) [32,45,50,64,80,[82], [83], [84],98,100].

Dry fecal weight

Total fiber intake

Across 20 trials (N = 1607), total fiber intake ranged from 4.5 to 65 g/d (median 18 g/d) with an overall mean dry fecal weight of 38 g/d [32,33,36,52,56,80,83,87,88,91,93,94,97,98,[100], [101], [102], [103], [104]]. Metaregression showed a statistically significant slope between total fiber intake and dry fecal weight (0.47 g/d dry fecal weight per g/d fiber, 95% CrI: 0.42, 0.51), indicating increasing dry fecal weight with increasing total fiber intake (Supplemental Table G-3.1.1). Spline analyses suggested steeper slopes at lower added fiber intake, with each additional g/d of added fiber associated with 0.8 to 1.2 g/d of dry fecal weight (≤25 g/d fiber) compared with 0.3 to 0.5 g/d of dry fecal weight at higher fiber doses, with no further statistically significant increase in dry fecal weight above 40 g/d total fiber intake (Supplemental Figure G-3.1.1).

Added fiber intake

Across 24 trials (N = 1863), added fiber in intervention arms ranged from 0.6 to 50 g/d [32,33,36,52,56,64,74,80,83,91,93,94,97,98,[100], [101], [102], [103], [104],[107], [108], [109], [110]]. Mean dry fecal weight was 33 g/d across no added fiber arms. Dry fecal weight increased with increasing added fiber intake (Supplemental Table G-3.2.1). Spline analyses suggested steeper slopes at lower added fiber intake, with each additional g/d of added fiber associated with 0.74 g/d of dry fecal weight (≤10 g/d fiber) compared with 0.18 g/d of dry fecal weight (beyond 10 g/d fiber). However, the association remained statistically significant at all intake levels (Supplemental Figure G-3.2.1).

Analyses of associations between added fiber and dry fecal weight for different categories of fiber based on solubility, viscosity, and fermentability showed distinct associations (Supplemental Table G-3.2.1 and Figure G-3.3.1.1), with low-solubility and low-fermentability fibers demonstrating substantially larger effect sizes (slopes 0.8 and 1.3, respectively) compared with higher solubility and fermentability fibers (slopes of about 0.2–0.3). However, analyses of added fiber by viscosity categories did not show substantial differences across the viscosity categories. Four trials that compared similar doses of fibers with different solubility, viscosity, and fermentability characteristics were mostly consistent across trials (Supplemental Material G-3.3.2 for further details) [64,80,83,98].

Fecal frequency

Total fiber intake

Across 40 trials (N = 3225), total fiber intake ranged from 6 to 75 g/d (median 18 g/d) with overall mean fecal frequency of 8.8 bowel movements/wk [[31], [32], [33], [34], [35],[37], [38], [39], [40], [41], [42], [43],[45], [46], [47], [48], [49], [50], [51], [52], [53],55,56,80,81,89,91,92,97,98,101,[111], [112], [113], [114], [115], [116], [117]]. Metaregression showed a statistically significant slope between total fiber intake and fecal frequency (0.053 additional bowel movements/wk per g/d fiber, 95% CrI: 0.042, 0.065), indicating increasing fecal frequency with increasing total fiber intake (Supplemental Table G-4.1.1). Spline analyses suggested consistent associations between total fiber intake and fecal frequency across most intake levels, with slopes of ∼0.05–0.07 bowel movements/wk per g/d of total fiber, except at extreme intakes below 10 g/d or above 40 g/d (Supplemental Figure G-4.1.1).

Added fiber intake

Across 58 trials (N = 1690), added fiber in intervention arms ranged from 0.6 to 60 g/d [[31], [32], [33], [34], [35],[37], [38], [39], [40], [41], [42], [43], [44], [45], [46], [47], [48], [49], [50], [51], [52], [53],55,56,[58], [59], [60], [61], [62], [63], [64][31–35,37–53,55,56,58–64,66,[68], [69], [70], [71], [72], [73], [74], [75],77,80,81,91,97,98,101,107,109,[111], [112], [113], [114], [115], [116], [117], [118], [119], [120], [121], [122], [123]. Mean fecal frequency was 8.8 bowel movements/wk across no added fiber arms. Fecal frequency increased with increasing added fiber intake (Supplemental Table G-4.2.1). Notably, spline analyses suggested stronger slopes at higher amounts of added fiber, with statistically significant and consistent slopes above each knot (0.06–0.08 bowel movements/wk per g/d added fiber) compared with nonsignificant or substantially smaller slopes below each corresponding knot (Supplemental Figure G-4.2.1).

Analyses of associations between added fiber and fecal frequency for different categories of fiber based on solubility, viscosity, and fermentability found variable associations in change in fecal frequency scores with increasing total daily intake across the different types of fibers (Supplemental Table G-4.2.1 and Figure G-4.3.1.1). Five trials that compared similar doses of fibers with different characteristics found no significant differences in fecal frequency between fiber types (Supplemental Material G-4.3.2 for further details) [45,50,65,80,98].

