Medical treatments for the maintenance therapy of reflux oesophagitis and endoscopic negative reflux disease (Review)

Medical treatments for the maintenance therapy of reflux oesophagitis and endoscopic negative reflux disease (Review) Donnellan C, Sharma N, Preston C, Moayyedi P This is a reprint of a Cochrane review, prepared and maintained by The Cochrane Collaboration and published in The Cochrane Library 2008, Issue 1 http://www.thecochranelibrary.com 1

T A B L E O F C O N T E N T S ABSTRACT...................................... BACKGROUND.................................... OBJECTIVES..................................... CRITERIA FOR CONSIDERING STUDIES FOR THIS REVIEW.................. SEARCH METHODS FOR IDENTIFICATION OF STUDIES................... METHODS OF THE REVIEW............................... DESCRIPTION OF STUDIES............................... METHODOLOGICAL QUALITY.............................. RESULTS....................................... DISCUSSION..................................... AUTHORS CONCLUSIONS............................... POTENTIAL CONFLICT OF INTEREST........................... ACKNOWLEDGEMENTS................................ SOURCES OF SUPPORT................................. REFERENCES..................................... TABLES....................................... Characteristics of included studies............................. Characteristics of excluded studies............................. ANALYSES...................................... Comparison 01. Healing dose PPI versus placebo........................ Comparison 02. Maintenance dose PPI versus placebo...................... Comparison 03. Healing dose PPI versus Maintenance dose PPI.................. Comparison 04. Healing dose PPI versus H2 receptor antagonist.................. Comparison 05. Maintenance dose PPI versus H2 receptor antagonist................ Comparison 06. H2 receptor antagonist versus placebo...................... Comparison 07. Prokinetic therapy versus placebo/no treatment.................. Comparison 08. Sucralfate versus placebo.......................... Comparison 09. Healing dose PPI versus Maintenance dose PPI 3yr follow up.............. Comparison 10. Endoscopic negative reflux disease - cisapride versus placebo.............. Comparison 11. Endoscopic negative reflux disease - omeprazole 10mg versus.placebo........... Comparison 12. One daily 15mg lansoprazole versus alternate days of 30mg.............. Comparison 13. Double dose PPI versus healing dose PPI..................... Comparison 14. Healing dose PPI type 1 versus healing dose of PPI type 2............... Comparison 15. Overall adverse effects........................... INDEX TERMS.................................... COVER SHEET.................................... GRAPHS AND OTHER TABLES.............................. Analysis 01.01. Comparison 01 Healing dose PPI versus placebo, Outcome 01 Relative risk of relapse (oesophagitis) Analysis 01.02. Comparison 01 Healing dose PPI versus placebo, Outcome 02 Relative risk of relapse (symptoms). Analysis 01.03. Comparison 01 Healing dose PPI versus placebo, Outcome 03 Adverse effects, overall..... Analysis 01.04. Comparison 01 Healing dose PPI versus placebo, Outcome 04 Adverse effects, headache.... Analysis 01.05. Comparison 01 Healing dose PPI versus placebo, Outcome 05 Adverse effects, diarrhoea.... Analysis 01.06. Comparison 01 Healing dose PPI versus placebo, Outcome 06 Adverse effects, abdominal pain.. Analysis 02.01. Comparison 02 Maintenance dose PPI versus placebo, Outcome 01 Relative risk of relapse (oesophagitis) Analysis 02.02. Comparison 02 Maintenance dose PPI versus placebo, Outcome 02 Relative risk of relapse (symptoms) Analysis 02.03. Comparison 02 Maintenance dose PPI versus placebo, Outcome 03 Adverse effects, overall.... Analysis 02.04. Comparison 02 Maintenance dose PPI versus placebo, Outcome 04 Adverse effects, headache... Analysis 02.05. Comparison 02 Maintenance dose PPI versus placebo, Outcome 05 Adverse effects, diarrhoea... Analysis 02.06. Comparison 02 Maintenance dose PPI versus placebo, Outcome 06 Adverse effects, abdominal pain Analysis 03.01. Comparison 03 Healing dose PPI versus Maintenance dose PPI, Outcome 01 Relative risk of relapse (oesophagitis).................................. 1 2 2 2 3 4 5 6 7 13 14 15 15 15 15 23 23 35 38 38 39 39 39 39 40 40 40 40 40 41 41 41 41 41 41 42 43 43 44 44 45 45 46 46 47 48 48 49 49 50 i

Analysis 03.02. Comparison 03 Healing dose PPI versus Maintenance dose PPI, Outcome 02 Relative risk of relapse (symptoms).................................. Analysis 03.03. Comparison 03 Healing dose PPI versus Maintenance dose PPI, Outcome 03 Adverse effects, overall Analysis 03.04. Comparison 03 Healing dose PPI versus Maintenance dose PPI, Outcome 04 Adverse effects, headache Analysis 03.05. Comparison 03 Healing dose PPI versus Maintenance dose PPI, Outcome 05 Adverse effects, diarrhoea Analysis 03.06. Comparison 03 Healing dose PPI versus Maintenance dose PPI, Outcome 06 Adverse effects, abdominal pain..................................... Analysis 04.01. Comparison 04 Healing dose PPI versus H2 receptor antagonist, Outcome 01 Relative risk of relapse (oesophagitis).................................. Analysis 04.02. Comparison 04 Healing dose PPI versus H2 receptor antagonist, Outcome 02 Relative risk of relapse (symptoms).................................. Analysis 04.03. Comparison 04 Healing dose PPI versus H2 receptor antagonist, Outcome 03 Adverse effects, overall Analysis 04.04. Comparison 04 Healing dose PPI versus H2 receptor antagonist, Outcome 04 Adverse effects, headache Analysis 04.06. Comparison 04 Healing dose PPI versus H2 receptor antagonist, Outcome 06 Adverse effects, abdominal pain..................................... Analysis 05.01. Comparison 05 Maintenance dose PPI versus H2 receptor antagonist, Outcome 01 Relative risk of relapse (oesophagitis)............................... Analysis 05.02. Comparison 05 Maintenance dose PPI versus H2 receptor antagonist, Outcome 02 Relative risk of relapse (symptoms)................................ Analysis 05.03. Comparison 05 Maintenance dose PPI versus H2 receptor antagonist, Outcome 03 Adverse effects, overall.................................... Analysis 05.04. Comparison 05 Maintenance dose PPI versus H2 receptor antagonist, Outcome 04 Adverse effects, headache................................... Analysis 05.05. Comparison 05 Maintenance dose PPI versus H2 receptor antagonist, Outcome 05 Adverse effects, abdominal pain................................. Analysis 06.01. Comparison 06 H2 receptor antagonist versus placebo, Outcome 01 Relative risk of relapse (oesophagitis) Analysis 06.02. Comparison 06 H2 receptor antagonist versus placebo, Outcome 02 Relative risk of relapse (symptoms) Analysis 06.03. Comparison 06 H2 receptor antagonist versus placebo, Outcome 03 Adverse effects, overall... Analysis 07.01. Comparison 07 Prokinetic therapy versus placebo/no treatment, Outcome 01 Relative risk of relapse (oesophagitis).................................. Analysis 07.02. Comparison 07 Prokinetic therapy versus placebo/no treatment, Outcome 02 Relative risk of relapse (symptoms).................................. Analysis 07.03. Comparison 07 Prokinetic therapy versus placebo/no treatment, Outcome 03 Adverse effects, overall Analysis 07.04. Comparison 07 Prokinetic therapy versus placebo/no treatment, Outcome 04 Adverse effects, diarrhoea Analysis 08.01. Comparison 08 Sucralfate versus placebo, Outcome 01 Relative risk of relapse (oesophagitis)... Analysis 08.02. Comparison 08 Sucralfate versus placebo, Outcome 02 Adverse effects, overall........ Analysis 09.01. Comparison 09 Healing dose PPI versus Maintenance dose PPI 3yr follow up, Outcome 01 Relative risk of relapse (oesophagitis).............................. Analysis 09.02. Comparison 09 Healing dose PPI versus Maintenance dose PPI 3yr follow up, Outcome 02 Relative risk of relapse (symptoms)............................... Analysis 09.03. Comparison 09 Healing dose PPI versus Maintenance dose PPI 3yr follow up, Outcome 03 Adverse effects, overall.................................. Analysis 09.04. Comparison 09 Healing dose PPI versus Maintenance dose PPI 3yr follow up, Outcome 04 Adverse effects, diarrhoea................................. Analysis 10.01. Comparison 10 Endoscopic negative reflux disease - cisapride versus placebo, Outcome 01 Relative risk of relapse (symptoms)............................... Analysis 11.01. Comparison 11 Endoscopic negative reflux disease - omeprazole 10mg versus.placebo, Outcome 01 Relative risk of relapse (symptoms).......................... Analysis 11.02. Comparison 11 Endoscopic negative reflux disease - omeprazole 10mg versus.placebo, Outcome 02 Adverse effects, overall............................... Analysis 12.01. Comparison 12 One daily 15mg lansoprazole versus alternate days of 30mg, Outcome 01 Relative risk of relapse (oesophagitis).............................. 51 52 52 53 53 54 55 55 56 56 57 57 58 58 59 59 60 60 61 61 62 62 63 63 64 64 65 65 66 66 67 67 ii

Analysis 12.02. Comparison 12 One daily 15mg lansoprazole versus alternate days of 30mg, Outcome 02 Relative risk of relapse (symptoms)............................... Analysis 12.03. Comparison 12 One daily 15mg lansoprazole versus alternate days of 30mg, Outcome 03 Adverse effects, overall.................................. Analysis 13.01. Comparison 13 Double dose PPI versus healing dose PPI, Outcome 01 Relative risk of relapse (oesophagitis).................................. Analysis 13.02. Comparison 13 Double dose PPI versus healing dose PPI, Outcome 02 Relative risk of relapse (symptoms).................................. Analysis 14.01. Comparison 14 Healing dose PPI type 1 versus healing dose of PPI type 2, Outcome 01 Relative risk of relapse (oesophagitis)............................... Analysis 14.02. Comparison 14 Healing dose PPI type 1 versus healing dose of PPI type 2, Outcome 02 Relative risk of relapse (symptoms)................................ Analysis 15.01. Comparison 15 Overall adverse effects, Outcome 01 Relative risk of adverse effect PPI versus placebo Analysis 15.02. Comparison 15 Overall adverse effects, Outcome 02 Relative risk of adverse effect PPI versus H2RA Analysis 15.03. Comparison 15 Overall adverse effects, Outcome 03 Relative risk of adverse effect H2RA versus placebo 68 68 69 69 70 70 71 71 72 iii

Medical treatments for the maintenance therapy of reflux oesophagitis and endoscopic negative reflux disease (Review) Donnellan C, Sharma N, Preston C, Moayyedi P This record should be cited as: Donnellan C, Sharma N, Preston C, Moayyedi P. Medical treatments for the maintenance therapy of reflux oesophagitis and endoscopic negative reflux disease. Cochrane Database of Systematic Reviews 2004, Issue 4. Art. No.: CD003245. DOI: 10.1002/14651858.CD003245.pub2. This version first published online: 18 October 2004 in Issue 4, 2004. Date of most recent substantive amendment: 04 August 2004 A B S T R A C T Background Gastro-oesophageal reflux disease (GORD) - reflux of stomach contents +/- bile into the oesophagus causing symptoms such as heartburn and acid reflux - is a common relapsing and remitting disease which often requires long-term maintenance therapy. Patients with GORD may have oesophagitis (inflammation of the oesophagus) or a normal endoscopy (endoscopy negative reflux disease or ENRD). Objectives To assess the effects of continuous maintenance therapy in adults with GORD (both ENRD and healed oesophagitis). Search strategy We searched Cochrane Central Register of Controlled Trials (The Cochrane Library Issue 2, 2003), MEDLINE (1966 to 2003), EMBASE (1980 to 2003), CINAHL (1982-2003), and the National Research Register (Issue 2, 2003) and reference lists of articles. We also contacted manufacturers and researchers in the field. Selection criteria Randomised controlled studies comparing PPIs, H2RAs, prokinetics, sucralfate and combinations either in comparison to another treatment regimen or to placebo in adults with reflux oesophagitis and ENRD. Data collection and analysis One author extracted data from included trials and a second author carried out an unblinded check. Two authors independently assessed trial quality. Study authors were contacted for additional information. Main results Maintenance of patients with healed oesophagitis: For a healing dose of PPI (generally the standard dose given by the manufacturer) versus placebo, the relative risk (RR) for oesophagitis relapse was 0.26 (95% confidence interval (CI) 0.19 to 0.36); versus H2RAs the RR was 0.36 (95% CI 0.28 to 0.46) and versus maintenance PPIs the RR was 0.63 (95% CI 0.55 to 0.73). However overall adverse effects were also more common and headaches were more common when comparing healing PPIs to H2RAs. For a maintenance dose of PPI (half of the standard dose) versus placebo, the RR for oesophagitis relapse was 0.46 (95% CI 0.38 to 0.57) and versus H2RAs the RR was 0.57 (95% CI 0.47 to 0.69). Overall adverse effects were more common. H2RAs were of marginal help but beneficial for symptomatic relief. Prokinetics and sucralfate were also more effective than placebo. For ENRD patients: Limited data with one showed benefit for omeprazole 10 mg once daily over placebo (RR 0.4; 95% CI 0.29 to 0.53). 1

