Appendix 1

Appendix 1: Details of searches for trials of Helicobacter pylori eradication in non-ulcer dyspepsia

The Cochrane Controlled Trials Register, Medline, EMBASE, CINAHL, AMED, SIGLE electronic databases were explored using broad search strategies to identify all randomised controlled trials evaluating H pylori eradication in non-ulcer dyspepsia. All searches were run from the earliest date available (1966 for Medline, 1988 for EMBASE, and 1982 for CINAHL) until March 1999. A final search of Medline and Embase was undertaken in May 2000. All languages and indexed journals were included and retrieved.

The Cochrane Collaboration has hand searched all the major general medical and gastroenterology journals with the exception of Alimentary Pharmacology and Therapeutics. We therefore hand searched this journal from 1994. In addition, we routinely reviewed the content of major gastroenterological and general medical journals for the year up until the end of May 2000.

The bibliographies of retrieved papers were also reviewed for relevant trials not identified by the database and hand searching. Experts in the field of dyspepsia known to the review team via the Cochrane Upper Digestive and Pancreatic Group were contacted. They were requested to supply any information about trials that they either knew about or were conducting. The editors of general medical and gastroenterology journals were contacted for information on any papers on H pylori eradication therapy and non-ulcer dyspepsia undergoing peer review. Pharmaceutical companies (Astra-Zeneca, Wyeth Laboratories, Knoll, Yamanouchi Pharma, Abbott Laboratories, Pfizer, and Rhone-Poulenc Rorer) were also contacted for any data on trials that had been published or were unpublished and in their archives.

Terms for H pylori eradication in non-ulcer dyspepsia in Cochrane controlled trial register, Medline, EMBASE, and CINAHL

