SEARCH

SEARCH BY CITATION

Keywords:

  • irritable bowel syndrome;
  • postinfectious;
  • functional gut disorder;
  • prospective cohort study

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References

Abstract  Postinfectious functional gastrointestinal disorders (FGIDs) may not be specific to gastroenteritis. This pilot study aimed to ascertain the 3- and 6-month incidence of functional gut disorders in people with non-gastrointestinal (GI) infection, gastroenteritis and healthy controls. This was a prospective study of three cohorts recruited from hospital (non-GI infections) and the community (others). FGIDs were diagnosed using self-completed Rome II modular questionnaires administered at baseline, 3 and 6 months. Thirty-six subjects with non-GI infection, 219 healthy subjects and 108 with bacterial gastroenteritis participated. No difference in incidence of FGID was detected between the GI and non-GI infection cohorts. Any FGID was more frequent in people who had a non-GI infection than in controls at both 3 [odds ratio: 4.34 (95% CI: 3.60–16.45)] and 6 months [4.76 (4.42–27.92)]. Irritable bowel syndrome (IBS) alone was more frequent in people with non-GI infections than in controls at 3 months (6.12 [1.30–29.12]) but did not quite reach statistical significance at 6 months (4.58 [0.79–26.46]). Our findings were unexpected. Postinfectious FGIDs may be related to non-GI and GI infection, although not all potential biases were controlled in study design. Further studies need to explore these preliminary findings and, if confirmed, the underlying mechanisms.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References

Functional gastrointestinal disorders (FGIDs), of which the irritable bowel syndrome (IBS) is the most common, are characterized by chronic symptoms referable to the gastrointestinal (GI) tract in the absence of a currently recognized biochemical or structural explanation.1 The aetiology of IBS and other FGIDs is complex, with physical, behavioural and psychological factors involved. A small but significant proportion of patients with IBS report an onset of symptoms following a bout of infectious gastroenteritis. Evidence in support of this ‘postinfectious’ aetiology is compelling2–7 although the pathogenesis of postinfectious IBS (PI-IBS) remains unclear.

It has been suggested that the association of IBS with intestinal infections implies that intestinal inflammation and the mucosal immune system play a role in symptom generation, and several studies have investigated the presence of intestinal inflammation in PI-IBS.8–12 It is also thought that low-grade inflammation is capable of sensitizing both intrinsic and extrinsic primary afferent neurones, leading to altered intestinal motor function, visceral hypersensitivity and symptom generation.13 A recent trial of prednisolone, however, failed to show any effect on symptoms of PI-IBS.14 Alternatively, the association between gastroenteritis and PI-IBS could be explained through behavioural and psychological factors, such as anxiety and depression,4 consultation behaviour or personal habits, such as smoking.15 We have previously shown that other functional gut disorders, especially functional diarrhoea, are also associated with GI infections.7

One important gap in our current knowledge is whether PI-IBS is specific to GI infections or whether other systemic bacterial infections might also cause PI-IBS or functional diarrhoea. This would have important implications for understanding of the aetiology of PI-IBS. We are not aware of any study that has investigated the incidence of IBS and other FGIDs following a non-GI infection.

Aims

The aim of this study was therefore to obtain preliminary information on the incidence of IBS and other FGIDs following a non-GI infection. Assuming that PI-IBS is mediated directly or indirectly through GI inflammation, our prior hypothesis was that the incidence of functional gut disorders following a non-GI infection would approximate to that of a control population, and would be lower than that seen after GI infection. This study was conceived as a pilot study in order to obtain better information to plan a definitive study and enable an accurate power calculation.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References

Setting and subjects

The non-GI infection cohort for this study was recruited from patients between the ages of 18 and 80 years admitted to North Tyneside General Hospital, a large district general hospital serving an estimated 250 000 people, between October 2003 and January 2004. Cases were identified from positive microbiology cultures and through review of patients admitted to the Medical Assessment Unit. Respiratory infections were defined according to British Thoracic Society guidelines, which require chest X-ray changes, and some other systemic evidence of infection.16 Septicaemia and urinary tract infection were diagnosed by positive culture. Skin and soft tissue infections were defined by combination of classical physical signs, raised C-reactive protein or erythrocyte sedimentation rate and consultant opinion. Exclusion criteria were inability to give informed consent, pregnancy, severe current psychiatric illness, history of abdominal surgery, inflammatory bowel disease and cancers that might be predicted to affect bowel symptoms, and alcoholism. We also excluded people with a prior functional gut disorder, diagnosed at entry to the study as described below using the Rome II questionnaire.