Authors conclusions The findings in this review support the long-term treatment of oesophagitis to prevent relapse, both endoscopically and symptomatically. Healing doses of PPIs are more effective than all other therapies, although there is an increase in overall adverse effects compared to placebo, and headache occurrence compared to H2RAs. H2RAs prevent relapse more effectively than placebo, demonstrating a role for PPI-intolerant patients. Prokinetics and sucralfate both show benefit over placebo, but the former is no longer licenced. There is only limited data for ENRD. B A C K G R O U N D Gastro-oesophageal reflux disease (GORD) is a common disease with an estimated prevalence of 20% in the adult American population (Sonnenberg 1999) and similar rates in Europe. GORD is primarily attributable to acid and bile refluxing through the lower oesophageal sphincter (Armstrong 1999) and produces symptoms of heartburn, acid regurgitation, and less commonly chest and abdominal pain. The main treatments available concentrate on reducing acid secretion by proton pump inhibitors (PPIs) and H2 receptor antagonists (H2RAs), increasing the pressure at the lower oesophageal sphincter, enhancing gastric emptying with prokinetics and protecting mucosa from acid damage with sucralfate. Endoscopic examination is a common investigation for patients with GORD (Navaratham 1998). When endoscopy is undertaken a majority of patients have normal findings (endoscopy negative reflux disease or ENRD) with the remainder having oesophageal inflammation (oesophagitis). The efficacy of medical therapy in acute management of oesophagitis has been the subject of other reviews and ENRD acute management has been evaluated in a previous Cochrane Review (van Pinxteren 2000). GORD is a chronic relapsing condition with significant impact on quality of life (Scott 1999). GORD symptoms recur rapidly on discontinuation of therapy therefore maintenance therapy is central to management of the condition (Williams 1997). There are currently only limited systematic reviews assessing maintenance therapy in the treatment of oesophagitis and none assessing ENRD. GORD affects a significant number of patients who present for treatment and place a burden on healthcare budgets (Chiba 1997). Primary and secondary care physicians are faced with a variety of treatment options. Randomised controlled trials suggest that PPIs are the most effective acute healing treatment (Earnest 1999) but the actual relative efficacy of each drug in the maintenance therapy for reflux oesophagitis and ENRD is uncertain. An accurate assessment of this would inform clinical decision making. Efficacy is not the only aspect of the decision making process as cost of therapy also needs to be considered (Dent 1999). Nevertheless, a more precise estimate of relative efficacy would help establish the most cost-effective therapy for maintenance of reflux oesophagitis and ENRD patients. This review is a systematic review of randomised controlled trials examining the efficacy of continuous PPIs, H2RAs, prokinetic therapy and sucralfate in the maintenance therapy of reflux oesophagitis and endoscopy negative reflux disease. While intermittent or on-demand therapy is increasingly being used, it is not covered by the scope of this review. O B J E C T I V E S (1) To assess the efficacy of PPIs (at different doses), H2RAs, prokinetic therapy, sucralfate and combinations compared to each other and placebo in preventing relapse of mucosal inflammation in patients with oesophagitis. (2) To assess the efficacy of PPIs (at different doses), H2RAs, prokinetic therapy, sucralfate and combinations compared to each other and placebo in and preventing relapse of symptoms in patients with ENRD and oesophagitis. (3) To compare the incidence of adverse effects associated with the different treatments. C R I T E R I A F O R C O N S I D E R I N G S T U D I E S F O R T H I S R E V I E W Types of studies Randomised controlled trials assessing maintenance therapy of reflux oesophagitis and ENRD. Those studies assessing patients with complicated reflux disease (i.e. those with oesophageal strictures or Barrett s oesophagus) were excluded. Types of participants All patients were adults who had had an endoscopy. In the case of ENRD, the main presenting symptoms were heartburn and/or acid reflux. Patients with reflux oesophagitis at endoscopy (mucosal inflammation and/or mucosal breaks of the oesophagus) required a repeat endoscopy to demonstrate healing after a course of therapy before entering into the maintenance phase of the trial. Any recognised grading of oesophagitis was acceptable and even if the oesophagitis was mild, patients were considered to have oesophagitis, rather than ENRD. 2

Types of intervention The tested drug had to fall within the following drug classes a- d, the comparison regimen was also one of a-e. (Combinations of these were admissible for either arm). (1) PPIs: esomeprazole, lansoprazole, omeprazole, pantoprazole, rabeprazole. We defined healing dose of a PPI as the manufacturers standard dose of drug (i.e. omeprazole 20mg, lansoprazole 30mg, pantoprazole 40mg and rabeprazole 20mg all once daily) whilst maintenance therapy was half this dose (i.e. omeprazole 10mg, lansoprazole 15mg, pantoprazole 20mg, rabeprazole 10mg). The exception was esomeprazole which is an S-enantiomer of omeprazole (a racemic mixture of S and R enantiomers). As esomeprazole 20mg should be at least as effective as omeprazole 20mg we defined a healing dose of esomeprazole as 20mg once daily and a maintenance dose as 10mg daily. This is half the manufacturers standard dose but this decision was taken to reduce heterogeneity between trials by keeping the same chemical entity at the same dose.) Double dose was defined as double the standard dose (i.e. omeprazole 40mg, esomeprazole 40mg). (2) H2RAs: cimetidine, famotidine, nizatidine, ranitidine (where results from different doses were pooled). (3) Prokinetic therapy: cisapride, domperidone, metoclopramide. (4) Sucralfate (5) Placebo +/- antacid +/- no treatment Patients must have had at least 12 weeks of continuous therapy. Types of outcome measures Trials were included if they reported evidence of assessment at 12 to 52 weeks. Primary outcomes varied according to the group evaluated. The primary outcomes in the reflux oesophagitis group were the proportion of patients in whom the oesophagitis had relapsed as well as the proportion of patients with significant symptoms. The primary outcome in the ENRD group was the proportion with significant symptoms. Secondary outcomes assessed quality of life (standard mean difference in scores) and adverse events/dropout rates (total number for each drug and individual symptoms of headache, diarrhoea and abdominal pain). S E A R C H M E T H O D S F O R I D E N T I F I C A T I O N O F S T U D I E S See: Cochrane Upper Gastrointestinal and Pancreatic Diseases Group methods used in reviews. We conducted a search to identify all published and unpublished randomised controlled trials (s). We searched the following electronic databases: the Cochrane Central Register of Controlled Trials (CENTRAL), The Cochrane Library (Issue 2, 2003), MEDLINE (1966 to 2003), EMBASE (1980 to 2003), CINAHL (1982-2003) and the National Research Register (Issue 2, 2003). We handsearched reference lists from trials selected by electronic searching to identify further relevant trials. We also handsearched published abstracts from conference proceedings from the United European Gastroenterology Week (published in Gut) and Digestive Disease Week (published in Gastroenterology). We designed the search strategies for the review by using a combination of subject headings specific to each database s syntax and text word phrases. The following search strategy was constructed by using a combination of subject headings and text words relating to the symptoms of gastro-oesophageal reflux disease (GORD) and pharmacological interventions used for alleviating symptoms of the disease. We applied the standard Cochrane search strategy filter for identifying randomised controlled trials to all searches. MEDLINE search strategy exp esophagus/ esophag$.tw. oesophag$.tw. exp gastroesophageal reflux/ gastroesophageal adj3 reflux.tw. gastro adj3 oesophageal adj3 reflux.tw. gastro adj3 esophageal adj3 reflux.tw. gord.tw. gerd.tw. exp duodenogastric reflux exp bile reflux acid adj3 reflux.tw. gastric adj3 acid adj3 secret$.tw. stomach adj3 acid secret$.tw. gastric adj3 eros$.tw. stomach adj3 eros$.tw. exp heartburn heartburn.tw. or indigestion.tw. exp esophagitis esophagitis.tw. oesophagitis.tw. Endoscopy/ or Endoscopy, digestive system/ or Endoscopy, gastrointestinal/ endoscopy adj3 negative adj3 reflux.tw. enrd.tw. low$ adj6 sphincter$ adj3 pressur$.tw. les.tw. exp gastric emptying/ exp gastroparesis/ exp gastritis/ gastr$ adj3 empt$ adj3 disorder$.tw. stomach adj3 empt$ adj3 disorder.tw. 3

or/30-60 exp proton pump/ proton adj3 pump adj3 inhibitor$.tw. ppi.tw. exp omeprazole/ omeprazole.tw. lansoprazole.tw. pantoprazole.tw. rabeprazole.tw. esomeprazole.tw exp histamine h2 antagonists/ cimetidine.tw. exp cimetidine famotidine.tw. exp famotidine nizatidine.tw. exp nizatidine ranitidine.tw. exp ranitidine esomeprazole.tw. prokinetic adj3 agent$.tw. exp domperidone domperidone.tw. exp metoclopramide metoclopramide.tw. exp sucralfate/ sucralfate.tw. exp anti ulcer agents anti adj3 ulcer adj3 agent$.tw. antiulcer adj3 agent$.tw. exp antacids antacid$.tw. aluminium adj1 hydroxide.tw. exp magnesium hydroxide magnesium adj1 hydroxide.tw. exp magnesium oxide magnesium adj1 oxide.tw. exp calcium carbonate calcium adj1 carbonate.tw. exp Magnesium Silicate magnesium adj1 silicate.tw. exp bismuth bismuth.tw. exp carbenoxolone carbenoxolone.tw. exp misoprostol misoprostol.tw. mucosa$ adj3 protect$ adj3 agent$.tw. hydrotalcite.tw. algicon.tw. alginates.tw. gaviscon.tw. maalox.tw. magnesium adj1 trisilicates.tw. sodium adj1 bicarbonate.tw. sodium adj1 carbonate.tw. 116 or/62-115 117 61 and 116 118 117 and 29 In addition, we contacted members of the Cochrane UGPD Group and pharmaceutical companies with an interest in gastroenterology and asked them to supply details of any outstanding clinical trials and relevant unpublished materials. Bibliographies from trials selected by electronic searching were then handsearched to identify further relevant trials. M E T H O D S O F T H E R E V I E W Selection of studies The lead reviewer screened titles and trial abstracts that had been identified by the search strategy. A sample of 200 of these was then assessed independently. Two reviewers then screened the full article of selected trials to confirm eligibility, using pre-designed eligibility forms to assess quality. An adjudicator assessed any discrepancies. Data extraction One of the reviewers extracted data. A second reviewer checked these unblinded. We recorded the following features: - Setting: primary or secondary care - country of origin - method of randomisation: true versus pseudo, stated versus not stated - adequacy of allocation concealment: stated versus not stated - details of blinding of patients and outcome assessors: stated versus not stated, masked versus not masked - inclusion and exclusion criteria used - baseline comparability between treatment groups - treatments compared and number of patients in each arm - dropouts reported and their reasons - dose of drugs - grading system for oesophagitis (e.g. Savary-Miller, Los Angeles) - validated GORD questionnaires - adverse events: the total number reported and individual symptoms of headache, abdominal pain and diarrhoea Data analysis All analysis was on intention to treat data and we considered reflux oesophagitis and ENRD patients separately. The main outcomes being assessed were oesophagitis relapse (reflux oesophagitis patients), symptom control (reflux oesophagitis and ENRD) and adverse effects. We had initially planned to assess these at 3, 6 and 12 months, but the individual s provided data at different timepoints, therefore the final time point was used (with a maximum of 52 weeks) as this gave us information with the 4

