Intestinal permeability in patients with irritable bowel syndrome after a waterborne outbreak of acute gastroenteritis in Walkerton, Ontario


Dr J. K. Marshall, Division of Gastroentrology (4W8), McMaster University Medical Centre, 1200 Main Street West, Hamilton, Ontario L8N 3Z5, Canada.


Background : Post-infectious irritable bowel syndrome is a common clinical phenomenon of uncertain aetiology.

Aim : To test the association between intestinal permeability and irritable bowel syndrome symptoms 2 years after a large waterborne outbreak of bacterial gastroenteritis.

Methods : Consecutive adults with Rome I irritable bowel syndrome and controls without irritable bowel syndrome attending a community clinic were enrolled. Intestinal permeability was measured as the ratio of fractional urinary excretions of lactulose and mannitol, and compared among cases vs. controls and predictors of abnormal intestinal permeability were assessed.

Results : A total of 218 subjects (132 irritable bowel syndrome cases and 86 non-irritable bowel syndrome controls) completed the study protocol. About 27 (12%) had been diagnosed with the irritable bowel syndrome before the outbreak and 115 (53%) had been ill during the outbreak. Lactulose–mannitol ratios were increased among cases vs. controls (Mann–Whitney mean rank 118.8 vs. 95.3, P = 0.007), and cases were more likely to have a ratio >0.020 (P = 0.007). Among cases, those with increased intestinal permeability were more likely to report increased stool frequency. Both irritable bowel syndrome symptoms and male gender, but not diarrhoeal illness during the outbreak, were significant predictors of abnormal permeability.

Conclusions : Irritable bowel syndrome symptoms are associated with a subtle increase in intestinal permeability irrespective of prior gastroenteritis. This may improve understanding of the aetiology of both sporadic and post-infectious irritable bowel syndrome.


Between 5 and 30% of patients who suffer an acute episode of infectious gastroenteritis develop chronic gastrointestinal symptoms despite clearance of the inciting pathogen.1–6 This common clinical phenomenon, called post-infectious irritable bowel syndrome (PI-IBS), was first described over five decades ago.7 However, its pathogenesis remains poorly understood and no specific therapy has been identified.

Both animal models and human observation have associated post-dysenteric gut dysfunction with persistent low-grade inflammation in the enteric submucosal and/or neuromuscular compartments.8 In patients with Crohn's disease, it has been suggested that increased permeability exposes the submucosa to luminal antigens and leads to chronic inflammation.9 In patients with PI-IBS, a similar increase in permeability could promote neuromuscular inflammation that disturbs enteric sensation and motility. Indeed, a landmark study by Spiller et al. observed persistent increases in intestinal permeability (IP) among a small cohort of patients with PI-IBS after acute bacterial gastroenteritis.10

In May 2000, contamination of the municipal water supply of Walkerton, Ontario with Escherichia coli 0157:H7, Campylobacter jejuni and other pathogens led to a large outbreak of acute bacterial gastroenteritis11 that affected at least 2300 local residents. Twenty-seven developed the haemolytic uraemic syndrome and six died. We studied IP among cohorts of Walkerton residents with and without persistent symptoms of IBS 2 years following the outbreak.


Walkerton Health Study

The Walkerton Health Study (WHS) was initiated to study the epidemiology and long-term health outcomes of the waterborne outbreak of acute gastroenteritis in May 2000, and to facilitate local residents’ access to specialty clinical care. The WHS is supervised by a multidisciplinary team of Walkerton E. coli Long-term (WEL) Investigators, and receives unrestricted funding from the Ontario provincial Ministry of Health and Long-term Care (MOHLTC). In 2002, the WHS established a special multidisciplinary clinic at the Walkerton Hospital to conduct semistructured assessments of all affected and unaffected residents of Walkerton and its surrounding area. To date, 4315 individuals have participated in the WHS including 2756 permanent residents of the Walkerton municipality (total population approximately 5000).

Permeability study participants

As part of their semistructured assessment, all adults assessed at the WHS clinic completed a modified version of the Bowel Disease Questionnaire, which allows diagnosis of IBS in accordance with the Rome I criteria.12, 13 For a 6-week enrolment period in July and August 2002, all consecutive patients with Rome I IBS were also invited to participate in a study of IP. During the same period, every fifth consecutive patient without IBS was also invited to participate as a control. Patients were excluded from participation if they: (i) were <16 years of age; (ii) resided outside the Walkerton postal code zone in May 2000; (iii) had a known history of inflammatory bowel disease, coeliac disease or diabetes mellitus; or (iv) were either unable or unwilling to refrain from use of alcohol or non-steroidal anti-inflammatory drugs (NSAIDs) as required by the protocol (see below).

