STUDY PROTOCOL

A multicentre, double blind, double dummy, randomised clinical trial with parallel groups was conducted according to Good Clinical Practice and the revised Declaration of Helsinki. The ethics committee at each centre approved the protocol and all patients gave written informed consent.

SELECTION OF PATIENTS

Recruitment took place between August 1994 and July 1996 and included 801 ambulatory patients of both sexes from 71 private gastroenterological practices in Germany, aged ⩾18 years, with chronic functional dyspepsia, with or without gastritis due toH pylori. Chronic functional dyspepsia was defined as epigastric symptoms in the absence of organic disease known to produce epigastric symptoms. Initial symptoms had to be severe enough to require treatment. Dyspeptic symptoms necessitating treatment had to be present for at least four weeks and be severe on at least three days of the seven day run in screening period.

We excluded patients with heartburn/acid regurgitation without concomitant epigastric symptoms and also those with symptoms suggesting irritable bowel syndrome (pain in the lower abdomen, flatulence, diarrhoea, or constipation) which were severe enough to require treatment or diagnostic tests. In addition, patients with any of the following were excluded: current or previously documented erosive or ulcerative oesophagitis, peptic ulcer, or previous abdominal surgery (except for inguinal hernia, appendicectomy, hysterectomy, or Caesarean section); treatment with proton pump inhibitors and/or antibiotic treatment within one month prior to screening or regular treatment within the previous week with other drugs which might interfere with the study outcome; symptoms indicative of serious disease (for example, unintended weight loss, haematemesis) during the previous three months; and conditions associated with poor study compliance (for example, drug addiction or alcoholism).

BASELINE INVESTIGATIONS

At baseline, a complete medical history was taken followed by physical examination, abdominal sonography, and gastroscopy with biopsy (only axial hernia, less than 10 gastric erosions and endoscopic signs attributed to H pylori induced gastritis were permitted at gastroscopy). Biopsy specimens were taken from the corpus (greater curvature) and from the antrum (3 cm proximal to the pylorus) for histological examination using the Sydney criteria7 for assessment of H pylori and gastritis. Rapid urease tests8 (HUT, Astra Chemicals, Wedel, Germany) were also performed and these were considered positive for H pylori if at least one sample of the gastric corpus or antrum caused red discolouration within 24 hours. A validated ¹³C urea breath test was also performed.8 H pylori infection was diagnosed when, in addition to a positive HUT test, histological examination showed Helicobacter-like organisms in the corpus and/or antrum. The investigators were aware of the HUT test results but were blinded to the histological results throughout the study. Blood samples were taken at entry for routine assessment of haematology and clinical chemistry (one minor transitory abnormality was permitted) and these were repeated after the two week double blind treatment. Additional investigations, for example colonoscopy or CT scan, were done at the investigator's discretion if clinically justified and all findings had to be normal for inclusion in the study.

STUDY TREATMENTS

Patients were given antacid tablets which could be taken up to three times a day during the seven day run in screening period. These were Maaloxan tablets (Rhône-Poulenc-Rorer, Germany) which contained 200 mg of aluminium hydroxide and 200 mg of magnesium hydroxide, with an acid binding capacity of 11 mmol HCl per tablet. Those patients who had severe dyspeptic symptoms on ⩾3 days during the run in screening period were eligible for study treatment. They were randomised and received omeprazole (Losec, Astra Pharmaceuticals Production AB, Sweden) capsules 10 mg (OM10) or 20 mg before breakfast (OM20), ranitidine (Zantac, Glaxo SpA, Italy) tablets 150 mg in the evening before retiring, or placebo tablets and capsules (matched for appearance, taste, and smell) for 14 days. A visit with assessment of symptoms (see below) was carried out at the end of this treatment and if no further management was required (that is, medical treatment other than liquid antacids and/or diagnostic procedures such as endoscopy), patients entered a follow up period of up to six months. With their consent, H pylori positive patients with symptoms requiring further treatment entered a second study evaluatingH pylori treatment (fig 1). ForH pylori negative patients with persistent symptoms at two weeks, the study was terminated, except for a personal visit or a telephone interview six months later.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Flow chart of the 974 dyspeptic patients who entered the run in week with low dose antacid treatment. A=not randomized after run in period for reasons other than response to antacid treatment (n=52). B=H pylori positive responders not entering follow up (n=17). C=H pylori negative responders not entering follow up (n=19). Responder=no dyspeptic symptoms requiring further management after the two week double blind treatment (main outcome criterion). Hp, Helicobacter pylori; recur.,recurrence of dyspeptic symptoms requiring management

FOLLOW UP PERIOD

Follow up of patients without symptoms requiring further treatment lasted for six months or until relapse of dyspepsia. During this period, no treatment for dyspepsia other than an antacid suspension (10 ml bags containing 3.13 g of aluminium hydroxide gel, 0.27 g of magnesium oxide, and 0.63 g of magnesium carbonate, with an acid binding capacity of 33 mmol/ml) for occasional dyspeptic symptoms was allowed. Unscheduled visits were encouraged at any time during the six month follow up period when dyspeptic symptoms recurred, when antacid treatment was needed for more than four consecutive days, and/or when antacid consumption exceeded 30 ml per day. Relapse was defined as reappearance of symptoms requiring management (treatment other than liquid antacid and/or diagnostic tests such as endoscopy).

