Efficacy of omeprazole in functional dyspepsia: double-blind, randomized, placebo-controlled trials (the Bond and Opera studies)

Authors


Talley Department of Medicine, University of Sydney, Nepean Hospital, Penrith, NSW 2751, Australia. E-mail: ntalley@blackburn.med.usyd.edu.au

Abstract

Background:

The efficacy of H2-receptor antagonists in functional dyspepsia is equivocal and the therapeutic place of proton pump inhibitors in functional dyspepsia is unknown.

Aim:

To evaluate the efficacy of proton pump inhibitor therapy in functional dyspepsia.

Methods:

Patients (n = 1262) with a clinical diagnosis of functional dyspepsia (persistent or recurrent epigastric pain or discomfort for at least 1 month and a normal upper gastrointestinal endoscopy) were randomized to receive omeprazole 20 mg, 10 mg or identical placebo, for 4 weeks. Symptoms were assessed using validated measures. Helicobacter pylori status was determined pre-entry by a 13C-urea breath test.

Results:

On an intention-to-treat analysis (n=1248), complete symptom relief was observed in 38% on omeprazole 20 mg, compared with 36% on omeprazole 10 mg and 28% on placebo (P = 0.002 and 0.02, respectively). Among those with ulcer-like and reflux-like dyspepsia, complete symptom relief was achieved in 40% and 54% on omeprazole 20 mg, and 35% and 45% on omeprazole 10 mg, respectively, compared with 27% and 23% on placebo (all P < 0.05, except omeprazole 10 mg in ulcer-like dyspepsia, P = 0.08). There was no significant benefit of omeprazole over placebo in dysmotility-like dyspepsia. Symptom relief was similar in H. pylori-positive and negative cases.

Conclusions:

Omeprazole is modestly superior to placebo in functional dyspepsia at standard (20 mg) and low doses (10 mg) but not in patients with dysmotility-like dyspepsia.

INTRODUCTION

Pain or discomfort in the upper abdomen is a common symptom complex in the general population; approximately 25% of adults in the community report such complaints and although only a minority see a doctor for dyspepsia in the UK and USA, it accounts for up to 5% of consultations in family practice.1[2][3]–4 Less than half of these patients when appropriately investigated have an underlying structural explanation for the symptoms such as peptic ulceration or reflux oesophagitis; the remainder are classified currently as having functional or non-ulcer dyspepsia.1, 5, 6 Functional dyspepsia is an important and costly entity because it is a significant cause of morbidity and time lost from work.7

The pathophysiology of functional dyspepsia remains inadequately understood. Between 30% and 60% of patients with functional dyspepsia have Helicobacter pylori gastritis but whether this is an incidental infection in these patients remains unclear.1, 8, 9 Basal and peak acid outputs do not differ in patients with functional dyspepsia compared with controls,10 but the acid response to gastrin-releasing peptide, which is considered to be a reflection of the postprandial state, may be abnormal in up to 50% of H. pylori infected patients with functional dyspepsia.11 Gastroduodenal sensation is disturbed in a subset12 but it is not clear whether the mucosa is more acid sensitive in functional dyspepsia.13

Standard treatment for functional dyspepsia has included H2-receptor antagonists;12, 14 the majority of patients in the community on long-term H2-receptor antagonists for dyspepsia are taking this medication for the management of functional dyspepsia rather than peptic ulcer disease.1 However, the efficacy of this class of compounds in functional dyspepsia is controversial.12, 14 Whether more effective acid suppression with proton pump inhibitors is efficacious in functional dyspepsia has not been adequately tested. As H. pylori appears to be linked to acid dysregulation in functional dyspepsia,11 it is conceivable that infected patients would have a better response to acid suppression but this has not been tested either.