longest follow-up. We also collected data on quality of life on the different interventions. We compared interventions, using placebo and combinations as described previously. We expressed the impacts of interventions as relative risks together with 95% confidence intervals. Analysis was repeated to obtain figures for risk difference and we calculated numbers of patients needed to treat (or harm) by 1/(risk difference) and expressed with 95% confidence intervals. Metaanalysis was only attempted if there were sufficient trials of similar comparisons reporting the same outcomes. Healing of oesophagitis was dichotomised into healed versus not healed and symptoms were dichotomised into no/minimal symptoms and significant symptoms. Results were synthesised using the relative risk of relapse of oesophagitis or symptoms, expressed with a 95% confidence interval. If significant heterogeneity was detected (p < 0.15), we investigated possible explanations by assessing the importance of the pre-defined factors listed in the protocol, using a random effects model. The following features were prospectively evaluated, by metaregression, to assess their effect on any observed outcomes as a cause for any heterogeneity: (1) Multi-centre versus single centre (2) Country of origin (3) Mean age of patients included in the study (4) Method of randomisation (5) Method of concealment of allocation (6) Masking versus no masking (7) Dose of drug (8) Different drugs in a specific class (e.g. omeprazole versus lansoprazole) (9) Duration of treatment (10) Severity of reflux oesophagitis included at study entry in the reflux oesophagitis group. The influence of the proportions of mild/ moderate and severe reflux oesophagitis in each study on the outcome was assessed. Mild/moderate was defined as Savary- Miller grade 1-2 or Los Angeles grade A-B, whilst severe was considered equivalent to Savary-Miller grade 3-5 or Los Angeles grade C-D. (11) Proportion Helicobacter pylori positive patients included in study (if known). (12) Patients allowed to have acid suppression or prokinetic therapy within four weeks of initial endoscopy (permitted versus not permitted). (13) Symptom severity at entry into the study. D E S C R I P T I O N O F S T U D I E S In total, we identified 9360 citations using the search strategy as outlined above. In all, 51 papers were included and 129 other trials were excluded as they did not meet the eligibility criteria or were abstracts of an already included trial. We were unable to extract data from a further five trials and correspondence with the authors is ongoing (these are all abstracts published between 1982 and 1997 which have never been published in full). All papers bar one, recruited participants from out-patients or open-access endoscopy lists. One paper provided data on participants randomised in Asia, 11 in America/Canada and the rest in Western Europe (including one multi-centre study carried out in both Europe and Australia). Some of the papers contained more than one comparison and were therefore included in more than one analysis as follows: For maintenance therapy of oesophagitis (1) PPI healing dose versus placebo Nine s (Bate 1995 a; Birbara 2000; Caos 2000; Johnson 2001; Robinson 1996; Sontag 1996; Sontag 1997; Staerk-Laursen 1995; Vakil 2001) There were eight multi-centre trials. A mean of 254 participants were randomised, the smallest included 163 participants (Sontag 1996) and the largest 406 participants (Sontag 1997). Data for six trials showed that 690 participants (47%) of those randomised had originally had severe oesophagitis (as defined previously). (2) PPI healing dose versus H2RA Ten s (Angelini 1993; Annibale 1998; Dent 1994; Endo 2000; Gough 1996a; Hallerback 1994; Lundell 1991; Metz 2003; Richter 2003; Vigneri 1995) There were nine multi-centre trials. A mean of 258 participants were randomised, the smallest included 63 participants (Lundell 1991) and the largest 721 participants (Richter 2003). Data for six trials showed that 471 participants (32.1%) of those randomised had originally had severe oesophagitis. (3) PPI healing dose versus maintenance dose Twenty two s (Baldi 1996; Bate 1995 a; Birbara 2000; Caos 2000; Endo 2000; Escourrou 1999; Gough 1996a; Hallerback 1994; Hatlebakk 1997b; Jansen 1998; Johnson 2001; Kristensen 2003; Lauritsen 2003; Metz 2003; Plein 2000; Poynard 1995; Richter 2003; Robinson 1996; Sontag 1996; Staerk-Laursen 1995; Thjodleifsson 2000; Vakil 2001) There were 19 multi-centre trials. A mean of 348 participants were randomised, the smallest included 97 participants (Endo 2000) and the largest 1224 participants (Lauritsen 2003). Data for 13 trials showed that 1527 participants (28.9%) of those randomised had originally had severe oesophagitis. (4) PPI maintenance dose versus placebo Ten s (Bardhan 1998; Bate 1995 a; Birbara 2000; Caos 2000; Johnson 2001; Pilotto 2003; Robinson 1996; Sontag 1996; Staerk-Laursen 1995; Vakil 2001) There were nine multi-centre trials. 5

A mean of 225 participants were randomised, the smallest included 105 participants (Pilotto 2003), the largest 375 participants (Vakil 2001). Data for seven trials showed that 615 participants (43.3%) of those randomised had originally had severe oesophagitis. (5)PPI Maintenance dose versus H2RA Seven s (Bate 1998; Endo 2000; Festen 1999; Gough 1996a; Hallerback 1994; Metz 2003; Richter 2003) There were six multi-centre trials. A mean of 324 participants were randomised, the smallest included 97 participants (Endo 2000), the largest 721 participants (Richter 2003) Data for four trials showed that 318 participants (24.6%) of those randomised had originally had severe oesophagitis. (6) H2RA versus placebo Five s (Armstrong 1989; Hegarty 1997; Kaul 1986; Sherbaniuk 1984; Simon 1995) There were three multi-centre trials. A mean of 148 participants were randomised, the smallest included 24 participants (Kaul 1986), the largest 279 participants (Hegarty 1997). Data for three trials showed that 181 participants (34.5%) of those randomised had originally had severe oesophagitis. (7) Prokinetics versus placebo Six s (Blum 1993; Hatlebakk 1997a; Kimmig 1995; Mc- Dougall 1997; Toussaint 1991; Tytgat 1992) There were five multi-centre trials. A mean of 266 participants were randomised, the smallest included 42 patients (McDougall 1997) and the largest 535 participants (Hatlebakk 1997a). Data for all six trials showed that 197 participants (12.4%) of those randomised had originally had severe oesophagitis. (8) Sucralfate versus placebo