- Top of page
Background : Oligofructose is metabolized by bifidobacteria, increasing their numbers in the colon. High bifidobacteria concentrations are important in providing ‘colonization resistance’ against pathogenic bacteria.
Aim : To reduce the incidence of antibiotic-associated diarrhoea in elderly patients.
Methods : Patients over the age of 65 taking broad-spectrum antibiotics received either oligofructose or placebo. A baseline stool sample was cultured for Clostridium difficile and tested for C. difficile toxin. A further stool sample was analysed for C. difficile if diarrhoea developed.
Results : No difference was seen in the baseline characteristics, incidence of diarrhoea, C. difficile infection or hospital stay between the two groups (n = 435). Oligofructose increased bifidobacterial concentrations (P < 0.001, 95% CI: 0.69–1.72). A total of 116 (27%) patients developed diarrhoea of which 49 (11%) were C. difficile-positive and were more likely to be taking a cephalosporin (P = 0.006), be female (P < 0.001), to have lost more weight (P < 0.001, 95% CI: 0.99–2.00) and stayed longer in hospital (P < 0.001, 95% CI: 0.10–1.40). Amoxicillin (amoxycillin) and clavulanic acid increased diarrhoea not caused by C. difficile (P = 0.006).
Conclusion : Oligofructose does not protect elderly patients receiving broad-spectrum antibiotics from antibiotic-associated diarrhoea whether caused by C. difficile or not. Oligofructose was well-tolerated and increased faecal bifidobacterial concentrations.
- Top of page
Antibiotic-associated diarrhoea (AAD) is a major problem, particularly in the over 65 age group where severe diarrhoea carries a significant morbidity and mortality. Some antibiotics, such as erythromycin, have a specific prokinetic action. However, in the majority of cases the mechanism of diarrhoea is unclear and may relate to alteration of normal intestinal flora homeostasis or a direct chemical irritation.1 An increasingly common problem is infection with Clostridium difficile.
Bifidobacteria constitute about 95% of the bacterial mass in the faeces of breast-fed infants, and it is believed that they contribute to the protection that breast-feeding provides against gut infections.2 In the human colon, numbers of bifidobacteria decrease with increasing age, such that they comprise about 25% of the total adult flora. In the elderly, their numbers decline even further, with a corresponding increase in numbers of Clostridium spp. and coliforms. Bifidobacteria are able to inhibit the growth of many pathogenic bacterial species.3, 4 This is thought to occur by the excretion of acidic end products of fermentation and the production of specific antimicrobial agents that have a wide spectrum of activity. Bifidobacteria influence faecal bacterial enzyme activity,5 reduce antibiotic-induced side-effects6 and inhibit induced mammary, liver and colonic tumours in animals.7, 8 It is likely that bifidobacteria contribute significantly towards colonization resistance, which the resident microflora offers against invading pathogens.9, 10
Probiotic microorganisms such as Lactobacilli GG and Saccharomyces boulardii have been fed to patients in an attempt to prevent AAD and diarrhoea caused by C. difficile.11–15 Opinion on their clinical effectiveness varies.16 As these organisms are taken orally, many will not survive the gastric acid barrier or may suffer adverse effects from bile and other secretions in the small bowel. Probiotic microorganisms are not able to colonize the colon and disappear rapidly after oral intake has ceased.17 Recently, interest has developed in the concept of prebiotics such as oligofructose and inulin which are fructo-oligosaccharides found in a variety of edible plants such as chicory, Asparagus and artichoke. Oligofructose is licensed in the UK as a food additive and is used commercially in many products such as yoghurt and margarine. Taken orally, oligofructose enters the large bowel intact where it is metabolized by bifidobacteria.18 This results in a marked increase in the numbers of bifidobacteria in healthy volunteers, such that these organisms become numerically predominant, while clostridia spp. decrease in number.19 Fructo-oligosaccharides have been shown to protect hamsters against C. difficile disease.20
The purpose of this study was to determine if the incidence of AAD in elderly patients could be reduced by the oral administration of oligofructose.
