Table 3. RCTs of specific diets excluded from the dietary intervention in IBS systematic review.

Reference	Design	Participants	Interventions	Methodology	Outcomes
King et al.13	UK RCT, single center	6 Female Rome I IBS and 6 female healthy controls. Tested for food serum IgG assay. Recruited from secondary care. Fecal excretion of fat, nitrogen, starch, and non-starch polysaccharide NSP was measured, and 24 h excretion of hydrogen and methane.	Restriction diet of exclusion of beef, all dairy products, yeasts, fruits, cereals, rice, caffeinated drinks, and tap water for 2 weeks vs. normal western diet in a crossover trial.	Method of randomization and concealment of allocation not stated. Patients not blinded—unclear if researchers masked).	For IB patients, the median symptom score was 4 (3–7) on the exclusion diet compared with 8 (IQR 5·25–10) for the standard diet (P=0.0001). Breath hydrogen production was greater in patients with IBS compared with controls, and in both groups gas production was reduced by an exclusion diet.
Ong et al.14	Australian RCT, single center	15 Rome III IBS patients and 15 healthy controls. Intolerant of gluten but celiac excluded. Healthy volunteers recruited from advertisement, IBS patients from secondary care clinic. 87% IBS patients female. Fourteen-hour excretion of hydrogen and methane measured.	High (50 g per day) vs. low (9 g per day) FODMAPs diet for 2 days in a crossover design.	Method of randomization and concealment of allocation not stated. Single-blind (patients not blinded).	Overall IBS symptom score was significantly higher for IBS patients during the high FOMAPs diet (median 6; range 2–9) compared with low FODMAPs diet (2; 0–7) (P=0.002). Breath hydrogen production was greater in patients with IBS than in controls. Breath hydrogen greater on high vs. low FODMAP diet in IBS and control patients.
Bentley et al.15	UK RCT, single center	27 Patients with clinically diagnosed of IBS. Recruited from secondary care; 81% female. All had skin prick tests for potential food allergy.	All patients asked to eat a diet of lamb, pears, and rice for an unspecified duration. Other foods then individually introduced at least 3 days apart, and if any caused symptoms, then the patient was exposed to a double-blind provocation trial with that food—three capsules with offending food and three with placebo in a random order. Patient needed to identify the challenge correctly on at least 5 occasions.	Method of randomization and concealment of allocation not stated. Double-blind.	3/19 Patients who completed the study correctly identified offending foods in placebo controlled trial. All had a history of atopy and a positive skin prick test to inhalant allergens, although only 1/3 had positive skin prick test to relevant foods.
Carroccio et al.16	Italian two center RCT	160 Patients with Rome II IBS, 40 patients with other GI diseases, and 50 healthy controls. Recruited from secondary care; 79% female. Fecal tryptase, ECP and calprotectin measured. Serum IgE (total and food specific) also measured.	All patients had a 4-week elimination diet excluding cow's milk, eggs, wheat, tomatoes, and chocolate. All completed a daily diary. Those who responded to the diet were given a double-blind placebo controlled trial of cow's milk and then wheat proteins using capsules containing these agents or placebo. Capsules given for 2 weeks with 1 week wash out and then crossed over and patient deemed food sensitive if symptoms returned when given the active capsule.	Method of randomization and concealment of allocation not stated. Double-blind.	70/160 (44%) IBS patients responded to open restriction diet. Forty of 70 (57%) were food sensitive according to the double-blind challenge (note with this design there is a 50% chance the patient will be “correct” if they guess with at least one of the food types so the importance of this finding is unclear. Interestingly, fecal ECP and tryptase (but not calprotectin) were higher in the food-sensitive IBS group compared with the nonsensitive IBS group.
Shepherd et al.17	Australian single center RCT	26 Patients with Rome II IBS and fructose malabsorption diagnosed with fructose hydrogen breath test. Symptoms responded to 3 months for a low FODMAPs diet. Recruited from secondary care dietetic practice.	Patients continued on a low FODMAPs diet. After run-in phase were blindly challenged with four drinks that were prepared to be high in fructose, fructans, fructans, and fructose mixed and glucose. The type of drink was administered randomly for 2 weeks. The dose of the agent was increased each 3 days from low to high and continued on the dose tolerated for 2 weeks. There was a washout period, and thus the patient was free of symptoms for 7 days before going on to the next phase in this crossover design.	Method of randomization and concealment of allocation not stated. Double-blind.	Overall IBS symptom score measured by visual analog scale was higher for fructose, fructans, and mixed groups compared with glucose group and this was reported to be statistically significant. Twenty to thirty percent felt their symptoms remained controlled on the fructose, fructans, and mixed groups compared with over 80% in the glucose group.
