Systematic review and meta-analysis: the clinical and physiological effects of fibre-containing enteral formulae

Authors


Dr M. Elia, Institute of Human Nutrition, University of Southampton, Southampton General Hospital, Mailpoint 113, Tremona Road, Southampton SO16 6YD, UK.
E-mail: m.elia@soton.ac.uk

Summary

Background  Enteral nutrition can be associated with gastrointestinal side effects and fibre supplementation has been proposed as a means to normalize bowel function.

Aim  To evaluate systematically the effects of fibre supplementation of enteral feeds in healthy volunteers and patients both in the hospital and community settings.

Methods  Electronic and manual bibliographic searches were conducted. Controlled studies in adults or children, comparing fibre-supplemented vs. fibre-free formulae given as the sole source of nutrition for at least 3 days, were included.

Results  Fifty-one studies (including 43 randomized-controlled trials), enrolling 1762 subjects (1591 patients and 171 healthy volunteers) met the inclusion criteria. Fibre supplementation was generally well tolerated. In the hospital setting, the incidence of diarrhoea was reduced as a result of fibre administration (OR 0.68, 95% CI: 0.48–0.96; 13 randomized-controlled trials). Meta-regression showed a more pronounced effect when the baseline incidence of diarrhoea was high. In both patients and healthy subjects, fibre significantly reduced bowel frequency when baseline frequency was high and increased it when it was low, revealing a significant moderating effect of fibre.

Conclusions  The review indicates that the fibre-supplemented enteral formulae have important physiological effects and clinical benefits. There is a need to use a consistent approach to undertake more studies on this issue in the community setting.

Introduction

Enteral nutrition (oral or tube feeding) is required when oral intake is insufficient or is likely to be absent for a period of more than 5–7 days.1 It is used in the in-patient and out-patient settings in a wide range of disease states, with the majority of patients requiring nutritional support for around 1 month.2 The duration of enteral feeding will depend upon the nature of the underlying condition. For example, patients recovering from surgery may require short-term enteral feeding whereas those in chronic-care facilities may depend on long-term enteral nutrition, usually tube feeding, over many months.

There is an ongoing debate regarding the definition of fibre, which includes either in its entirety, or as a major component, the polysaccharides of plant cell walls.3 The properties of fibres depend on their chemical as well as physical characteristics. Physiological approaches to definitions have been based on the consideration of nondigestibility of carbohydrates in the small intestine, with the plant cell wall polysaccharide being the dominant component. The solubility of fibre determines to some extent its physiological properties such as water-holding capacity. Nevertheless, this categorization may be too simplistic as it neglects the metabolism in the colon.4 The degree of fermentability may be equally relevant for the health effects of fibres. In general, well-fermented fibres are soluble, whilst less well-fermented fibres are insoluble. However, there are some exceptions, e.g. insoluble soy polysaccharides, especially when finely ground, may be well-fermented. In addition, certain fibre types, the so-called prebiotics, have the ability to stimulate the growth of specific types of colonic bacteria.5 An attempt has also been made to link the definition of fibre to health benefits, and as a healthy diet encompasses a large variety of fibres, it seems reasonable to consider the use of a mixture of fibres in enteral feeds.

Dietary fibre was omitted from the early commercial enteral formulae, mainly on account of its effects on increasing viscosity and sedimentation. These technological issues have largely been overcome and several fibre-supplemented enteral formulae are now available, although doubts remain about their role, tolerability and efficacy.

Gastrointestinal symptoms are observed in enterally fed patients, which can be influenced by various factors such as the nature and application of the feed (composition, temperature and flow rate), the disease state and/or the medication. Lack of fibre in enteral formulae has been suggested to be a cause of impaired bowel function.6 For example, in the acute hospital setting, especially critical care units, diarrhoea is reported to be a frequent complication of enteral feeding,7 and the aim of fibre supplementation in this situation would be to reduce diarrhoea. However, in patients receiving long-term enteral feeding, constipation is probably the most frequently reported gastrointestinal problem.8 The principal role of fibre supplementation in this situation is to prevent constipation, which may otherwise require large amounts of laxatives or even manual evacuation. Fibre can speed up gastrointestinal transit time, increase faecal bulk and reduce constipation.9, 10 Fibre may also have other beneficial effects, such as improvement in gut barrier function, which prevents translocation of bacteria and toxins from the gut into the systemic circulation,11 an increase in turnover or regeneration of epithelial cells,12 and effects on fluid and electrolyte absorption.13 Conversely, fibre is sometimes perceived as a factor leading to tolerance problems including bloating, flatulence and diarrhoea.14, 15

Clinical trials that have examined the effects of fibre-supplemented formulae on gastrointestinal function in patients have shown variable results.16 This inconsistency might be on account of the differences in patient types, types of intervention, end point definitions and lack of evaluation of the confounding factors. Studies performed in healthy volunteers could help establish overarching principles that apply to both health and disease, avoiding the confounding effects of disease, drug administration, and immobility, all of which may be present in a clinical setting.

The effects of fibre supplementation in enteral nutrition have been evaluated in two recently published systematic reviews.17, 18 The first by Yang et al. included only seven randomized-controlled trials (RCTs) published prior to 2003 (two of which included additional supplementation with probiotics) and did not include an analysis of the effects of fibre supplementation in healthy subjects.17 The outcome parameters examined were restricted to the occurrence of diarrhoea, infection and length of hospital stay in hospitalized patients. The review by del Olmo et al. (English abstract only) included 25 studies published before the end of 2002,18 focussing on incidence of diarrhoea and constipation, use of laxatives and frequency of bowel movements. The meta-analyses combined findings from patient groups with those from healthy volunteers. Neither of the reviews examined whether the effects of fibre were dependent on the dose and type of fibre administered.

In view of the various limitations of the previously published work, we felt a need to undertake a more comprehensive and up-to-date systematic review, applying more extensive statistical methods (including meta-analysis and meta-regression). The present review was undertaken to examine the potential clinical benefits, physiological effects and tolerance of fibre-containing feeds compared with standard fibre-free feeds. Studies performed in healthy subjects and patients with acute and/or chronic conditions were considered. The decision was taken a priori not to combine patients and healthy subjects in the same meta-analysis, although any concepts emerging from healthy volunteers would be considered in the light of results emerging from patient populations. The final objective was to identify knowledge gaps in the published literature and to propose areas requiring further research.

Methods

Research design and methods

The main aim of the systematic review was to evaluate the effect of fibre-supplemented enteral feeds on objective and subjective measures of gastrointestinal function in both healthy volunteers and patient populations. The review was planned, conducted and reported following published guidelines. These include those issued by the Cochrane Collaboration19 and the QUORUM guidelines.20

Identification and retrieval of studies

Potentially relevant studies were identified by searching the electronic databases PubMed, Embase and Biosys, accessed in July 2006. Two separate searches were performed focussing on different search terms. The first search concentrated on ‘fibre and enteral nutrition’ and search terms included fibre (any spelling), enteral nutrition, clinical nutrition, artificial nutrition, medical food, tube feed, sip feed and oral nutritional supplement. The second search focussed on fibre types (‘fibre subgroup’ search) and search terms comprised soy polysaccharides, resistant starch, oligofructose, oligosaccharides, inulin, pectin, arabic gum, guar gum, acacia, cellulose, pea* and oat*.

Bibliographies of identified trials and relevant articles were checked and experts consulted for any additional studies.

Study selection criteria, data extraction and outcome measures

Studies were deemed eligible for inclusion in the review if they conformed to predetermined inclusion and exclusion criteria. Subjects eligible for inclusion were healthy volunteers or patients >1 year of age of any nutritional status (well nourished and malnourished) and based in any setting (e.g. hospital, care home and at home). Studies in which patients did not receive oral- or tube feeding as their main source of nutrition were excluded. Studies including a self-selected (SS) diet arm were included to allow comparison with fibre-free/fibre-containing enteral feeds. However, studies involving infant formulae and studies that did not involve a fibre-free period were excluded. Although priority was given to RCTs, non-RCTs and observational cohort studies were also admissible. Epidemiological/dietary intake studies, animal or in vitro studies, literature reviews, observational case studies and acute studies (<3 days) were excluded. No other restrictions were placed on studies with regard to year of publication, format, language (providing an English abstract was available) and source.

Following the identification of potentially relevant studies based on title and abstract, full articles were obtained and evaluated by one researcher. A second independent assessor verified the validity of inclusion/exclusion decisions. Disputes as to eligibility were referred to the authors’ panel. A predetermined data extraction table was designed to capture study characteristics and outcome data and allow the assimilation of data from all study designs. Both subjective (incidence of diarrhoea/constipation, stool consistency and gastrointestinal symptoms) and objective (transit time, stool weight and bowel frequency) outcome measures were considered.

Quality assessment

The methodological quality of individual studies was assessed by one researcher and verified by a second assessor using the two scales which are commonly adopted by health technology appraisal groups worldwide. The first method was the one developed by Jadad et al.,21 an objective grading system which awards a score of 0–5 and assesses randomization (0–2 points), double blinding (0–2 points), and withdrawals and dropouts (0–1 point). The second was a six-point scale adapted from the Quality of Evidence Quality Assessment (AHCPR) scale.22 A high JADAD score is deemed to be of better quality. The converse is true for the AHCPR scale, where a lower score is indicative of higher quality.