The decision to recruit hospital patients with infection for this study was taken because of essentially practical reasons. It would also have been desirable to compare community-based patients with non-GI infections to our community GI infections and controls. However, in primary care in the UK, patients are often treated empirically with no radiological or microbiological evidence for infection and we felt that recruitment of cases would be problematic. As a pilot study, it was also important to get further data on outcomes before using resources to attempt to recruit in primary care, where recruitment was likely to be more difficult.

The recruitment methods used for the cohorts for GI infection and control subjects have been described in detail.7 Briefly, patients with a culture-proven GI infection (mainly Campylobacter species) were recruited from the community. Controls were age- and sex-matched from the same general practice. Entry and exclusion criteria were similar to the current study and FGIDs were defined in exactly the same way using the Rome II questionnaire.

The Newcastle and North Tyneside Local Research Ethics Committee granted ethical approval and all participants gave written informed consent. The patient information leaflet was carefully worded to make sure that the participants were not biased by knowledge of the hypothesis being tested.

This was a prospective parallel cohort study to investigate a potential association between non-GI infections and the development of FGIDs. FGIDs were diagnosed according to the modular components of the Rome II questionnaire relating to IBS, functional dyspepsia and functional diarrhoea, as these are the most frequent FGIDs. The Rome II questionnaire is widely used and has been validated for use in both clinical practice and epidemiological surveys.17 Demographic data were obtained at baseline.

At the outset the Rome II questionnaire was administered to identify those subjects with a pre-existing FGID who were then excluded from further participation. Follow-up postal questionnaires were sent to eligible, consented subjects at 3 and 6 months to examine the incidence of development of new cases of IBS, functional dyspepsia and functional diarrhoea. It was not thought practical to collect other data (such as current drug therapy) with the use of a self-completed postal questionnaire, particularly as questionnaire burden is likely to negatively influence participation rate. Results for the subjects with a non-GI infection were compared to those with gastroenteritis and control subjects from the previous study.7

The study was designed to detect a difference in the incidence of any FGID at 6 months between previously studied gastroenteritis cases, in which the observed rate of FGID development at 6 months in the gastroenteritis cases was 25%,7 and the new non-GI infection cohort. We had no information on which to base our power calculation for the study but given the prior null hypothesis we assumed that the incidence of FGIDs following non-GI infection should approach that of the control group (2.9%). The distribution of functional gut disorders between the three subject groups was explored from a 3 × 2 table using an online χ2 calculator (http://www.georgetown.edu/faculty/ballc/webtools/web_chi.html). Odds ratios were calculated using spss statistical package version 10 (SPSS Inc., Chicago, IL, USA).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References

During the recruitment period, 126 non-GI infections in people between 18 and 80 were identified with 32 patients excluded because of inability to give informed consent, 22 because of co-existing malignancy, nine because of pre-existing FGID and 12 for miscellaneous other reasons including death and discharge before we were able to approach them or before infection was confirmed. Thus, there were 51 eligible cases of whom 36 (70%) consented to participate. The response rates at 3 and 6 months were 27 (75%) and 24 (66%) respectively. For comparison, in the GI infection cohort, we identified 500 cases but only 128 eligible cases participated after exclusions and those not wishing to participate and 108 returned their 3- and 6-month questionnaires.

The bacterial infections diagnosed in the patients were respiratory (n = 14), cellulitis (n = 7), urinary tract infection, including pyelonephritis (n = 8), septicaemia (n = 2), wound infections (n = 2), osteomyelitis (n = 1), empyema (n = 1) and skin abscess (n = 1). Patients received a wide variety of antibiotics, often more than one and frequently by more than one route. Numbers for each antibiotic were too small for meaningful analysis.