OUTCOME CRITERIA

The main outcome criterion, tested at two weeks, was lack of dyspeptic symptoms requiring further management (as defined above). In addition, the severity of individual dyspeptic and other abdominal symptoms (specifically epigastric pain/burning, epigastric pressure/fullness, heartburn, acid regurgitation, nausea, vomiting, pain in the lower abdomen, flatulence, diarrhoea, and constipation) during the previous week was graded according to a four point scale (0=no complaints, 1=complaints not interfering with daily activities and not requiring treatment, 2=complaints requiring treatment but not interfering with daily activities, and 3=complaints interfering with daily activities and requiring treatment). Patients answered the question “How were your symptoms in the area of the oesophagus and stomach” and “How was your general condition during the last seven days” by making a mark on a 10 cm visual analogue scale (10 cm, best possible condition; 0 cm, worst condition). Quality of life (QoL) during the previous week was assessed using a validated questionnaire adapted to German lifestyles.9 It contained 40 general items relating to physical strength, ability to enjoy and relax, positive mood, absence of negative mood, social contacts, and social well being. In addition, nine questions validated in Germany, relating to impairment of QoL by dyspeptic symptoms,10 were asked. They assessed the effect of dyspeptic symptoms on eating, other daily activities, social contacts, sleep, and fears of serious disease. Finally, the patient's time spent off work and/or in hospital was recorded.

DATA MANAGEMENT AND STATISTICS

Data management was performed according to Good Clinical Practice and data were transferred to an independent statistical institute (Institut für numerische Statistik, Cologne, Germany). The treatment code was broken after decisions on the allocation of individual patients to intention to treat (ITT) and per protocol (PP) analyses. All analyses were based on SAS11 and SPSS7.5 (for Windows).

The response rate based on the primary outcome variable was estimated to be 60% after placebo and 80% after omeprazole therapy. Thus accepting a β error of 0.20, the necessary number of patients per group in an ITT analysis would be 75 (Fisher's exact test, two sided), with an α error adjusted to 0.017 according to Bonferroni to compensate for comparisons between the three active treatment groups and placebo. This number was increased to 90 per group to provide for a valid PP analysis. The primary response rate was tested for centre effects using the Breslow-Day test.

Secondary outcome measures included effects of treatment on gastrointestinal symptoms and on QoL. These analyses were exploratory and used standard significance levels (p<0.05). A life table analysis of relapses during the six month follow up period was performed.

POSSIBLE PROGNOSTIC FACTORS

The following factors were included in a logistic regression analysis using the main outcome criterion as the dependent variable: treatment, age, gender, smoking, alcohol consumption, dyspepsia, gastric erosions present, and positive urea breath test.

PATIENTS

A flow chart showing the fate of the 974 patients entering the one week run in period is shown in fig 1. The response rate to low dose antacid was 13% and was independent of the predominant symptom recorded at entry. Patients who were found to have a cause for their symptoms other than functional dyspepsia were excluded. The remaining 792 patients were symptomatic for at least three days of the run in week and were considered to require treatment. H pylori positive and negative patients (422 and 370, respectively) were randomised to study treatments and data from those patients were included in an ITT analysis. PP analysis was performed on data from 354 and 315 patients (reasons for exclusion of 123 patients are summarised in table 1). There was a similar distribution of patient characteristics across the four treatment groups in bothH pylori positive and negative cohorts (table 2). Comparing the cohorts, H pyloripositive patients were older and more frequently male thanH pylori negative patients.

EFFECT OF TWO WEEK ANTISECRETORY TREATMENT

Patient response rates (ITT) at the end of two weeks of active or placebo treatment are illustrated in fig 2 for the two outcomes: “no need for management” and “no dyspeptic symptoms”. For comparisons, Fisher's exact test was used. Compared with placebo, a significant therapeutic gain was obtained only with OM20, the beneficial effect being both in terms of management (17.6%; p=0.014 using the Bonferroni adjusted p level of 0.017) and abolition of symptoms (19.6%; p=0.0005) in the H pylori positive group. In the H pylori negative group, there were no significant changes in the primary response (need for management) compared with placebo but in significantly more patients symptoms were abolished after OM20 compared with placebo (16.7%; p=0.023). All changes (compared with placebo) are presented in table 3 and indicate that treatment with OM20 produced the greatest therapeutic benefit, especially in the H pylori positive group. All treatments were more effective at reducing the intensity of dyspeptic symptoms below the “need for treatment” threshold than in abolishing symptoms altogether (fig 2). The results of the ITT analysis for the primary variable were confirmed using a PP approach.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Effect of two week treatment with placebo, ranitidine, omeprazole 10 mg (OM10), or omeprazole 20 mg (OM20) on dyspeptic symptoms according to percentage of patients who became completely symptom free (no symptoms) and those with no need for further management (main outcome criterion) (means and 95% two sided confidence intervals; exact test). For assessment of significance levels, see text. The number of patients per group is given in table 1. Hp, Helicobacter pylori.