Vazquez Roque et al.18	US single center RCT	Forty-five patients with Rome II IBS-D recruited from tertiary care clinic; 96% female. All took gluten before entering trial and no evidence of celiac disease on serology or duodenal biopsy. HLA genotyping was performed and large and small bowel transit was measured with scintigraphy. Lactulose/mannitol test for small and large bowel permeability was also measured. Tight junctions were quantified using real-time PCR from small bowel and rectal biopsies.	Twenty-three patients randomized to gluten-free diet, 22 randomized to gluten-containing diet for 4 weeks. Stool form was measured using Bristol Stool chart. Overall IBS symptoms were not measured (confirmed by the authors).	Method of randomization and concealment of allocation adequate. Single-blind (patients not blinded).	Patients on a gluten-free diet had marginally statistically significant lower frequency of bowel movements—~0.25 less bowel movements per day (P=0.04) with no difference in stool form or ease of passage. There was no significant effect on transit time, but there was a statistically significant increase in small (but not large) bowel permeability HLA-DQ2/8-positive patients.
Halmos et al.19	Australian single center RCT	Thirty patients with Rome III IBS recruited by advertisement and eight healthy controls. Seventy percent IBS patients female.	Twenty-one days of a low FODMAPs diet or 21 days of typical Australian diet randomized in a crossover design with a 21-day washout period. Gastrointestinal symptoms were measured on a visual analog scale.	Method of randomization adequate and concealment of allocation not stated. Patients said to be blinded but over 80% IBS patients correctly identified the active diet.	Overall gastrointestinal symptoms less with the low FODMAPs diet (22.8 (95% CI=16.7–28.8) compared with the control diet (44.9 (95% CI=36.6–53.1), P<0.001. Effects were noted in IBS-C and IBS-D. The other IBS groups were too small to analyze.
Biesiekierski et al.20	Australian single center RCT	Thirty-seven patients with Rome III IBS recruited by advertisement. Patients with normal duodenal biopsy and HLA-DQ2 and HLA-DQ8 negative. All thought their symptoms were gluten related and had adhered to a gluten-free diet for at least 6 weeks.	Initial 7-day run-in of low FODMAPS diet and gluten-free diet. Patients then given three diets—high gluten (16 g per day), low gluten (2 g per day), or placebo in random order for 1 week with a 2-week washout period in a crossover design. Symptoms were measured by visual analog scale.	Method of randomization and concealment of allocation adequate. Double-blind.	Significant reduction in symptoms scores during the open 7-day run-in of low FODMAPs diet (P<0.001, actual numbers are not given). No change in overall symptoms on either gluten-containing diets compared with no gluten—all continued on low FODMAPs diet.
Berg et al.21	Norwegian three center RCT	202 with Rome II IBS recruited from secondary care. All had a fructose hydrogen breath test but patients continued in the trial whatever the result; 75% female.	Two-week run-in with “IBS” diet—not clear what this was. Patients were then randomized to either carry on with their diet or also add a fructose-reduced diet with guidance from a study nurse.	Method of randomization adequate and concealment of allocation not stated. Not blinded.	One hundred and eighty-two patients completed the study. Significant reduction in overall symptoms in the fructose-reduced diet compared with no change in the control group. The effect was independent of the results of the fructose hydrogen breath test (63% positive).
Jones et al.22	UK single center RCT	Twenty-five patients with clinical diagnosis IBS. Unclear how many were female. Plasma immune complexes, histamine, eosinophil count, and breath hydrogen excretion measured. Rectal PGE2 was measured.	All patients were asked to limit their diet to one meat, a single fruit, and distilled or spring water for 1 week. Those who responded were admitted to hospital for 4 days and asked to continue on the same foods, but were given breakfast via a nasogastric tube with or without offending foods in a randomized order for each day. No wash out.	Method of randomization not stated and concealment of allocation adequate. Double-blind.	Four refused to try the diet and 14/21 responded to diet. Six patients were admitted for the crossover trial for 4 days. Rectal PGE2 was significantly increased in patients with offending food challenges compared with controls. Symptoms not measured.

CI, confidence interval; ECP, eosinophil cationic protein; FODMAPs, Fermentable Oligo-Di-Monosaccharides and Polyols; GI, gastrointestinal; HLA, human leukocyte antigen; IBS, irritable bowel syndrome; IgG, immunoglobulin G; IQR, interquartile range; NSP, non-starch polysaccharides; PGE₂, prostaglandin E2; RCT, randomized controlled trial.