Permeability measurement

Intestinal permeability was measured by timed urine collection after ingestion of an oral load of lactulose and mannitol as described by Meddings.14 Subjects were required to refrain from ingestion of alcohol, aspirin (ASA) and NSAIDs for at least 5 days prior to the test. At bedtime on the evening of the test, subjects fully emptied their bladders, then ingested a standard oligosaccharide solution of sucrose (100 g), lactulose (5 g), mannitol (2 g) and flavoured drink crystals (1.5 g) in 500 mL of tap water. All urine produced overnight was collected in a plastic container with 5 mL of thymol solution as a preservative.

Urine containers were returned immediately to the study centre, where the total urine volume was measured and 10 mL aliquots were frozen at −20 °C. Urinary concentrations of sucrose, lactulose and mannitol were measured by high-pressure liquid chromatography (HPLC). The fractional excretion of each sugar was then estimated as the ratio of its total urinary excretion (urine concentration multiplied by total urine volume) to the total oral dose. For each subject, the lactulose–mannitol ratio was calculated as the fractional excretion of lactulose divided by that of mannitol (normal: <0.025).

Data analysis

The characteristics of participants with and without IBS were analysed descriptively. Proportions of subjects with abnormal lactulose–mannitol ratios were compared between groups using chi-square tests. In the primary analysis, lactulose–mannitol ratios >0.025 were considered to be abnormal but other thresholds (0.020, 0.030, 0.035, 0.040) were explored in secondary analyses. The lactulose–mannitol ratios of IBS cases and non-IBS controls were also compared with continuous outcomes using a non-parametric approach (Mann–Whitney test) because their distribution was not normal.

The primary analysis included all subjects, regardless of their IBS symptoms before the outbreak or their enteric symptoms during the outbreak. Subgroup analyses then excluded participants with premorbid IBS before the outbreak, and those who experienced no acute diarrhoeal illness during the outbreak. Individual subject characteristics were tested in univariate and multivariate logistic regression analyses for their association with increased gut permeability. Characteristics considered in this model included age, gender, history of acute diarrhoea in May 2000 (defined as passing at least 4 stools/day for 3 days), characteristics of the acute illness (weight loss, bloody stools, abdominal pain, duration of diarrhoea, maximum number of daily stools) and use of specific medications during the acute illness (antisecretory agents, antibiotics or systemic corticosteroids).

Sample size determination

We estimated that, with 120 IBS cases and 80 non-IBS controls, the study design would provide 80% power (β: 0.20) to detect a 15% difference in the prevalence of abnormal IP assuming that IP is abnormal in 5% of non-IBS controls.

Ethics and consent

The study protocol was approved by the Research Ethics Board of Hamilton Health Sciences (Hamilton, ON). All subjects provided full written and informed consent to participate.


Study participants

A total of 220 eligible subjects completed the protocol and returned their timed urine specimens. Two were subsequently excluded because of a prior diagnosis of inflammatory bowel disease, leaving 218 participants (132 IBS cases and 86 non-IBS controls). Characteristics of the participants are summarized in Table 1. Twenty-seven subjects (12.4%) reported a clinical diagnosis of IBS made prior to the outbreak of May 2000 and 115 (52.8%) recalled an episode of acute diarrhoea during the outbreak (defined as 3 days of diarrhoea with a peak of at least 4 stools/day). Because local health resources were severely strained during the outbreak, only a minority of affected residents was able to submit stool specimens for culture. Accordingly, only six participants in this study had bacterial gastroenteritis that was confirmed by culture (three with C. jejuni and three with E. coli 0157:H7).

Table 1.  Characteristics of study participants
 IBS cases (N = 132)Non-IBS controls (N = 86)P-value
  1. IBS, irritable bowel syndrome.

Age in years, mean (range)44.7 (16–84)48.9 (17–84)0.039
Male, n (%)51 (38.6)34 (39.5)0.894
IBS before May 2000, n (%)22 (16.7)5 (5.8)0.017
Acute diarrhoea May 2000, n (%)82 (62.1)33 (38.4)0.001

Intestinal permeability

The lactulose–mannitol ratios of IBS cases were significantly higher than those of non-IBS controls (Mann–Whitney mean rank 118.8 vs. 95.3, P = 0.007). Forty-seven IBS cases (35.6%) had lactulose–mannitol ratios >0.020, compared with 16 non-IBS controls (18.6%, P = 0.007) (Table 2 and Figure 1). The prevalence of abnormal IP using a lactulose–mannitol ratio cut-point of 0.025 was numerically higher among IBS cases than among non-IBS controls (15.9% vs. 8.1%), but the difference was not statistically significant (P = 0.09).