Synthesis of data and statistical methods

Meta-analyses were conducted using a fixed effects model, which combines treatment estimates of effect size assuming that there is no heterogeneity between study results. Heterogeneity was investigated using the I2 statistic.23, 24 A mean treatment difference was considered to be statistically significant if the 95% confidence interval (CI) did not span the value of zero. Forest plots were used to present the effect size of each meta-analysis. Funnel plots to investigate publication bias are deemed inappropriate when a small number of studies are involved. Therefore, such plots and associated statistics (Begg and Mazumdar’s rank correlation and Egger’s regression intercept method) were only used when 10 or more studies were included in the meta-analyses,25 although 10 studies are still considered to be an inadequate number by many investigators.26 Linear meta-regression was undertaken. Like ordinary regression, meta-regression examines the extent to which a putative explanatory variable accounts for the variability or heterogeneity in results. Unlike ordinary regression, where the unit is an individual subject, in meta-regression the unit is an individual study. Studies are weighted so that the more precise ones have more influence in meta-regression analysis. In the figures of meta-analyses and meta-regressions, the area of the symbols is proportional to the weight of the studies.

All meta-analyses and meta-regressions were performed using comprehensive meta analysis Version 2.0 (Biostat, Englewood, NJ, USA). A P-value of <0.05 (two-tailed) was considered to be statistically significant. In the forest plots the squares indicate point estimates of treatment effect, with the size of the square representing the weight attributed to each study and the horizontal bars indicating 95% CI.

Results

Overall search findings

In total, 339 references were identified by systematic search of literature: 276 references from the ‘fibre and enteral nutrition search’ and 63 references from the ‘fibre subgroup’ search. Upon examination of the abstracts and/or full texts, 247 and 61 papers were excluded, respectively, leaving 31 papers. Cross-referencing and expert consultations identified a further 20 papers, giving a total of 51 studies, which complied with the inclusion criteria and were therefore included in the review (Figure 1).

Figure 1.

 Protocol of included and excluded papers in the literature search.

Of the 51 studies included in the review, 38 were available as full papers and 13 as abstracts. RCTs accounted for 43 of 51 (84%) studies, scoring the highest grade of 1 according to the Quality of Evidence Scale.22 However, the methodology of individual RCTs in individual cases was often poorly described (with regard to method of randomization, blinding and recording the fate of withdrawals) with only one study15 scoring the top grade of 5 on the Jadad scale. The remaining RCTs scored 4,10, 27–34 3,8, 35–40 2,9, 41–50 114, 51–62 or 0.63, 64 Eight studies were either nonrandomized or there was insufficient information to conclude whether they were randomized from the details available in the publication.65–72

Patient subgroups

The 51 studies were divided into subgroups, with studies involving healthy volunteers (13 studies, 171 subjects) considered separately from those in patients (38 studies, 1591 patients). The patient studies were further subdivided on the basis of the setting, whether adult or paediatric populations were considered, and based on underlying condition. The details of the subgroups are reported in Table 1.

Table 1.   Demographics of study groups
Study subgroupsMean number of patients/study (range)Mean treatment duration, days (range)
  1. * Including studies with mainly ICU patients and septic patients.

  2. RCT, randomized-controlled trial.

Adults
 Healthy volunteers (n = 13): RCTs9, 10, 15, 34, 43–47, 61, 62
 Non-RCTs67, 72
13 (8–22)10 (7–21)
 Intensive care unit (ICU)/critically ill* (n = 11): RCTs30–33, 39, 40, 48, 49, 64
 Non-RCTs68, 69
51 (9–100)11 (4–18)
 Surgical/post-operative (n = 8): RCTs28, 29, 38, 50, 55, 57, 59, 6047 (16–100)9 (5–15)
 Medical (n = 6): RCTs37, 42, 53, 56, 58
 Non-RCT65
56 (11–155)14 (5–28)
 Chronic care (n = 6): RCTs8, 14, 35, 36, 52, 5316 (7–28)24 (7–49)
Paediatrics
 Acute/medical (n = 2): RCTs51, 6372 (31–112)18 (15–21)
 Chronic care (n = 5): RCTs27, 41
 Non-RCTs66, 70, 71
14 (11–20)93 (14–300)

Route of formula administration

In the 13 studies conducted in healthy volunteers, subjects were fed via the oral route.9, 10, 15, 34, 43–47, 56, 61, 62, 72 In the patient studies (n = 38), enteral nutrition was administered via various routes:

Reported end points

Diarrhoea and constipation were the most frequently reported clinical end points. Stool/bowel frequency, transit time, stool weight, stool consistency, short-chain fatty acids (SCFA) and stool microflora were the most commonly reported physiological parameters. Tolerance was reported as a range of subjective end points and laxative use was also reported as it is often regarded as an indicator of bowel function. The frequency of the main end points reported in the 51 studies that have been included, is reported in Table 2.

Table 2.   End points reported in included studies
End point reportedNumber of studies reporting end point (not all study results were eligible for further analysis)
  1. * Tolerance was defined as the incidence of gastrointestinal symptoms and included nausea, vomiting, flatulence, abdominal cramps and bloating. The incidence of diarrhoea and constipation were considered as separate end points.

  2. SCFA, short chain fatty acids.

Diarrhoea (n = 23)
 Healthy volunteersn = 3/1334, 43, 46
 Patients n = 20/388, 28–32, 37–41, 53, 56–58, 60, 63, 64, 68, 69
Constipation (n = 17)
 Healthy volunteersn = 4/1310, 43, 46, 61
 Patientsn = 13/388, 14, 28–30, 33, 35, 38, 41, 50, 56, 58, 60
Stool/bowel frequency (n = 29)
 Healthy volunteersn = 10/139, 10, 15, 34, 43, 45–47, 61, 62
 Patientsn = 19/388, 14, 27, 35, 36, 41, 42, 49–56, 60, 65, 66, 70
Transit time (n = 14)
 Healthy volunteersn = 9/139, 34, 43, 45, 46, 61, 62, 67, 72
 Patientsn = 5/388, 27, 41, 52, 53
Faecal mass (n = 18)
 Healthy volunteersn = 10/139, 10, 15, 34, 45, 46, 61, 62, 67, 72
 Patientsn = 8/388, 39, 42, 52, 53, 55, 58, 66
Stool consistency (n = 24)
 Healthy volunteersn = 5/1310, 34, 43, 45, 47
 Patientsn = 19/3814, 27, 32, 33, 36, 37, 39, 40, 49–52, 54–56, 65, 66, 69, 70
Laxative use (n = 10)
 Patientsn = 10/388, 27, 35, 49, 52, 54, 56, 63, 66, 70
Microflora (n = 4)
 Healthy volunteersn = 1/1315
 Patients n = 3/3814, 35, 65
SCFA (n = 8)
 Healthy volunteersn = 5/139, 10, 15, 34, 47
 Patients n = 3/3814, 35, 65
Tolerance (n = 26)*
 Healthy volunteersn = 5/1315, 34, 43, 46, 62
 Patientsn = 21/388, 14, 27, 29, 33, 35, 38, 40–42, 48, 51–53, 55–58, 63, 68, 70

Several other end points, such as plasma glucose levels,10, 33, 50 mineral retention44, 71 or cholesterol levels33, 53, 54 were reported in a small number of studies only and were therefore not considered further.

Amount and type of fibre sources in the studies included

Over 15 different fibre sources administered either as a sole fibre source or as part of a mixture, were employed in the studies included (Table 3). Soy polysaccharide was the most frequently studied fibre source, followed by a mixture of six different fibres (Nutricia Multi Fibre, Nutricia, Zoetermeer, The Netherlands; used in 16% of all studies). In 65% of the studies a single fibre source was used, in 8% of studies a mixture of two sources, in 4% a mixture of five sources and in 6% of studies insufficient details were reported. A number of studies investigated the effects of different fibre sources.9, 32, 34

Table 3.   Fibre sources used in studies
Fibre source*Studies evaluating fibre source (%)Number of fibresMean fibre intake (range; g/day)†
  1. * Some studies examined more than one fibre source.

  2. † As reported in the articles, amount of total dietary fibre was used whenever available.

  3. § Majority of patients in study received Nutricia Multi Fibre.

  4. ‡ Initial dose given.

