Six-month trial of on-demand rabeprazole 10 mg maintains symptom relief in patients with non-erosive reflux disease


Dr P. Bytzer, Department of Medical Gastroenterology, Glostrup University Hospital, DK-2600 Glostrup, Copenhagen, Denmark.


Background : Compliance studies have shown that patients with reflux symptoms generally take their medication only when experiencing these symptoms.

Aim : To evaluate the efficacy of on-demand rabeprazole maintenance therapy in patients with non-erosive reflux disease.

Methods : This multicentre, randomized, double-blind, placebo-controlled, withdrawal study compared 6 months of on-demand treatment with rabeprazole 10 mg vs. placebo. Adults with a history of reflux symptoms, a negative endoscopy, and ≥3 days of moderate to very severe heartburn in the 7 days before enrolment (N = 535) entered 4 weeks of open-label, acute treatment with rabeprazole 10 mg once daily. Patients with complete symptom relief then entered the on-demand phase. The primary end-point was discontinuation due to lack of heartburn control during the on-demand phase.

Results : Eighty-three percent (432 of 523) of patients reported complete symptom relief at the end of the acute phase. During on-demand treatment, rates of discontinuation because of inadequate heartburn control were 20% (28 of 139) for placebo vs. 6% (16 of 279) for rabeprazole (P < 0.00001). Antacid use was twofold higher in the placebo group vs. the rabeprazole group (P = 0.0009).

Conclusions : Rabeprazole 10 mg once daily is highly effective in acute symptom relief and as on-demand long-term maintenance therapy in non-erosive reflux disease patients.


Gastro-oesophageal reflux disease (GERD) is a common, chronic medical disorder. Approximately 50–70% of patients with GERD symptoms have normal oesophageal mucosa upon endoscopy1– a condition known as non-erosive reflux disease (NERD). Whilst these patients are typically prescribed daily proton-pump inhibitor therapy, compliance studies have shown patients with reflux symptoms generally take their medication only when experiencing these symptoms2. On-demand acid-suppressing therapy has proved effective for maintaining symptom relief in patients with NERD3, 4 and has the potential to substantially lower drug-acquisition costs.5 Because therapy is not continuous, this approach would seem to dictate that the agent used provide rapid and profound acid suppression and symptom relief with a single dose.

Proton-pump inhibitors are highly effective for the treatment of NERD,3, 4, 6 but differences among these drugs can influence their suitability for on-demand therapy. Rabeprazole results in more rapid and profound acid suppression than other proton-pump inhibitors7–11 and has been shown to provide first-dose symptom relief in a majority of patients.12 Studies have also demonstrated that rabeprazole is effective for treating symptoms in patients with NERD.13 The present study was undertaken to assess the effectiveness of rabeprazole 10 mg for on-demand maintenance therapy in patients with NERD.


Study design

This was an international, multicentre, randomized, double-blind, placebo-controlled, withdrawal study. Following screening, all patients received open-label, acute treatment with 4 weeks of rabeprazole 10 mg once daily. Patients with complete absence of heartburn during the final 7 days were randomized (2 : 1) to rabeprazole 10 mg or placebo once daily on-demand for an additional 6 months. The study was conducted between August 2001 and October 2002 in accordance with the Declaration of Helsinki and subsequent revisions and with Good Clinical Practice (GCP) as outlined in the International Conference on Harmonization-GCP guidelines. Patients were enrolled at sites in Greece, Italy, the Netherlands, Spain, France, Portugal, Sweden, Denmark, Ireland, Belgium, United Kingdom, Russia, Poland, and Lithuania (a list of investigators that enrolled patients appears at the end of the Discussion section). The study protocol was approved by the appropriate Independent Ethics Committees, and all patients provided written, informed consent.


Men and women aged ≥18 years with a ≥12-month history of NERD, which included ≥2 symptomatic episodes, were enrolled. Patients had a negative endoscopy and ≥3 days of moderate to very severe heartburn in the 7 days prior to enrolment in the acute treatment phase. Women of child-bearing potential had a negative urine pregnancy test and were required to use an accepted method of birth control.