Table 2.  Prevalence of abnormal permeability among cases and controls using various lactulose–mannitol ratio cut-points among all subjects (N = 218), subjects with no recalled diagnosis of irritable bowel syndrome (IBS) before the outbreak (N = 191), and subjects with no recalled diagnosis of irritable bowel syndrome before the outbreak who reported acute diarrhoea during the outbreak (N = 105)
PopulationThreshold Lac-Man ratioAll subjects, N (%)IBS cases, N (%)Non-IBS controls, N (%)P-value
  1. * Chi-square test (cases vs. controls).

  2. † Fisher's exact test (cases vs. controls).

All subjects (N = 218)>0.02063 (29)47 (36)16 (19)<0.01*
>0.02528 (13)21 (16)7 (8)0.09*
>0.03019 (9)14 (11)5 (6)0.22*
>0.03510 (5)8 (6)2 (2)0.32†
>0.0406 (3)5 (4)1 (1)0.41†
No premorbid IBS (N = 191)>0.02056 (29)40 (36)16 (20)0.01*
>0.02523 (12)16 (15)7 (9)0.22*
>0.03014 (7)9 (8)5 (6)0.60*
>0.0357 (4)5 (5)2 (2)0.70†
>0.0405 (3)4 (4)1 (1)0.40†
No premorbid IBS with acute diarrhoea illness during outbreak (N = 105)>0.02030 (29)26 (35)4 (13)0.03†
>0.02514 (13)11 (15)3 (10)0.75†
>0.03010 (10)7 (10)3 (10)1.00†
>0.0355 (5)3 (4)2 (7)0.63†
>0.0403 (3)2 (3)1 (3)1.00†
Figure 1.

Distributions of lactulose–mannitol ratios of study subjects with and without irritable bowel syndrome by Rome I diagnostic criteria.

In univariate logistic regression, male gender was also positively associated with abnormal permeability using a cut-point of 0.020 (P = 0.02) but not 0.025 (P = 0.09). Thirty-six of 85 males (42.4%) had a lactulose–mannitol ratio >0.020, compared with 27 of 133 females (20.3%). In multiple logistic regression analysis, both current IBS (OR: 2.99, 95% CI: 1.46–6.11, P = 0.003) and male gender (OR: 3.09, 95% CI: 1.64–5.83, P < 0.001) persisted as significant independent predictors of a lactulose–mannitol ratio >0.020. None of the other patient characteristics, including history of acute diarrhoeal illness during the outbreak, was significantly associated with abnormal permeability in univariate analysis using any lactulose–mannitol ratio threshold. IP among the six subjects with confirmed bacterial gastroenteritis did not differ significantly from that observed among other subjects.

Among the 191 participants who had not received a clinical diagnosis of IBS prior to May 2000 (110 IBS cases and 81 non-IBS controls), results were similar to those reported from the primary analysis (Table 2). In this subgroup, weight loss during the outbreak also emerged as a predictor of increased IP at lactulose–mannitol ratio cut-points of 0.030 and higher (i.e. not 0.020 or 0.025). Among the 105 participants with no diagnosis of IBS before the outbreak who recalled an acute diarrhoeal illness during the outbreak, results were similar; a lactulose–mannitol ration >0.020 was seen in 26 of 74 IBS cases (35%) vs. four of 31 non-IBS controls (13%; P = 0.03) (Table 2).

Among the 132 IBS cases, responses to each question on the Bowel Disease Questionnaire were tested for their association with abnormal IP. Cases with lactulose–mannitol ratios >0.020 were more likely to report increased stool frequency than those with lactulose–mannitol ratios <0.020 (63.8% vs. 44.7%, P = 0.035). No other significant relationships were identified.


This study is the largest reported analysis of IP in patients with IBS of any aetiology. Its results suggest that IP is increased among subjects with persistent IBS symptoms 2 years after a waterborne outbreak of acute bacterial gastroenteritis, and corroborate the earlier findings of Spiller et al.10 This important observation provides evidence of an organic disturbance in patients with functional gut disease, contributes to a conceptual model of the pathogenesis of IBS, and could lead to novel therapies for this challenging syndrome.

Our study was conducted 2 years after a large waterborne outbreak of acute bacterial gastroenteritis. The majority of subjects with IBS recalled an acute diarrhoeal illness during the outbreak and could be classified clinically as suffering from PI-IBS. To further isolate incident cases of PI-IBS, secondary analyses excluded subjects who reported a prior clinical diagnosis of IBS. However, an unexpected finding was that no analysis associated a self-reported acute diarrhoea during the outbreak with abnormal IP. In part, this may reflect the limitations of a retrospective design that required subjects to recall details of their acute diarrhoeal illness and relied on prior diagnoses of IBS that would not necessarily satisfy standard diagnostic criteria. Alternatively, it may imply that disturbances in IP are found in both sporadic and post-infectious IBS. Further studies are clearly needed to pursue these hypotheses.