  5. FOS, fructooligosaccharide.

Soy polysaccharides (n = 20)8, 9, 27–30, 34, 36, 39, 44, 45, 52, 55, 58, 59, 64, 66, 67, 69, 7139123 (6–35)
Soy polysaccharides, alpha-cellulose, arabic gum, inulin, oligofructose, resistant starch (Nutricia Multi Fibre; n = 8)35, 41, 46, 54§,57, 59, 63, 7016630
Guar gum, hydrolysed (n = 7)10, 31, 33, 34, 38, 43, 4814130 (15–42)
Soy polysaccharides/oat (n = 2)50, 53427‡–28.8
Pectin (n = 2)32, 474123–40
Psyllium (n = 2)37, 494114
Oat, soy, gum arabic, carboxymethylcellulose, FOS (n = 2)32, 51457–32
Inulin (n = 1)142133.8
Cellulose (n = 1)612130
Oat (n = 1)92115
Galactomannan (n = 1)652128
FOS (n = 1)562118.5
Carrot (pectin/cellulose) (n = 1)422224
FOS/pea (n = 1)152226
Soy oligosaccharides (n = 1)92115
Ispaghula husk (n = 1)40217
Oat, soy, gum arabic, carboxymethylcellulose, FOS, pectin (n = 1)322629
Not reported (n = 3)62, 68, 726 20

Overall, there was considerable variability between the studies in the amount of fibre administered daily (Table 3). The actual daily fibre intake was not explicitly reported in 18 studies,10, 29–31, 35, 36, 41, 48, 55, 57, 59, 60, 63, 64, 68–70, 72 however, it was possible in several cases to calculate the actual intake based on information reported in the article. Wherever available, the amount of fibre intake was based on actual fibre analysis by various methods, e.g. by the methods of AOAC, and Southgate.4, 73 In all other cases, data on fibre intake was used as stated in the article. Fructooligosaccharides (FOS and inulin) were assumed to be part of the total fibre intake.

Of the 51 studies which reported the fibre source, 26 studies administered fibre as an integrated constituent of the formula8, 15, 29–32, 35, 36, 39, 41, 46, 50,52–54, 56–60, 63, 64, 66, 69–71 compared with 20 studies where the fibre source was added to the feed.9, 10, 14, 27, 28, 33, 34, 37, 38, 40, 42–44, 47, 49, 51, 55, 61, 65, 67 In the study by Slavin et al., patients received three different fibre regimens over the course of the study, one of which was an integrated formula compared with two, where the fibre was added to the feed.45

In seven studies (four conducted in healthy volunteers44, 45, 61, 72 and three in patients29, 66, 71) subjects received different amounts of fibre during different treatment phases, allowing the effect of the amount of fibre on bowel function to be studied. Considering the studies enrolling healthy volunteers, they compared: 30 and 60 g of soy polysaccharides/day,45 20, 30 and 40 g/day of either cellulose/day61 or soy polysaccharide44 and low/moderate/high levels of fibre.72 Patient-enrolled studies compared high (14 g/L) with moderate (7 g/L) levels of soy polysaccharide29 or administered an increase in daily fibre intake from 15 to 21 g during the final study period.66, 71

Bowel-related outcomes

Diarrhoea.

There was widespread inter-study variation in the quantification of diarrhoea in the studies that were reviewed. Twenty-two studies defined diarrhoea and the definitions reported can be grouped into four categories:

Some studies reported a diarrhoea-related end point without explicitly providing a definition for diarrhoea.41, 55, 56, 68

Only two of the studies conducted in healthy volunteers reported the incidence of diarrhoea. Increased watery stools were observed in the fibre-free and the guar gum groups compared with subjects treated with feeds supplemented with soy polysaccharide34 or Nutricia Multi Fibre.46

A meta-analysis was undertaken of the incidence of diarrhoea (defined as number of patients with diarrhoea) in the hospitalized patients participating in RCTs (16 data sets from 13 RCTs; Table 2; n = 338 in fibre group and n = 345 in fibre-free group). Incidence of diarrhoea was found to be significantly reduced as a result of fibre administration [odds ratio (OR) 0.68; 95% CI: 0.48–0.96; P = 0.03, Figure 2, overall results]. There was no evidence of publication bias (Begg and Mazumdar’s rank correlation, P = 0.39; Eggers’s regression intercept method, P = 0.71), although there was significant heterogeneity between the studies (test of overall heterogeneity, I2 = 38%, P = 0.05). This heterogeneity resulted from the ICU studies (I2 = 44%; P = 0.09) rather than the non-ICU studies (five surgical, one medical and one paediatric; I2 = 0%, P = 0.50). Thus the incidence of diarrhoea in the ICU studies was variable, ranging from 9–92% in the fibre-free groups. Subgroup analyses (Figure 2) revealed a significant reduction in the incidence of diarrhoea in the non-ICU studies (OR 0.42, 95% CI: 0.25–0.72; P = 0.001; eight data sets from seven RCTs, n = 185 in fibre group and n = 183 in fibre-free group) and in the surgical studies alone (data not reported). However, this effect was attenuated when ICU patients were analysed as a distinct group (OR 0.98, 95% CI: 0.62–1.56; P = 0.93; eight data sets from six RCTs, n = 153 in fibre group and n = 162 in fibre-free group).

Figure 2.

 Meta-analysis (fixed effect model) of the incidence of diarrhoea (proportion of patients with diarrhoea/proportion of patients without diarrhoea) in tube-fed hospitalized patients participating in randomized-controlled trials. *Fibre mixture: oat, soy, gum arabic, carboxymethylcellulose, fructose oligosaccharides; **fibre mixture: soy polysaccharide, alpha-cellulose, arabic gum, inulin, oligofructose, resistant starch (Nutricia Multi Fibre). OR of <1.0 (favouring fibre) indicates that administration of a fibre-containing feed reduces the incidence of diarrhoea.

The 16 datasets were divided into eight studies with the highest (>30%) and lowest (<30%) incidence of diarrhoea in the fibre-free group. A significant reduction was found in the incidence of diarrhoea in the high incidence group (OR 0.61, 95% CI: 0.42–0.92; I2 = 49%, P = 0.06), but not in the low incidence group (OR 0.88, 95% CI: 0.46–1.71; I2 = 38%, P = 0.06).

A meta-regression of the 16 data sets indicated that the extent to which the incidence of diarrhoea was reduced by fibre supplementation was related to the incidence of diarrhoea in the group receiving a fibre-free diet [Figure 3: intercept, 0.62 log OR; slope (log OR/proportion of patients with diarrhoea) −3.4 (S.E. 0.91); z = −2.2, P = 0.0002]. Thus, beneficial effects were more likely to occur when the incidence of diarrhoea was high. This effect was maintained when the ICU [intercept, 1.3 log OR; slope −3.5 (S.E. 1.3); P = 0.008] and non-ICU subgroups [intercept, 0.37 log OR; slope −2.9 (S.E. 1.3); P = 0.001] were analysed separately. This type of meta-regression is subject to a certain degree of bias74 (regression to the mean, which increases the strength of the relationship), as a measurement error in the fibre-free group also appears to directly influence the dependent variable (treatment effect). However, the large variation in risk across trials (9–92% of patients had diarrhoea) would be expected to reduce this type of error.

Figure 3.

 Meta-regression of proportion of patients with diarrhoea in fibre-free group on log odds ratio for development of diarrhoea (proportion with diarrhoea/proportion without diarrhoea). The horizontal dotted line indicates that fibre-containing feeds have no effect on the incidence of diarrhoea in comparison with fibre-free feeds, while values below the dotted line indicate that fibre decreases the proportion of patients with diarrhoea.

The level of fibre intake was reported in six of these studies, ranging from 14.0 to 34.9 g/day.28, 30, 31, 37–39 However, no significant relationship was observed between the level of fibre intake during feeding and incidence of diarrhoea [intercept (log OR) −0.33; slope 0.0009; z = −0.24; P = 0.81; n = 135 in fibre group and n = 135 in fibre-free group].

A meta-analysis was also undertaken of seven studies, where the incidence of diarrhoea was defined as the percentage of days with diarrhoea, as opposed to the percentage of patients with diarrhoea as discussed above. These studies included three in the ICU,31, 40, 64 two medical,37, 58 one paediatric41 and one surgical28 (n = 173 in fibre group, n = 170 in fibre-free group). This analysis showed that fibre reduced the duration of diarrhoea compared with the fibre-free group but the effect was not significant (OR 0.60, 95% CI: 0.34–1.08; P = 0.09; test of heterogeneity, I2 = 0%, P = 0.93). All but one of these studies were performed in adults with acute conditions and when this study41 (enrolling children with chronic conditions) was excluded, meta-analysis of the remaining six studies (n = 157 in fibre group and 154 in control group), confirmed that fibre reduced the frequency of diarrhoea compared with the fibre-free group, although the effect was still not significant (OR 0.61, 95% CI: 0.33–1.11; P = 0.11; test of heterogeneity, I2 = 0%, P = 0.86).

Antibiotic use.

Antibiotic use was reported in a total of 15 studies,28–30, 32, 33, 37, 39, 40, 48–50, 54, 55, 63, 64 although antibiotics may have been prescribed in other studies without reporting. The route of antibiotic administration was not explicitly stated in any of the studies. Nine studies reported the effect of antibiotic usage on diarrhoea incidence. While there was no effect on incidence of diarrhoea in four studies,28, 32, 50, 54 five reported an increased incidence of diarrhoea in patients receiving antibiotic therapy.29, 39, 40, 55, 64 Regarding any relationship between the particular antibiotic prescribed and the incidence of diarrhoea, the study by Frankenfield et al. reported that every subject who developed diarrhoea received a combination of clindamycin and gentamicin (plus ranitidine).39 In a further study, subjects receiving cefazolin and metronidazole experienced a lower incidence of diarrhoea compared with subjects receiving other antibiotics.29

Constipation.