Two s (Schotborgh 1989; Tytgat 1995 a) Both were multi-centre trials. Schotborgh 1989 included 26 participants Tytgat 1995 a included 184 participants Schotborgh 1989 did not present data on oesophagitis severity and Tytgat 1995 a did not randomise any patients with severe oesophagitis. (9) Lansoprazole 15 mg oral doseonce daily (OD) versus 30 mg alternate days Two s (Baldi 2002; Houcke 2000) Both were multi-centre trials Houcke 2000 included 52 participants Baldi 2002 included 131 participants Data for both trials showed that 24 participants (13.3%) of those randomised had originally had severe oesophagitis. (10) Double dose PPI versus Healing dose PPI Two s (Johnson 2001; Vakil 2001) Both were multi-centre trials Johnson 2001 included 318 participants Vakil 2001 included 375 participants Vakil 2001 did not present data on oesophagitis severity, for Johnson 2001 86 participants (27%) had originally had severe oesophagitis. (11) Healing dose PPI (omeprazole) versus 2nd healing dose PPI Two s (Baldi 1996; Carling 1998) compared healing doses of omeprazole with lansoprazole and one (Thjodleifsson 2000) compared omeprazole with rabeprazole. All were multi-centre trials with a mean of 466 participants who were randomised. The smallest included 243 participants (Thjodleifsson 2000) and the largest 906 participants (Baldi 1996). Data for all three trials showed that 425 participants (30.4%) of those randomised had originally had severe oesophagitis. (12) Maintenance therapy for longer than one year One (Thjodleifsson 2003) which was a follow-up to Thjodleifsson 2000 with follow-up at three years. (13) Other comparisons for oesophagitis maintenance In addition, one trial compared treatment with a five different therapies, including combination treatments (Vigneri 1995). There was also one trial which compared two doses of H2RA without placebo (Pace 1990). For maintenance therapy of ENRD The number of studies assessing patients with ENRD was small and they examined different drug therapies. These studies have therefore been described, rather than formal analysis being carried out. The three s were: Bate 1998, comparing omeprazole 10 mg and cimetidine 800 mg nocte (participants with both oesophagitis (81%) and ENRD (19%) but data combined, so presented in the oesophagitis section). Hatlebakk 1997a, comparing cisapride and placebo in 535 participants (with both oesophagitis and ENRD, but data presented separately) which was a multi-centre trial. Venables 1997, comparing omeprazole 10 mg and placebo in 495 participants which was also a multi-centre trial (ENRD patients only). M E T H O D O L O G I C A L Q U A L I T Y Two reviewers independently undertook an assessment of the quality of each eligible study. of randomisation, concealment, and masking were assessed. Sixteen trials stated the method of randomisation (Baldi 2002; Bardhan 1998; Blum 1993; Carling 1998; Escourrou 1999; Festen 1999; Lundell 1991; 6

Plein 2000; Sherbaniuk 1984; Staerk-Laursen 1995; Thjodleifsson 2000; Thjodleifsson 2003; Toussaint 1991; Tytgat 1992; Vakil 2001 and Vigneri 1995) and six (Baldi 2002; Bardhan 1998; Lauritsen 2003; Metz 2003; Robinson 1996; Vakil 2001) reported the method of concealment. All papers from which provided data to this review were randomised controlled trials of good quality and the majority had similar dropout rates between the 2 comparative groups. R E S U L T S For maintenance therapy of oesophagitis (1) PPI healing dose versus placebo (a) Maintenance of healed oesophagitis Nine s (Bate 1995 a; Birbara 2000; Caos 2000; Johnson 2001; Robinson 1996; Sontag 1996; Sontag 1997; Staerk-Laursen 1995; Vakil 2001) reported a dichotomous outcome for healing dose of PPI versus placebo in the maintenance therapy of oesophagitis; evaluating 1385 participants between 26 and 52 weeks. Overall 21.7% of participants relapsed in the group taking a healing dose of PPIs, compared to 78.8% in the placebo group. There was statistically significant heterogeneity between the trial results (chi squared 37.84, df = 8, p < 0.00001). Meta-regression exploring type of PPI, duration of follow-up, age, country, method of masking, randomisation and concealment of allocation failed to explain this heterogeneity. There was a significant benefit of taking healing dose PPI therapy compared to placebo to maintain remission of oesophagitis (relative risk of relapse (RR) 0.26; 95% confidence interval (CI) 0.19 to 0.36 with a number needed to treat to gain benefit (NNT) of 1.7 (95% CI 1.6 to 1.8). (b) Maintenance of symptom relief Nine s (Bate 1995 a; Birbara 2000; Caos 2000; Johnson 2001; Robinson 1996; Sontag 1996; Sontag 1997; Staerk-Laursen 1995; Vakil 2001) reported a dichotomous outcome for symptom relief, evaluating 1334 participants between 26 and 52 weeks. Three trials (Caos 2000; Robinson 1996; Vakil 2001) presented results in terms of heartburn only, three trials (Birbara 2000; Sontag 1996; Sontag 1997) in terms of global symptom reflux scores (GSRS) and three trials (Bate 1995a; Johnson 2001; Staerk-Laursen 1995) with both (where heartburn was used in preference to GSRS). Overall 29.4% had had significant symptoms in the group taking a healing dose of PPI compared to 76.3% in the placebo group. There was statistically significant heterogeneity between trial results (chi squared 46.29, df = 8, p < 0.00001). However as no cause had been found for the heterogeneity in assessment of oesophagitis, this was not explored further. There was a significant benefit of taking healing dose PPI therapy compared to placebo to maintain remission of symptoms (RR of relapse 0.34; 95% CI 0.25 to 0.46) with a NNT of 2.0 (95% CI 1.7 to 2.7). (c) Overall adverse effects Five trials (Bate 1995 a; Birbara 2000; Caos 2000; Johnson 2001; Vakil 2001) presented data for overall adverse effects on 806 participants. In total, 54.1% of participants taking a healing dose of PPI were reported to experience at least one side effect, compared to 41.6% in the placebo group. There was no significant heterogeneity between the results (chi squared 4.94, df = 4, p = 0.29). There was significant harm in taking a healing dose of PPI compared to a placebo (RR 1.32; 95% CI 1.15 to 1.52) with the number of participants needed to treat before an adverse effect occurred (number needed to harm or NNH) of 7.7 (95% CI 5.3 to 14.3). (d) Headache Two trials (Johnson 2001; Vakil 2001) presented data for headache occurrence on 348 participants. Overall 3.4% of participants taking a healing dose of PPI were reported to experience a headache, compared to 3.0% in the placebo group. There was no significant heterogeneity in the results (chi squared 0.26, df = 1, p = 0.61). Headache occurrence was not significantly different between the two groups (RR 1.13; 95% CI 0.35 to 3.62). (e) Diarrhoea Two trials (Robinson 1996; Vakil 2001) presented data for diarrhoea occurrence on 305 participants. Overall 5.2% of participants taking a healing dose of PPI were reported to experience diarrhoea, compared to 2.0% in the placebo group. There was significant heterogeneity in the results (chi squared 2.51, df = 1, p = 0.11) but this was not explored due to the small number of trials. Diarrhoea occurrence was not significantly different between the two groups (RR 2.52; 95% CI 0.2 to 31.8). (f) Abdominal pain One trial (Vakil 2001) presented data for occurrence of abdominal pain on 190 participants. Overall 3.1 % of participants taking a healing dose of PPI were reported to experience abdominal pain, compared to 2.2% of participants in the placebo group. This difference was not statistically significant (RR 1.41; 95% CI 0.24 to 8.24). (2) PPI healing dose versus H2RA (a) Maintenance of healed oesophagitis Ten s (Angelini 1993; Annibale 1998; Dent 1994; Endo 2000; Gough 1996a; Hallerback 1994; Lundell 1991; Metz 2003; Richter 2003; Vigneri 1995) reported a dichotomous outcome for healing dose of PPI versus H2RA in the maintenance therapy of oesophagitis, evaluating 1583 participants between 24 and 52 weeks. Overall 22.5% of participants relapsed in the group taking a healing dose of PPIs, compared to 58.4% in the H2RA group. There was statistically significant heterogeneity between the trial results (chi squared 23.91, df = 9, p = 0.004). Metaregression ex- 7

ploring type of PPI, dose of H2RA, method of masking, randomisation and concealment of allocation, duration of follow-up and country of origin did not explain this heterogeneity. Removing Dent 1994 and Endo 2000, both of which showed a greater effect of PPIs over H2RAs, reduced the heterogeneity (so that chi squared 11.5, df = 7, p = 0.13). There was a significant benefit of taking a healing dose of PPI compared to H2RAs to maintain remission of oesophagitis (RR of relapse 0.36; 95% CI 0.28 to 0.46) with a NNT of 2.5 (95% CI 2.0 to 3.4). (b) Maintenance of symptom relief Five s (Annibale 1998; Dent 1994; Gough 1996a; Hallerback 1994; Vigneri 1995) reported a dichotomous outcome for symptom relief evaluating 797 participants between 26 and 52 weeks. Two trials (Dent 1994; Gough 1996a) presented results in terms of heartburn only, two trials (Hallerback 1994; Vigneri 1995) in terms of GSRS and one trial (Annibale 1998) with both (where heartburn was used in preference to GSRS). Overall 21.6% had had significant symptoms in the group taking a healing dose of PPI compared to 44.3% in the H2RA group. There was no statistically significant heterogeneity between trial results (chi squared 5.01, df = 4, p = 0.29). There was a significant benefit of taking a healing dose of PPI therapy compared to H2RA therapy to maintain remission of symptoms (RR of relapse 0.48; 95% CI 0.39 to 0.60) with a NNT of 4.3 (95% CI 3.4 to 5.9). (c) Overall adverse effects Three trials (Annibale 1998; Gough 1996a; Lundell 1991) presented data for overall adverse effects on 469 participants between 26 and 52 weeks. In total 18.9% of participants taking a healing dose of PPI were reported to experience at least one side effect, compared to 15% in the H2RA group. There was no significant heterogeneity between the results (chi squared 2.75, df = 2, p = 0.25). There was no significant harm in taking a healing dose of PPI compared to H2RA therapy (RR 1.26; 95% CI 0.89 to 1.80). (d) Headache One trial (Metz 2003) presented data for headache occurrence on 189 participants. Overall 19.1% of participants taking a healing dose of PPI were reported to experience a headache, compared to 8.4% in the H2RA group. This represented a significant difference in headache occurrence between the two groups (RR 2.27; 95% CI 1.04 to 4.97) with a NNH of 9.1 (95% CI 5 to 100). (e) Diarrhoea No trials in this group presented extractable data for diarrhoea occurrence. (f) Abdominal pain One trial (Hallerback 1994) presented data for occurrence of abdominal pain on 259 patients. Overall 9.2% of patients taking a healing dose of PPI were reported to experience abdominal pain, compared to 10.2% of patients in the H2RA group. This difference was not statistically different (RR 0.90; 95% CI 0.43 to 1.90). (3) PPI healing dose versus maintenance dose (a) Maintenance of healed oesophagitis Twenty two s (Baldi 1996; Bate 1995 a; Birbara 2000; Caos 2000; Endo 2000; Escourrou 1999; Gough 1996a; Hallerback 1994; Hatlebakk 1997b; Jansen 1998; Johnson 2001; Kristensen 2003; Lauritsen 2003; Metz 2003; Plein 2000; Poynard 1995; Richter 2003; Robinson 1996; Sontag 1996; Staerk-Laursen 1995; Thjodleifsson 2000; Vakil 2001) reported a dichotomous outcome for oesophagitis, comparing a healing dose of PPI versus a maintenance dose, in the maintenance therapy of oesophagitis, evaluating 5964 participants between 24 and 52 weeks. Overall 17.5% of participants relapsed in the group taking a healing dose of PPIs, compared to 29.1% in the group taking a maintenance dose. There was statistically significant heterogeneity between the trial results (chi squared 37.25, df = 21, p = 0.02). Meta-regression exploring number of centres, country, age, duration of follow-up, method of concealment of allocation, randomisation or masking, type of PPI or severity of oesophagitis failed to explain this heterogeneity. If data were removed for Johnson 2001 (which showed a greater effect of healing dose PPI over maintenance dose) and Sontag 1996 (which showed no significant difference between the two doses), there was no significant heterogeneity (chi squared 16.83 (df = 19), p = 0.60). There was a significant benefit of taking a healing dose of PPI compared to a maintenance dose to maintain remission of oesophagitis (RR of relapse 0.63; 95% CI 0.55 to 0.73) with a NNT of 9.1 (95% CI 6.7 to 14.3). (b) Maintenance of symptom relief Eighteen s (Baldi 1996; Bate 1995 a; Birbara 2000; Caos 2000; Escourrou 1999; Gough 1996a; Hallerback 1994; Hatlebakk 1997b; Jansen 1998; Johnson 2001; Kristensen 