- Top of page
The Local Research Ethics Committees of the University Hospital of Wales, Cardiff, Addenbrooke's Hospital, Cambridge and Derriford Hospital, Plymouth approved the study. Consecutive in-patients, who were over the age of 65 and who had been prescribed a broad-spectrum antibiotic within the preceding 24 h, were invited to participate. Patients were excluded if they had received antibiotics within the previous 6 weeks, were immunocompromised, had gastrointestinal disease, were diabetic or were unable to give informed consent. All subjects were able to eat and were not receiving tube feeding.
The study was performed using a double-blind randomized placebo-controlled design. Subjects were randomly allocated to receive either oligofructose (12 g/day) or placebo (sucrose, 12 g/day). The study was divided into three phases (Figure 1). The study powders were taken while the subjects received antibiotics (phase 1) and continued for a further 7 days following cessation of the antibiotics (phase 2). Both were administered by ward nursing staff concomitantly with their other prescribed medication. Subjects were followed up for a further 7 days (phase 3). Previous work shows that 92% of C. difficile-associated diarrhoea occurs within 14 days of antibiotic exposure.21 The randomization codes were generated by computer and held in sealed envelopes.
On entry into the study and before discharge from hospital, subjects were weighed. Continual assessments were made which included a record of stool form on a 4-point scale (1 – hard, 2 – lumpy, 3 – mushy and 4 – loose) and defecation frequency. Participants were asked to report any abdominal symptoms such as bloating. These recordings were performed by the study subjects, or if this was not possible, by the nursing staff. Stool form and interdefecatory intervals were averaged over the three study periods. For the purposes of analysis significant diarrhoea (grade 4) was defined as at least three loose stools in a 24-h period. Non-significant diarrhoea was defined as one or two loose stools within a 24-h period. This was done for each of the three study periods. If a patient experienced significant diarrhoea then they were withdrawn from the study (as oligofructose is potentially a laxative) and given specific treatment.
On entry into the study a stool sample was sent from all patients for culture of C. difficile and for detection of C. difficile toxin.
If during the study, a subject reported diarrhoea, a further stool sample was tested for the presence of enteric pathogens and for the presence of C. difficile toxin as above.
Stool samples were collected at entry into the study and at the end of each of the three study periods from 49 consecutive patients. The samples were quantitatively cultured for total aerobes, anaerobes and bifidobacteria.
Faecal slurries were made with deoxygenated half strength Ringers solution (5%, w/v) to which cetyl trimethylammonium bromide (0.01 mg/L) had been added. Serial dilutions in anaerobic half strength Ringers solution at 37 °C were then made. About 0.1 mL volumes were plated onto neomycin Fastidious Anaerobe Agar (FAA) supplemented with 6% horse blood (Lab M Ltd, Bury UK), a modified Columbia agar selective for bifidobacteria22 and blood agar. Clostridium difficile culture was done using the alcohol shock method and culture on Braziers media.23 All the media were incubated at 37 °C aerobically or in an anaerobic cabinet as appropriate and were read at 48 h. Colony identification was by traditional techniques. Bacterial counts were expressed as log10cfu/g of faeces (wet weight). Clostridium difficile toxin A and B were detected using immunoassay (Meridian Premier Assay; Launch diagnostics, Ash House, New Ash Green, Kent, UK) or using faecal cytotoxin detection using vero cell culture and detection of cytopathic effect and neutralization by antiserum to C. sordelli (Colidale, UK).
Historical data of acute admissions to the hospitals participating in the study suggested that by selecting patients over the age of 65 the incidence of AAD is 30% in patients receiving broad-spectrum antibiotics. Thus, a study size of 225 patients in each group (active and placebo) will enable (P = 0.8, α = 0.5) the study to detect a 23% or greater reduction in incidence of diarrhoea. Assuming an incidence of C. difficile-associated diarrhoea of 20% then a total study size of 450 patients would allow the detection of 29% or greater reduction in diarrhoea.
Analysis was done on an ‘intention-to-treat’ basis and performed blind to the intervention each patient received. Data were assessed as parametrically or non-parametrically distributed by Ryan Joiner tests and bar histograms. Bacterial counts were logarithmically transformed and analysed using parametric tests. This data was expressed as geometric mean and 95% CI of the geometric mean values. As data was predominantly non-parametrically distributed, comparisons between groups and changes from baseline were assessed for significance using Fishers exact tests, Mann–Whitney tests or logistic regression as appropriate.