Figure 3 shows the proportions of H pyloripositive and negative patients with symptoms of epigastric pain/burning, epigastric pressure/fullness, or heartburn before and after treatment. Placebo treatment caused disappearance of symptoms in approximately one third of patients in each group. OM20 significantly reduced all three symptoms compared with placebo in theH pylori positive patients but only epigastric pressure/fullness was significantly reduced inH pylori negative patients by all three active treatments.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Effect of treatment with placebo, ranitidine, omeprazole 10 mg (OM10), or omeprazole 20 mg (OM20) on individual dyspeptic symptoms. The horizontal bars show the percentages of patients with each individual symptom before and after each two week double blind treatment. *p<0.05; **p<0.025, ***p<0.001 compared with placebo in the same Helicobacter pylori (HP) group.

QoL parameters and impairment caused by dyspepsia were analysed inH pylori positive and negative cohorts before and after treatment. Baseline QoL was similar across the treatment groups for both H pylori positive and negative cohorts and there was significant improvement within all groups (including placebo) following treatment (Wilcoxon signed rank test). However, when improvement was compared between treatments, OM20 improved QoL to a significantly greater extent than placebo on seven out of nine parameters measured in H pyloripositive patients (Wilcoxon rank sum test). In H pylori negative patients, OM20 was significantly better than placebo on only one of nine items of the QoL. With the other treatments, no significant differences from placebo were found for either H pylori positive or negative patients.

ADVERSE EVENTS

The adverse events noted during the study are described below for the four treatment groups and numbers in parentheses refer toH pylori positive and H pylori negative status, respectively. Eight patients (placebo 3 (2/1); ranitidine 4 (3/1); OM10 1 (0/1)) stopped treatment early because of worsening symptoms (they were included in the PP analyses as “treatment failures”) and three patients because of other adverse events: ranitidine 1 (1/0) vascular disorder; OM10 1 (1/0) taste disorder; OM20 1 (1/0) scarlet fever (these three patients were excluded from PP analyses). One OM20 (1/0) patient who developed borreliosis completed the study. All other adverse events (placebo 21 (11/10); ranitidine 28 (14/14); OM10 28 (12/16); OM20 26 (14/11)) were not serious and had no effect on the course of the study or analyses.

LOGISTIC REGRESSION ANALYSIS OF FACTORS AFFECTING THE PRIMARY OUTCOME RESPONSE

In a logistic regression analysis of data from all patients (n=792) using the factors previously described, antisecretory treatment and H pylori infection had a statistically significant influence on treatment outcome. The odds ratios for the two factors were 1.39 (95% CI 1.00–1.93) and 0.46 (95% CI 0.34–0.62), respectively. When, instead of “antisecretory treatment”, each of the four treatment groups was entered separately into the logistic regression analysis, the only statistically significant factor wasH pylori infection. WhenH pylori positive and negative patients were analysed separately, excluding H pyloristatus from the analysis, a significant influence of antisecretory treatment was observed only in H pylori infected patients (odds ratio 1.57, 95% CI 1.01–2.42). No other factors proved to be significant in either group.

RESULTS OF FOLLOW UP

A total of 427 patients (placebo n=97, ranitidine n=108, OM10 n=106, OM20 n=116) entered follow up. Forty nine per cent of patients remained symptom free and 34% had minor symptoms below the need for treatment threshold during the six month follow up period. Only 18% of patients (25% of the H pylori positive group and 12% of the H pylori negative group) had a relapse as defined by the protocol—that is, reappearance of symptoms requiring management. Thirty four patients (total followed by H pylori positive/H pylori negative: placebo 6 (6/0); ranitidine 10 (8/2); OM10 7 (5/2); and OM20 11 (7/4)) were re-endoscoped. Two (2/0) ranitidine patients had erosive duodenitis; all others had normal endoscopic findings. Life table analysis of symptomatic relapse is shown in fig 4. The symptom pattern of relapses closely resembled that observed on admission to the study. Interestingly, OM20 was the most effective treatment for the prevention of relapse in the H pylori positive group while relapse rates were similar with OM20 and ranitidine 150 mg (and not significantly different from placebo) in the H pylori negative group.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Life table analysis of patients without sufficient symptoms to require specific active management after successful treatment in the four treatment groups (placebo, ranitidine, omeprazole 10 mg (OM10), or omeprazole 20 mg (OM20)). At time 0 the proportion of patients with successful two week treatment is given; these patients entered follow up. The number of patients per group is given in the text. Hp, Helicobacter pylori.