Transit time

Total fiber intake

Across 16 trials (N = 1081), total fiber intake ranged from 2.8 to 60 g/d (median 18 g/d), with an overall mean transit time of 36.8 h [34,36,43,44,46,52,80,83,84,91,[97], [98], [99], [100],102,124]. Metaregression showed the statistically significant slope between fiber intake and transit time (−0.24 h per g/d fiber, 95% CrI: −0.36, −0.12), indicating decreasing transit time with increasing fiber intake (Supplemental Table G-5.1.1). Spline analyses suggested that transit time decreases per additional g/d of total fiber at doses above the tested knots compared with lower doses, but there is no clear threshold (Supplemental Figure G-5.1.1).

Added fiber intake

Across 19 trials (N = 1225), added fiber in intervention arms ranged from 0.6 to 37 g/d [34,36,43,44,46,52,64,80,83,84,91,[97], [98], [99], [100],102,109,110,124]. Mean transit time was 44.4 h across no added fiber arms. Transit time decreased with increasing fiber intake (Supplemental Table G-5.2.1). In contrast with the total dietary intake, spline analyses suggested the most substantial effects between added fiber and change in transit time at lower added fiber intake (<5 g/d fiber: −1.1 h per g/d added fiber, totaling 5.7 h reduction with 5 g/d added fiber) compared with progressively smaller and nonsignificant associations at higher added fiber intake (15–25 g/d fiber: −0.2 to −0.3 h per g/d added fiber) (Supplemental Figure G-5.2.1).

Analyses of associations between added fiber and transit time for different categories of fiber based on solubility, viscosity, and fermentability did not yield substantial differences in change in transit time with increasing total daily intake across the different types of fibers (Supplemental Table G-5.2.1 and Figure G-5.3.1.1). Five trials that compared similar doses of fibers with different characteristics were mostly consistent across trials (Supplemental Material G-5.3.2 for further details) [64,80,83,84,100].

Sensitivity analyses

Analyses excluding high risk of bias studies yielded mostly similar results (Supplemental Material H). The slope of BSS with total fiber intake was slightly weaker, but still statistically significant (0.009 compared with 0.013 units per g/d fiber; Supplemental Table H-1-1-1.s); the association with added fiber was nearly identical. Analyses of fiber intake thresholds and specific fiber characteristics were very similar. Analyses of total and dry fecal weight and fecal frequency were substantially similar. Excluding high risk of bias studies yielded somewhat stronger associations between total and added fiber and transit time (e.g., −0.39 compared with −0.24 h per g/d fiber; Supplemental Table H-5-1-1.s). In addition, estimates of slopes were stronger for low-solubility fibers (−1.3 compared with −0.3 h per g/d fiber), low-viscosity fibers (−0.8 compared with −0.1 h per g/d fiber), and low-fermentability fibers (−1.4 compared with −0.4 h per g/d fiber) (Supplemental Table H-5-2-1.s). It is not readily apparent why studies that did not report blinding or account for period effects (in crossover studies) would preferentially increase transit time with some fiber types. SoE assessments were unchanged by the sensitivity analyses.

Discussion

Across 113 RCTs, models identified linear associations between increasing fiber intake (per g/d) and laxation outcomes, with moderate SoE, related to their being based on indirect evidence across trials. Notably, no study effectively addressed the research question of this review, the overall effect of dietary fiber on laxation across a wide range of fiber intake amounts and in different types of fibers. Thus, the main findings are based on regression models that evaluated the association between change in fiber amount and change in outcome across studies. The effects of total and added fiber intakes on stool consistency and of added fiber on fecal frequency were strongest at higher fiber doses. However, the effects of total and added fiber on both transit time and dry fecal weight, and of added fiber on total fecal weight, were strongest at relatively low fiber doses. The associations between total fiber intake and both total fecal weight with fecal frequency were stable across fiber doses. However, conclusions about the specific thresholds (e.g., whether there was a change in slope at 5 g/d) are based on multiple statistical analysis without controlling for multiplicity and were somewhat subjective and open to interpretation. The analyses of specific fiber characteristics (solubility, viscosity, and fermentability) yielded conclusions with low or moderate SoE, with some inconsistencies in the findings (e.g., that medium-viscosity fibers had strong associations but both low- and high-viscosity fibers did not). Low-solubility fibers were more strongly associated (steeper slopes) than medium- or high-solubility fibers with stool consistency, dry fecal weight, and fecal frequency (each low SoE), but not total fecal weight (moderate SoE). Both low- and medium-solubility fibers were more strongly associated with transit time than high-solubility fibers (moderate SoE). Associations with outcomes did not consistently vary by fiber viscosity for most outcomes (moderate SoE), but medium- and high-viscosity fibers were more strongly associated with fecal frequency than low-viscosity fibers (moderate SoE). Low-fermentability fibers were associated with stronger associations than medium- or high-fermentability fibers with stool consistency (low SoE), dry fecal weight and fecal frequency, but not transit time (moderate SoE for latter outcomes). Both medium- and low-fermentability fibers were more strongly associated with transit time than high-fermentability fibers (moderate SoE).