MeSH search terms

Dyspepsia and H pylori related

Dyspepsia

Eructation

Flatulence

Heartburn

Gastroparesis

Gastric emptying

Duodenogastric gastric reflux

Hiatal hernia

Epigastric pain

Stomach pain

Regurgitation

Indigestion

Helicobacter pylori

Campylobacter Pyloridis

H pylori therapy related

Omeprazole

Lansoprazole

Pantoprazole

Amoxycillin

Azithromycin

Bismuth

Bismuth citrate

Bismuth compound

Bismuth salicylate

Clarithromycin

Doxycycline

Erythromycin

Metronidazole

Oxytetracycline

Ranitidine Bismuth Citrate

Tetracycline

Tinidazole

Textword search terms

Dyspepsia and H pylori related

Dyspep$

Acid adj5 reflux

Belch$

Bloat$

Burp$

Heartburn

Indigestion

Nausea

Vomiting

Pyrosis

Hiatus hernia

Flatu$

Stomach paresis

Abdominal adj5 distension

Stomach adj5 distension

Postprandial adj5 fullness

Early satiety

Nausea

Vomiting

Gastritis

Gastric acid adj5 secretion

Gastric adj5 erosion$

Stomach adj5 erosion$

Gastric emptying adj5 disorder$

Gastroparesis

Helicobacter pylori

campylobacter pylori

campylobacter pyloridis

helicobacter pylori adj5 eradication

campylobacter pylori adj5 eradication

H pylori therapy related

Omeprazole

Lansoprazole

Pantoprazole

Amox?cillin

Azithromycin

Bismuth

Bismuth citrate

Bismuth compound

Bismuth salicylate

Clarithromycin

Denol

De-Noltab

Doxycycline

Erythromycin

Metronidazole

Oxytetracycline

Ranitidine Bismuth Citrate

Pylorid

Tetracycline

Tinidazole

Tripotassium bismuthate

Tripotassium citrate

Experts contacted for systematic review

Dr N Ahluwalia, Stockport, UK

Dr A Andren-Sandberg, Lund, Sweden

Dr M Asante, Surrey, UK

Professor ATR Axon, Leeds, UK

Dr C Bardhan, Rotherham, UK

Dr C Bassi, Rome, Italy

Ms H Bastian, Blackwood, Australia

Professor A Blum, Lausanne, Switzerland

Dr S Boesby, Copenhagen, Denmark

Dr N Broutet, Bordeaux, France

Ms J Bruce, Aberdeen, UK

Dr P Bytzer, Glostrup, Denmark

Dr F Carballo, Guadalajara, Spain

Dr N Chiba, Guelph, Canada

Dr AR Dar, London, Canada

Dr E de Koster, Brussels, Belgium

Dr M Delvaux, Toulouse, France

Sister J DeSilva, Rotherham, UK

Dr J Dixon, Middlesex, UK

Dr JE Dominguez-Munoz, Santaigo de Compostela, Spain

Dr C Gluud, Copenhagen, Demark

Professor C Hawkey, Nottingham , UK

Dr E Hentschel, Vienna, Austria

Professor R Hunt, Hamilton, Canada

Ms E Jonsson, Molndal, Sweden

Mr JD Kirby, Solihull, UK

Dr K Krogsgaard, Copenhagen, Denmark

Dr E Kuipers, Amsterdam, Netherlands

Dr R Laheij, Nijmegen, Netherlands

Professor L Laine, Los Angeles, USA

Dr J Lambert, Frankston, Australia

Professor M Langman, Birmingham, UK

Dr M Larvin, Leeds, UK

Professor J Lennard-Jones, London, UK

Dr R Logan, Nottingham, UK

Professor J Malagelada, Barcelona, Spain

Dr R Malthaner, London, Canada

Mr I Martin MD, FRCS, Leeds , UK

Dr W Matthews, London, Canada

Dr P Matzen, Hvidovre, Denmark

Professor J McDonald, London, Canada

Professor F Megraud, Bordeaux, France

Mr S Meisner, Copenhagen , Denmark

Dr JEM Midgeley, Ilkley, UK

Dr G Misiewicz, London, UK

Dr H Moller, Copenhagen, Denmark

Dr M Numans, Utrecht, Netherlands

Dr O Nyren, Uppsala, Sweden

Dr A Oxman, Oslo, Norway

Dr J Penston, Dundee, Scotland

Dr H Persson, Stockholm, Sweden

Professor T Poynard, Paris, France

Professor A Price, Edinburgh, UK

Dr E Rauws, Amsterdam, Netherlands

Dr A Rostom, Ottawa, Canada.

Dr S Rune, Copenhagen, Denmark

Ms D Saddler, Chicago, USA

Dr E Saperas, Barcelona, Spain

Professor O Schaffalitzy de Muckadell, Odense, Denmark

Dr K Shenoy, Trivandrum, India

Dr L Stewart, London, UK

Professor N Talley, Penrith, Australia

Dr A Thomson, Edmonton, Canada

Dr J Tierney, London, UK

Dr P Unge, Gävle, Sweden

Dr S Veldhuyzen van Zanten, Halifax, Canada

Dr N Waugh, Aberdeen, UK

Dr P Webb, Herston, Australia

Professor S Wessely, London, UK

Dr P Wille Jorgensen, Copenhagen , Denmark

Dr C Williams, Oxford, UK

List of journals contacted for unpublished papers

Alimentary Pharmacology and Therapeutics

American Journal of Gastroenterology

American Journal of Medicine

Annals of Internal Medicine

British Journal of Clinical Practice

British Journal of General Practice

British Medical Journal

Canadian Journal of Gastroenterology

Current Therapeutic Research

Digestion

Digestive Diseases and Sciences

Digestive Endoscopy

European Journal of Gastroenterology and Hepatology

European Journal of General Practice

Family Practice

Gastroenterology

Gastroenterology International

Gut

Helicobacter

International Journal of Gastroenterology

Italian Journal of Gastroenterology

JAMA

Journal of Clinical Gastroenterology

Journal of Gastroenterology

Journal of Gastroenterology and Hepatology

Lancet

New England Journal of Medicine

Scandinavian Journal of Gastroenterology

Appendix 2: Structure of Markov model used to evaluate the effectiveness of H pylori eradication in non-ulcer dyspepsia patients

Strategies compared in the model

The model compared H pylori eradication with one month of antacid therapy in non-ulcer dyspepsia patients. Antacid therapy was assumed to act as an inexpensive placebo, and the impact of these interventions was assessed over one year. Patients with continuing symptoms despite these interventions were given lifestyle advice and reassurance on consultation with their general practitioner. The model did not compare H pylori eradication therapy with a "do nothing" strategy as clinicians usually feel obliged to give some form of therapy to non-ulcer dyspepsia patients.

Costs and benefits identified in the model

The model evaluated the impact of H pylori eradication from a health service perspective over a one-year time frame. The cost of medication and visits to the general practitioner were assessed (table 2). The robustness of the results was explored in one-way sensitivity analyses. The main uncertainty in the model was the number of times a patient with continuing dyspepsia would visit their GP. Data from recent primary care trials show that patients visit their GP three times in one year for dyspepsia, but this assumption was evaluated over a wide range of values in a sensitivity analysis. The costs and benefits were not discounted as the assessment was being made over one year.

The benefit of therapy was measured in terms of the number of months of minimal or no dyspeptic symptoms over one year. The mean placebo response rate seen in the H pylori eradication therapy was incorporated into the Markov model (fig A) and patients receiving antacids were assumed to follow this pattern. The probability that a patient remained free from symptoms over the next 12 months was calculated from the mean placebo response rate seen at one year in the trials. The risk of dyspepsia relapse was then calculated for each one month cycle of the Markov model, converting from rates to probabilities using the formula (1- e^-CER/12), where CER=placebo response rate. The corresponding figure for patients receiving H pylori eradication was derived from the relative risk obtained in the systematic review using the formula (1- e^{-(CER/12) x rr}) where rr=relative risk of having dyspepsia.