We aimed to evaluate, in two identically designed randomized controlled trials (the Bond [Based on omeprazole in nonulcer dyspepsia] and Opera [Omeprazole and placebo—effect on relieving abdominal pain/discomfort] studies), the efficacy of the proton pump inhibitor omeprazole compared with placebo, all given once daily in the morning. We hypothesized that omeprazole at a standard dose of 20 mg would, compared with placebo, provide complete relief of epigastric pain and discomfort in twice as many patients after a 4 week course of treatment. A secondary aim was to determine whether the presence or absence of H. pylori infection influenced the symptom response to omeprazole. We hypothesized that the symptom relief would be greater in H. pylori infected patients with functional dyspepsia. Finally, we aimed to evaluate whether subgroups of dyspepsia based on ranking of the most bothersome complaint would identify responders to acid suppression. We hypothesized that those with predominant epigastric pain rather than discomfort would best respond to proton pump inhibitor therapy.

METHODS

Study population

Investigators were instructed to enrol consecutive patients presenting with a clinical diagnosis of functional dyspepsia into two identical studies. The trials were approved by the local ethics committee in each centre, and informed consent was obtained from all patients. Both general practitioners and specialist gastroenterologists recruited patients from their practices.

Patients with persistent or recurrent epigastric pain and/or epigastric discomfort that was experienced on at least one of the 3 days immediately prior to randomization were eligible to be enrolled.1 Discomfort was defined as a subjective, negative feeling that did not reach the level of pain according to the patient and which included symptoms such as postprandial fullness, bloating or early satiety.1 Patients were also required to have at least a 1 month history of dyspeptic symptoms and symptoms had to occur on a minimum of 25% of days during that month.

Patients underwent upper endoscopy in the period 5–14 days before the inclusion visit. Those with any erosive change in the oesophagus or duodenum, oesophageal strictures, Barrett’s oesophagus, duodenal deformity or chronic gastric or duodenal ulcer were excluded. Patients with five or fewer gastric erosions were considered eligible for entry if they fulfilled the other inclusion criteria, because there is no convincing evidence that gastric erosions cause a distinct symptomatic entity.1, 5, 6 Patients with a previous history of documented peptic ulcer disease by endoscopy or radiology, or a past history of gastro-oesophageal reflux disease documented by endoscopy or 24 h oesophageal pH monitoring, were excluded. Furthermore, patients with classical heartburn or acid regurgitation as their only symptom without an epigastric component were not considered eligible for inclusion in the study, so as to minimize including patients with undiagnosed gastro-oesophageal reflux disease.1

The presence of any alarm symptoms (including unintentional weight loss, vomiting, dysphagia, haematemesis, melaena, fever, jaundice or other symptoms or signs suggesting serious or malignant disease) led to exclusion. Those patients who were clinically diagnosed as having irritable bowel syndrome were also not eligible; this was defined conservatively as the presence of two or more of the six standard Manning symptom criteria (pain relief by defecation, more frequent stools at the onset of pain, looser stools with the onset of pain, visible abdominal distension, rectal passage of mucus or a sensation of incomplete rectal evacuation).15, 16 Similarly, patients with chronic severe constipation were excluded as it was considered conceivable that this group may have their upper abdominal pain or discomfort associated with their constipation. Patients who had been treated with bismuth-containing compounds, prokinetics or ulcer-healing doses of antisecretory agents for more than 7 days in the month prior to endoscopy were also considered ineligible. The remaining exclusion criteria were presence of significant medical disease that would complicate the evaluation of outcome (e.g. unstable diabetes mellitus or malignancy), pregnancy or lactation, alcohol or drug abuse, and age below 18 or above 80 years.

Assignment

Patients entered a parallel group, double-blind, randomized, placebo-controlled trial. In the period 1 week before endoscopy and in the period between endoscopy and randomization, only antacid use on an as needed basis was allowed (in order to avoid the contaminating effects of other drugs on treatment outcome). Randomization of eligible patients was performed in the proportions 1:1:1 according to a computer-generated randomization list. This was carried out within blocks of three consecutive patients. Patients either received omeprazole 20 mg, omeprazole 10 mg or placebo, all given once daily in the morning. The patients were treated with medication for 4 weeks (±4 days) with the first doses taken the morning following the day after randomization.