2003; Lauritsen 2003; Plein 2000; Robinson 1996; Sontag 1996; Staerk- Laursen 1995; Thjodleifsson 2000; Vakil 2001) reported a dichotomous outcome for symptom relief, comparing a healing and maintenance dose of PPI. This evaluated 5116 participants over a period of between 26 and 52 weeks. Seven trials (Caos 2000; Gough 1996a; Jansen 1998; Kristensen 2003; Lauritsen 2003; Robinson 1996; Vakil 2001) presented results in terms of heartburn only, eight trials (Baldi 1996; Birbara 2000; Escourrou 1999; Hallerback 1994; Hatlebakk 1997b; Plein 2000; Sontag 1996; Thjodleifsson 2000) in terms of GSRS and three trials (Bate 1995 a; Johnson 2001; Staerk-Laursen 1995) with both (where heartburn was used in preference to GSRS). Overall 30.7% had significant symptoms in the group taking a healing dose of PPI compared to 35.8% in the group taking a maintenance dose of PPI. There was statistically significant heterogeneity between trial results (chi squared 33.65, df = 17, p = 0.009). However as no cause 8

had been found for the heterogeneity in assessment of oesophagitis, this was not explored further. There was a significant benefit of taking a healing dose of PPI compared to a maintenance dose to maintain remission of symptoms (RR of relapse 0.78; 95% CI 0.68 to 0.88) with a NNT of 12.5 (95% CI 8.3 to 25). (c) Overall adverse effects Ten trials (Baldi 1996; Bate 1995 a; Birbara 2000; Caos 2000; Escourrou 1999; Gough 1996a; Hatlebakk 1997b; Johnson 2001; Plein 2000; Vakil 2001) presented data for overall adverse effects on 2812 participants. In total, 41.5% of participants taking a healing dose of PPI were reported to experience at least one side effect, compared to 41.4% in the maintenance group. There was no significant heterogeneity between the results (chi squared 2.81, df = 9, p = 0.97) and no significant difference in adverse effects between the two doses (RR 1.01; 95% CI 0.93 to 1.09). (d) Headache Four trials (Johnson 2001; Lauritsen 2003; Metz 2003; Vakil 2001) presented data for headache occurrence on 1764 participants. Overall 5.6% of participants taking a healing dose of PPI were reported to experience a headache, compared to 4.3% in the maintenance PPI group. There was no significant heterogeneity in the results (chi squared 2.82, df = 3, p = 0.42) and no significant difference in headache occurrence between the two groups (RR 1.29; 95% CI 0.86 to 1.94). (e) Diarrhoea Four trials (Baldi 1996; Lauritsen 2003; Robinson 1996; Vakil 2001) presented data for diarrhoea occurrence on 2441 participants. Overall 5.8% of participants taking a healing dose of PPI were reported to experience diarrhoea, compared to 5.2% in the maintenance group. There was significant heterogeneity in the results (chi squared 5.81, df = 3, p = 0.12) but this was not explored further. There was no significant difference in diarrhoea occurrence between the two groups (RR 1.30; 95% CI 0.67 to 2.53). (f) Abdominal pain Four trials (Baldi 1996; Hallerback 1994; Lauritsen 2003; Vakil 2001) presented data for occurrence of abdominal pain on 2590 participants. Overall 4.1% of participants taking a healing dose of PPI were reported to experience abdominal pain, compared to 3.7% of participants in the maintenance group. There was no significant heterogeneity in the results (chi squared 2.29, df = 3, p = 0.51) and this difference was not statistically significant (RR 1.12; 95% CI 0.76 to 1.64). (4) PPI maintenance dose versus placebo (a) Maintenance of healed oesophagitis Ten s (Bardhan 1998; Bate 1995 a; Birbara 2000; Caos 2000; Johnson 2001; Pilotto 2003; Robinson 1996; Sontag 1996; Staerk-Laursen 1995; Vakil 2001) reported a dichotomous outcome for a maintenance dose of PPI versus placebo in the maintenance therapy of oesophagitis, evaluating 1465 participants between 26 and 52 weeks. Overall 36.1% of participants relapsed in the group taking a maintenance dose of PPI, compared to 75.4% in the placebo group. There was statistically significant heterogeneity between the trial results (chi squared 36.36, df = 9, p < 0.0001) but metaregression exploring type of PPI, duration of follow-up, age, country, method of masking, randomisation and concealment of allocation failed to explain this heterogeneity. There was a significant benefit of taking a maintenance dose of PPI compared to placebo to maintain remission of oesophagitis (RR of relapse 0.46; 95% CI 0.38 to 0.57) with a NNT of 2.4 (95% CI 2.1 to 2.9). (b) Maintenance of symptom relief Ten s (Bardhan 1998; Bate 1995 a; Birbara 2000; Caos 2000; Johnson 2001; Pilotto 2003; Robinson 1996; Sontag 1996; Staerk-Laursen 1995; Vakil 2001) reported a dichotomous outcome for symptom relief evaluating 1426 participants between 24 and 52 weeks. Three trials (Caos 2000; Robinson 1996; Vakil 2001) presented results in terms of heartburn only, three trials (Bardhan 1998; Birbara 2000; Sontag 1996) in terms of GSRS and four trials (Bate 1995 a; Johnson 2001; Pilotto 2003; Staerk- Laursen 1995) with both (where heartburn was used in preference to GSRS). Overall 44.4% had significant symptoms in the group taking a maintenance PPI compared to 73.4% in the placebo group. There was statistically significant heterogeneity between trial results (chi squared 69.02, df = 9, p < 0.00001). However as no cause had been found for the heterogeneity in assessment of oesophagitis, this was not explored further. There was a significant benefit of taking a maintenance dose of PPI compared to placebo to maintain remission of symptoms (RR of relapse 0.54; 95% CI 0.42 to 0.69) with a NNT of 3.0 (95% CI 2.2 to 4.8). (c) Overall adverse effects Six trials (Bate 1995 a; Bardhan 1998; Birbara 2000; Caos 2000; Johnson 2001; Vakil 2001) presented data for overall adverse effects on 1057 participants. Overall 42.4% of participants taking maintenance PPI therapy were reported to experience at least one side effect, compared to 33.9% in the placebo group. There was no significant heterogeneity between the results (chi squared 6.36, df = 5, p = 0.27) and there was significant harm in taking maintenance PPI therapy compared to placebo (RR 1.25; 95% CI 1.09 to 1.43) with a NNH of 11.1 (95% CI 7.1 to 25). (d) Headache Two trials (Johnson 2001; Vakil 2001) presented data for headache occurrence on 336 participants. Overall 4.2% of participants taking maintenance PPI therapy were reported to experience a headache, compared to 3.0% in the placebo group. There was no 9