(F1) Fig A

Structure of the Markov model

Appendix 3: Details of trials of H pylori eradication in non-ulcer dyspepsia identified in searches

Table 1 Trials excluded from systematic review

References	Participants	Intervention	Outcomes and results	Reasons for exclusion
Berstad 1988	Norway. RCT. Double-blind placebo-controlled trial. 100 with NUD and erosive prepyloric changes. Cp was found in 25% of the patients.	4 weeks. Antacids (aluminium-magnesium) 1 tablet qid vs placebo I, Pirenzepine 50mg bid vs placebo II.	Neither NUD nor erosive prepyloric changes are strongly associated with antral Cp changes. Aluminium-magnesium antacids may suppress antral Cp infection without healing the gastritis or relieving symptoms.	No predefined Hp eradication regimen. No dyspeptic scores or quality of life.
Holcome 1992	UK. RCT. 130 NUD and Hp infection.	4 weeks. Gelusil 1 tablet qid vs De-Noltab 240mg qid for 28 days together with amoxycillin 500mg qid for the first 14 days.	Bismuth and amoxycillin were significantly better at achieving symptom resolution than antacid but symptomatic improvement did not relate to clearance of Hp.	No predefined Hp eradication regimen.
Goh 1991	Malaysia. RCT, Double-blind placebo-controlled trial. 71 NUD +/ Hp infection. 40 patients with Hp +ve: 21 on treatment and 19 on placebo. 31 Hp ve: 17 on treatment and 14 on placebo. Non-erosive duodenitis included. 84.5% completed trial.	4 weeks. CBS. 2 tablets bid vs placebo	All groups reported improvement in symptom scores. In the Hp +ve group, CBS treated group recorded a significantly higher improvement than the other groups. 12 of 16 patients relapsed 1 month after withdrawal of CBS.	No predefined Hp eradication regimen.
Kang 1990	Singapore. RCT, double-blind placebo-controlled trial. 73 total – 21 on treatment, 19 on placebo. with food related abdominal pain	8 weeks. CBS 1 tablet qid versus placebo.	CBS benefited those with gastritis but not those without.	No predefined Hp eradication regimen.
Kazi 1990	India. RCT. 52 total. 26 in each arm. Dyspepsia and Hp gastritis.	3 weeks. Bismuth salicylate 500mg tid vs placebo.	Resolution of gastritis and improvement of symptoms were significantly better in patients where Hp was cleared as compared to those where Hp persisted.	No predefined Hp eradication regimen.
Kumar 1996	India. RCT, Double-blind placebo-controlled trial. 81 total – 18 on CBS, 15 on placebo I, 15 on sucralfate, 15 on placebo II. NUD and Hp	CBS 240mg bid vs placebo I to sucralfate and sucralfate 2 gm bid vs placebo II to sucralfate.	CBS is more effective than sucralfate in inducing endoscopic and histology healing of Hp related gastritis among NUD patients.	No predefined Hp eradication regimen.
Lambert 1989	Australia. RCT, double-blind placebo-controlled trial. 82 with NUD +/ Cp infection: 48 Cp +ve: 22 on treatment, 26 on placebo. 30 Cp ve: 16 on treatment and 14 on placebo.	4 weeks. Bismuth subcitrate 4 tablets daily vs placebo	Clearance of Cp and histological improvement was associated with significant decrease in symptoms. In Cp ve patients, improvement occurred in both placebo and treatment groups.	No predefined Hp eradication regimen.
Lanza 1989	USA. RCT. Blinding not mentioned. 20 with a variety of upper gastrointestinal complaints and Cp gastritis.	3 weeks. BSS 525mg qid vs placebo.	BSS provides short-term clearance of Cp in antral mucosa and this clearance is associated with an improvement in histological diagnosis.	No predefined Hp eradication regimen. No dyspeptic scores or quality of life.
Loffeld 1989	Netherlands. RCT, double-blind placebo-controlled trial. 57 total - 26 on treatment, 24 on placebo. NUD with Cp gastritis.	4 weeks. CBS 240mg daily vs placebo.	Subjective complaints improved in both groups except for nausea and meteorism that improved more in the CBS group. Significant reduction in Cp colonisation and gastritis score.	No predefined Hp eradication regimen.
Marshall 1993	USA. RCT. Double-blind placebo-controlled trial. 2 week placebo run-in period. 50 with severe dyspepsia and Hp gastritis.	3 weeks. BSS 512mg qid vs placebo.	No significant change in level of dyspeptic symptoms. BSS suppresses but does not eradicate Hp. Results reported as number of days with symptoms in a week.	No predefined Hp eradication regimen.
Rokkas 1988	UK. RCT. Double-blind placebo-controlled trial. 66 total. 25 on Denol. 27 on placebo. NUD including some patients with Cp infection.	8 weeks. Denol two tablets bid vs placebo.	In Denol group, gastritis improved and symptomatic response was better than the placebo group. Cp clearance in Denol group was 83.3%, none in the placebo group.	No predefined Hp eradication regimen.