Determination of H. pylori status

At baseline, a standardized 13C-urea breath test was obtained using a well validated protocol.17 An increase in the patient’s breath 13C of more than 5 p.p.m. over the baseline was considered positive for H. pylori infection.17

Blinding

Patients, investigators and study centres maintained strict blinding throughout the study. The centre recruitment codes were placed in opaque envelopes. They were all returned and none had been opened during the trial. The omeprazole and placebo capsules were identical in appearance.

Compliance

Patient compliance was checked by counting the returned study medication. A priori, non-compliance was defined as an intake of less than 75% of the medication during the study period. The administration of all drugs during the study period was recorded.

Symptom and quality of life assessment

This was undertaken at baseline and at the end of the trial, at the 4 week visit.

Symptoms.

The primary outcome assessment was measurement of gastrointestinal symptoms. Epigastric pain and/or discomfort during the 3 days prior to the first and last visits in the trial was recorded by the physician. Symptoms were graded by the investigator at interview on a four-point Likert scale as follows: 0(none), no symptoms; 1(mild), awareness of the symptom but easily tolerated; 2(moderate), symptoms sufficient to cause an interference with normal activities; 3(severe), incapacitating symptoms with an inability to perform normal activities. This scale has been shown to be responsive and valid in reflux disease18 and functional dyspepsia.19, 20 Complete absence of epigastric pain and discomfort on each of the 3 days was the primary end-point.

At the final visit, the patient was interviewed by a physician and asked whether the study medication had provided sufficient control of symptoms, to estimate global patient satisfaction with the treatment.

At the first visit, the patient also ranked his or her three most bothersome symptoms from the following choices: epigastric pain, heartburn, acid regurgitation, bloating, belching, rectal flatus, nausea, early satiety, and postprandial fullness, as defined by the Rome criteria.1 The patients were then subdivided into the following a priori symptom subgroups based on symptom predominance:

Ulcer-like dyspepsia—predominant epigastric pain.

Dysmotility-like dyspepsia—predominant discomfort (postprandial fullness, early satiety, bloating or belching).

Reflux-like dyspepsia—predominant reflux symptoms (heartburn or acid regurgitation).

Other—predominant nausea or rectal flatus.

Patients recorded their epigastric pain or discomfort on daily diary cards during the 4 week treatment period, where symptoms were self-rated as present or absent.

Quality of life.

Quality of life was assessed using two standard and validated self-report questionnaires:

(a) The Psychological General Well Being Index (PGWB). This measures subjective well-being or distress.21, 22 It includes 22 items which can be combined into a global score that ranges from a maximum of 132 to a minimum of 22. Six-point Likert scales comprise the response format, with higher values denoting better well-being. The PGWB has been applied in studies of dyspepsia and gastro-oesophageal reflux disease; there are extensive normative data available.21, 22

(b) The Gastrointestinal Symptom Rating Scale (GSRS). This has 15 items that measure gastrointestinal symptoms on seven-point Likert scales over the prior 2 weeks, and can be combined into a total score.21, 22 The lower the score the better the symptom status. The GSRS is valid and responsive, and substantial normative data are available.21, 22

Statistical analysis

The primary efficacy variable, i.e. complete relief of epigastric pain/discomfort, was defined as having no symptoms during the last 3 days of the 4 weeks of treatment. The proportion of patients with complete relief of epigastric pain/discomfort was compared between omeprazole 20 mg and 10 mg vs. placebo using both an intention-to-treat (ITT) and a per protocol (PP) approach. Secondary efficacy variables were analysed using an ITT approach only.

The number needed to treat (the number of patients who need to be treated to prevent a poor outcome) and the relative risk reduction (the proportional reduction in event rates between control and experimental patients) were calculated.

A Mantel–Haenzel chi-squared test with stratification based on countries was used to analyse the difference between the two treatment groups with respect to the primary efficacy variable. The significance level was adjusted for two comparisons using the Bonferroni inequality. Confidence intervals were computed for the proportions and the differences between the treatment groups.