In the open-label acute phase, patients took rabeprazole 10 mg once daily for 4 weeks. During the 6-month, double-blind, on-demand phase, patients took either rabeprazole 10 mg or placebo when first experiencing recurrent heartburn and then one tablet each subsequent day before breakfast. Medication was discontinued only after the patient experienced no symptoms for a full 24 h. Concomitant antacids (Maalox TC) were provided for unbearable heartburn. Dosing habits were measured by a medical event monitoring system (MEMS; Aardex Ltd, Zug, Switzerland) device in the container caps, which recorded the date and time of day that the bottle was opened.

Patients enrolled in the on-demand phase of the trial were randomly allocated (in a 2 : 1 ratio) to rabeprazole or placebo. Subjects were randomized in consecutive order, starting with the lowest available medication number. Blinding in the on-demand phase was accomplished by providing the investigator with a sealed envelope containing coded treatment details for each patient. This code was broken only in an emergency in which further treatment of the patient depended on knowing the trial medication he or she had been receiving.


Heartburn severity was evaluated upon enrolment in the acute phase, and heartburn severity and level of control were assessed at randomization and at 1, 3, and 6 months or at discontinuation of double-blind treatment. The primary efficacy variable was the proportion of patients discontinuing treatment in the on-demand phase because of inadequate heartburn control. Secondary measures included discontinuation rates for any reason, time to discontinuation, heartburn severity (scored on a 5-point Likert scale: 0 = none; 4 = very severe), assessment of dosing habits, and antacid consumption. Quality of life was assessed at the start of both the acute and on-demand treatment phases and again at 6 months using the 22-item Psychological General Well-being Index (PGWBI).14 This index ranges from 22 to 132, with a mean of 102.9 (95% confidence interval [CI]: 102.1–103.8) representing the general population norm. A linear transformation of the PGWBI scores was calculated ([102.9 − 22]/110 × 100) to obtain a standardized score between 0 and 100, with the general population norm of 73.5 (95% CI: 72.8–74.4). Safety assessments included monitoring of adverse events and vital signs.

Statistical analysis

Statistical analyses were carried out using the SAS 8.1 (SAS Institute Inc., Cary, NC, USA) statistical package. To detect a difference between rabeprazole 10 mg and placebo for the primary end-point with a power of 90% and a 0.1% two-sided type-I error, approximately 208 and 104 patients, respectively, were required (Casagrande, Pike, and Smith method). The intent-to-treat (ITT) efficacy population included patients who used trial medication at least once during the acute trial and had baseline efficacy (open-label) assessments as well as all randomized patients who had data for the primary efficacy variable on ≥1 postrandomization visits.

Pearson's chi-squared test was used to compare discontinuation rates for rabeprazole vs. placebo. Sensitivity subanalyses were performed to assess the influence of Helicobacter pylori status (determined via histological testing), heartburn severity and frequency at screening, percentage of days with intake of trial medication, weekly average antacid consumption, body mass index (BMI), heartburn duration, use of drugs for peptic ulcer or flatulence, investigator qualification, and country on the primary efficacy end-point. Each time, a Cochran-Mantel-Haenszel test for general association (controlling for the mentioned strata) was used to compare both treatment groups, together with a Breslow-Day test for homogeneity of odds ratios across the strata. Time from randomization to discontinuation was analysed using survival techniques (Kaplan–Meier product-limit estimation). Gehan-generalized Wilcoxon and log-rank tests were used to test the effect of treatment on the risk of discontinuation. Comparisons of the two groups for the secondary efficacy parameters were analysed using Pearson's chi-squared and Wilcoxon's rank sum tests.

Within-group changes from baseline were assessed with Wilcoxon's signed ranks tests. For efficacy parameters, statistical tests were interpreted at the 0.1% two-tailed significance level (except for the Breslow-Day test for homogeneity and interaction terms: 10% two-sided significance level). Frequency of adverse events during the on-demand phase was adjusted for time spent in the study.



A total of 535 patients were enrolled in the open-label acute phase, and 418 were randomized to on-demand treatment with placebo (n = 139) or rabeprazole (n = 279) (Figure 1). There were no significant between-group differences in patient demographics or vital signs (Table 1).

Figure 1.

Randomization protocol and patient disposition. *Sponsor's decision to withdraw patient, error in treatment distribution, medication too old, false-positive pregnancy test, patient noncompliance, and concomitant nonsteroidal anti-inflammatory drug use.

Table 1.  Baseline demographic and clinical characteristics of patients
 Acute phase only* (n = 117)On-demand phaseTotal (N = 535)
Rabeprazole 10 mg (n = 279) Placebo (n = 139)
  1. BMI, body mass index; SE, standard error.