Vaira 1992	Italy. RCT, double-blind placebo-controlled trial. 80 total. 40 in each arm. Hp associated NUD. Follow-up at 4 weeks post treatment was 97.5%. Eradication rate 54%.	4 weeks. CBS 240mg bid vs placebo.	CBS is effective treatment for Hp associated NUD with improved gastric antral histological appearances and has a beneficial effect on symptoms.	No predefined Hp eradication regimen.
Frazzoni 1993	Italy. 2 RCT. Double-blind placebo-controlled trial. 53 with dyspepsia. Treatment regimen depends on Hp status.	For Hp +ve group: CBS 240mg bid for 28 days plus metronidazole 500mg tid for 10 days vs placebo. For Hp ve group: cisapride 10mg tid for 28 days vs placebo.	Symptomatic remission rates following a 1 month wash-out period in both treatment groups were no higher than that in controls. Bulbar and antral biopsies are not useful in clinical management.	No predefined Hp eradication regimen.
Humphreys 1988	Ireland. RCT. 135 patients with peptic ulcer disease and Cp gastritis.	6 weeks. Cimetidine 400mg bid vs CBS 5mls qid.	Cp was strongly associated with the presence of histological gastritis, which was decreased by CBS.	No prior endoscopy excluding PUD and oesophagitis. No predefined Hp eradication regimen. No dyspeptic scores or quality of life.
McNulty 1986	UK. RCT, single-blind placebo controlled trial. 50 with Cp associated gastritis with various aetiologies.	BSS 30ml qid for 3 weeks vs placebo I vs erythromycin ethylsuccinate 10mls qid vs placebo II.	Significantly greater improvement in endoscopic appearances in patients cleared of Cp compared with those with persistent infection. Symptoms improved with Cp clearance though not statistically significant.	No prior endoscopy excluding PUD and oesophagitis. No predefined Hp eradication regimen.
McNulty 1990	UK. RCT. Single-blind placebo controlled trial. 50 upper gastrointestinal symptoms with Cp gastritis.	BSS 30ml qid for 3 weeks vs placebo I vs erythromycin ethylsuccinate 10mls qid vs placebo II.	Significantly greater improvement in endoscopic appearances in patients cleared of Cp compared with those with persistent infection. Heartburn improved in 50% with BSS and 17% with placebo.	No prior endoscopy excluding PUD and oesophagitis. No predefined Hp eradication regimen.
Nafeeza 1992	Malaysia. RCT. Double-blind controlled trial. 48 with NUD and Hp gastritis.	CBS 480mg bid for 28 days and ampicillin 500mg qid for the first 10 days vs CBS for 28 days and placebo matched to ampicillin vs ampicillin 500mg qid for first 10 days and placebo matched to CBS for 28 days.	There was suppression of Hp on combined therapy but none in the single therapy. Suppression of Hp was associated with both histological and symptomatic improvement.	No predefined Hp eradication regimen.
Patchett 1991	Ireland. Not RCT. 90 with NUD and Hp gastritis.	CBS 120mg qid for 4 weeks vs metronidazole 400mg tid and amoxycillin 500mg tid for 1 week vs CBS 120mg qid for 4 weeks and metronidazole 400mg tid for 1 week.	Gastritis score improved with Hp eradication. However, mean symptom scores improved whether or not gastritis improved. Antral infection with Hp does not have an important aetiological role in NUD	No predefined Hp eradication regimen. Not RCT.
Labenz 1993	Germany. Open clinical trial. 180 with Hp associated NUD or PUD. Only 17 patients had NUD	5 groups where 4 groups with various dosages of combination of omeprazole and amoxycillin in treatment periods vs omeprazole alone.	Omeprazole-enhanced amoxycillin therapy is a simple and effective approach to the eradication of Hp colonisation.	No dyspeptic scores or quality of life. Not RCT.
Gad 1989	Sweden. Uncontrolled pilot trial. 186 with NUD and Cp.	10 days. Erythromycin 500mg bid vs Cavedess (bismuth subnitrate and antacids combination).	The combination of antibiotic and bismuth/antacid compound resulted in improvement of the histological picture, disappearance of Cp and amelioration of symptoms.	No predefined Hp eradication regimen. Not RCT.
Gad 1989	Italy. RCT, double-blind placebo-controlled multicentre trial. 128 with NUD and Cp gastritis. 66 on Pirenzepine, 62 on placebo. 84% completed trial.	4 weeks. Pirenzepine 50mg bid vs placebo.	There was improvement of the endoscopic and clinical findings but no change of the degree of the mucosal inflammation or the extent of colonisation by Cp.	No predefined Hp eradication regimen.
Unge 1989	Sweden. Randomised double-blind pilot study. 24 Cp positive patients.	2 weeks. a) Omeprazole 40mg/day plus amoxycillin 750mg bid vs b) omeprazole 40mg/day vs c) amoxycillin 750mg bid.	Omeprazole as a powerful inhibitor of gastric acid secretion has been identified as a promising therapeutic means of combating Cp infection.	No prior endoscopy excluding PUD and oesophagitis. No dyspeptic scores or quality of life. Not RCT.
Parente 1998	Italy. RCT. 38 functional dyspepsia and Hp gastritis.	Omeprazole 40mg /day, clarithromycin 250mg bid and tinidazole 500mg bid for 1 week vs ranitidine 300mg/day for 3 weeks.	Eradication of Hp in the long run significantly reduces gastrin and pepsinogen I release as a result of improvement in the underlying antral gastritis, but this is not accompanied by modifications of gastric emptying of solids or acid secretion.	No dyspeptic scores or quality of life.