For the secondary efficacy variable, i.e. sufficient control of symptoms, confidence intervals were computed for the proportions and the differences between the treatment groups. The change in total score of the PGWB and mean item score of the GSRS from baseline to the 4-week visit were evaluated. Differences between treatments regarding the quality of life questionnaires were analysed with the baseline scores as a covariate.

RESULTS

The progress through the trials is summarized in Figure 1, which shows the number of eligible patients who were randomized, the numbers withdrawn, and the numbers who completed the trial in each arm.

Baseline comparisons and compliance

The three treatment groups were well balanced regarding all baseline characteristics in both trials (Table 1). Compliance was excellent; more than 90% of capsules were taken by 86, 87 and 83% of patients, respectively, in the omeprazole 20 mg, omeprazole 10 mg and placebo arms.

Relief of dyspepsia

In the intention-to-treat analysis, complete symptom relief was observed in 38.2% (161/421) on omeprazole 20 mg and 36.0% (146/405) on omeprazole 10 mg, compared with 28.2% (119/422) on placebo in the combined studies. The difference between omeprazole 20 mg and placebo, and between omeprazole 10 mg and placebo, was statistically significant (95% CI: 3.7–16.4% and 1.4–14.3%, respectively). The per protocol analysis yielded essentially the same results (Table 2). Comparing omeprazole 20 mg with placebo, the number needed to treat (NNT) was 10 (95% CI: 6–27) and the relative risk reduction was 14.0% (95% CI: 5.3–21.9%). Details from the individual studies are given in Table 2.

Sufficient control of symptoms by an intention-to-treat analysis was reported by 61% on omeprazole 20 mg (95% CI: 55.7–65.3%) compared with 59% on omeprazole 10 mg (95% CI: 54.3–64.1%) and 51% on placebo (95% CI: 46.3–56.0%) in the combined studies. Corresponding figures for the Bond study were 57% on omeprazole 20 mg (95% CI: 50.2–63.7%), 63% on omeprazole 10 mg (95% CI: 56.2–69.9%) and 44% on placebo (95% CI: 37.6–51.1%) and for the Opera study, 64% on omeprazole 20 mg (95% CI: 57.3–71.0%), 55% on omeprazole 10 mg (95% CI: 48.1–62.2%) and 59% on placebo (95% CI: 51.5–65.5%).

Omeprazole 20 mg and 10 mg provided a significantly higher proportion of symptom-free days (52.3 and 50.3%) than placebo (45.5%) over the treatment period in the combined studies. The difference between placebo and omeprazole 20 mg was 6.85% (95% CI: 2.64–11.07%) and for omeprazole 10 mg was 4.84% (95% CI: 0.59–9.10%). Symptoms improved to their maximum level by 7 days; the placebo response did not decrease over the 4 weeks of therapy.

Symptom subgroups

In patients on omeprazole 20 mg, omeprazole 10 mg and placebo, those with ulcer-like dyspepsia (708 patients) had complete symptom relief in 40, 35 and 27% of cases, respectively (P = 0.006 omeprazole 20 mg vs. placebo and P = 0.08 omeprazole 10 mg vs. placebo), while those with reflux-like dyspepsia (143 patients) had complete relief in 54, 45 and 23% of cases, respectively (P = 0.002 omeprazole 20 mg vs. placebo and P = 0.02 omeprazole 10 mg vs. placebo). In contrast, those with dysmotility-like dyspepsia (291 patients) had complete relief on omeprazole 20 mg, omeprazole 10 mg and placebo in 32, 37 and 31% of cases, respectively (P = 0.92 omeprazole 20 mg vs. placebo and P = 0.33 omeprazole 10 mg vs. placebo) and for patients with other symptoms (i.e. nausea, flatus) (106 patients) the corresponding figures were 25, 30 and 38%, respectively (P = 0.25 omeprazole 20 mg vs. placebo and P = 0.46 omeprazole 10 mg vs. placebo).