- Top of page
Of 3226 patients who were approached (over the age of 65 and taking broad-spectrum antibiotics), 1327 were suitable to be entered into the study. Of those unable to be entered into the study, 27% were diabetic, 53% were unable to give informed consent (memory impairment or too ill), 13% had gastrointestinal disease and the remainder had other exclusion criteria. About 450 patients agreed to enrol the study, of these 15 were excluded from analysis (seven withdrew from the study before taking any intervention and eight had gastrointestinal bleeds).
The baseline characteristics for the two groups (oligofructose and placebo) were similar (Table 1). A total of 116 (27%) patients experienced diarrhoea, of which 73 patients (17%) developed significant diarrhoea and 43 (10%) non-significant diarrhoea. In those taking oligofructose, there was no reduction in diarrhoea however defined, or whether or not caused by C. difficile infection. Oligofructose intake was not associated with any increase in bloating, stool form or reduction in interdefecatory interval.
Table 1. Baseline characteristics and study outcomes for those subjects receiving oligofructose and placebo (median, interquartile range)
|Age (years)||76 (70, 83)||78 (72, 84)|
|Length of stay in hospital (days)||10 (6, 15)||9 (7, 15)|
|Time antibiotics taken (days)||7 (6, 8)||7 (5, 8)|
|Number of antibiotics taken||2 (1, 2)||2 (1, 2)|
|Antibiotic to oligofructose* (h)||4.5 (2.0, 8.5)||4.5 (2.0, 7.5)|
|Presence of Cdt at entry||6||4|
|Growth of Clostridium difficile at entry||31||23|
|Number who died||4||2|
|Significant diarrhoea, Cdt+||19||21|
|Significant diarrhoea, Cdt−||17||16|
|Non-significant diarrhoea, Cdt+||3||6|
|Non-significant diarrhoea, Cdt−||17||17|
|Proton-pump inhibitor use||29||31|
|Interdefecatory interval (h)||26 (22, 52)||24 (21, 35)|
|Stool form||2.2 (2, 2.9)||2.6 (2.0, 3.0)|
|Amoxicillin + clavulanic acid||92||94|
|Change in weight (kg)||0.0 (−0.4, 0.1)||0.0 (−0.6, 0.0)|
Forty-nine patients (11%) developed C. difficile toxin-positive diarrhoea of which 40 patients (9%) developed significant diarrhoea (Table 2). Patients who developed significant C. difficile toxin-positive diarrhoea and were receiving oligofructose were more likely to be taking amoxicillin and clavulanic acid (P = 0.003).
Table 2. Subjects who developed significant diarrhoea caused by Clostridium difficile (median, interquartile range)
|Age (years)||76 (69, 86)||77 (69, 80)|
|Growth of C. difficile at entry||8||7|
|Change in weight (kg)||−1.5 (−1.5, 0)||−2.0 (−3.0, −1.0)|
|Length of stay in hospital (days)||17 (13, 22)||15 (11, 18)|
|Length of antibiotics (days)||8 (6, 8)||8 (5, 8)|
|Number of antibiotics taken||2 (2, 2)||2 (1, 2)|
|Number who died||2||0|
|Proton-pump inhibitor use||3||7|
|Ampicillin or amoxicillin||1||4|
|Amoxicillin + clavulanic acid||13*||4|
Compared with those patients who did not develop diarrhoea, those patients who developed significant C. difficile toxin-positive diarrhoea were more likely to be female, be culture-positive for C. difficile on entry into the study, taking cephalosporins (but not ampicillin or amoxicillin), to have lost more weight and stayed in hospital longer (Table 3). Compared with patients not developing diarrhoea, patients developing non-significant C. difficile toxin-positive diarrhoea were more likely to lose weight and receive antibiotics for a shorter period of time. Amoxicillin and clavulanic were the only antibiotics associated with C. difficile-negative diarrhoea. The onset of development of diarrhoea is given in Table 4.