Tham 1996	Ireland. RCT. 80 patients with dyspepsia of various aetiology including peptic ulcer disease, oesophagitis and those with normal endoscopy and Hp infection. Subgroup analysis not possible.	2 weeks of (1) omeprazole 20mg daily, amoxycillin 500mg tid and metronidazole 400mg tid; (2) ranitidine 600mg bid, amoxycillin 500mg tid and metronidazole 400mg tid; (3) omeprazole 20mg daily and placebo; or (4) omeprazole 20mg daily and clarithromycin 500mg tid.	Hp eradication rates with high dose Ranitidine plus amoxycillin and metronidazole may be similar to that of low dose omeprazole in combination with the same antibiotics or omeprazole with clarithromycin. Overall eradication rates were low due to high incidence of metronidazole resistance.	No prior endoscopy excluding PUD and oesophagitis. No dyspeptic scores or quality of life.
Hazell 1997	Australia. RCT. Double-blind double dummy study. 101 with peptic ulcer disease and NUD and Hp infection.	Lansoprazole 30mg/day for 4 weeks plus placebo tid vs lansoprazole 30mg/day for 4 weeks plus amoxycillin 500mg tid.	Inflammation improved in patients treated with lansoprazole plus amoxycillin. Both duodenal ulcer and NUD patients showed improvement in the symptoms irrespective of the treatment arm. (Data not shown in article). Author contacted with no reply.	No dyspeptic scores or quality of life.
Glupczynski 1988	Belgium. RCT. Double-blind placebo-controlled trial. 45 Cp gastritis.	8 days. Amoxycillin 1gm bid vs placebo.	No significant improvement was observed in gastritis and clinical symptoms. Reappearance of Cp and significant worsening of gastritis was seen after 2 weeks in all patients.	No predefined Hp eradication regimen.
Xiao 1990	Republic of China. RCT. 72 with dyspepsia and Hp gastritis.	3 weeks. Furazolidone 500mg tid vs metronidazole 200mg tid vs placebo.	Furazolidone is effective in the clearance of Hp and providing marked improvement in both the inflammatory infiltration in the gastric mucosa and symptoms.	No predefined Hp eradication regimen.
Catalano 1999	Italian RCT. 126 elderly NUD patients with Hp	2 weeks. Bismuth, amoxycillin, metronidazole vs. omeprazole, amoxycillin	Both regimens eradicated Hp in over 60% of patients with no significant difference between treatments in reducing dyspepsia symptoms.	No predefined Hp eradication regimen.
Morgan 1988	Peru. RCT. Double-blind placebo-controlled trial. 106 with upper gastrointestinal symptoms and Cp gastritis.	2 weeks. Furazolidone 400mg qid vs nitrofurantoin 400mg qid vs placebo.	No significant relief in symptoms found between the three treatments.	No predefined Hp eradication regimen.
van Zanten 1995	Canada. RCT. 53 Hp positive NUD patients	Bismuth subsalicylate 302 mg qds, amoxycillin 500 mg tds, metronidazole 500 mg tds vs. placebo	No statistically significant difference between mean dyspepsia symptom score in eradication and control groups	Abstract; no additional data available.
David 1996	USA. RCT. 41 patients with NUD and Hp infection confirmed by rapid urease test	2 weeks of ranitidine 300 mg bd, amoxycillin 500 mg tds and metronidazole 250 mg tds vs. ranitidine 300 mg bd and placebo antibiotics.	Trend for Hp eradication to resolution symptoms at six weeks compared with placebo (82% vs. 62%) but no statistically significant difference.	Abstract.
Mitty 1997	USA. RCT. 15 Hp positive patients with dyspepsia and normal endoscopy.	2 weeks omeprazole 40 mg od, clarithromycin 500 mg tds vs. placebo	Statistically significant reduction in mean abdominal pain score in eradication group compared with placebo group at 6 months.	Abstract; no additional data available.
Passos 1998	Brazil. RCT. 81 Hp positive patients with functional dyspepsia	5 days of amoxycillin 500 mg tds, metronidazole 250 mg tds, furazolidone 200 mg tds vs. placebo	Mean dyspepsia symptom score at 36 months similar between the two groups with no statistically significant differences.	Abstract; no additional data available.
Hsu 1999	Taiwan. RCT. 71 patients with NUD and Hp infection	"Lansoprazole quadruple therapy" or placebo	Patients with successful eradication had a significant improvement in symptoms at the end of 12 months compared with control group	Abstract; data not analysed according to randomisation.
Florent 2000	France. RCT. 121 with epigastric pain and Hp gastritis but no ulcer disease.	10 days of lansoprazole 30 mg od, amoxycillin 1g bd, clarithromycin 500 mg bd vs. placebo	31% of the eradication group were asymptomatic at 12 months compared with 22% of the placebo group. This difference was not statistically significant.	Abstract; no additional data available.

RCT=randomised controlled trial. NUD=non-ulcer dyspepsia. Cp=Campylobacter pylori. Hp=Helicobacter pylori. CBS=colloidal bismuth subcitrate. BSS=bismuth subsalicylate. PUD=peptic ulcer disease.

Table 2 Trials included in the systematic review

Reference	Methods	Participants	Intervention	Outcomes
Sheu 1996²⁷	Blinding: patients not masked; outcome assessments masked, but no method stated. Randomisation: no description of method of randomisation or concealment. Drop-outs: full follow-up until 2 months when those still Hp +ve in the treatment arm (25%) received additional eradication therapy and not followed further.	Single-centre trial: Taiwan. Sample: 41 patients. Patient selection: consecutive referrals for dyspepsia with normal endoscopy and +ve for Hp on UBT, serology and histology, dyspepsia score >2/10. Reflux-like disease: patients with symptoms of heartburn, acid regurgitation and vomiting were excluded.	Duration: mixed. Eradication arm: 4 weeks of colloidal bismuth subcitrate 120mg tid and 2 weeks of metronidazole 500mg tid plus amoxycillin 500mg tid. Control arm: 2 months of H₂RA. Concomitant treatments: Short courses of antisecretory agents permitted. Hp eradication rates: 75% for the eradication arm, not stated for the control arm.	(a) Mean symptom score (own scale based on assessment of 5 symptom groups). (b) Mean serological titres of Hp. (c) Grade of gastritis. Length of follow-up: 2 months in randomised groups, 6 months in total.
Blum 1998¹⁸	Blinding: patients and investigators masked by use of matching placebos. Randomisation: random allocation used, but no method of concealment stated. Drop-outs: 6% of participants excluded from intention to treat analysis	Multi-centre trial: Austria, Canada, Germany, Iceland, Ireland, South Aftrica, Sweden. Sample: 348 patients. Patient selection: >6 month history of dyspepsia with normal endoscopic findings, and +ve for Hp on RUT. Moderate to very severe dyspepsia for >3 days in previous week. Reflux-like disease: excluded by definition of dypepsia and previous history	Duration: 1 week. Eradication arm: omeprazole 20mg bid, amoxycillin 1g bid and clarithromycin 500mg bid. Control arm: omeprazole 20mg bid and matching placebos. Concomitant treatments: Not stated. Hp eradication rates: 79% for the eradication arm and 2% for the control arm	(a) Treatment success defined as no symptoms or only minimal pain in previous 7 days. (b) Symptoms rated according to GSRS. (c) Quality of life measured with PGWB. (d) Healing of gastritis. Length of follow-up: 12 months
McColl 1998¹⁷	Blinding: patients and investigators masked by use of matching placebos. Randomisation: concealed random allocation used . Drop-outs: 4% of participants excluded from intention to treat analysis.	Single-centre trial: Scotland. Sample: 318 patients. Patient selection: >4 month history of dyspepsia with normal endoscopy and +ve for Hp on UBT. All dyspeptics referred from primary care were eligible. Reflux-like disease: included	Duration: 2 weeks. Eradication arm: omeprazole 20mg bid, amoxycillin 500mg tid (or tetracyline 500mg tid) and metronidazole 400mg tid. Control arm: omeprazole 20mg bid and matching placebos. Concomitant treatments: antisecretory agents permitted. Hp eradication rates: 88% for the eradication arm and 5% for the control arm	(a) Treatment success defined as a GDSS rating of zero in previous 6 months. (b) Mean GDSS scores. (c) Quality of life measured using the SF-36. (d) Use of H₂RA and PPI medication. Length of follow-up: 12 months
Talley 1999²⁰.	Blinding: patients and investigators masked by use of matching placebos. Randomisation: random allocation used, but no method of concealment stated. Drop-outs: 25% excluded as found to be Hp-ve by UBT at randomisation. Additional 1% excluded from intention to treat analysis.	Multi-centre trial: Australia, Denmark, Finland, France, Hungary, Netherland, New Zealand, Norway, Spain, UK. Sample: 370 patients. Patient selection: >3 month history of dyspepsia with normal endoscopy and +ve for Hp on serology. Moderate to very severe dyspepsia for >3 days in previous week. Reflux-like disease: excluded by definition of dypepsia and previous history	Duration: 1 week. Eradication arm: omeprazole 20mg bid, amoxycillin 1g bid plus clarithromycin 500mg bid. Control arm: omeprazole 20 mg bid and matching placebos. Concomitant treatments: use of weak antacid permitted. Hp eradication rates: 85% for the eradication arm and 4% for the control arm	(a) Treatment success defined as no symptoms or only minimal pain in previous 7 days. (b) Symptoms rated according to GSRS. (c) Quality of life measured with PGWB. (d) Healing of gastritis. Length of follow-up: 12 months