In total, 24 studies reported constipation as a clinical outcome (includes studies reporting laxative use). A variety of definitions was used in eight studies out of above, which reported:

  • 1‘Hard stool retains its shape, with difficulty in defecation’;10
  • 2‘The development of infrequent hard stools’;58
  • 3‘Absence of any bowel movement over a period of 72 h’;8, 28
  • 4‘Zero bowel frequency over 10-day study period’;50
  • 5‘Intestinal evacuation less frequent over every 2–3 days, or increased flatus in combination with abdominal distension’;38
  • 6‘Absence of bowel movements for more than 3 days or the inability to pass a bowel movement after straining or pushing for more than 10 min’;33
  • 7‘Patients were classified as constipated if they required use of laxatives, suppositories or enemas’.29

Some studies reported a constipation-related end point without explicitly providing a definition for constipation.30, 35, 41, 56, 60, 61

A meta-analysis of the incidence of constipation (reported in various ways) was possible on data from seven RCTs in the acute setting (two ICU,30, 33 three surgical28, 50, 60 and two medical;38, 56n = 144 in fibre group, of which 23 had constipation, and 149 in fibre-free group, of which 36 had constipation). There was a nonsignificant trend for fibre to reduce the percentage of patients reporting constipation (test of overall effect, OR 0.57, 95% CI: 0.30–1.10; P = 0.09; test of heterogeneity, I2 = 0%; P = 0.51).

The effect of fibre on laxative use for the treatment of constipation was reported in four studies in the acute setting.49, 54, 56, 63 Following fibre supplementation, use of elimination aid was either significantly reduced54, 56, 63 or unchanged.49 Constipation was also considered to be a problem in the study of Vandewoude et al. (longer term medical patients) where patients were prescribed laxatives as required.54

Due to the paucity of data, it was not possible to meta-analyse the effect of fibre-supplemented feeds on constipation in patients with chronic conditions. For example, Schneider et al. reported that 53% of stable patients considered themselves constipated, although a comparison between intervention and fibre-free groups was not reported.35 Two studies8, 29 showed no differences between those receiving the fibre-containing and fibre-free formulae with the study by Shankardass et al. being confounded by the use of laxatives.8

Five studies in the chronic setting reported the use of laxatives/enemas for the treatment of constipation,8, 27, 35, 52, 66 with all but one study35 reporting the effect of fibre supplementation on elimination aid use. Following fibre supplementation, use of elimination aid was either significantly reduced,8 minimally reduced52 or unchanged.27, 66

Bowel frequency.

Healthy volunteers: Bowel frequency during administration of fibre-free and fibre-containing feeds was reported in 14 data sets derived from nine RCTs (n = 154, fibre-containing feed; n = 153, fibre-free feed).9, 10, 15, 34, 43, 46, 47, 61, 62 In a meta-analysis (Figure 4), the fibre-containing feed significantly increased mean bowel frequency [test of overall effect, 0.14 (S.E. 0.05) times/day, z = 2.8, P = 0.005; test of heterogeneity, I2 = 26%, P = 0.17]. A funnel plot and associated statistical tests [Begg and Mazumdar’s rank correlation (P = 0.17) and Egger’s regression intercept method (P = 0.14)] did not suggest publication bias. The results remained significant when the study of Kies et al. was excluded from the analysis [test of overall effect, 0.12 (S.E. 0.05) times/day, z = 2.2; P = 0.03]. The study of Kies et al. is unique in that four of the 18 healthy subjects given a feed-containing cellulose (Sulka-floc), developed severe constipation with impaction, which required digital evacuation.61 It is not clear if the reported frequency of bowel action included these evacuations. No such impactions were observed in subjects receiving a fibre-free feed.

Figure 4.

 Meta-analysis (fixed effect model) comparing bowel frequency (times/day) in healthy subjects receiving a fibre-containing feed or a fibre-free feed. *Fibre mixture: soy polysaccharides, alpha-cellulose, arabic gum, inulin, oligofructose, resistant starch (Nutricia Multi Fibre); FOS: fructooligosaccharides. A value >0 (favouring fibre) indicates that administration of a fibre-containing feed increases bowel frequency compared with fibre-free feeds.

Meta-regression of these nine RCTs indicated that fibre supplementation increased bowel frequency in those in whom the frequency was low, and decreased it in those in whom it was high, and had little or no effect in those with an average frequency that was close to one bowel movement per day. This relationship was of borderline significance if the study by Kies et al.61 was included [intercept (mean difference), 0.38 times/day, slope −0.27 (S.E. 0.15); z = −1.8; P = 0.08]. However, a highly significant relationship was reported when this study was excluded from the analysis [Figure 5: intercept, 0.72 times/day; slope 0.73 (S.E. 0.17); z = −3.8, P = 0.0002; 11 data sets from eight RCTs, n = 100 in fibre group and n = 99 in fibre-free group]. However, despite the observed strong relationships, these results should be regarded carefully, as they are subject to bias associated with regression to the mean, which tends to increase the strength of the relationships.

Figure 5.

 Meta-regression of bowel frequency (fibre-free group) on difference in mean values (bowel frequency on fibre-containing feed − bowel frequency in fibre-free feed) in healthy subjects. The horizontal dotted line indicates no difference in bowel frequency when a fibre-containing feed is compared with a fibre-free feed. Values above the dotted line (subjects having low bowel frequency) indicate that fibre-containing feeds increase bowel frequency and values below the line (subjects having a high frequency) report a decreased bowel frequency compared with a fibre-free feed.

Patients: Data were available from six studies (five RCTs) in patients (five adult and one paediatric patient population) requiring chronic care (n = 80 both in fibre-supplemented and fibre-free group).8, 14, 41, 52, 53, 66 Supplementation of the enteral feed with fibre significantly increased bowel frequency [test of overall effect, 0.27 (S.E. 0.08) times/day, z = 3.2, P = 0.001; test of heterogeneity, I2 = 0%, P = 0.68]. When the meta-analysis was repeated following exclusion of the non-RCT study66 (five RCTs; n = 69 both in fibre and fibre-free groups), the effect of fibre in increasing bowel frequency remained significant [test of overall effect, 0.25 (S.E. 0.1) times/day, z = 2.5, P = 0.01; test of heterogeneity, I2 = 0%, P = 0.57]. Exclusion of the paediatric study41 did not affect the result [test of overall effect, 0.25 (S.E. 0.10) times/day, z = 2.6, P = 0.009; test of heterogeneity, I2 = 0.00, P = 0.52].

Three RCTs enrolled patients with acute medical conditions (n = 95 in fibre group and 105 in fibre-free group).42, 54, 65 Fibre supplementation led to a significant decrease in bowel frequency compared with fibre-free enteral feeds [overall test of effect, −0.22 (S.E. 0.08) times/day, z = −2.7, P = 0.006], although the inter-study results showed significant heterogeneity (test of heterogeneity, I2 = 72%, P = 0.03). An additional study in patients with head and neck cancer,55 which could not be included in the above analysis due to lack of relevant information in the abstract, reported that fibre supplementation increased bowel frequency by 0.33 times/day compared with the fibre-free control group (z = 2.3, P = 0.02). The authors considered that results for both stool consistency and bowel frequency were confounded by antibiotic treatment.

Although meta-analysis of combined results from the acute and chronic studies (nine RCTs8, 14, 41, 42, 52–55, 65 and one non-RCT66) reported that fibre-containing feeds had no significant overall effect on bowel frequency because of substantial variability between studies (test of heterogeneity, I2 = 71%; P = 0.001; overall difference of 0.05 times/day; z = −0.98; P = 0.33), meta-regression demonstrated a highly significant moderating effect of fibre on bowel frequency, analogous to that found in normal subjects. Fibre supplementation decreased bowel frequency in those with high bowel frequency, and increased it in those patients with low bowel frequency. The highly significant relationship was observed in analyses involving RCTs and non-RCTs of adults and children [intercept, 0.39 times/day, slope −0.71 (S.E. 0.18), z = −3.9, P < 0.0001], only RCTs of adults and children [n = 175 in fibre group and n = 189 in control group; intercept, 0.36 times/day; slope −0.68 (S.E. 0.19); z = −3.6; P < 0.0001], and only RCTs of adults [n = 159 in fibre group and n = 173 in fibre-free (control) group; intercept, 0.47 times/day; slope −0.91 (S.E. 0.23); z = −3.9, P = 0.0001].

Transit time.

Healthy volunteers: A meta-analysis of 14 data sets from seven RCTs9, 34, 43, 45, 46, 61, 62 (n = 175 both in the fibre-containing and fibre-free group) investigated the effect of fibre supplementation on whole gut transit time in healthy volunteers. Administration of a fibre-supplemented enteral feed resulted in significantly faster gastrointestinal transit (Figure 6: test of overall effect, −9.3 (S.E. 2.29) h, z = −4.0, P < 0.001]. The analysis showed no statistical indication of publication bias [Begg and Mazumdar’s rank correlation (P = 0.13); Egger’s regression intercept method (P = 0.67)]. There was significant heterogeneity between studies (I2 = 55%, P = 0.007), which may reflect the variety of fibre sources used in the enteral feed between studies.

Figure 6.

 Meta-analysis (fixed effect model) comparing transit time (hours) in healthy subjects receiving fibre-containing feed or fibre-free feed. *Fibre mixture: soy polysaccharides, alpha-cellulose, arabic gum, inulin, oligofructose, resistant starch (Nutricia Multi Fibre). A negative value (favouring fibre-containing group) indicates that administration of a fibre-containing feed reduces transit time compared with a fibre-free feed.