Table 3. Subjects who developed diarrhoea (significant and non-significant) vs. those who did not (median, interquartile range)
| ||Significant diarrhoea (>3 loose stools/day)||Non-significant diarrhoea (<3 loose stools/day)||No diarrhoea|
|N||40|| ||33|| ||9|| ||34|| ||319|| |
|M/F||8/32a***|| ||13/20|| ||4/5|| ||18/16|| ||170/149|| |
|Oligofructose/placebo||19/21|| ||17/16|| ||3/6|| ||17/17|| ||159/160|| |
|Clostridium difficile cultured at entry||15a***|| ||–|| ||4|| ||–|| ||35|| |
|Change in weight||−1.5a***||−2.3 to −0.4||0.0||−6 to 0||−0.4a**||−3.5 to 0.1||0.0b***||−0.5 to 0.1||0.0||−0.4 to 0.1|
|Length of hospital stay (days)||16a***||13–21||9b**||7–17||8b*||7–17||8b**||7–17||9||6–14|
|Time antibiotics taken (days)||7||6–8||7||5–10||5a**b**1||3–7||7||6–8||7||5–8|
|Number of antibiotics taken||2||1–2||2||1–2||2||2–2||2||1–2||2||1–2|
|Antibiotic to oligofructose†||6.0||3.0–11.0||3.0||2.0–8.0||3.0||1.5–5.5||5.3||3.0–7.6||5.9||2.0–7.5|
|Deaths||2|| ||0|| ||0|| ||1|| ||3|| |
|Laxative use||3|| ||8|| ||0|| ||8|| ||49|| |
|PPI use||10a*|| ||3|| ||2|| ||7|| ||38|| |
|Ampicillin||5a*|| ||12b*|| ||3|| ||5|| ||94|| |
|Amoxicillin + clavulanic acid||17|| ||11|| ||2|| ||23a**|| ||133|| |
|Clarithromycin||18|| ||15|| ||4|| ||10|| ||141|| |
|Cephalosporin||18a**|| ||6b*|| ||0|| ||5b**|| ||75|| |
|Ciprofloxacin||2|| ||4|| ||3a*|| ||2|| ||26|| |
|Doxycycline||2|| ||4|| ||0|| ||0|| ||15|| |
Table 4. Phase in the study at which subjects developed diarrhoea (n oligofructose/placebo with percentage for both oligofructose and placebo)
|Diarrhoea||Phase 1||Phase 2||Phase 3|
|Significant Cdt+||11/20 (78)||4/1 (13)||4/0 (10)|
|Significant Cdt−||13/16 (88)||4/0 (12)||0/0 (0)|
|Not-significant Cdt+||3/4 (78)||0/0 (0)||0/2 (22)|
|Not-significant Cdt−||14/11 (74)||1/4 (15)||2/2 (12)|
Fifty-four (12%) patients were culture-positive for C. difficile on entry into the study (Table 5); these patients were more likely to develop significant C. difficile-related diarrhoea (P < 0.001). Clostridium difficile was cultured from the faeces of all the patients with detectable toxin.
Table 5. Comparison of subjects whose stools were culture-positive for Clostridium difficile at entry into the study
|Developed significant diarrhoea||8||7|
|Developed non-significant diarrhoea||0||4|
|Did not develop diarrhoea||23||12|
Changes in faecal bacterial concentrations in those patients who did not develop diarrhoea (n = 40) are shown in Figure 2. In those taking oligofructose bifidobacterial concentrations rose at the end of phase 2 (P < 0.001, 95% CI: 0.69–1.72), total aerobe concentrations were lower at the end of phase 1 (P < 0.001, 95% CI: −0.88 to −0.35) and 3 (P < 0.001, 95% CI: −0.97 to −0.46). For those taking placebo, total anaerobes fell at the end of phase 1 (P < 0.001, 95% CI: −1.80 to −0.98) and 2 (P =0.025, 95% CI: −0.74 to −0.06), bifidobacterial concentrations were reduced at the end of phase 1 (P < 0.001, 95% CI: −1.39 to −0.81). Faecal total aerobes were reduced at the end of phase 1 (P < 0.001, 95% CI: −1.43 to −0.60), 2 (P < 0.001, 95% CI: 1.41 to −0.66) and 3 (P < 0.001, 95% CI: −1.20 to −0.63). Comparing bacterial concentrations between