H. pylori

A total of 41% of patients (507/1222) were H. pylori-positive by the 13C-urea breath test. Complete relief of dyspepsia in the active treatment arms was not significantly different between the H. pylori infected and uninfected patients (Table 3).

Quality of life

The total GSRS score improved from 2.67 to 2.02 on omeprazole 20 mg, from 2.74 to 2.08 on omeprazole 10 mg and from 2.75 to 2.17 on placebo. The difference between omeprazole 20 mg and placebo was statistically significant (P = 0.02, with baseline GSRS as a covariate), but the difference between omeprazole 10 mg and placebo was not (P = 0.08).

The total PGWB score improved similarly in all groups by the last visit. There was no statistically significant difference in the total score or in the subscale scores between the three treatment arms (data not shown).

Treatment response in the Bond vs. Opera studies

The difference in treatment response between active treatment and placebo was greater in the Bond than in the Opera study (Table 4). This difference was statistically significant when comparing active treatment (omeprazole 20 mg or omeprazole 10 mg) with placebo (P = 0.006) as well as when comparing each active drug with placebo (P = 0.04 for omeprazole 20 mg vs. placebo and P = 0.005 for omeprazole 10 mg vs. placebo). The distribution of the dyspepsia subgroups was similar in the Bond and Opera studies. However, in patients recruited by a general practitioner the difference in treatment response between active treatment and placebo was considerably greater than in patients recruited by a specialist. In the Bond study, 34% of the patients were from family practice but in the Opera study only 8% were from this setting. When adjusting for the type of investigator (family physician or specialist) the differences between the studies in treatment response between active treatment and placebo was no longer statistically significant.

Adverse events

The number of adverse events was low and was similar across the three treatment arms. A total of 34 patients discontinued treatment due to adverse events (10 on omeprazole 20 mg, 9 on omeprazole 10 mg and 15 on placebo).

DISCUSSION

Drugs that reduce gastric acid secretion are commonly prescribed for patients with functional dyspepsia, but their efficacy has been questioned.12, 14 Randomized placebo-controlled trials of H2-receptor antagonists in functional dyspepsia have produced conflicting results; in the positive studies the benefit over placebo was small based on the symptom scores applied and therefore of questionable clinical significance,14, 23, 24 while the negative trials were often underpowered to detect an important difference between drug and placebo.14, 25 A meta-analysis suggested that H2-blockers produced a therapeutic effect of 20% over placebo but the report has been criticized for only including some of the available trials.26 Thus, it remains to be definitely established that this class of antisecretory compounds is superior to placebo in functional dyspepsia. On the other hand, the role of more potent acid suppression has not until now been adequately investigated. We found in the present study that omeprazole was superior to placebo in relieving the symptoms of functional dyspepsia, although the benefit was very modest. We combined the results of the two trials because they were of identical design, with the exception that different countries recruited the patients. Rather than relying on an arbitrary symptom score which may not translate into a clinically meaningful number, the main outcome measure in this trial chosen a priori was the most rigorous possible, namely absence of epigastric pain and discomfort. Importantly, all the symptom measures improved in parallel and both doses of omeprazole were superior to placebo. The evidence is consistent with the observed small therapeutic gain with omeprazole being clinically meaningful.

It has been suggested that patients with functional dyspepsia can usefully be divided into subgroups based on symptoms. In the original proposal, ulcer-like, dysmotility-like, reflux-like and non-specific dyspepsia were identified based on clusters of symptoms,27 and subsequently the Rome criteria were developed along similar lines, although reflux-like dyspepsia was discarded.1 However, symptom clusters have appeared to be a dismal failure because of substantial symptom overlap and a lack of correlation with pathophysiological disturbances.2, 3, 8 Hence, in the present study we a priori applied a new forced choice classification based on symptom predominance. We also did not discard reflux symptoms because these are so common in patients with any upper gastrointestinal tract disease including peptic ulceration.28 We noted that the symptom benefit with omeprazole was greatest in patients with ulcer-like or reflux-like dyspepsia, and was not observed in those with dysmotility-like symptoms. The results suggest that symptom subgrouping based on symptom predominance has clinical utility because it predicts treatment response.