With respect to the effects of fibre source on transit time, a significant reduction was reported with the administration of a feed containing six different types of fibres (Multi Fibre) compared with a fibre-free feed (50.4 ± 6.8 h vs. 75.7 ± 12.6 h; P < 0.05, paired t-test).46 Furthermore, a significant reduction in transit time was reported for studies administering soy polysaccharide (test of overall effect, −19.3 h, z = −5.5, P < 0.001; five data sets from three RCTs9, 34, 45 with n = 66 in both fibre-free and fibre-supplemented group) or cellulose [test of overall effect, −12.7 h, z = −2.4, P = 0.017; three data sets from one RCT61 (n = 54 in both fibre-free and fibre-supplemented group)]. The effects of oat, soy oligosaccharides and hydrolysed guar gum on transit time were not significant.9, 43 A meta-regression vs. fibre dose showed that increased intake had a tendency to shorten transit time, however, results were not significant [intercept, −0.32 (S.E. 0.18), z = −1.7, P = 0.09].

Patients: Whole gut transit time was investigated in three RCTs (two in adults and one in a paediatric population27, 41, 53), each of which administered a different fibre type (soy, oat and soy and fibre mixture). A combined meta-analysis of all studies reported no significant effect [test of overall effect, 1.7 (S.E. 6.4) h, z = 0.26, P = 0.79; test of heterogeneity, I2 = 52%; P = 0.12; very similar tests of significance were obtained when standardized differences were used in the analysis]. On account of the limited number of studies, meta-regression was not performed.

Faecal mass.

Healthy volunteers: An analysis of 12 data sets from seven RCTs in healthy volunteers (n = 129 in fibre-supplemented and fibre-free groups),9, 10, 15, 34, 45, 46, 62 reported that mean faecal mass was significantly increased following treatment with a fibre-containing feed compared with fibre-free feed: 109 (S.E. 26) g vs. 74 (S.E. 21) g, respectively [Figure 7: test of overall effect, 35 g, z = 6.2, P < 0.001; test of heterogeneity, I2 = 4% (P = 0.41)]. An increase in faecal mass of 1.1 g/g (s.d. 0.5) of supplemented fibre was observed when the weighted mean results from each study were used.

Figure 7.

 Meta-analysis (fixed effect model) comparing faecal mass (g/day) in healthy subjects receiving a fibre-containing feed or a fibre-free feed. *Fibre mixture: soy polysaccharides, alpha-cellulose, arabic gum, inulin, oligofructose, resistant starch (Nutricia Multi Fibre); FOS: fructooligosaccharides. A positive value (favouring fibre) indicates that administration of a fibre-containing feed was associated with greater faecal mass (g/day) compared with a fibre-free feed.

Using meta-regression, a statistically significant positive relationship was reported between faecal mass and fibre intake [intercept, 3.9 g; gradient, 0.98 (S.E. 0.41) g/g fibre; z = 2.4, P = 0.02]. The increase in faecal mass was not significantly related to either study duration or the baseline faecal mass of the fibre-free control group.

In a meta-analysis of five data sets from two RCTs34, 61 (n = 76 in fibre-supplemented and fibre-free groups), fibre supplementation resulted in a variable increase in faecal dry weight compared with the fibre-free group [test of overall effect, 16.5 (S.E. 1.4), z = 11.8, P < 0.001; test of heterogeneity, I2 = 92%; P < 0.001]. Three doses of cellulose (20, 30 or 40 g) were administered in one study61 while the second study compared guar gum with soy polysaccharide (15 g in both treatment arms).34 The observed increase in faecal dry weight was significantly related to the dose (15–40 g) of fibre administered (z = 6.8, P < 0.0001). The weighted mean increase in dry faecal mass was 0.59 (s.d. 0.25) g/g of fibre added [the corresponding nonweighted value was 0.53 (s.d. 0.30) g/g of fibre added].

In four RCTs a comparison of fibre-free feed with SS diet was performed.9, 34, 45, 46 A meta-analysis of five data sets from these four RCTs reported that faecal mass was twofold greater in subjects ingesting a SS diet (n = 50), compared with those ingesting a fibre-free feed (n = 50) [weighted mean mass, 159 g/day (s.d. 69) vs. 79 g/day (s.d. 33)]. This corresponded to an approximate 5 g increase in faecal mass per gram of supplemented fibre (weighted value, 5.0 g/g fibre; nonweighted mean value, 4.6 g/g fibre).

A meta-analysis of nine data sets (n = 99 in both treatment arms) based on these four RCTs9, 34, 45, 46 reported that faecal mass was also increased following ingestion of a SS diet [154 (S.E. 6) g/day] compared with a fibre-containing feed [112 (S.E. 6) g/day; Figure 8: test of overall effect (difference in mean values), 42 (S.E. 8.4) g/day, z = 5.0; P < 0.001; test of heterogeneity I2 = 27%; P = 0.21]. This increase was observed even though the SS diets were reported to contain approximately half the amount of fibre compared with the fibre-containing feeds (15.1 vs. 31.5 g/day). Inter-study variation in the difference in fibre intake between treatment arms was observed and as this difference became smaller, a corresponding decrease in faecal mass was reported [meta-regression intercept, 70.9 g; gradient −1.7 (S.E. 0.60) g faeces/g fibre; P = 0.003].

Figure 8.

 Meta-analysis (fixed effect model) comparing faecal mass in healthy subjects receiving a fibre-containing feed and a self-selected (SS) diet. Fibre-type refers to the fibre-containing feed. A positive value (favouring SS diet) indicates that SS diet increases faecal mass compared with a fibre-containing feed.

Patients: Faecal mass was reported in seven studies: six RCTs (three chronic care,8, 52, 53 one ICU39 and two medical populations42, 58) and one non-RCT (chronic-care paediatric population;66 total n = 75 in fibre group and n = 75 in fibre-free group). Mean faecal mass was greater in the fibre group compared with the fibre-free group in all of these studies with the exception of the ICU study where faeces were reported to be ‘watery’ and faecal mass decreased significantly over time.39 A meta-analysis of all studies showed a significant increase in faecal mass following fibre supplementation of the feed [test of overall effect, 21.2 (S.E. 6.4) g/day, z = 4.7, P < 0.001; test of heterogeneity, I2 = 0%, P = 0.50]. A meta-analysis of the six RCTs only (n = 64 in fibre-free group; n = 64 in fibre group) confirmed the significant effect of fibre in increasing faecal mass [test of overall effect, 16.7 (S.E. 5.2) g/day, z = 3.2, P = 0.001; test of heterogeneity, I2 = 0%, P = 0.87].

Two RCTs reported a significant increase in dry faecal mass following administration of a fibre-containing feed.53, 58 Medical patients in the study by Heymsfield et al.58 received 8 g soy fibre/day [dry faecal weight: 27 (s.d. 9) g/day vs. 12 (s.d. 5) g/day; P < 0.01], whereas the chronic-care patients in the study by Zarling et al.53 received 28.8 g of oat and soy fibre/day [dry faecal weight, 57 (s.d. 31) g/day (faecal moisture content of 72%) vs. 32 (s.d. 25) g/day (faecal moisture content of 74%), P < 0.05].

The effect of fibre supplementation on faecal moisture and pH showed considerable inter-study variation. Studies reported a significant increase,34, 66 decrease61, 65 or no effect15, 52, 53 on faecal moisture levels. The effect was observed to be dependent on both the type and amount of fibre administered. In the study by Lampe et al.34 a significant increase in faecal moisture was reported for soy, but not guar gum and Kies et al.61 reported a significant decrease in faecal moisture compared with fibre-free treatment, but only following treatment with the higher doses of cellulose (30 and 40 g). Treatment with fibre-containing formula was reported to have no effect34 or to significantly decrease15, 65 faecal pH compared with fibre-free formula.

Microflora.

The effects of fibre supplementation on microflora levels were reported in four studies. These comprised one study in healthy volunteers receiving a combination of FOS and pea fibre,15 medical patients receiving galactomannans,65 and chronic-care patients receiving either inulin14 or a mixture of six fibres.35 Bacterial counts (measured by cells or colony-forming units per gram of faeces) were assessed via plating techniques,35, 65 fluorescence in situ hybridization15, 35 or measurement of the level of indican.14

Compared with fibre-free formula, fibre supplementation was associated with either no change14, 65 or a significant increase15 (11.0 log10 cells/g dry faeces vs. 11.2 log10 cells/g dry faeces, respectively; P = 0.005) in total bacterial counts. Whelan et al. also reported significantly increased concentrations of bifidobacteria following fibre treatment.15 Although the study by Nakao et al. reported no change in total bacterial levels, a significant decrease in the number of aerobic bacteria was reported following soluble fibre supplementation.65

Finally, the study by Schneider et al. reported a significant increase in the numbers of enterococci at the end of the fibre-free treatment compared with baseline and a significant increase in the number of Bacteroides at the end of the fibre-supplemented treatment compared with the fibre-free liquid diet.35

Total short-chain fatty acids.