those taking oligofructose and those taking placebo, total anaerobes were increased at the end of phase 1 (P < 0.001, 95% CI: 0.69–1.57) and 2 (P = 0.019, 95% CI: 0.08–0.78). Bifidobacterial concentrations were higher at the end of phase 1 (P < 0.001, 95% CI: 0.62–1.23), 2 (P < 0.001, 95% CI: 0.75–1.87) and 3 (P = 0.048, 95% CI: 0.00–0.92). Total aerobes were higher at the end of phase 2 (P = 0.001, 95% CI: 0.43–1.32). In those nine patients who developed diarrhoea, baseline stool cultures were similar to those patients who did not develop diarrhoea [bifidobacterial 8.8 (8.2, 9.5) log10cfu/g, total anaerobes 9.9 (9.4, 10.4) log10cfu/g, total aerobes 6.3 (6.0, 6.6) log10cfu/g]. Five patients developed significant C. difficile toxin-positive diarrhoea during phase 1 [four placebo (from diarrhoea stool, bifidobacteria 8.5 (7.8, 9.3) log10cfu/g, anaerobes 8.5 (7.0, 10.0) log10cfu/g and aerobes 5.3 (4.2, 6.4) log10cfu/g), one oligofructose (from diarrhoea stool, bifidobacteria 9.3 log10cfu/g, anaerobes 9.31 log10cfu/g and aerobes 5.43 log10cfu/g)]. Stool culture for bifidobacteria, anaerobes and aerobes were no different to those done in patients without diarrhoea at the end of phase 1 and were prescribed metronidazole, further stool culture was not done. Three patients [one placebo (from diarrhoea stool, bifidobacteria 7.6 log10cfu/g, anaerobes 7.9 log10cfu/g and aerobes 5.4 log10cfu/g), two oligofructose (from diarrhoea stool, bifidobacteria 9.7 log10cfu/g, anaerobes 10.2 log10cfu/g and aerobes 5.0 log10cfu/g)] developed significant C. difficile toxin-negative diarrhoea during phase 1 and one (placebo) non-significant C. difficile toxin-negative diarrhoea during phase 2 (from diarrhoea stool, bifidobacteria 8.5 log10cfu/g, anaerobes 9.5 log10cfu/g and aerobes 4.7 log10cfu/g). In these four patients, faecal bacterial concentrations were no different from those patients who did not develop diarrhoea.
Figure 2. Change if faecal bacterial concentrations in patients who did not develop diarrhoea (n = 40, log10cfu/g and 95% confidence).
Download figure to PowerPoint
There was an association (P = 0.04) between the use of proton-pump inhibitors and significant diarrhoea caused by C. difficile.
In hospital, 87% of doses of trial powder were taken. After discharge from hospital compliance was reported by the patients or carers as being 91%.
Of the 3226 patients approached, 13.6% developed significant diarrhoea caused by C. difficile. Of those excluded, because of memory impairment or being too ill to consent to be in the study, 17.4% (14.3% if include diabetics) developed significant diarrhoea caused by C. difficile. We did not collect data on non-significant diarrhoea or diarrhoea not caused by C. difficile.
- Top of page
Oligofructose was well-tolerated and increased faecal bifidobacterial concentrations; no increase in problems such as bloating or diarrhoea was noted. This study found no evidence that the co-administration of oligofructose with broad-spectrum antibiotics reduced the likelihood of patients developing significant or non-significant diarrhoea related or not related to the presence of Clostridium difficile.
