Table.

Studies included in systematic review

Reference	Type of study	Participants	Intervention	Outcome	Comments or conclusions
Werdmuller and Loffeld 199710	Descriptive, prospective	Consecutive patients undergoing endoscopy of upper gastrointestinal tract for upper abdominal symptoms or reflux symptoms. Cases (n=240, of which 118 patients with proved gastro-oesophageal reflux disease included). Rest with hiatus hernia and no reflux oesophagitis or with Barrett's oesophagus excluded. Reference group (n=399): normal endoscopy and presumed absence of typical reflux symptoms	Endoscopy of upper gastrointestinal tract, H pylori testing by histology (haematoxylin and eosin stain), culture, quick urease test, and serology (not all tests in every patient)	H pylori prevalence in patients with gastro-oesophageal reflux disease (29%) and reference group (51%)	We assumed from details that patients in reference group do not have reflux disease
Koike et al 199935	Case-control, prospective	Patients were self referred and referred by doctor. Cases (n=175): patients with reflux oesophagitis. Controls: age-sex matched, randomly selected, who visited hospital, were asymptomatic, and had normal endoscopy results	Endoscopy of upper gastrointestinal tract. H pylori testing by histology, rapid urease test, and serology. Atrophic gastritis assessed by updated Sydney system, and serum pepsinogen measured	H pylori prevalence in patients with gastro-oesophageal reflux disease (34%) and controls (72%)
Csendes et al 199736	Case-control, prospective, prevalence study	Cases (n=136): patients with chronic gastro-oesophageal reflux disease (reflux oesophagitis, negative reflux disease on endoscopy) symptoms of at least three years' duration. Controls (n=190): patients needing endoscopy, none of whom had symptoms of gastro-oesophageal reflux disease	Endoscopy of upper gastrointestinal tract in cases and controls, H pylori testing by histology, pH-metry in all cases of gastro-oesophageal reflux disease, no pH-metry in controls	H pylori prevalence in patients with reflux oesophagitis, reflux disease negative on endoscopy, Barrett's oesophagus, and controls. No significant difference in H pylori prevalence between patients with patients with reflux oesophagitis (32%), reflux disease negative on endoscopy (25%), and controls (29%). Also no difference in age and sex distribution between reflux patients and controls	Exclusion of peptic ulcer not clearly stated
El-Serag et al 199937	Descriptive, prospective	Patients referred for elective endoscopy of upper gastrointestinal tract. Cases (n=154, of which 116 patients were included, 38 excluded because of Barrett's oesophagus): all patients with erosive oesophagitis. Controls (n=148): Patients with normal endoscopy result and absence of symptoms of gastro-oesophageal reflux disease	Endoscopy of upper gastrointestinal tract in cases and controls, H pylori testing by haematoxylin and eosin stain	H pylori prevalence in patients with gastro-oesophageal reflux disease (31%) and controls (43%)	This study looked at protective effect of corpus gastritis against reflux oesophagitis. We excluded Barrett's oesophagus from our analysis
Fallone et al 200038	Descriptive, prospective.	Patients scheduled for endoscopy of upper gastrointestinal tract. Cases (n=327, of which 81 patients with gastro-oesophageal reflux disease included). Rest were classified into non-ulcer disease, duodenal ulcer, gastric ulcer, and therefore excluded. Patients with gastro-oesophageal reflux disease had reflux oesophagitis or negative reflux disease on endoscopy. Comparator group (n=78): patients in whom there were no symptoms of gastro-oeosophageal reflux disease and in whom indications for endoscopy were multiple. All had normal oesophagus or findings unrelated to gastro-oesophageal reflux disease	Endoscopy of upper gastrointestinal tract; H pylori testing by histology and culture; detection of specific genes or gene sequence within H pylori and detection of CagA antibodies	H pylori prevalence in patients with gastro-oesophageal reflux disease (33%) and comparator group (48%). Prevalence of CagA, CagE, vacA S1 genotypes, and CagA antibody determined in cases and comparator group	Some patients with reflux disease negative on endoscopy but reflux not proved may have been included in our prevalence data. This study concluded that gastro-oesophageal reflux disease was associated with a significantly lower rate of vacA S1 genotype than in controls