Measuring the fractional urinary excretion of lactulose and mannitol after ingestion of a standard oral load is a well-accepted, non-invasive and inexpensive means of assessing small IP.14 Uptake of mannitol follows a transcellular route and correlates with intestinal surface area. Lactulose absorption occurs through a paracellular route and is enhanced by intestinal damage. Therefore, the ratio of fractional lactulose to mannitol excretion measures intestinal injury normalized to total surface area. Other common factors that increase lactulose permeability, such as inflammatory bowel disease, coeliac disease, alcohol consumption and NSAID, were excluded by our protocol. For the most part, lactulose–mannitol ratios measure small IP with only small influences from changes in gastric or colonic permeability.

Previous studies have used lactulose–mannitol ratio cut-points between 0.025 and 0.030 to distinguish healthy controls from disease states-like Crohn's disease.15 Using a lactulose–mannitol ratio cut-point of 0.025, we observed a trend towards increased permeability in IBS cases but no significant difference relative to non-IBS controls, perhaps because of insufficient statistical power. Our findings were statistically significant using a lower lactulose–mannitol cut-point (0.020) and analysing lactulose–mannitol ratios as a continuous outcome. Arguably, a lower diagnostic threshold may be more appropriate to distinguish a more subtle disorder-like PI-IBS. However, this observation requires further validation in other similar populations.

In this cohort, increased IP was associated with male gender. To our knowledge, this is a novel observation that has not been described previously. However, uptake of single probes (such as sucrose for gastric permeability or mannitol for small bowel permeability) depends on surface area, and gender differences in measures of permeability are plausible. Among those who were ill during the outbreak and who had no prior history of IBS, we also observed an association between IP and weight loss during the outbreak. Here, it is likely that weight loss serves as a surrogate marker for acute illness severity. Hence, severe gastroenteritis may be more likely to increase IP and more likely to result in post-infectious IBS. All of these observations require confirmation in other prospective series.

The aetiology of PI-IBS remains poorly understood, but has been postulated to reflect persistent gut inflammation with disturbed neuromuscular function.8 Increased IP could enhance neuromuscular exposure to luminal antigens, promote immune activation and maintain gut dysfunction. A subset of patients with sporadic IBS also appears to be genetically predisposed to chronic inflammation because of reduced expression of counter-inflammatory cytokines.16 Spiller et al. have observed increased enteroendocrine and T-cell numbers in the rectal biopsies of patients with PI-IBS after acute Campylobacter gastroenteritis,10 and Chadwick et al. have demonstrated immune activation (with increased CD3+ and CD25+ lymphocytes) in patients with sporadic IBS.17 However, a recent controlled trial of systemic corticosteroids in patients with PI-IBS was negative,18 suggesting that anti-inflammatory therapy would have to be started early in order to avert and/or alter its natural history.

Increased gut permeability is also thought to occur early in the pathogenesis of inflammatory bowel disease,9 but it remains unclear whether this phenomenon is primary or secondary. Although the aetiology of inflammatory bowel disease is complex and likely multifactorial, observational studies have linked it to prior episodes of bacterial enteritis19 and to recent diagnoses of IBS.20 IBS and inflammatory bowel disease may also share a genetic predisposition that impairs the down-regulation of inflammation.16, 21 A common defect in IP would provide further evidence of a common predisposition to both PI-IBS and IBD, and suggest that patients who develop PI-IBS after enteric infection are at an increased future risk of IBD. This intriguing but unproven hypothesis requires careful prospective validation.

Antecedent psychological stress has been identified as a risk factor for the development of PI-IBS.3 Stress has also been demonstrated to increase IP and promote the trans-epithelial passage of luminal antigens in rat models.22 Thus, changes in IP could be one biological mechanism by which stress leads to chronic gastrointestinal disturbance after enteric infection.

It has been observed that PI-IBS is more likely than sporadic IBS to follow a diarrhoea-predominant symptom pattern,23 suggesting that these represent distinct clinical disorders. In our cohort, 68 of 132 subjects with IBS (51.5%) reported increased stool frequency at least 25% of the time. Subjects with lactulose–mannitol ratios >0.020 were more likely than others to report frequent stool passage (63.8% vs. 44.7%, P = 0.035). This finding is consistent with results from animal models of PI-IBS wherein increased in permeability after enteric infection triggers neuromuscular inflammation with smooth muscle hypercontractility and increased stool frequency.

In conclusion, we have demonstrated subtle increases in IP in a cohort of patients with IBS 2 years after a local waterborne outbreak of acute gastroenteritis. This observation contributes to a conceptual model of IBS that requires further validation and study.


This study was funded by an arms-length grant from the Ontario Ministry of Health and Long-term Care. Drs Marshall and Garg received Clinician Scientist Awards from the Canadian Institutes of Health Research (CIHR). Other members of the WEL Investigators include Drs John Howard, Doug Matsell and Marina Salvadori.