Background The cause of irritable bowel syndrome (IBS) is unknown.It may follow gastroenteritis and be associated with an abnormal gut flora and with food intolerance. Our study was designed to assess whether these factors were associated with colonic malfermentation.
Methods We carried out a crossover controlled trial of a standard diet and an exclusion diet matched for macronutrients in six female IBS patients and six female controls.During the final 72 h on each diet, faecal excretion of fat, nitrogen, starch, and non-starch polysaccharide NSP was measured, and total excretion of hydrogen and methane collected over 24 h in a purpose-built 1.4 m3 whole-body calorimeter. Breath hydrogen and methane excretion were then measured for 3 h after 20 g oral lactulose.
Findings The maximum rate of gas excretion was significantly greater in patients than in controls (2.4 mL/min IQR 1.7-2.6 vs 0.6, 0.4-1.1). Although total gas production in patients was not greater than in controls (median 527 mL/24 h IQR 387-660 vs 412, 234-507), hydrogen production was higher (332, 318-478 vs 162, 126-217, p=0.009). In patients, the exclusion diet reduced symptoms and produced a fall in maximum gas excretion (0.5 mL/min IQR 0.3-0.7). After lactulose, breath hydrogen was greater on the standard than on the exclusion diet.
Interpretation Colonic-gas production, particularly of hydrogen, is greater in patients with IBS than in controls, and both symptoms and gas production are reduced by an exclusion diet. This reduction may be associated with alterations in the activity of hydrogen-consuming bacteria. Fermentation may be an important factor in the pathogenesis of IBS.
Lancet 1998; 352: 1187-89
Irritable bowel syndrome (IBS) is common, but of unknown cause.  It is not a single condition, but a collection of disorders causing similar symptoms of abdominal pain and abnormal bowel habit. Food intolerance is a recognised exacerbating factor of IBS and may affect as many as 50% of patients. [2,3] However, there is no evidence of food allergy.  A range of fermentable substrates may provoke gastrointestinal symptoms, [5,6] as may inhibitors of carbohydrate digestion.  IBS frequently follows gastroenteritis  or use of antibiotics,  and the gut flora has been shown to be unstable.  We postulated that abnormal colonic fermentation could be a factor in the pathogenesis of IBS. We investigated this hypothesis by assessing rates and patterns of colonic-gas production on standard and exclusion diets.
Patients with IBS fulfilling the Rome criteria  were recruited from outpatients. Other gastrointestinal diseases were excluded by blood and stool analysis, sigmoidoscopy, lactose-hydrogen breath test, and, in patients older than 40 years, by barium enema. Controls came from a group of symptom-free individuals recruited for dietary studies. Exclusion criteria were previous gastrointestinal disease, and the use of laxatives, antibiotics, or other medication within the preceding 6 weeks. All patients gave informed consent. The study was approved by the Cambridge Local Research ethics committee and the Dunn Nutrition Centre ethics commitee.
Diets provided from a metabolic kitchen were each taken for 2 weeks with a 2-week washout. The standard diet contained normal western foods. The exclusion diet was that routinely used in this department. Fish and meat, except beef, were allowed, but dairy products were replaced by soya products and cereals other than rice were forbidden. There were also restrictions on yeast, citrus fruits, caffeinated drinks, and tap water.  Each diet provided 8 MJ per day as three isocaloric meals adjusted to achieve energy balance calculated from Schofield's prediction Equation bythe addition of supplements free from starch and fibre. Both diets contained 45% of energy as carbohydrate, 40% fat, and 15% protein, and also 12.5 non-starch polysaccharide (Englyst), 120 g starch, and 3.5 g resistant starch, daily. Compliance was assessed by daily contact and food diary.
Throughout the final 24 h on both diets, indirect calorimetry was done in a purpose built 1.4 m3 polythene canopy supported over a bed by a metal frame. Air was withdrawn at a rate of 100 L/min and replaced from the surrounding room. Exhaust gases were vented outside the room. Recovery of carbon dioxide infused at 0.2 L/min was 98.2% (95% CI 97.8-98.6), and apparent consumption of oxygen, following infusion of nitrogen at the same rate, was 100% (95% CI 99.9-100.1) of predicted values.