Gisbert et al 200139	Descriptive, prospective, prevalence	Consecutive patients undergoing 24 hour oesophageal pH monitoring in motility unit because of symptoms suggestive of gastro-oesophageal reflux disease. Cases (n=56): typical symptoms of gastro-oesophageal reflux disease and positive pH findings. Controls (n=44): symptoms of gastro-oesophageal reflux disease but negative pH findings	Endoscopy of upper gastrointestinal tract, 24 hour oesophageal pH monitoring and H pylori testing by histology and rapid urease test	H pylori prevalence in patients with gastro-oesophageal reflux disease (57%) and controls (52%)	Comparator group may represent patients with non-ulcer dyspepsia
Goldblum et al 199840	Case-control, prospective	Cases (n=58): patients with classic symptoms of gastro-oesophageal reflux disease enrolled into study. Controls (n=27): patients undergoing endoscopy for reasons other than symptoms of gastro-oesophageal reflux disease, Barrett's oesophagus, peptic ulcer disease, or dyspepsia	Endoscopy of upper gastrointestinal tract in cases and controls; H pylori testing by histology (haematoxylin and eosin and Giemsa stain) and serology	H pylori prevalence in patients with gastro-oesophageal reflux disease (41%) and controls (48%). Prevalence of carditis and intestinal metaplasia of cardia in cases and controls also determined	This study also concluded that cardia inflammation and cardia intestinal metaplasia are associated with H pylori infection
Hacklesberger et al 199841	Case-control, prospective	Cases (130 of 171 included, remaining 41 had associated peptic ulcer disease): consecutive Caucasian patients undergoing elective endoscopy. Controls (n=227): asymptomatic volunteers or patients attending for other reasons and without any symptoms of gastro-oesophageal reflux disease	Endoscopy of upper gastrointestinal tract in cases only. H pylori testing by histology and rapid urease test in cases, 13-carbon urease breath test	H pylori prevalence in patients with gastro-oesophageal reflux disease (38%) and controls (39%)	Different methods of H pylori testing in cases and controls. No endoscopy in controls
Haruma et al 200042	Retrospective case-control	Of 6205 patients undergoing endoscopy of upper gastrointestinal tract between defined periods, 229 were defined as having reflux oesophagitis. Of these, 95 met authors' inclusion criteria. Controls (n=190): healthy, asymptomatic, age-sex matched selected from among 608 healthy individuals who had undergone routine healthcare check for gastric cancer	Endoscopy of upper gastrointestinal tract in cases and controls; H pylori testing by Giemsa stain and serology. Inflammation, atrophy, and intestinal metaplasia were evaluated using updated Sydney system. Serum gastrin and pepsinogen concentrations determined	H pylori prevalence in patients with gastro-oesophageal reflux disease (41%) and controls (76%)	The authors found significant low prevalence of H pylori in patients over 60 but not under 59 with reflux oesophagitis, when compared with age-sex matched controls
Liston et al 199643	Descriptive, prospective, prevalence	Consecutive patients admitted for gastroscopy recruited regardless of reasons for procedure. Main reasons were anaemia, reflux symptoms, and epigastric pains. Cases (n=37): reflux oesophagitis (macroscopic or microscopic). Comparator group (n=33): normal endoscopy result and no evidence of histological oesophagitis	Endoscopy of upper gastrointestinal tract; H pylori testing by histology, rapid urease test, serology, and 13-carbon urease breath test	H pylori prevalence in patients with reflux oesophagitis (76%) and comparator group (82%). Patterns of gastritis described in the two groups	Although exclusion of patients with peptic ulcer disease had not been clearly stated, on reading the paper, we assumed this to be the case