Our findings align with results from previous systematic reviews that had narrower scopes. A systematic review published in 2015 focused on cereal fibers and reported similar directional effects but with larger effect sizes, likely due to their inclusion of observational studies and different analytical approaches [125]. The current review's restriction to RCTs and the Bayesian regression modeling provide a more rigorous analysis of the evidence. The inclusion of spline analyses and evaluation of fiber characteristics provides new insights into optimal dosing and fiber selection.

This review was primarily designed to aid a planned rereview of DRI macronutrient values. Although not definitive, its findings provide the best evidence for policymakers, clinicians, nutritionists, and others to determine dietary advice regarding fiber intake, both total and supplemental, that considers the effects of the fiber’s characteristics. However, it is important to avoid overinterpretation of the data. Findings, such that each g/d of fiber softens stool by a mean of 0.013 units on the BSS, may be statistically significant, but are likely of little importance to most individuals. It will be important for policymakers to consider the likely impact on health that changes in the measured and analyzed laxation outcomes may have on the health of the general population (e.g., the health implications of shorter gut transit time or increased fecal frequency).

The eligible trials were extremely heterogeneous, mostly precluding high SoE conclusions since major conclusions could be drawn only through indirect regression analysis across studies that had very few similar fiber comparisons. Studies were commonly poorly reported, with fiber descriptions often limited to names and consumption amounts. When fibers were incorporated into foods (muffins, oatmeal), it was frequently unclear either how much actual fiber was added to participants’ diets or what their background fiber intake was. Very few studies characterized fibers in terms of their solubility, viscosity, and fermentability. The evidence base is primarily applicable to generally healthy adults, with limited evidence specific to children, older adults, pregnant females, or comparing effects in females and males (our full review describes the few trials conducted in specific populations [17]).

Several additional limitations must be considered. One of the biggest challenges was the categorization of fiber characteristics (solubility, viscosity, fermentability), which required expert judgment due to insufficient reporting in primary studies. We addressed this by having 2 experienced fiber researchers characterize the 32 fibers using prespecified criteria. Very few of the trials reported information about their evaluated fibers to allow more specific fiber categorization. We did not account for possible confounding of each characteristic (e.g., fermentability) by the other characteristics (e.g., solubility and viscosity) because of data loss from requiring categorizations for all 3 fiber characteristics.

Researchers conducting further evaluations of the effects of fiber on gut health should consider the laxation outcomes evaluated in this review, which were prioritized by a panel of stakeholders with clinical and research perspectives. To provide more actionable dietary recommendations, future studies should be specifically designed to evaluate fibers within the food matrix, assess the effects of incrementally adding fiber to existing diets of known fiber content, better characterize fiber properties in terms of solubility, viscosity, and fermentability, and comprehensively describe participants’ background and total fiber intake to allow more actionable dietary recommendations. More trials are needed in understudied populations, particularly children and adolescents, older adults, and perinatal females. Subgroup analyses by participant sex could help elucidate if differential effects exist for females and males.

Overall, the included trials all had different underlying research questions than the key question of this systematic review, as well as across studies. Nevertheless, RCTs support small to modest effects of dietary fiber on laxation in the general noninfant population. The effect of increasing dietary fiber intake on laxation tends to be greatest at lower doses for total and dry fecal weight and transit time (e.g., below about 35 g/d total fiber), but at higher doses for stool consistency and fecal frequency (e.g., above about 40 g/d total fiber). The models found that low-solubility and low-fermentability fibers had the strongest associations with most laxation outcomes, but the degree of fiber viscosity was not consistently associated with laxation outcomes. The evidence primarily applies to generally healthy adults with normal bowel function, with insufficient data for other life stages or populations.

Author contributions

The authors’ responsibilities were as follows – EMB, NJT, AHL: designed research; EMB, EC, HJM, YC, GPA, GK, ELC, TAT, GW, KD: conducted research; EMB, HJM, YC, TAT: analyzed data; EMB, EC, HJM, GPA, GK, ELC: wrote the article; EMB: had primary responsibility for final content; and all authors: read and approved the final manuscript.

Disclaimer

This project was funded under Contract No. 75Q80120D00001 Task Order 75Q80124F32011 from the Agency for Healthcare Research and Quality (AHRQ), U.S. Department of Health and Human Services (HHS). The authors of this manuscript are responsible for its content. Statements in the manuscript do not necessarily represent the official views of or imply endorsement by AHRQ or HHS.

Data availability

Data described in the manuscript, codebook, and analytic code will be made available on request pending application and approval.

Funding

This project was funded under contract no. 75Q80120D00001 from the Agency for Healthcare Research and Quality.

Conflict of interest

The authors report no conflicts of interest.

Appendix A. Supplementary data

The following is the Supplementary data to this article: