The effectiveness of omeprazole, clarithromycin and tinidazole in eradicating Helicobacter pylori in a community screen and treat programme

Transcription

1 Aliment Pharmacol Ther 2000; 14: 719±728. The effectiveness of omeprazole, clarithromycin and tinidazole in eradicating Helicobacter pylori in a community screen and treat programme P. MOAYYEDI*, R. FELTBOWER, W.CROCOMBEà, S.MASONà, P.ATHA,J.BROWNà, A. C. DOWELL±, I. D. G. RICHARDS & A. T. R. AXON* ON BEHALF OF THE LEEDS HELP STUDY GROUP *Centre for Digestive Diseases, The General In rmary at Leeds, Paediatric Epidemiology Group, ànorthern & Yorkshire Clinical Trials and Research Unit, Institute of Epidemiology & Health Services Research, University of Leeds, Leeds, UK; and ±Department of General Practice, Wellington School of Medicine, New Zealand Accepted for publication 7 February 2000 SUMMARY Introduction: Helicobacter pylori screening and treatment has been proposed as a cost-effective method of preventing gastric cancer. Aim: To assess, in a randomized controlled trial, the ef cacy of therapy in eradicating H. pylori as part of a screening programme, and to report the adverse events associated with this strategy. Methods: Subjects between the ages of 40±49 years were randomly selected from the lists of 36 primary care centres. Participants attended their local practice and H. pylori status was determined by 13 C-urea breath test. Infected subjects were randomized to receive omeprazole 20 mg b.d., clarithromycin 250 mg b.d. and tinidazole 500 mg b.d. for 7 days (OCT) or identical placebos. Eradication was determined by a 13 C-urea breath test 6 months and 2 years after the rst visit. Successful eradication was de ned as two negative 13 C-urea breath tests or one negative and one missing test. Adverse events and compliance were assessed at the 6-month visit. Results: A total of subjects were invited to attend, 8407 were evaluable, and 2329 (28%) of these were H. pylori-positive. A total of 1161 subjects were randomized to OCT and 1163 to placebo; over 80% returned for a repeat 13 C-urea breath test on at least one occasion. The eradication rates in those allocated to OCT were as follows: intention-to-treat, 710 out of 1161 (61%; 95% con dence interval: 58±64%); evaluable 710 out of 967 (73%; 95% CI: 71±76%); took all medication 645 out of 769 (84%; 95% CI: 81±87%). Adverse events occurred in 45% of the treatment group and in 18% of the placebo group (relative risk 2.5; 95% CI: 2.1±2.9). Compliance, male gender, no antibiotic prescription in the subsequent 2 years and experiencing a bitter taste with the medication were independently associated with treatment success. Conclusions: The OCT regimen has an eradication rate of 61% in intention-to-treat analysis and is therefore less successful in treating H. pylori as part of a screening programme compared with hospital studies in dyspeptic patients. INTRODUCTION Correspondence to: Dr P. Moayyedi, Centre for Digestive Diseases, The General In rmary at Leeds, Great George Street, Leeds LS1 3EX, UK. paulmo@ulth.northy.nhs.uk Helicobacter pylori is associated with more premature mortality in developed countries than any other infectious disease. 1 H. pylori infection is also the main cause of peptic ulcer disease, which is a major clinical and Ó 2000 Blackwell Science Ltd 719

2 720 P. MOAYYEDI et al. nancial burden to Western societies. 2 Non-invasive tests for H. pylori are available and treatment has become cheaper, simpler and more effective with proton pump inhibitor triple therapies. 3 H. pylori screening and treatment of the normal population is therefore practicable, and could potentially reduce mortality from gastric cancer and peptic ulcer disease. A Markov model suggested that this approach is cost-effective, assuming H. pylori therapy decreases the risk of gastric cancer. 4 Decision analysis models assume that therapy will be 80±100% effective in eradicating H. pylori. 4 These gures are based on eradication rates achieved in secondary care patients with peptic ulcer disease or dyspepsia requiring hospital investigation. 3 H. pylori therapy may be less effective in the normal population where most have no symptoms or only mild dyspepsia. It is important to establish the effectiveness of eradication therapy in this setting as it will have implications for the ef cacy of an H. pylori screen-and-treat public health policy. The H. pylori eradication regimen used in a screening programme should be acceptable to the population and cause few side-effects. The combination of omeprazole, clarithromycin and tinidazole uses low dose antibiotics for 1 week and can be used in the 10% of subjects who claim an allergy to penicillin. This regimen is associated with adverse events although these are usually mild. 5 These data, however, are again based on secondary care patients and this may not re ect the experience of subjects in the general population. Public health measures attempting to treat this infection should identify factors that reduce the effectiveness of therapy. Compliance is an important determinant of treatment success. 6 Studies suggest that therapy is less likely to be successful in smokers but results have been con icting. 7, 8 Patients with peptic ulcer disease are reported to have a higher success rate but this represents only a minority of the population receiving treatment. 9 There are no other factors commonly associated with treatment failure, although studies investigating this are usually small and do not address the problem from a community perspective. In particular antibiotic prescription is common in primary care and may in uence the success of treatment. We have investigated the effectiveness of proton pump inhibitor triple therapy in a large cohort randomly selected from the general population, evaluated adverse events and assessed factors associated with treatment failure. METHODS Study sample selection This paper reports one of the outcomes of a larger study assessing the medical bene ts and health economics of H. pylori eradication in the community. 10 The trial was conducted in the cities of Leeds and Bradford in the UK, which have a combined population of approximately Primary care centres in the area were approached to participate in the study and 36 out of 37 consented. Individuals between the ages of 40±49 years were randomly selected from the databases of these primary care centres and were invited by letter to attend their local practice. A trained researcher interviewed subjects attending their primary care centre; those taking antibiotics or bismuth salts within the last 2 weeks or proton pump inhibitors within the last 1 week were excluded from the study. Exclusion criteria also included: allergy to macrolides, 5-nitroimidazoles and proton pump inhibitors; unwillingness to abstain from alcohol for 1 week; and concomitant antihistamines, anti-convulsants, theophyllines, digoxin or warfarin. Written informed consent was obtained from participants and the study was approved by Leeds and Bradford Ethics Committees. Subject assessment and intervention A trained researcher interviewed participants about smoking habits, alcohol and coffee intake, ethnic origin, marital status, and childhood and adult socio-economic factors. Social class was categorized according to the UK Of ce of National Statistics de nitions. 11 The presence and severity of dyspepsia was measured using the validated Leeds Dyspepsia Questionnaire. 12 H. pylori status was evaluated by a non-fasting 13 C-urea breath test. Participants were permitted to have two slices of toast with jam or honey together with a cup of tea or coffee in the preceding 4 h. Baseline breath samples were obtained in two 10 ml exetainer tubes and subjects were given 100 mg of 13 C-labelled urea (99% pure Boston Isotopes, Boston, US) and 4 g of citric acid to delay gastric emptying. Further breath samples were obtained at 30 min and the 13 CO 2 was measured using a mass spectrometer (Europa Scienti c, Brentford, UK). Subjects with an increase in 13 CO 2 excretion of > 5/mL at 30 min were de ned as H. pylori-positive. This protocol has

3 SCREEN AND TREAT H. PYLORI IN THE COMMUNITY % sensitivity and 96% speci city for diagnosing H. pylori in the population from which the study sample was taken. 13 H. pylori-positive cases were allocated to receive omeprazole 20 mg b.d., clarithromycin 250 mg b.d., and tinidazole 500 mg b.d. or identical placebos for 1 week using computer generated random number tables. A central clinical trials unit, not involved in collecting the data, performed the randomization to ensure masking. The medication was prepared by a hospital pharmacy and delivered to the local general practices. Participants were informed by letter that their medication was ready for collection from their primary care centre. Subjects were also given a prepaid postcard to complete if they experienced any adverse events. The primary care notes were reviewed on all H. pyloripositive participants to evaluate antibiotic prescriptions over the 2-year duration of the trial. Past medical history of peptic ulcer disease diagnosed at endoscopy or barium meal, and non-steroidal prescriptions were also recorded. The general practice records were also used to determine visits to the primary care physician for dyspepsia, H 2 -receptor antagonist prescriptions, and proton pump inhibitor prescriptions over the previous 2 years. This was compared with the primary care notes review of a random selection of non-responders to evaluate any differences between those who did and those who did not participate in the study. Outcome measures Subjects were given an appointment to return to their local primary care centres 6 months and 2 years after recruitment for repeat 13 C-urea breath tests. Treatment success was de ned as two negative 13 C-urea breath test results or one negative and one missing test. If either the 6-month or 2-year 13 C-urea breath test was positive the subject was de ned as a treatment failure. Adverse events were recorded at the 6-month interview using speci c questions on 14 symptoms. These were classi ed as mild or moderate/severe depending on whether they interfered with activities of daily living. Compliance was also assessed at the 6-month interview and this was veri ed by recording whether subjects collected their medication from their local practice. The primary care notes were reviewed to determine any hospital admission that could be attributed to taking antibiotics. Statistical analysis and sample size The effectiveness of the treatment was recorded in the treatment group using an intention-to-treat analysis of evaluable subjects. A `best case' eradication rate was calculated by excluding participants who did not take all their medication. A `worst case' eradication rate was estimated by assuming that all those failing to attend follow-up were treatment failures. Adverse events reported in the treatment group were compared with the placebo group using Fisher's exact test. The association between the presence of dyspepsia, smoking, alcohol intake, coffee consumption, socioeconomic conditions, previous antibiotic use, past history of peptic ulcer disease and treatment failure were evaluated using Pearson's v 2 -test. The correlation between age and dyspepsia severity was analysed using the Student's t-test. Independent risk factors for treatment failure were determined by backward elimination logistic regression. Randomly selected non-responders were approached for permission to review their primary care notes. The number of primary care dyspepsia visits, H 2 -receptor antagonist prescriptions, and proton pump inhibitor prescriptions were compared between participants and non-participants to evaluate to what extent the results from the study sample could be generalized. A two-sided P < 0.05 was considered statistically signi cant and analyses were carried out using STATA version 5 (Stata corporation, Texas, USA). Sample size considerations for this trial were based on clinically signi cant reductions in dyspepsia rather than the effectiveness of treatment. Nevertheless, assuming 10% of the placebo group report adverse events and a drop-out rate of 25%, 2400 subjects randomized to treatment or placebo would allow a ve percentage point difference in adverse events to be detected between treatment and placebo groups at the 80% power and 5% signi cance level. RESULTS Recruitment The proportion of subjects responding to the invitation was 25% and 2329 out of 8407 (28%) were H. pylori-positive (Figure 1). Infected subjects were reviewed at 6 months and 2 years with over 80% returning for a repeat 13 C-urea breath test on at least one occasion.

4 722 P. MOAYYEDI et al. Figure 1. Progress of subjects invited to participate in the trial. The primary care notes of responders were compared with 624 randomly selected non-responders to determine the differences between the two groups. Non-responders were less likely than responders to have visited their primary care physician with dyspepsia in the previous 2 years (15.1% vs. 8.2%, P < 0.01, v 2 -test). There was, however, a similar frequency of H 2 -receptor antagonist and proton pump inhibitor prescriptions between participants and nonparticipants (H 2 -receptor antagonists 4.0% vs. 3.8%, P ˆ 0.95, v 2 -test; proton pump inhibitor prescriptions 3.4% vs. 2.0%, P ˆ 0.1, v 2 -test). There was no clinically signi cant difference in age between the two groups (mean standard deviation age for participants was years; non-participants years). Eradication rate There were 1161 subjects allocated to receive OCT, 967 attended at 6 months and/or 2 years (Figure 1) and 769 completed the therapy. The overall eradication rate was 61% if all that did not attend were assumed to have failed therapy, 73% in evaluable cases and 84% in those completing therapy (Table 1). Subjects taking previous eradication therapy were not excluded from the trial but this occurred in only three cases. Two of these subjects had successful therapy and the third was lost to followup. These subjects were included in all analyses. Thirty-six subjects discontinued therapy early and the H. pylori eradication rate was signi cantly associated with duration of treatment. Subjects only completing 1± 2 days of therapy had an eradication rate of 20% (two

5 SCREEN AND TREAT H. PYLORI IN THE COMMUNITY 723 Table 1. H. pylori eradication rates Type of analysis No. of subjects No. successfully treated % eradication 95% CI Evaluable subjects % 71±76% Intention-to-treat % 58±64% Compliance Completed % 81±87% No data available* % 45±70% Did not complete % 14±28% * These are subjects that only attended the 2 year visit (compliance was assessed at 6 months). out of 10) compared with 53% (eight out of 15) in those taking medication for 3±4 days, and 73% (eight out of 11) in subjects taking 5±6 days of treatment (v 2 for linear trend P ˆ 0.02). Eradication of H. pylori was observed in 28 out of 972 evaluable cases (eradication rate 3%: 95% CI: 2±4%) allocated to receive placebo. Adverse events Adverse events occurred more frequently in subjects receiving OCT compared with placebo with a 27% increase in absolute risk (Table 2); subjects allocated to OCT were particularly at increased risk of experiencing a bitter taste (Table 2). Participants given OCT were also signi cantly more likely to complain of vomiting, skin rash and itching than those given placebo but the absolute increase in risk was less than 1% (data not shown). Adverse events that interfered with activities of daily living occurred in 122 out of 909 (13%) of the OCT group and 90 out of 920 (10%) of the placebo group (relative risk 1.37, 95% CI: 1.06±1.77 v 2 P ˆ 0.02). This represents a 4% (95% CI: 1±7%) increase in absolute risk of developing moderate to severe adverse events in subjects taking OCT. Factors predicting treatment success Factors predicting treatment success were explored. In a univariate analysis OCT was less successful in women, alcohol drinkers, subjects not reporting a bitter taste as an adverse event and subjects not completing the course of medication (Table 3). There was no signi cant difference in age between treatment failure and success (mean standard deviation of treatment failure, years; treatment success, years, P ˆ 0.55) and dyspepsia scores were similar between the two groups (treatment failure ; treatment success , P ˆ 0.44). A total of 467 subjects received at least one antibiotic course during the course of the study and 325 of these (70%) had successful treatment compared with 385 out 500 (77%) of subjects not prescribed antibiotics (P ˆ 0.01 Fisher's exact test). The ef cacy of treatment was reduced with increasing number of scripts over the 2-year period (Table 4). Women were nearly twice as likely to be Table 2. Adverse events in subjects receiving treatment compared to placebo (only adverse events with > 1% absolute risk included) Adverse event Number experiencing event in treatment group (n = 813) Number experiencing event in placebo group (n = 839) Absolute risk (%) Relative risk 95% CI P-value Bitter taste 140 (17%) 20 (2%) 15% ±11.4 < Nausea 115 (14%) 52 (6%) 8% ±3.1 < Diarrhoea 92 (11%) 37 (4%) 7% ±3.7 < Sore mouth 87 (11%) 14 (2%) 9% ±11.1 < Flatulence 74 (9%) 36 (4%) 5% ± Headache 40 (5%) 19 (2%) 3% ± Tiredness 36 (4%) 11 (1%) 3% ± Thrush 33 (4%) 3 (0.3%) 4% ±36.7 < Any adverse event 367 (45%) 153 (18%) 27% ±2.9 <

6 724 P. MOAYYEDI et al. Variable Number* % H. pylori eradication Relative risk for treatment failure 95% CI P-value Table 3. Factors associated with H. pylori therapy failure Dyspepsia Yes % 1.00 No % ± Previous peptic ulcer disease Yes 52 85% 1.00 No % ± Took medication Yes % 1.00 No % ±31.15 < Gender Male % 1.00 Female % ± Smoker Yes % 1.00 No % ± Drinks coffee Yes % 1.00 No % ± Drinks alcohol Yes % 1.00 No % ± Social class Nonmanual % 1.00 Manual % ± Bitter taste with OCT Yes % 1.00 No % ± Nausea with OCT Yes % 1.00 No % ± Diarrhoea with OCT Yes 92 80% 1.00 No % ± Sore mouth with OCT Yes 87 84% 1.00 No % ± * Numbers do not always add up to 967 due to missing data. prescribed 5-nitroimidazoles in the previous 2 years compared with men (31 out of 573 [5.4%] vs. 17 out of 588 [2.9%] P ˆ 0.04). Independent predictors of treatment success were explored using backward elimination multiple logistic regression of all factors outlined in Tables 3 and 4 and variables with P > 0.05 were excluded from the nal model. Gender, compliance, bitter taste and antibiotic prescription remained independent predictors of treatment success (Table 5). This model explained 9% of the variance in the data. DISCUSSION This is the rst trial to evaluate the ef cacy of H. pylori eradication therapy in the general population. The

7 SCREEN AND TREAT H. PYLORI IN THE COMMUNITY 725 Table 4. Antibiotic prescriptions during study and OCT treatment failure Antibiotic prescriptions Number % H. pylori eradication Relative risk for treatment failure 95% CI No prescriptions % prescription % ± prescriptions % ± prescriptions 54 67% ±1.45 > 3 prescriptions 70 60% ±1.60 Pearson v 2 = Linear-by±linear association = Table 5. Multiple logistic regression of factors independently predicting OCT treatment success* Variable Adjusted odds ratio** for treatment failure 95% CI P-value Compliant Yes 1.00 No ±15.51 < Gender Male 1.00 Female ± Experienced bitter taste Yes 1.00 No ± Antibiotics prescribed None 1.00 One prescription ± Two prescriptions ± Three prescriptions ± Four or more prescriptions ± * Backward elimination method starting with all variables listed in Tables 3 and 4. ** Odds ratio mutally adjusted for factors in Table 5. results indicate that OCT cures 73% of evaluable participants. This may be an overestimate as the response rate was only 25% and those who attended may be more compliant than the general population. The intention-to-treat cure rate of 61% may therefore be a more realistic gure of the success rate of OCT. These cure rates are much lower than achieved in hospital trials and may re ect poorer compliance in the general population. 14, 15 This is supported by the H. pylori eradication rate of 84% in those completing the course of medication, which is similar to that reported in secondary care patients. 3 These data suggest that cure rates of 60±85% should be used in decision analysis models evaluating the cost-effectiveness of H. pylori screening and treatment to prevent gastric cancer. Adverse events were common when OCT was used in this H. pylori screen-and-treat programme. These data suggest that 45% of those taking OCT experience adverse events, although only reported by 27% of the study population, are attributable to active treatment. Early studies suggested that OCT rarely caused sideeffects as low doses of antibiotics were being used for only 1 week. 16 Recent trials have reported adverse 17, 18 events at a frequency compatible with our data. Bitter taste, diarrhoea and nausea were the most common events recorded, as reported in other studies; the frequency of bitter taste was higher than others have reported. 5, 17, 18 This may re ect the method of data collection as researchers asked about speci c symptoms using a prede ned questionnaire rather than making general enquiries about any event that occurred whilst taking medication. An H. pylori screening programme would need to treat many individuals to prevent one gastric cancer so it is important that the therapy should be safe and associated with few severe side-effects. Although adverse events were common in this trial these were usually mild; 13% of subjects experienced problems that interfered with activities of daily living and in 4% this could be attributed to active medication. We cannot exclude the possibility that side-effects occurred more frequently in subjects who did not attend follow-up although no severe event was recorded in the general practitioner notes in these individuals. A number of factors have been reported to in uence the effectiveness of H. pylori therapy. Treatment has been less effective in smokers, but this mainly relates to omeprazole and amoxycillin dual therapy, and results with other triple therapies are con icting. 7, 8, 19 This large cross-sectional study of H. pylori eradication in the general population does not support the hypothesis that smoking alters the ef cacy of OCT. Labenz et al. suggested that activity of gastritis and gastric ulcer were predictors of treatment success. 9 This is supported by the observation that therapy is particularly successful against more virulent strains of H. pylori. 20 We found only a marginal association between peptic ulcer disease and success of therapy in univariate analysis and this was not a signi cant factor in the logistic regression model. There were only a small number of

8 726 P. MOAYYEDI et al. peptic ulcer cases in this community sample and a minor association may have been missed. Nevertheless this trial emphasizes that peptic ulcer disease is not an important determinant for OCT success in the context of a community screen-and-treat programme. Several studies have highlighted the importance of compliance in determining treatment success and this is supported by these data. 6, 7 Clarithromycin and metronidazole are actively secreted in the saliva and this can lead to taste disturbance in individuals taking OCT. 21 Subjects reporting taste perversion were more likely to become H. pylorinegative and this was an independent predictor of success in a logistic regression model. This may be due to the success of treatment being partially dependent on the dose of antibiotic found in the saliva or alternatively taste disturbance may be a marker of compliance. Hospital H. pylori eradication study protocols usually exclude the use of antibiotics during follow-up as this could theoretically in uence the results; this would be impractical in evaluating population H. pylori therapy over 2 years. We therefore monitored antibiotic prescribed during the course of the study to evaluate whether this in uenced the results. The quantity of antibiotics prescribed during the study was comparable to of cial data on the amount of antimicrobial use in the area. 22 We hypothesized that subjects taking antibiotics may have a slightly increased eradication rate as single antimicrobials occasionally successfully treat H. pylori. 23 Subjects were, however, less likely to have successful therapy if they were prescribed antibiotics during follow-up. Patient expectation is an important determinant of antibiotic prescription and subjects who receive one course of antibiotics are more likely to receive further courses. 24, 25 Subjects receiving multiple courses of antibiotics during follow-up may have also been given many courses of antibiotics before the start of the study and developed resistant strains of H. pylori. The ef cacy of OCT is reduced against macrolide and 5-nitroimidazole resistant H. pylori strains. 26 This emphasizes the importance of minimizing antibiotic prescription and highlights the possible risks of widespread eradication of H. pylori from the community. Therapy was signi cantly less successful in women; we have reported this previously but the reasons for this nding are unclear. 19 This may relate to an increased prevalence of 5-nitroimidazole resistant organisms in women; strains of H. pylori resistant to metronidazole are more prevalent in women than men in our area (D. S. Tompkins, personal communication). Women were also more likely to have been prescribed 5- nitroimidazoles in this study. Alternatively there may be gender differences in acid output and gastric blood ow that in uence treatment success. 27 This nding requires evaluation in prospective studies. The decision to screen-and-treat populations for H. pylori depends on the effect of this on reducing gastric cancer mortality. There is no evidence that H. pylori screening and treatment will be of bene t and randomized trials are needed to investigate this. These data suggest that the sample size needed for such a trial may be larger than expected as the ef cacy of proton pump inhibitor based triple therapy in eradicating H. pylori is lower in the community than that achieved in the hospital setting. ACKNOWLEDGEMENTS This trial was funded by a Research and Development grant from the Northern and Yorkshire region and by ASTRA Hassle. Dr Paul Moayyedi is funded by a UK Medical Research Council Training Fellowship. We are grateful to Miss Julie Mackintosh for her help in preparing the manuscript. The Leeds Helicobacter pylori Eradication in GeneraL Practice (HELP) Study Group consists of the following: Study Supervisors: Professor A. T. R. Axon a, Professor I. D. G. Richards b, Professor A. C. Dowell c, Professor P. Heywood d, Professor A. Walan e. Clinical Co-ordinator: Dr P. Moayyedi a. Trial co-ordinators and data management: Dr S. Mason f, Mr W. Crocombe f, Mrs R. Muthukumar f, Miss J. Norton f. Data collection: Mrs S. Duffett b, Mrs P. Atha b, Mrs M. Liptrott b, Mrs J. Nathan b, Mrs C. Youings b, Mrs R. Hall b, Mrs J. Greatrex b, Mrs J. Hammacott b, Miss A. Zilles b, Mrs J. Welsby b, Miss C. Walton b. Statistical support: Dr D. Braunholtz g, Mrs J. Brown f, Dr P. McKinney h, Mr R. Feltbower h, Dr S. Eriksson e. Health Economic support: Professor M. Drummond i, Dr J. Mason i, Dr N.-O. Stalhammar e. Data monitoring committee: Dr R. Spiller j, Dr M. Jones k, Professor D. Forman. l 13 C-urea breath test analysis: Mr M. Clough a. Trial pharmacist: Ms C. Bedford m. Finance: Mrs A. Starkey n. Addresses a Centre for Digestive Diseases, The General In rmary at Leeds, UK; b Institute of Epidemiology and Health Services Research, University of Leeds, UK; c Department

9 SCREEN AND TREAT H. PYLORI IN THE COMMUNITY 727 of General Practice, Wellington School of Medicine, New Zealand; d Centre for Research in Primary Care, Nuf eld Institute for Health, Leeds, UK; e ASTRA Hassle AB, Molndal, Sweden; f Northern and Yorkshire Clinical Trials and Research Unit, University of Leeds, UK; g The Department of Public Health and Epidemiology, University of Birmingham, UK; h Paediatric Epidemiology Group, University of Leeds, UK; i Centre for Health Economics, University of York, UK; j Reader in Gastroenterology, Queens Medical Centre, Nottingham, UK; k ICRF, St James' & Seacroft University Hospitals, Leeds, UK; l Centre for Cancer Research, University of Leeds, UK; m Outpatient Pharmacy, The General In rmary at Leeds, UK; n Research School of Medicine, University of Leeds, UK. Participating general practices Meanwood Health Centre; Kippax Health Centre; The Croft Surgery; Windsor House Surgery; Woodsley Health Centre; Bridge Street Surgery; Marsh Street Surgery; High Field Surgery; Lingwell Croft Surgery; Beeston Hill Health Centre; Woodhouse Medical Centre; Leigh View Medical Practice; Fountain Medical Centre; Dib Lane Practice; Hunslet Health Centre; Grange Medical Centre; Crossland Surgery; St Martin's Practice; Robin Lane Medical Centre; Manor Park Surgery; Carlton Surgery; Burton Croft Surgery; Ridge Medical Practice; West Lodge Surgery; Windmill Health Centre; Yeadon Health Centre; The Medical Centre, 30 Buttershaw Lane, Bradford; Garforth Medical Centre; The Chapeloak Practice; The Health Centre, King's Road, Wrose, Bradford; Burley Park Medical Centre; The Street Lane Practice; Silver Lane Surgery; New Wortley Health Centre; Cullingworth Medical Centre; Westcliffe Medical Centre. REFERENCES 1 Axon ATR, Forman D. Helicobacter gastroduodenitis: a serious infectious disease. Br Med J 1997; 314: 1430±1. 2 Moayyedi P, Axon ATR. Is there a rationale for eradication of Helicobacter pylori? Cost-bene t: the case for. Br Med Bull 1998; 54: 243±50. 3 Unge P. What other regimens are under investigation to treat Helicobacter pylori infection? Gastroenterology 1997; 113: S131±48. 4 Parsonnet J, Harris RA, Hack HM, et al. Modelling costeffectiveness of Helicobacter pylori screening to prevent gastric cancer: a mandate for clinical trials. Lancet 1996; 348: 150±4. 5 Lind T, Veldhuyzen vz, Unge P, et al. Eradication of Helicobacter pylori using one-week triple therapies combining omeprazole with two antimicrobials: the MACH I Study. Helicobacter 1996; 1: 138±44. 6 Graham DY, Lew GM, Malaty HM, et al. Factors in uencing the eradication of Helicobacter pylori with triple therapy. Gastroenterology 1992; 102: 493±6. 7 Cutler AF, Schubert TT. Patient factors affecting Helicobacter pylori eradication with triple therapy. Am J Gastroenterology 1993; 88: 505±9. 8 Unge P, Gad A, Eriksson K, et al. Amoxicillin added to omeprazole prevents relapse in the treatment of duodenal ulcer patients. Eur J Gastroenterol Hepatol 1993; 5: 325±31. 9 Labenz J, Leverkus F, Borsch G. Omeprazole plus amoxicillin for cure of Helicobacter pylori infection. Factors in uencing the treatment success. Scand J Gastroenterol 1994; 29: 1070±5. 10 Moayyedi P. Eradication of Helicobacter pylori in general practice: medical bene ts and health economics. PhD thesis, University of Leeds, Of ce of National Statistics Census. London: HMSO, Moayyedi P, Duffett S, Braunholtz D, et al. The Leeds Dyspepsia Questionnaire: a valid tool for measuring the presence and severity of dyspepsia. Aliment Pharmacol Ther 1998; 12: 1257± Moayyedi P, Braunholtz D, Heminbrough E, et al. Do patients need to fast for a 13 C-urea breath test? Eur J Gastroenterol Hepatol 1997; 9: 275±7. 14 Unge P, Berstad A. Pooled analysis of anti-helicobacter pylori treatment regimens. Scand J Gastroenterol 1996; 220(Suppl.): 27± van der Hulst RW, Keller JJ, Rauws EA, et al. Treatment of Helicobacter pylori infection: a review of the world literature. Helicobacter 1996; 1: 6± Bazzoli F, Zagari RM, Fossi S, et al. Short-term low-dose triple therapy for the eradication of Helicobacter pylori. Eur J Gastroenterol Hepatol 1994; 6: 773±7. 17 Moayyedi P, Sahay P, Tompkins DS, et al. Ef cacy and optimum dose of omeprazole in a new 1-week triple therapy regimen to eradicate Helicobacter pylori. Eur J Gastroenterol Hepatol 1995; 7: 835± Schwartz H, Krause R, Siepman N, et al. Seven-day triple therapy with lansoprazole, clarithromycin, and metronidazole for the cure of Helicobacter pylori infection: a short report. Helicobacter 1996; 1: 251±5. 19 Moayyedi P, Chalmers DM, Axon ATR. Patient factors that predict failure of omeprazole, clarithromycin, and tinidazole to eradicate Helicobacter pylori. J Gastroenterol 1997; 32: 24±7. 20 van der Hulst RW, van der Ende A, Dekker FW, et al. Effect of Helicobacter pylori eradication on gastritis in relation to caga: a prospective 1-year follow-up study. Gastroenterology 1997; 113: 25± Goddard AF, Jessa MF, Barrett DA, et al. Effect of omeprazole on distribution of metronidazole, amoxycillin, and clarithromycin in human gastric juice. Gastroenterology 1996; 111: 358±67.

10 728 P. MOAYYEDI et al. 22 Leeds Health Care PACT data Axon ATR. The role of acid inhibition in the treatment of Helicobacter pylori infection. Scand J Gastroenterol 1994; 201(Suppl.): 16± Macfarlane J, Holmes W, Macfarlane R, et al. In uence of patients' expectations on antibiotic management of acute lower respiratory tract illness in general practice: questionnaire study. Br Med J 1997; 315: 1211±4. 25 Macfarlane J, Prewett J, Rose D, et al. Prospective case-control study of role of infection in patients who reconsult after initial antibiotic treatment for lower respiratory tract infection in primary care. Br Med J 1997; 315: 1206± Lind T, Megraud F, Unge P, et al. The MACH2 Study: Role of omeprazole in eradication of Helicobacter pylori with 1-week triple therapies. Gastroenterology 1999; 116: 248± Prewett EJ, Smith JT, Nwokolo CU, et al. Twenty-four hour intragastric acidity and plasma gastrin concentration pro les in female and male subjects. Clin Sci 1992; 80: 595±6.