Air was continuously drawn from the canopy and from the adjacent room into two Douglas bags that were changed every 30 min. Hydrogen was measured by an electrochemical cell (GMI, Renfrew, UK) and methane by dedicated flame-ionisation detector (Model 3000 hydrocarbon analyser, Signal, Camberley, UK). At an infusion rate of 1 mL/min, recovery of hydrogen was 94% (95% CI 87-100) and of methane 107% (99-115). Recovery of a 5 mL bolus was 103% (97-110) for hydrogen and 107% (86-128) for methane with a measured time-constant of 4 min.
End-expiratory breath samples were also collected by each participant every 30 min during waking hours without leaving the calorimeter. Hydrogen concentration was measured as above, methane by gas chromatography and flame-ionisation detector (Phillips, Cambridge, UK), and the rate of excretion of each gas in breath was estimated. 
Immediately after leaving the calorimeter, having fasted overnight, participants ingested 20 g lactulose (66% solution BP, APS, Leeds, UK) dissolved in 200 mL water. End-expiratory breath samples were collected for a further 3 h and analysed immediately for hydrogen and methane. A validated composite symptom score  and whole gut transit time  were calculated.
All faeces passed during the final 72 h of each diet was frozen immediately in a thermos flask containing solid-phase carbon dioxide, and radiographed to assess recovery of radio-opaque faecal markers. Freeze-dried faeces were ground in a centrifugal mill and analysed for starch and non starch polysaccharide content  and for fat.  A 3-day sample of pooled foods from each diet was similarly analysed.
The study had a 90% power to detect a difference of 600 mL between the median total gas excretions of each group. We compared patient and control groups using the Mann-Whitney U-test, and paired data from each diet using Wilcoxon matched pairs.
Six female patients and six female controls took part. The median age of the patients was 45 years (IQR 42-58) of the controls 53 (51-54). All patients experienced stool frequency and looseness with abdominal pain on a daily basis. None showed evidence of anxiety before entry into the study, as assessed by the Eysenck personality questionnaire. One of each group admitted single violations of their diets, both more than 1 week before calorimetry.
On the standard diet, IBS patients excreted significantly more hydrogen than did controls, though the total volume of hydrogen and methane was not different ( Table 1). Rates of gas excretion were substantially greater in patients ( Table 1, Mann-Whitney U-test, p=0.015). In four of six patients, symptoms occurred when gas excretion was rapid. In IBS patients, but not controls, the total gas excretion (mainly hydrogen) was reduced substantially by the exclusion diet ( Table 1, Figure 1; Wilcoxon matched pairs p=0.031).
Our findings are consistent with the suggestion that the IBS may be caused by abnormal colonic fermentation. Maximum rates of gas excretion (mainly hydrogen) were much greater in IBS patients than in controls. On an exclusion diet, which significantly improved symptoms, gas excretion (hydrogen and methane) fell dramatically, but no such change occurred in controls.
Changes in hydrogen excretion do not result from differences in the rate of expulsion of gas after production, because the relation of breath excretion of hydrogen and methane to total excretion was the same in IBS patients and controls. We attempted to keep differences in delivery of fermentable substrate to the caecum to a minimum by matching the diets closely for nutrient content. The total amount of substrate entering the caecum can only be calculated accurately by invasive methods. There was, however, no evidence of malabsorption, and differences in gas excretion were not associated with differences in the quantity of non-starch polysaccharide excreted or fermented ( Table 2). Furthermore, the differences between the diets in relation to hydrogen and methane excretion on the breath after the standard challenge of 20 g lactulose strongly suggest that the changes in gas excretion were the result of a change in the fermentation process rather than of any differences in the availability of substrate.
Increased rates of gas excretion alone are unlikely to be the cause of symptoms, since controls never developed any symptoms, even at the most rapid rates of excretion, which on occassion, overlapped with those in IBS patients. Other factors, such as those regulating sensory input by the enteric and central nervous system  and the local effects of other fermentation products-such as short-chain fatty acids  or bioactive amines -may be important.
Studies with intubation  or radiography  have revealed no quantitive differences in colonic gas between IBS patients and controls. However, in contrast to our study, such methods do not provide a dynamic picture of gas production. Since our findings relate to small numbers of people, they should not be expected to apply to every patient with IBS. However, they are consistent with the hypothesis that some people with IBS may have disturbances in bacterial fermentation and colonic gas production.
Calorimetry was done by T S King. All authors took part in the design of the study and the drafting of the typescript.
We thank H Englyst of the Dunn Clinical Nutrition Centre, UK, for carrying out carbohydrate analysis, Elaine Collard for food preparation, and Jenny Woolner, State-Registered Dietitian for diet design.