A total of eight RCTs examined the effect of fibre supplementation on total SCFA concentrations in faeces.9, 10, 14, 15, 34, 35, 47, 65 None of these studies explicitly reported the use of antibiotics. Five studies were performed on healthy volunteers, three of which reported no differences in SCFA production following fibre supplementation.10, 34, 47 In contrast, two studies reported significant increases in either total SCFA, acetate and propionate levels following supplementation with FOS/pea15 or total SCFA and butyric acid levels for one of the three fibre sources being studied (soy oligosaccharide vs. soy polysaccharide or oat).9 In patients, one study reported no difference in SCFA concentrations compared with fibre-free treatment periods14 with the remaining two studies reporting a significant increase in total SCFA levels following fibre administration.35, 65 However, many of the studies were associated with large interindividual differences and the conflicting results may be a consequence of the various fibre source used in the enteral feeds and also the sample collection method used.

Stool consistency.

Stool consistency was reported as an end point in 24 of 51 (47%) studies (five healthy volunteers, 19 patient studies; Table 2). In 17 of 24 (71%) studies, stool consistency was measured using a subjective rating scale ranging from liquid to hard stools.10, 14, 27, 34, 36, 37, 39, 43, 45, 49–51, 54, 56, 65, 66, 70 Other definitions used included the frequency of liquid/formed stools/day32, 33, 47, 52, 55, 69 and a diarrhoea score.40

Healthy volunteers: Five of the 13 RCTs involving healthy volunteers reported stool consistency as an end point.10, 34, 43, 45, 47 In three studies, addition of fibre in the form of pectin,47 guar gum10 or soy fibre45 had a favourable effect, resulting in a more formed/normal stool compared with a hard/dry or liquid stool. In the remaining two studies, no significant effect on stool consistency following fibre supplementation was found (either guar gum34, 43 or soy fibre).34

Patients: In the chronic-care setting, three of the six studies in adult patient populations14, 36, 52 and three of the five studies in paediatric populations reported stool consistency as an end point.27, 66, 70 Although none of the studies in the adult population reported a significant change in stool consistency following fibre supplementations, one study reported ‘a trend towards more soft unformed and soft formed’ stools in this constipated population following addition of soy fibre for a 2-week period.52 In the paediatric population, a variable effect of fibre supplementation on stool consistency was reported. In the study by Grogan et al. in which a mixture of six fibres (Nutricia Multi Fibre) was administered for 4 weeks, three children with loose bowel movements showed a clinical improvement in stool consistency.70 A year-long study in which patients received soy fibre from day 61, a significant improvement (P = 0.01) in stool consistency, compared with the initial fibre-free period, was reported only for those patients not experiencing loose stools at the outset of the study.66 Interestingly, a 50% increase in the amount of fibre administered in the final 60-day study period had no effect on stool consistency. In the third study, no significant difference in stool consistency was reported between treatment groups.27

No significant effect on stool consistency following treatment with a mixture of five fibres (P = 0.17) was reported in a paediatric medical population.51

Stool consistency was reported in six of the 11 studies performed in the ICU population.32, 33, 39, 40, 49, 69 Significantly improved stool consistency (more formed stool) was reported in four RCTs, employing psyllium,49 soy fibre,39 guar gum33 or an unknown fibre69 as the fibre source. The remaining two studies reported no significant difference in stool consistency (as determined by diarrhoea score or number of patients with formed stool) between fibre-free and fibre-supplemented treatments (pectin/fibre mixture32 or Ispaghula husk40).

Stool consistency was reported for one of the seven studies in the surgical population but no significant difference was reported for consistency following fibre (soy fibre/oat) supplementation.50 Considering the adult medical population, stool consistency was reported for five of the seven studies.37, 54–56, 65 A statistically significant improvement in stool consistency was reported in two of these studies following fibre supplementation.54, 55 Vandewoude et al. reported a reduction in the incidence of hard faeces and a change to more soft pasty faeces following fibre (Nutricia Multi Fibre) supplementation54 and Frascio et al. reported an increased frequency of solid stools in patients administered a soy-enriched formula.55 Although there was no significant difference between the derived stool consistency score in the study by Belknap et al., patients treated with psyllium reported both a significantly lower number of liquid stools and a higher number of normal stools.37 While an improvement in stool consistency (less liquid) was noted following a 4-week administration of galactomannan as a result of a significant decrease in faecal water content,65 no significant difference between treatment groups was reported in the final study which administered FOS as the sole source of fibre.56

Tolerance.

For the purpose of the present review, symptoms of nausea, vomiting, flatulence, abdominal cramps and bloating were considered indicators of tolerance. The issues of diarrhoea and constipation have been addressed separately.

Overall the fibre-containing feeds were well tolerated. The incidence of intolerance symptoms was mentioned in 26 studies (five in healthy volunteers, 21 in patients; Table 2) with no difference between fibre-supplemented and fibre-free formulae reported in 21 studies: healthy volunteers34, 43, 46, 62 and patients.8, 27, 29, 33, 35, 41, 42, 48, 51–53, 55–58, 63, 70 A small number of studies in patients reported an increase in flatulence,14, 38 abdominal distension,40 vomiting,68 whereas Whelan et al.15 reported an increased incidence of all three symptoms in healthy subjects following fibre supplementation.

In a chronic-care population, Sobotka et al. reported that supplementation with inulin led to a significant increase in the number of days with flatulence [9.9 (s.d. 7.4) days vs. 1.4 (s.d. 1.1) days; P < 0.05] for six of nine patients.14 Other studies reporting this outcome reported no significant effect of fibre supplementation on flatulence in a variety of patient populations: critical care (hydrolysed guar),48 medical (FOS56 or soy fibre55, 58), paediatric medical (fibre mixture)63 and paediatric chronic care (soy fibre27).

A meta-analysis was performed on three RCTs (four data sets), which enrolled patients from various clinical settings: ICU,33 surgical29 and medical.38 No statistically significant difference in the percentage of patients reporting flatulence was observed between the fibre-supplemented (27%) and fibre-free (23%) populations (test of overall effect, OR 1.5, 95% CI: 0.61–3.35; P = 0.4).

Gastrointestinal intolerance may have a detrimental effect on the daily intake of enteral feed that can be administered to the patient. The mean volume of actual formula intake was reported in 28 studies (six in healthy volunteers15, 34, 43, 45–47 and 22 in patients8, 14, 27, 28, 30–33, 37–40, 48, 51, 52, 54, 56, 58, 65, 66, 68, 71). Of these, 22 of 28 studies (three healthy volunteers and 19 patient studies) reported no difference in mean formula volume intake between fibre-supplemented and fibre-free groups, two of 28 (7%) studies (both studying patient groups) reported increased mean formula volume in fibre-supplemented groups33, 68 and two of 28 (7%) studies (one in healthy volunteers and one in patients) reported increased mean volume in the fibre-free group.15, 39

Gastric emptying was assessed in two RCTs which enrolled patients with chronic conditions.27, 53 Gastric emptying was assessed either by addition of a radioactive marker with 50 mL of feed53 or by measuring the percentage of radioactivity remaining in the stomach 1 h postingestion.27 Neither study reported a significant effect of fibre supplementation.

Effect of varying amounts of fibre administered

All four studies in healthy subjects in which more than one fibre quantity was administered44, 45, 61, 72 reported increased stool weight, increased stool frequency and decreased transit time with increased fibre intake. However, differences were reported for stool consistency. While the study by Slavin et al.45 reported that stools became softer with increased fibre (soy polysaccharide) intake, Kies et al.61 reported a decrease in faecal moisture with increased fibre (cellulose) leading to abnormal, hard stools in several subjects receiving 40 g cellulose/day, which required digital mechanical removal. In the study by Taper et al. an increase in the concentration of soy fibre above 30 g/day was accompanied by a significant decrease in mineral retention.44

In patients, no significant difference in the incidence of diarrhoea or gastrointestinal discomfort was reported in a study comparing 7 g/L with 14 g/L soy fibre in surgical patients.29 In a year-long chronic paediatric study, divided into four treatment periods (no fibre, 14, 15 and 21 g/day), the higher fibre intake in the final period was associated with no change in stool consistency, an increase in stool wet weight, a significant increase in bowel frequency and significantly improved nitrogen and phosphorus retention.66, 71

DISCUSSION

Diarrhoea and constipation, representing the two extremes of bowel function, continue to be the most common problems associated with enteral tube feeding.6 Diarrhoea is a notable feature in the acute care setting.7 Apart from its major impact on quality of life, it can have additional clinical consequences such as dehydration and increased infection risk, with a subsequent negative impact on resource costs including nursing time, cleaning costs and prolonged length of stay in hospital. Constipation is a less overt clinical problem, but can also lead to impaired quality of life, and need for nursing and pharmaceutical interventions, e.g. manual evacuations and laxative prescription.75 Although in both cases the causes are multiple and often poorly understood, the absence of fibres in enteral feeds has been implicated as a cause for these impairments in bowel function.