Talley 1999²¹	Blinding: patients and investigators masked by use of matching placebos. Randomisation: random allocation used, but no method of concealment stated. Drop-outs: 13% of participants excluded from intention to treat analysis.	Multi-centre trial: USA. Sample: 337 patients. Patient selection: >3 month history of dyspepsia with normal endoscopy and +ve for Hp on UBT. Moderate to very severe dyspepsia for >3 days in previous week. Reflux-like disease: Patients with predominant heartburn were excluded	Duration: 2 weeks. Eradication arm: omeprazole 20 mg bid, amoxycillin 1g bid, clarithromycin 500 mg bid. Control arm: omeprazole 20 mg bid and matching placebos. Concomitant treatments: Use of weak antacid permitted. Hp eradication rates: 90% for the eradication arm and 2% for the control arm	(a) Treatment success defined as no symptoms or only minimal pain in previous 7 days, and no use of medication other than antacids in the previous 30 days. (b) Symptoms rated according to GSRS. (c) Quality of life measured with SF-36. (d) Healing of gastritis. Length of follow-up: 12 months
Dhali 1999²⁸	Blinding: patients and investigators not masked. Randomisation: random allocation used but no method of concealment stated. Drop-outs: unclear	Centre: India. Sample: 62. Patient selection: >4 week history of dyspepsia with normal endoscopy and +ve for Hp on biopsy or rapid urease test. Reflux-like disease: Patients with predominant heartburn excluded	Duration: Mixed. Eradication arm: bismuth subcitrate 120 mg qid, tetracycline 500 mg qid, metronidazole 400 mg tid for 2 weeks. Control arm: sucralfate 1 g qds for 4 weeks. Concomitant treatments: antacids. Hp eradication rates: 88% for the eradication arm and 0% for the control arm	Patients subjective improvement in dyspepsia symptoms. Changes in dyspepsia questionnaire score. Length of follow-up: 3 months
Greenburg 1999²²	Blinding: patients and investigators were masked using identical placebos. Randomisation: method of randomisation and concealment not stated. Drop-outs: 16% of patients excluded from ITT analysis	Single centre study: USA. Sample: 84. Patient selection: >4 week history of epigastric pain not responsive to H₂RA with normal endoscopy and +ve for Hp on biopsy. Reflux-like disease: Patients with predominant heartburn excluded	Duration: 2 weeks. Eradication arm: omeprazole 20 mg bd and clarithromycin 500 mg tds. Control arm: identical placebos. Concomitant treatments: Patients allowed any dyspepsia medication. Hp eradication rates: 71% for the eradication arm, 8% of the placebo arm	Change in dyspepsia score. Questionnaire evaluated 5 symptoms using a visual analogue scale (possible scores 0-500). Unclear whether this questionnaire validated. Length of follow-up: 12 months
Miwa 2000²³	Blinding: study described as double blind. The use of identical placebos not explicitly stated. Randomisation: method of randomisation and concealment not stated. Drop outs: 6% patients excluded from ITT analysis	Single centre: Japan. Sample: 90. Patient selection: >4 week history of dyspepsia with normal endoscopy and +ve for Hp on biopsy or rapid urease test. Reflux-like disease: Patients with predominant heartburn excluded	Duration: 1 week. Eradication arm: omeprazole 20 mg bid, amoxycillin 500 mg tds, clarithromycin 200 mg bd. Control arm: placebos. Concomitant treatments: antacids and H₂ receptor antagonists. Hp eradication rates: 85% in the eradication arm and 0% in the placebo arm	Treatment success defined as none or minimal symptoms (score 0 or 1) in the previous 7 days. Symptoms rated according GSRS. Length of follow-up: 3 months