Oligofructose has been shown to increase bifidobacterial numbers in normal volunteers taking 15 g/day.19 In a pilot study, we found larger (>12 g) doses of oligofructose caused bloating, abdominal pain, were laxative and were not well-tolerated in elderly patients. Dividing the daily dose of oligofructose did not alter its tolerability. Faecal bifidobacterial counts did not fall with antibiotics when taken with oligofructose and increased during phase 2, whereas they fell in patients taking the placebo. Faecal cultures for bifidobacteria done in patients who developed diarrhoea were not different from those who did, but numbers were too small (the majority of patients developing diarrhoea did so within phase 1 to draw meaningful conclusions. Patients who developed significant diarrhoea were withdrawn from the study and prescribed metronidazole or loperamide as appropriate (no further stool cultures were obtained). It is not clear if quantitative bacterial cultures done from a diarrhoea stool would reflect any prebiotic effects of oligofructose, as faster intestinal transit increases bifidobacterial concentrations and the disease process itself may alter the bacterial flora. It is possible that in those patients who developed diarrhoea the oligofructose did not increase bifidobacterial concentration. To examine this question, we would need to have done cultures just prior to the development of diarrhoea. The delay between starting antibiotics and taking the first dose of trial powders was small and there was no difference between oligofructose and placebo. Total anaerobe (at the end of phase 1 and 2) and total aerobe (at the end of phase 1) bacterial concentrations were higher in those patients taking oligofructose compared with those taking placebo, which raises the possibility that oligofructose is not as selectively metabolized by bifidobacteria as previously described. Studies in rats suggest that oligofructose and inulin increase the animal's intestinal permeability and susceptibility to bacterial translocation from Salmonella enteritidis,24, 25 which contrasts to studies in hamsters where protection against C. difficile infection is seen.20 Indeed, the ability of oligofructose to abrogate the effect of antibiotic on faecal anaerobic bacteria may have countered any benefit from increased bifidobacterial concentrations. Orrhage et al.26 found that oligofructose did not prevent a reduction in faecal anaerobes in volunteers given cefpodoxime proxetil, even when taken with oral bifidobacterial supplements but did cause an increase in aerobes mainly because of an overgrowth of enterococci. We found no evidence that oligofructose increased diarrhoea of any cause, length of hospital stay or mortality in patients. Interestingly, a study in 244 healthy travellers showed no reduction in diarrhoea from fructo-oligosaccharides.27
The overall incidence of diarrhoea in the placebo group was 28%, thus the study was able to detect a 24% or greater reduction in the incidence of diarrhoea and a 39% or greater reduction in C. difficile-related diarrhoea. The lower than expected incidence of diarrhoea raises the possibility that oligofructose may reduce the incidence of C. difficile-related diarrhoea, but the study was not large enough to detect any reduction (type two error), however, the negligible reduction in significant diarrhoea caused by C. difficile in those taking oligofructose makes this unlikely. Patient activity figures, for the hospitals in the study, showed a much higher incidence of C. difficile-associated diarrhoea in the overall population of over 65 year olds receiving broad-spectrum antibiotics. The most likely reason for the low incidence of C. difficile in this study was due to patient selection. The highest risk patients were not recruited because of disease severity, confusion or diabetes. It is unclear if this altered our findings. We cannot be certain that the cause of diarrhoea was always directly attributable to C. difficile toxin, rather than another pathogen or laxative use. All stool samples were cultured for the common bacterial pathogens but viruses were not routinely looked for.
The cause of diarrhoea in C. difficile-negative patients was not clear. Antibiotics such as erythromycin stimulate intestinal motilin receptors to cause diarrhoea. Other antibiotics, by reducing the intestinal bacterial flora metabolisms of carbohydrates, may cause an osmotic diarrhoea, which may be exacerbated by laxatives.28 Antibiotics may also cause diarrhoea by allowing the establishment of a pathogen such as C. difficile. We did not look for other pathogens such as C. perfringens or Norwalk virus. Augmentin was associated with diarrhoea, while laxative use was not associated with development of diarrhoea of any cause or severity.
Patients who developed diarrhoea caused by C. difficile were more likely to be culture-positive on entry to the study but did not receive more or longer courses of antibiotics than those who did not develop diarrhoea. The study was underpowered to show a protective effect for oligofructose between those patients who were C. difficile culture-positive on entry into the study and those who were not. Broad-spectrum cephalosporins (principally cefuroxime) were the only antibiotics significantly associated with the development of significant diarrhoea caused by C. difficile, whilst patients receiving ampicillin or amoxicillin were less likely to develop significant diarrhoea. Patients receiving amoxicillin with clavulanic acid were more likely to develop non-significant, C. difficile-negative diarrhoea.
Patients developing significant diarrhoea caused by C. difficile were clearly disadvantaged in that they had weight loss and stayed longer in hospital. Patients with significant but C. difficile toxin-negative diarrhoea did not appear to be that disadvantaged.
It has been suggested that the increase in gastric pH increases the chance of ingested C. difficile survival and thus its chances of developing into a symptomatic infection.29 The majority of patients were taking proton-pump inhibitors for gastro-oesophageal reflux or protection against non-steroidal anti-inflammatory drugs. We recorded a weak association between the use of proton-pump inhibitors and significant diarrhoea caused by C. difficile.