1. Maxwell PR, Mendall MA, Kumar D. Irritable bowel syndrome. Lancet 1997; 350: 1691-95. [Fulltext Link] [Medline Link] [Context Link]
2. Nanda R, James R, Smith H, Dudley CR, Jewell DP. Food intolerance and the irritable bowel syndrome. Gut 1989; 30: 1099-104. [Medline Link] [Context Link]
3. Parker TJ, Naylor SJ, Riordan AM, Hunter JO. Management of patients with food intolerance in irritable bowel syndrome: the development and use of an exclusion diet. J Hum Nutrit Diet 1995; 8: 159-66. [Context Link]
4. Hunter JO. Food allergy-or enterometabolic disorder? Lancet 1991; 338: 495-96. [Medline Link] [Context Link]
5. Ravich RJ, Bayless TM, Thomas M. Fructose: incomplete intestinal absorption in humans. Gastroenterology 1983; 84: 26-29. [Medline Link] [Context Link]
6. Phillips J, Muir JG, Birkett A, et al. Effect of resistant starch on faecal bulk and fermentation dependent events in humans. Am J Clin Nutr 1995; 62: 121-30. [Medline Link] [Context Link]
7. Reuser AJ, Wisselaar HA. An evalulation of the potential side effects of alpha glucosidase inhibitors used for the management of diabetes mellitus. Eur J Clin Invest 1994; 24 (suppl 3): 19-24. [Medline Link] [Context Link]
8. Gwee KA, Graham JC, McKendrick MW, et al. Psychometric scores and persistence of irritable bowel syndrome after infectious diarrhoea. Lancet 1996; 347: 150-53. [Fulltext Link] [Medline Link] [Context Link]
9. Alun Jones V, Wilson AJ, Hunter JO, Robinson RE. The aetiological role of antibiotic prophylaxis with hysterectomy in irritable bowel syndrome. J Obstet Gynecol 1984; 5 (suppl 1): 22-23. [Context Link]
10. Balsari A, Ceccarelli A, Dubini F, Fesce E, Pol G. The faecal microbial population in the irritable bowel syndrome. Microbiologica 1982; 5: 185-94. [Medline Link] [Context Link]
11. Christl SU, Gibson GR, Murgatroyd PR, Scheppach W, Cummings JH. Impaired hydrogen metabolism in pneumatosis cystoides intestinalis. Gastroenterology 1993; 104: 392-97. [Medline Link] [Context Link]
12. King TS, Hunter JO. Do symptoms improve with quality of life in IBS?: British Society of Gastroenterology. Gut 1997; 40 (suppl 1): 64 abstr. [Context Link]
13. Fotherby KJ, Hunter JO. Idiopathic slow transit constipation: whole gut transit times, measured by a new simplified method, are not shortened by opiate antagonists. Alimen Pharmacol Therapeut 1987; 1: 331-38. [Context Link]
14. Englyst HN, Quigley ME, Hudson GJ. Determination of dietary fibre as non-starch polysaccharides with gas-liquid chromatographic, high-performance liquid chromatographic or spectrophotometric measurement of constituent sugars. Analyst 1994; 119: 1497-509. [Medline Link] [Context Link]
15. Wade GN, Jennings G, Trayhurn P. Energy balance and brown adipose tissue thermogenesis during pregnancy in Syrian hamsters. Am J Physiol 1986; 250: R845-50. [Medline Link] [Context Link]
16. Serra J, Azpiroz F, Malagelada J-R. Perception and reflex responses to intestinal distension in humans are modified by simultaneous or previous stimulation. Gastroenterology 1995; 109: 1742-49. [Medline Link] [Context Link]
17. Argenzio RA, Meuten DJ. Short-chain fatty acids induce reversible injury of porcine colon. Dig Dis Sci 1991; 36: 1459-68. [Medline Link] [Context Link]
18. Murray KE, Adams RF, Early JW, Shaw KJ. Studies of the free faecal amines of infants with gastroenteritis and of healthy infants. Gut 1986; 27: 1173-80. [Medline Link] [Context Link]
19. Lasser RB, Bond JH, Levitt MD. The role of intestinal gas in functional abdominal pain. N Engl J Med 1975; 293: 524-26. [Medline Link] [Context Link]
20. Chami TN, Schuster MM, Bohlman ME, Pulliam TJ, Kamal N, Whitehead WE. A simple radiological method to estimate the quantity of bowel gas. Am J Gastroenterol 1991; 86: 599-602. [Medline Link] [Context Link]