There are very few other trials that have addressed the clinical benefit of proton pump inhibitor therapy in functional dyspepsia. Lauritsen and co-workers randomized 197 patients with functional dyspepsia to omeprazole 20 mg twice daily or placebo for 2 weeks in Denmark and Sweden; complete relief of dyspepsia in the last 2 days was observed in 35% on omeprazole and 13% on placebo, which was a significant difference.19 On the other hand, a German multicentre study randomized 801 patients with functional dyspepsia to omeprazole (10 mg and 20 mg daily), ranitidine (150 mg nightly) or placebo for 2 weeks.29 The investigator judged whether the patient required further investigation or treatment, and using this as the primary end-point there was no significant difference between the groups.

There are other studies that have evaluated proton pump inhibitor therapy in uninvestigated dyspepsia in primary care. Meineche-Schmidt & Krag reported a randomized trial of omeprazole (20 mg daily) and placebo for 2 weeks in 536 Danish patients with no documented history of ulcer or gastro-oesophageal reflux disease but who had ulcer-like or reflux-like dyspepsia; 50% on omeprazole compared with 35% on placebo had no symptoms at the end of the trial.30 Another trial enrolled 674 patients with uninvestigated dyspepsia and/or heartburn and in an open-label study randomized them to omeprazole 10 mg daily or Gaviscon four times daily for 4 weeks; symptom abolition at 4 weeks was observed in 41% on omeprazole and 16% on Gaviscon, but although statistically significant, the results are very difficult to interpret because of the lack of blinding.31 Jones & Baxter in a randomized trial compared lansoprazole 30 mg daily with ranitidine 150 mg twice daily; the proton pump inhibitor provided significant symptom relief compared with the H2-antagonist in those with epigastric pain and/or heartburn, but patients with and without a history of structural disease were enrolled and no placebo control group was included.32

A recent systematic review of all published randomized controlled trials in functional dyspepsia evaluated a total of 52 eligible studies;33 the majority of trials suffered from serious weaknesses in study design and execution, and only five studies used previously validated outcome measures. In this trial careful attention was given to avoiding previously identified methodological concerns. In particular, validated outcome measures were utilized, strict blinding was maintained, an adequate placebo control was included and the study was sufficiently powered to detect modest but clinically significant differences. Utilizing the hard end-point of symptom absence, the placebo response was minimized. Hence our results suggest that a subgroup of patients with functional dyspepsia are responsive to acid suppression, and we conclude that the findings are likely to be accurate and applicable in clinical practice.

If proton pump inhibitors are efficacious in some patients with functional dyspepsia, by what mechanism do they reduce symptoms? Modulation of gastric acid secretion is one consideration. Earlier reports demonstrated that both basal and peak acid output were similar in patients with functional dyspepsia compared with appropriate control groups.10 More recent work has shown that acid secretion in response to gastrin-releasing peptide (GRP) was significantly increased in H. pylori infected patients with functional dyspepsia compared to H. pylori-positive healthy volunteers.11 Overall, approximately 50% of patients with functional dyspepsia had a disturbance of GRP-stimulated acid secretion and in this group the disturbance was similar to that found in patients with duodenal ulcer.11 The data suggest that a subset of patients with functional dyspepsia and. H. pylori infection have acid dysregulation that potentially may respond to acid suppression. However, in the present trial, H. pylori status did not predict response to therapy so that this mechanism appears unlikely to explain the results.