The present systematic review, with application of meta-analysis and meta-regression where feasible, represents the most comprehensive examination of the role of fibre in enteral formulae to date. Although it is clearly desirable to examine the impact of fibre supplementation on the end points of relevance (i.e. diarrhoea and constipation) in well-defined patient groups in the clinical setting, it was recognized a priori that there were limitations in the evidence base.17, 18 For this reason, it was considered appropriate to include a review of additional bowel-related end points which would support or refute a possible role for fibre in preventing or alleviating diarrhoea and constipation. Moreover, to establish the effects of fibre in enteral formulae under carefully controlled conditions in the absence of potential confounding factors such as medications and to allow a better interpretation of patient data, it was deemed relevant to include a separate review of studies of fibre supplementation of enteral formulae in healthy volunteers. Although the range of normal values for a particular gastrointestinal end point will be different for healthy subjects fed via the oral route compared with those for patients fed via a nasoenteric tube, studies with healthy subjects can provide useful information. These studies not only provide insights into the amount and type of fibre that should be considered for use in enteral feeds in patients, but they also provide information that is relevant to the examination of a potential moderating effect of fibre on gastrointestinal function.

This review demonstrates that fibre in enteral formulae is well tolerated and has clinical benefits in patients, most pronounced in diarrhoea but with trends in constipation, and in terms of acute and chronic healthcare settings. These clinical findings are substantiated by the measured effects of fibre on objective physiological parameters in both patients and healthy volunteers. Furthermore, fibre supplementation has been demonstrated for the first time to have a moderating effect on bowel function, most notable when impairments are most pronounced at either end of the bowel function spectrum.

Diarrhoea

In the acute hospital setting (medical, surgical and ICU studies combined or surgical alone), overall there was a significant reduction in the percentage of patients with diarrhoea (Figure 2: OR 0.68, 95% CI: 0.48–0.96; 16 data sets) and a tendency for a reduction in percentage of days with diarrhoea. This is consistent with findings on incidence of diarrhoea in critically ill and surgical patients from previous meta-analyses. Del Olmo et al.18 combined results of nine studies (OR 0.66, 95% CI: 0.46–0.95), while Yang et al.17 included five studies, just missing statistical significance (OR 0.61, 95% CI: 0.36–1.05).

In the present analysis, the effect of fibre on diarrhoea was variable, which was further examined by subgroup analyses (Figure 2) and meta-regression (Figure 3). Results suggested a strong effect of fibre in reducing the incidence of diarrhoea in patient groups with a high incidence, and a smaller or absent effect in those with a low incidence of diarrhoea (Figure 3). Although the strength of this relationship is increased statistically by regression to the mean, the large variability in risk of diarrhoea across trials is likely to make this less biased. A likely mechanism underlying this effect of fibre in reducing diarrhoea is the stimulation of colonic water and electrolyte absorption by SCFA that are produced as a consequence of anaerobic metabolism of fibre. SCFA production depends on the number and types of bacterial species present in the colon, substrate source and gut transit time.76

Some, but not all, studies showed a significant increase in faecal SCFA concentrations and this may reflect the fibre source used, the sample collection method, or the influence of antibiotics which may inhibit bacterial fermentation. However, faecal SCFA concentrations might only be poor indicators of SCFA production as more than 95% of SCFA formed in the colon are absorbed.77, 78

Apart from SCFA production, the indigenous gut microflora per se has an important role to play in maintaining intestinal homeostasis and in patients receiving enteral feeding; alterations to this balance may predispose the patient to diarrhoea.35, 64 Nakao et al.65 reported an increase in the ratio of anaerobic:aerobic bacteria with fibre supplementation, which may be beneficial, as anaerobic bacteria are thought to protect against the overgrowth of potential pathogens,79 some of which may cause diarrhoea. On the other hand, consumption of enteral formulae with low fibre content can adversely affect the balance of faecal microflora in healthy subjects (reviewed in Ref.80).

Some of the variable effects of fibre on diarrhoea may be explained by the confounding effects of antibiotics, which were used in many of the clinical studies. Antibiotics have an adverse effect on the normal colonic microflora,81 allowing overgrowth of micro-organisms that induce diarrhoea, in particular Clostridium difficile which is responsible for 10–25% of cases of antibiotic-associated diarrhoea.82 Diarrhoea is a common adverse effect of antibiotic treatment,83 with antibiotic-associated diarrhoea occurring in about 5–25% of patients either early during antibiotic therapy or up to 2 months after the end of therapy.82 Risk factors for antibiotic-associated diarrhoea include age over 65 years, immunosuppression, being in an ICU and prolonged hospitalization. Although almost all antibiotics can cause diarrhoea, the risk is greater following treatment with broad spectrum antibiotics and those that target the anaerobic flora.84 Such antibiotics include aminopenicillins, a combination of aminopenicillins and clavulanate, cephalosporins and clindamycin.

Our findings indicate that not all studies in which antibiotics were used reported an effect on diarrhoea incidence. Although certain antibiotics appear to be more strongly associated with diarrhoea than others, data were too limited to draw any clear conclusions from synthesis of information from the studies examined. This seems an area worthy of further evaluation.

Constipation

At the other end of the spectrum of bowel function, constipation is a common finding in the elderly, especially among those receiving multiple medications and who are bed-ridden.75 Surprisingly, there were more studies reporting on constipation in the hospital setting compared with the chronic setting. As there were only limited data available to assess constipation, a meta-analysis was not possible in the chronic-care setting. This may reflect the difficulties in undertaking carefully controlled long-term studies in the community.

In the acute hospital setting, there was an overall tendency for the percentage of patients with constipation to be reduced by fibre supplementation, but the results did not reach statistical significance. The most likely explanation is the enormous variance in definitions of constipation used in the studies, which ranged from absence of bowel action over a 3-day or 10-day period to descriptors of stool consistency. In surgical patients, the presence of post-operative ileus, which has a different aetiology and significance as medical causes of constipation, may influence interpretation. Reese et al. reported that constipation was present in 33% of surgical patients and fibre supplementation resulted in no significant difference between groups.29 Another reason for the variable effects of fibre supplementation on constipation concerns the confounding effects of laxatives, which were used in some of the studies. In a few studies, laxative use was used as an indicator of constipation, but as the protocol for their use was rarely described, it is difficult to draw clear conclusions on how successfully fibre supplementation impacted on laxative use.

In view of the scarcity of studies using comparable definitions of diarrhoea and constipation, and in light of the paucity of clinical data, particularly in relation to constipation, there is merit in examining objective parameters that may play a role in diarrhoea and actual or perceived constipation. These include bowel frequency, whole gut transit time and faecal mass. Stool consistency, although rather more subjective, may also be a useful indicator.

Bowel frequency

The analysis shows wide interindividual variation in bowel frequency in both health and disease states. The effect of fibre supplementation on bowel frequency in the clinical setting was also variable and generally greater in patients, compared with healthy subjects (Figure 4). This is probably due to the confounding effects of disease and its treatment, especially drug treatment, and the greater variability in age and dose of fibre in the patient population compared with the healthy population. In a previous systematic review,18 an increase in bowel frequency was demonstrated, but patient and healthy volunteer data were combined (mean difference 0.07; 95% CI: 0.03–0.1) and no assessment was made whether extremes in bowel frequency on either side of the mean were ‘normalized’ by administration of fibre. Bowel frequency can be regarded as an objective indicator of bowel function, whereby an excessive frequency may be linked to diarrhoea and a reduced frequency to constipation. In both patient and healthy volunteer studies, a significant moderating effect of fibre was observed (Figure 5), suggesting that fibre can play a role in reducing bowel frequency when it is high, and increasing it when it is low, thereby potentially ameliorating both diarrhoea and constipation. This moderating effect may help to answer the frequently posed question of how fibre can have a beneficial role in such different conditions.

It is likely that there is more than one mechanism mediating the effects of fibre on bowel frequency, which might also explain the outcome seen in those patients with very high as well as those with very low bowel frequency. On the one hand, administration of fibre (in particular insoluble fibre) leads to faecal bulking because of its own mass, water retaining properties, and the increase in bacterial biomass resulting from fermentation. This increase in faecal bulking is expected to increase bowel frequency by speeding up intestinal transit. On the other hand, colonic fermentation of fibre (in particular soluble fibre) produces SCFA, which stimulate salt and water absorption. This effect is expected to reduce faecal output, especially when the stool is watery or has high faecal moisture. Furthermore, animal studies have shown that an absence of fibre reduces colonic motility,85 which in turn could also influence bowel frequency.

Transit time

The addition of fibre to enteral formulae was generally found to accelerate whole gut transit time, suggesting that fibre should be advantageous in enterally fed individuals suffering from constipation as a result of slow transit time. Most of data were available from studies performed on healthy subjects, with significant overall results reported (Figure 6). This analysis suggests that different types of fibres have different effects on bowel frequency. Studies employing soy polysaccharide, cellulose or a mixture of six fibre sources were consistent in speeding up transit time in comparison with fibre-free feeds. In contrast, studies with guar gum, oat and soy oligosaccharide showed a nonsignificant trend to a prolongation of transit time. This variable effect, which has also been observed in healthy subjects ingesting food,6 is probably related to the type and level of fibre used as well as interactions with other dietary constituents. In subjects ingesting a SS diet, transit time was significantly faster compared with an enteral formula containing similar or greater amounts of a single fibre source.9, 34 However, when using a fibre mixture (Nutricia Multi Fibre), transit time normalized close to that of a SS diet.46

The absence of significant effects of fibre on whole gut transit time in patients is to be expected considering the use of a number of different methodologies in small number of patients suffering from a variety of diseases. The limited data prevented an examination of the moderating effect of fibre on whole gut transit time, which may have been expected based on the findings reported in healthy subjects.