Malfertheiner 2000²⁴ (abstract; extra details given by authors)	Blinding: patients and investigators masked using identical placebos. Randomisation: adequate randomisation and concealment. Drop outs: 22% of participants excluded from ITT analysis	Multi-centre trial: Germany. Sample: 860 patients. Patient selection: >4 week history of dyspepsia with normal endoscopy and +ve for Hp on biopsy or rapid urease test. Reflux-like disease: Unclear whether patients with predominant heartburn excluded	Duration: 1 week. Eradication arm: lansoprazole 30/15mg bd, clarithromycin 500 mg bd and amoxycillin 1g bd for 7 days. Control arm: lansoprazole 15 mg od and matching placebos. Concomitant treatments: antacids and H₂ receptor antagonists. Hp eradication rate 80% in the eradication arm and 7% in the placebo arm	Treatment success defined as no or minimal symptoms in previous week (dyspepsia score 1). Validated German dyspepsia questionnaire evaluating 9 dyspeptic symptoms. Length of follow up: 12 months
Bruley des Varannes 2000²⁶ (abstract; extra details given by authors)	Blinding: patients and investigators masked using identical placebos. Randomisation: adequate randomisation, method of concealment not stated. Drop outs: all enrolled patients included in ITT analysis	Multi-centre trial: France. Sample: 253 patients. Patient selection: >3 months of epigastric pain with normal endoscopy and H pylori positive on biopsy or rapid urease test. Reflux-like disease: Excluded by the definition of dyspepsia	Duration: 1 week. Eradication arm: ranitidine 300 mg bd, amoxycillin 1g bd, clarithromycin 500 mg bd. Control arm: matching placebos. Concomitant treatments: unclear. Hp eradication rate: 70% in the eradication arm and 9% in the placebo arm	Treatment success defined as no epigastric pain in the previous week (score=0). Likert scale used, unclear whether the dyspepsia questionnaire was validated. Length of follow up: 12 months
Froehlich 2000²⁵ (abstract; extra details given by authors)	Blinding: patients and investigators masked using identical placebos. Randomisation: adequate randomisation and concealment. Drop outs: 9% excluded from ITT analysis	Multi-centre trial: Switzerland. Sample: 158 patients. Patient selection: >3 months of epigastric pain with normal endoscopy and H pylori positive on biopsy or rapid urease test. Reflux-like disease: excluded by the definition of dyspepsia	Duration: 1 week. Eradication arm: lansoprazole 30 mg bd, clarithromycin 500 mg bd, amoxycillin 1g bd. Control arm: lansoprazole 30 mg bd plus matching placebos. Concomitant treatments: any dyspepsia medication allowed during follow-up. Hp eradication rate: 72% in the eradication group and 2% in the placebo arm	Treatment success defined as a dyspepsia score of less than 10 (defined as a sum of 5 dyspeptic symptoms, minimum score=5, maximum score=25). Validated dyspepsia questionnaire used (van Zanten et al.). Quality of life using SF-12. Length of follow-up: 12 months
Koelz 1998¹⁹ (abstract; full paper made available)	Blinding: patients and investigators masked by use of matching placebos. Randomisation: no description of method of randomisation or concealment. Drop-outs: all participants included in intention to treat analysis	Multi-centre trial: Germany. Sample: 181 patients. Patient selection: >1 month history of dyspepsia with normal endoscopy and +ve for Hp on RUT and UBT. Patients were all resistant to PPI or H₂RA (ascertained in previous trial). Dyspepsia was severe enough to require treatment. Reflux-like disease: specifically excluded	Duration: 2 weeks. Eradication arm: omeprazole 40mg bid and amoxycillin 1 g bid. Control arm: omeprazole 20mg/day alone and matching placebos. Concomitant treatments: occasional use of antacid. Hp eradication rates: 52% for the eradication arm and 10% for the control arm	(a) Treatment success defined as no dyspepsia symptoms in previous week. (b) subjective grading of individual symptoms and overall symptoms in previous week. (c) quality of life measured using a German lifestyle questionnaire (results not available). (d) time off work/in hospital. Length of follow-up: 6 months

Hp=Helicobacter pylori. UBT=urea breath test. H₂RA=H₂ receptor antagonist. PPI=proton pump inhibitor. GSRS=Gastrointestinal symptoms rating scale. PGWB=psychological general well being index. GDSS=Glasgow dyspepsia severity scale. SF-36=short form 36