A further possible explanation for the efficacy of omeprazole in functional dyspepsia may relate to undiagnosed gastro-oesophageal reflux disease (GERD). Heartburn is a very common complaint in the general population;34 furthermore, the majority of patients with functional dyspepsia have co-existent heartburn if specific enquiries are made.35 It is conceivable that some patients labelled by their physician as having functional dyspepsia in this trial actually had atypical GERD and this may explain any benefit of acid suppression.36, 37 Indeed, while the majority of patients enrolled had ulcer-like dyspepsia (n = 708 or 56%), 143 patients (11%) were classified at entry as having predominant reflux symptoms and it is likely that true pathological acid reflux was present in a subset of these patients. The rate of complete symptom relief observed in reflux-like dyspepsia in this trial was similar to the symptom relief of heartburn observed in patients with endoscopy-negative GERD.38 On the other hand, in the present study patients with endoscopically diagnosed oesophagitis, a past history of documented GERD or heartburn alone were excluded. Moreover, patients had to have a diagnosis of functional dyspepsia based on the investigators, judgement in order to enter the trial. While 24 h oesophageal pH testing would have provided additional information to address how many patients had true gastro-oesophageal reflux, it is invasive and would have resulted in a highly selected patient population being enrolled because of refusals, and hence was considered impractical. Moreover, other work suggests that acid exposure as measured by 24 h oesophageal pH testing does not predict the response to proton pump inhibitor therapy in dyspepsia.19 In practice this is not a crucial issue as our results show that a subset of patients with a clinical and endoscopic diagnosis of functional dyspepsia will respond to omeprazole. However, more research is required to determine if atypical GERD explains the benefit of acid suppression in some patients with functional dyspepsia.

We did observe a lower treatment effect in Study 2 compared to the Bond study. This could not be explained by baseline differences and for this reason it was justifiable to combine the studies. More general practitioners participated in the Bond study whereas a higher proportion of gastroenterologists participated in the Opera study. Interestingly, the placebo response was higher amongst those patients seeing a gastroenterologist (32%) compared to those enrolled by a family practitioner (13%), although the response to active treatment was reasonably similar in both groups (Table 4). Conceivably, those seeing a specialist may feel more reassured that they do not have a serious underlying disease, which may promote a higher placebo response.

In conclusion, our results have demonstrated that the proton pump inhibitor omeprazole at a dose of 20 mg and 10 mg was modestly superior to placebo in relieving the symptoms of functional dyspepsia, particularly in those with ulcer-like or reflux-like dyspepsia, but symptom relief was similar in those with and without H. pylori infection.

ACKNOWLEDGEMENTS

Financial support for this study was provided by Astra Hässle. The assistance of P. Jerndal, B. Hermenius, O. Junghard and Elisabeth Bolling-Sternevald from Astra Hässle is gratefully acknowledged. The following investigators contributed to this study by patient recruitment and this is also gratefully acknowledged. Greece: co-ordinator A. Archimandritis aided by 10 principal investigators and 16 co-investigators. UK/Ireland: co-ordinator C. A. O’Morain aided by 15 principal investigators and 2 co-investigators. Belgium: co-ordinator D. Urbain aided by 8 principal investigators and 7 co-investigators. Finland: co-ordinator P. Jauhonen aided by 10 principal investigators and 4 co-investigators. Portugal: co-ordinator D. de Freitas aided by 8 principal investigators and 31 co-investigators. Canada: co-ordinator P. Pare aided by 12 principal investigators and 17 co-investigators. Norway: co-ordinator K. Vetvik aided by 16 principal investigators and 1 co-investigator. Denmark: co-ordinator K. Lauritsen aided by 8 principal investigators and 17 co-investigators. France: co-ordinator M. A. Bigard aided by 9 principal investigators and 10 co-investigators. Germany: co-ordinator N. Schindlbeck aided by 9 principal investigators and 9 co-investigators. Holland: co-ordinator H. van der Heide aided by 9 principal investigators and 4 co-investigators. Hungary: co-ordinator G. Mózsik aided by 6 principal investigators and 13 co-investigators. Poland: co-ordinator L. Hryniewiecki aided by 6 principal investigators and 20 co-investigators.

Ancillary