The mean age of the 36 patients with non-GI infections was 56.4 years (SD: 19.16, range: 18–79), and 17 (47%) were male. The mean age and sex distribution of the patients with GI infections was 46.6 (SD: 14.25) with 43% male, and for the controls the mean age was 49 (SD: 12.79) and 35% were male.7

Table 1 shows the proportions of individuals from each cohort who, at 3 and 6 months, had either any functional gut disorder studied (i.e. functional dyspepsia, functional diarrhoea, IBS) or IBS alone. The distribution of the cases between the three groups differed from expected in all four categories. Odds ratios (with 95% confidence intervals) for any functional GI disorder at 3 months for non-GI infection group against controls and 4.34 (3.60–16.45) and against infectious gastroenteritis was 1.77 (0.62–5.10). At 6 months these odds ratios are 2.33 (0.65–8.44) and 4.76 (4.42–27.92) respectively. For end point of IBS at 3 months, odds ratios for non-GI infection against control and GI infection, respectively, were 6.12 (1.30–29.12) and 1.6 (0.44–5.89) and at 6 months they were 4.58 (0.79–26.46) and 2.20 (0.48–10.20). In summary therefore, at 3- and 6-month follow-up, FGIDs in general were more frequent in the non-GI infection group than the controls but no difference was found from the GI infection cohort and IBS was significantly more frequent 3 months after entry in the non-GI infection group from the controls but no difference was found between the GI and non-GI infection group in incidence of IBS.

Table 1.   Incidence of functional gut disorders at 3 and 6 months
FGID (months)ControlsGI infectionNon-GI infectionP-value for chi-square test for 3 × 2 table
  1. FGID, functional gastrointestinal disorder; IBS, irritable bowel syndrome.

Any at 310/201 (5%)31/108 (28.7%)5/27 (18.5%)<0.001
Any at 66/206 (2.9%)27/108 (25%)3/24 (12.5%)<0.001
IBS at 34/201 (1.9%)18/108 (16.7%)4/27 (14.8%)<0.001
IBS at 64/206 (1.9%)18/108 (16.7%)2/24 (8.3%)<0.001

The four cases of IBS at 3 months were in people with cellulitis (n = 3) and pneumonia (n = 2; one patient had pneumonia and cellulitis).

The numbers with functional diarrhoea and functional dyspepsia were too small in the controls and non-GI infection cohorts to allow statistically meaningful comparison.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References

This is the first report of FGIDs after non-GI bacterial infections and we found completely the opposite result from that which we expected. Because of the limitations of the study, described below and acknowledged at the design phase, no definitive conclusions can be drawn at this stage. However, the study does give useful information for the design of further studies. Unexpectedly, this study shows that patients who had a confirmed non-GI bacterial infection are more likely than a community control group to have symptoms of FGIDs as diagnosed by the Rome II criteria at 3 and 6 months following their infection. There was no detectable difference in the incidence of FGIDs between GI and non-GI infections but because the frequency of PI-IBS in the non-GI infection group was higher than expected and because of difficulties in obtaining the planned sample, we may not have had sufficient power to detect a small true difference.

The strengths of the study lie in the fact that the cohorts were recruited prospectively, we carefully excluded people with pre-existing FGIDs and we utilized validated, standardized questionnaires to define the outcome measures of FGIDs, and strict criteria for diagnosing bacterial infections.

It is never possible to eliminate all sources of bias in a study and this one is no exception. For a number of reasons we chose a cohort of patients with non-GI infections recruited as hospital inpatients to compare with our previous cohorts of GI infection and controls who were recruited in the community. In that previous study we had identified 500 cases of community-treated gastroenteritis to recruit just 128 actual cases to the study. There were many reasons for drop outs including the two-layer approach to the patient [first contact had to come from the patient's General Practitioner (GP), not the research team] that was required by our local ethics committee and people who were ineligible because of a pre-existing functional gut disorder or who simply did not wish to participate. We also knew that it would be difficult to identify patients with definite non-GI infection in primary care, where many people are treated empirically. So we initially designed a pilot study to clarify the risk of FGID after non-GI infection, which would help in the power calculation for a definitive study and would help to establish the practicalities of such a study in the community. There are insufficient numbers of hospital inpatients with bacterial gastroenteritis to recruit a hospital GI infection cohort.

The control cohort are healthy community controls. It could be argued that a hospital control cohort would have been more suitable and would have reduced some sources of bias. However, hospital controls without infection might introduce other sources of bias, such as additional medications, behaviours, such as alcohol consumption or smoking, various diseases known to affect bowel function and admissions with mental health problems or somatization disorders that are known to be associated with functional gut disorders. It would not be sensible to assume that different categories of patients admitted to hospital can be used uncritically as controls for each other. They may well not share the same behavioural and epidemiological factors and so new sources of bias might be introduced. It would be no doubt be interesting to compare the rate of FGID in cohorts after admission for non-GI infection or for non-infective illnesses and that would be the focus of further study. Such a study would, however, need to be large and would optimally include several different control categories to try and reduce and explore the sources of bias that influenced the findings.

A second weakness of the study was limited power to be sure that there was no true difference in rates of subsequent FGID between GI and non-GI infection. We had originally planned to recruit approximately 45 patients but we had to exclude nine consented subjects who were willing to participate but in whom the Rome II questionnaire revealed a pre-existing FGID. There was also a small drop out rate in the 6-month follow-up phase of the study. This study does, however, give an interesting estimation of the frequency of FGIDs after bacterial infections in a hospital cohort. This information should be helpful in the design of adequately powered studies to further investigate the incidence and aetiology of postinfectious IBS. It is clear that the size of the groups in such a study will have to be larger than the one that we planned to recruit for this study. Our study will help in the planning of further studies of functional gut disorders after non-GI infections, which should compare hospitalized non-GI infections with appropriate hospitalized controls, and community non-GI infections with community controls. The lack of power to distinguish between GI and non-GI infections does not, however, weaken our finding that there was a difference between the controls and the non-GI infection cohorts.

Notwithstanding these limitations, we do believe that our finding is novel and suggests the need for further investigation. There may be a number of explanations for the differences seen between the control and non-GI infection cohorts in this study.

Firstly, they may be an artefact due either to the lack of comparability between the three cohorts (discussed above) or it maybe a type II error due to the small numbers or it may arise from the drop out rate. A type II error is unlikely because the P-values for the 3 × 2 table are very small and the confidence intervals for the ‘any FIGD’ end point do not approach 1. Our results need to be replicated in more and preferably larger studies with both inpatients and outpatients with non-GI infections, and appropriate controls. The drop out rate in the three cohorts was similar, so bias resulting from this source (for example, more participants with bowel symptoms responding) is unlikely to be significant.

Secondly, it may be a true finding but the link between systemic bacterial infections and FGIDs may be due to another factor, such as having been a recent inpatient, the use of antibiotics, other medication or co-existing illnesses. In this context we acknowledge that the choice of cohorts has not eliminated all bias. However, early studies of new concepts are often imperfect in their methods, particularly when they have found a new and unexpected finding. For example, the first highly acclaimed paper on postinfectious IBS used a highly selected and small inpatient group primarily with Salmonella infection and had no control group, but is widely acknowledged as having established the presence of PI-IBS as a distinct disease.3 It is very unlikely that infection with hospital-acquired infections, such as Clostridium difficile could be responsible for our findings as most patients develop diarrhoea and would have had appropriate stool testing. IBS was diagnosed with the highly specific Rome criteria that define IBS predominantly in terms of abdominal pain rather than diarrhoea.17

Thirdly, there may be a causative link between systemic bacterial infections and FGIDs, particularly IBS, but this remains unproven and further studies are needed. These studies should not only seek to clarify the true risk, but also pursue possible mechanisms. There is some suggestion that the duration of symptoms of gastroenteritis is a predictor for PI-IBS3,9 and smoking and psychological factors, particularly depression, and life events have also been suggested as important factors.15 The role of the immune system in PI-IBS has also been highlighted.8–12 The site-specific bacterial infections that we have seen in this study will; however, have systemic effects on the immune system with increased circulating levels of proinflammatory cytokines, activation of clotting cascades and activation of vascular endothelium amongst many other systemic effects. It is therefore quite possible that the activation of these immune pathways in both systemic bacterial infections and specific GI infections could have similar effects on bowel function, including local sensitization of somatic afferent nerve function.13 The role of the specific infections involved and the antibiotics given would be an interesting further study in a larger cohort.

In conclusion, in this study FGIDs were more frequent in patients 3 and 6 months after non-GI bacterial infection than a control group but we did not detect a difference between GI and non-GI infection. Our findings need further confirmation but it does throw up further interesting questions about the pathogenesis of postinfectious functional gut disorders. In particular, there is an intriguing suggestion that it may be the systemic rather than the local effects of bacterial infection that give rise to PI-IBS.

Acknowledgment

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References

The study was funded by a grant from Northumbria Healthcare NHS Trust.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. References
  • 1
    Halder SL, Locke GR, Talley NJ et al. Impact of functional gastrointestinal disorders on health-related quality of life: a population-based case-control study. Aliment Pharmacol Ther 2004; 19: 23342.
  • 2
    McKendrick MW, Read NW. Irritable bowel syndrome – post-Salmonella infection. J Infect 1994; 29: 13.
  • 3
    Gwee KA, Graham JC, McKendrick MW et al. Psychometric scores and persistence of irritable bowel after infectious diarrhoea. Lancet 1996; 347: 1503.
  • 4
    Gwee KA, Leong YL, Graham C et al. The role of psychological and biological factors in postinfective gut dysfunction. Gut 1999; 44: 4006.
  • 5
    Rodriguez LA, Ruigomez A. Increased risk of irritable bowel syndrome after bacterial gastroenteritis: cohort study. Br Med J 1999; 318: 5656.
  • 6
    Neal KR, Hebden J, Spiller R. Prevalence of gastrointestinal symptoms six months after bacterial gastroenteritis and risk factors for development of the irritable bowel syndrome: postal survey of patients. Br Med J 1997; 314: 77982.
  • 7
    Parry SD, Stansfield R, Jelley D et al. Does bacterial gastroenteritis predispose people to functional gastrointestinal disorders? A prospective, community-based, case-control study. Am J Gastroenterol 2003; 98: 19705.
  • 8
    Hiatt RB, Katz L. Mast cells in inflammatory conditions of the gastrointestinal tract. Am J Gastroenterol 1962; 37: 5415.
  • 9
    Wang LH, Fang XC, Pan GZ. Bacillary dysentery as a causative factor of irritable bowel syndrome and its pathogenesis. Gut 2004; 53: 1096101.
  • 10
    Spiller RC, Jenkins D, Thornley JP et al. Increased rectal mucosal enteroendocrine cells, T lymphocytes, and increased gut permeability following acute Campylobacter enteritis and in post-dysenteric irritable bowel syndrome. Gut 2000; 47: 80411.
  • 11
    Gwee KA, Collins SM, Read NW et al. Increased rectal mucosal expression of interleukin 1beta in recently acquired post-infectious irritable bowel syndrome. Gut 2003; 52: 5236.
  • 12
    Collins SM. Irritable bowel syndrome could be an inflammatory disorder. Eur J Gastroenterol Hepatol 1994; 6: 47881.
  • 13
    Barbara G, De Giorgio R, Stanghellini V et al. A role for inflammation in irritable bowel syndrome? Gut 2002; 51: 414.
  • 14
    Dunlop SP, Jenkins D, Neal KR et al. Randomised, double blind, placebo-controlled trial of prednisolone in post-infectious irritable bowel syndrome. Aliment Pharmacol Ther 2003; 18: 7784.
  • 15
    Parry S, Barton JR, Welfare MR. Factors associated with the development of post-infectious functional gastro-intestinal diseases (FGID's): does smoking play a role? Eur J Gastroenterol Hepatol 2005; 17: 10715.
  • 16
    British Thoracic Society Standards of Care. BTS guidelines for the management of community acquired pneumonia in adults. Thorax 2001; 56: 164.
  • 17
    Drossman DA, Corazziari E, Talley NJ et al. Rome II Modular Questionnaire – Investigator and Respondent Forms. In Drossman DA, Corazziari E, Talley NJ et al. (Eds) Rome II: the Functional Gastrointestinal Disorders. Diagnosis, Pathophysiology and Treatment: a Multinational Consensus, 2nd edn. USA: Allen Press, 2000.