Faecal mass

Although there is no generally agreed definition of constipation or diarrhoea, existing definitions may include a quantification of faecal mass (or stool weight). Unfortunately, the nature of this end point means that it is rarely realistic to undertake such measurements in clinical practice, or even in clinical trials. This explains why most data available on faecal mass comes from studies in healthy volunteers. The present review indicates that in healthy subjects ingesting a fibre-free feed, faecal mass is low, but it is consistently increased when fibre is added to the feed (Figure 7). The increase in faecal mass is to be expected, as it is well established that increasing the fibre intake in the normal diet of healthy subjects results in increased faecal mass, compared with ingestion of a low-fibre diet.6

Faecal mass and dry stool weights were also increased in patients receiving fibre-supplemented enteral feeds. These findings support the concept that fibre added to enteral formulae may increase faecal mass in patients where it is low. Only one study reported that fibre supplementation resulted in a reduction in faecal mass. In this case, the majority of patients had watery faeces with high faecal mass, which was reduced following supplementation with fibre.39 A key feature here may also be an improvement of stool consistency. Several studies suggest that fibre not only improved consistency to more normal/formed stools when they were ‘liquid’ and unformed, but also improved consistency to softer, more normal stool when the faeces were ‘hard’ and ‘dry’.10, 33, 39, 45, 47, 49, 52, 70 Different mechanisms are probably involved in changing stool consistency from hard to soft stool and from liquid, unformed stool to formed stool, and in reducing the diarrhoea and constipation as explained above. Although the data are limited, these findings again suggest a moderating effect of fibre on bowel function.

The analysis of the healthy volunteer data has revealed a remarkable difference in the effect of fibre on faecal mass if included in enteral formula vs. its effects if consumed as part of a normal SS diet.9, 34, 45, 46 This was seen despite the fact that the supplemented formulae were reported to contain far more fibre (up to 30 g/day) compared with the SS normal diets (15 g/day) (Figure 8). Calculations showed that in subjects receiving enteral formulae there was only a 1.1 g increase in faecal mass per gram of added fibre. This is substantially less than the increase, which occurs when fibre is added to a normal diet. Nevertheless, Cummings6 reported a large variation in increments in faecal weight per gram of naturally occurring fibre ingested (g/g) in subjects ingesting food, ranging from 1.2 for pectin (soluble, well-fermented fibre) to 8.9 for methylcellulose. Most fibres produced values between 3.3 and 5.4 g/g (fruit and vegetables, corn, oats, gums and mucilages, cellulose and carboxymethylcellulose).

There are at least four possible explanations for these differences. First, in studies reporting the intake of fibre in subjects ingesting food, calculations of fibre intake were typically based on published food composition tables, which historically did not include all components which are currently recognized as fibre (e.g. inulin, FOS and resistant starch) on account of the nature of the analytical methods used. For example, the Englyst method measures only nonstarch polysaccharides and does not capture resistant starch, FOS or inulin, whereas several AOAC methods only partially capture these components.4 Therefore, the reported amount of fibre ingested in the normal diet may considerably underestimate the total amount reaching the colon.

Secondly, the type of fibres administered in studies may differ from that used in food studies. Most studies using fibre-containing formulae have involved soy polysaccharide, which has not been extensively studied in healthy subjects ingesting food. Those that have investigated the effects of adding soy fibre to food reported a smaller increment in stool weight per gram soy fibre added (g/g) compared with most other types of fibre (e.g. for soy pulp, never dried, 1.3; for purified soy pulp 2.4 and for soy bean hulls, 2.6).6

Thirdly, the fibre used in formulae is often modified (e.g. hydrolysed guar gum or finely ground soy polysaccharide) to reduce feed viscosity and minimize the risk of feeding tube occlusion. This may alter its biological characteristics, such as water-holding capacity, fermentability (which increases when soy fibre is finely ground), and the pathways by which they are metabolized.86 Highly fermentable substrates generally have less faecal bulking effect than poorly fermentable substrates, such as cellulose and bran.6 Particle size is a key determinant of stool bulking46, 87 and small particle size may be one of the most important factors limiting the efficacy of the stool bulking property of fibre in enteral nutrition.

Finally, different dietary fibres within a mixture may interact to produce a variety of physiological effects compared with an enteral feed containing a single fibre source.

A relevant question to pose is how relevant is faecal mass per se from a physiological standpoint. It is generally accepted that fibre induces laxation by increasing faecal bulk.4 However, an increase in faecal bulk does not always produce laxation or alleviate constipation. The effect seems to depend on the type of fibre, which can exert effects independently of faecal mass.88

Quality and quantity of fibre and tolerance

One of the key issues of importance in relation to the design of enteral formulae concerns the type and amount of fibre that should be added to the feeds. One approach is simply to mimic the desirable intake (quality and quantity) of fibre in a healthy population, taking into account that fibre intakes in normal diets may be under-reported for the reasons outlined earlier. An alternative approach is to provide sufficient fibre(s) to reproduce specific physiological effects, which are most relevant to clinical needs. The two approaches do not necessarily yield the same results, at least when the outcome variable is faecal bulking.

One of the caveats of the latter approach is the recognition that use of a single fibre source may have adverse effects. Although studies reported that fibre is generally well tolerated in enteral formulae with no impairment of gastric emptying or formula intake,8, 27, 34, 52, 53 high doses of highly fermentable soluble fibres, such as hydrolysed guar gum or inulin caused increased flatulence,14, 38 resulting from gaseous production (e.g. hydrogen, methane and CO2) during anaerobic fermentation of fibre by bacterial enzyme systems in the right colon.89 Conversely, use of a single insoluble fibre such as cellulose led to constipation and faecal impaction in four of the 18 healthy subjects, which required digital evacuation.61 Mean faecal moisture content was 67% and 63% in those administered 30 or 40 g of cellulose, respectively, which was significantly lower compared with those receiving a fibre-free formula (74%) and lower than all values reported in a review of normal subjects ingesting food with and without added fibre.6 These findings caution against the use of certain single fibre sources with very high or very low degrees of solubility and fermentability. An additional concern with the use of single fibre sources relates to the fact that colonic bacterial polysaccharide enzyme systems are substrate inducible, thus with time the use of a single fibre may increase the degree of fermentability and this adaptation might reduce any stool bulking effects.46

In vitro studies can be used to select the type of fibres and fibre combinations for inclusion in enteral feeds (e.g. on the basis of a high production rate of butyrate, which has beneficial effects on the mucosa, and low rate of gas production to avoid flatulence90–92) but this was not the focus of the present review. Further examination of the effects of fibre mixtures vs. single fibre sources was not attempted in this review, but this would have been difficult due to the effects of other confounding variables and a general lack of studies assessing this comparison.

Limitations of current evidence base and future research

The present review has identified a number of areas that need to be addressed further. One of the most obvious is concerned with the potential value of fibre in long-term care. In many countries, most tube-fed patients are found in the community, e.g. care homes and at home, but paradoxically there is little information available from these nonhospital settings. Thus, although constipation is a common problem in long-term care, there were insufficient data for meta-analysis to properly evaluate the effect of fibre supplementation. Examination of the associated underlying mechanisms for constipation and diarrhoea, potentially requiring biochemical and bacteriological investigations in tandem with the clinical studies, will help to direct future innovations.

It is also of concern that several studies did not specify a clear primary end point and lacked sample size calculation. Many studies also used different criteria to define constipation and diarrhoea. Future studies would benefit from the use of consistent definitions of diarrhoea and constipation and use of clinically relevant markers of gastrointestinal function. This would allow a more thorough evaluation of different types and amounts of fibres in different patient groups and healthcare settings, and allow a more robust comparison between trials.

The use of laxatives in long-term, tube-fed patients deserves more careful examination. Anecdotal evidence suggests that these are used widely but without clear protocols or monitoring. A reduction of laxative use may have a significant budgetary impact, and this could potentially be achieved through the application of fibre-containing feeds.

Conclusion

In conclusion, the current systematic review:

  • 1Represents the most comprehensive examination of the role of fibres in enteral formulae to date;
  • 2Demonstrates that fibre-containing enteral formulae are well tolerated, especially when given as fibre mixtures;
  • 3Demonstrates significant clinical benefits of fibre-supplemented enteral feeds in patients suffering from diarrhoea, with a positive trend also observed for patients with constipation. The findings were relevant in both acute and chronic healthcare settings and across all age ranges;
  • 4Reports a moderating effect of fibre supplementation on bowel function, which is most pronounced in patients at either end of the bowel function spectrum;
  • 5Has highlighted the need for both consistent reporting of clinical end points in future studies and further studies in the community setting;
  • 6Concludes that first-line treatment with fibre-containing feeds should be considered as an important modality of clinical care.

Acknowledgements

Declaration of personal interests: The authors would like to thank Dr Stephen Mitchell from Abacus International for his excellent assistance in the preparation and submission of this article. Marinos Elia has served as a speaker and/or consultant for Fresenius Kabi, Nestle, Novartis and Numico. Meike Engfer and Ceri Green are employees of Numico. David Silk has served as a speaker and consultant for Nestle and Numico. Declaration of funding interests: The preparation of this paper was funded in part by Numico.

Ancillary