  2. * Received rabeprazole in the acute phase, but not randomized into the on-demand phase.

Gender (male : female, %)28 : 7244 : 5641 : 5940 : 60
Mean age (years, SE)46 (1.44)47 (0.81)48 (1.23)47 (0.62)
Mean BMI (kg/cm, SE)26 (0.39)26 (0.25)27 (0.35)26 (0.18)
Heartburn duration (mean years, SE)5.7 (0.624)7.1 (0.471)6.1 (0.514)6.5 (0.312)
Heartburn duration (year [n, %])
 <111 (9)12 (4)11 (8)34 (6)
 1–568 (58)162 (58)76 (55)306 (57)
 >538 (32)105 (38)52 (37)195 (36)
Heartburn severity (n, %)
 Moderate78 (67)183 (66)81 (58)342 (64)
 Severe36 (31)89 (32)49 (35)174 (33)
 Very severe3 (3)7 (3)9 (7)19 (4)
Heartburn frequency (days/week [n, %])
 1–21 (1)0 (0)0 (0)1 (<1)
 3–451 (44)122 (44)58 (42)231 (43)
 5–629 (25)67 (24)33 (24)129 (24)
 736 (31)90 (32)48 (35)174 (33)
Positive Helicobacter pylori test (n, %)35 (30)100 (36)52 (37)187 (35)


Acute phase.

Complete relief of symptoms was experienced by 83% (n = 432) of patients treated with once-daily rabeprazole for 4 weeks. For all enrolled patients, mean heartburn severity and frequency scores decreased from 2.4 (± 0.02) to 0.3 (± 0.03) and 2.9 (± 0.04) to 0.4 (± 0.04), respectively (P < 0.0001 for both).

On-demand phase.

Rabeprazole was significantly superior to placebo as on-demand therapy (Figure 2). Rates of discontinuation due to inadequate heartburn control in the ITT population were 20% (28 of 139) for placebo vs. 6% (16 of 279) for rabeprazole (P < 0.00001). Inadequate heartburn relief was the cause of the majority of discontinuations for placebo-treated patients (82%; 28 of 34) but accounted for only 40% (16 of 37) of withdrawals for those who received rabeprazole. Reasons for discontinuation can be found in Figure 1. The between-group difference in dropout rates remained significant after adjustments for H. pylori status, heartburn severity and frequency, BMI, heartburn duration, concurrent use of antacids, consumption of medications for peptic ulcer or flatulence, investigator qualification, and country (P ≤ 0.00004 for all).

Figure 2.

Discontinuation due to inadequate heartburn control during on-demand treatment (intention-to-treat patients) (rabeprazole n = 279, placebo n = 139). *P < 0.00001.

Placebo-treated patients who discontinued therapy because of inadequate heartburn control or for any reason did so earlier than those who received rabeprazole (Figure 3a,b). Of those placebo patients who discontinued because of inadequate heartburn control, 57% did so within 1 month of randomization vs. 31% in the rabeprazole group (P < 0.0001). Of those patients who discontinued for any reason, 50% in the placebo group left the study within 1 month vs. 16% for rabeprazole (P = 0.0012).

Figure 3.

Kaplan–Meier plots for time to discontinuation for patients due to inadequate heartburn control (a) and for any reason (b). *P < 0.0001, **P ≤ 0.0029.

Results for other secondary study parameters appear in Table 2. Rabeprazole was significantly superior to placebo in reducing symptom severity and the proportion of patients with insufficient heartburn control. Although mean duration of heartburn-free periods and time to first dose were similar for the two treatment groups (12.4 and 17 days, respectively, for both groups), time required for symptom resolution during a heartburn episode was significantly shorter with rabeprazole vs. placebo. In the rabeprazole group, 30% of patients achieved complete 24-h heartburn control after only 1 or 2 days of treatment, compared with only 18% in the placebo group (P = 0.0106). Moreover, 59% of patients in the rabeprazole group required only four or less days of medication to reach 24-h heartburn relief for each symptom episode during the on-demand phase compared with 45% of patients in the placebo group (P = 0.0096). Patients took medication on 26% of days. Mean weekly antacid consumption was approximately twofold higher among patients who received placebo vs. those treated with rabeprazole (3.9 vs. 2.0; P = 0.0009).

Table 2.  Summary of secondary efficacy parameters (on-demand phase)
 Rabeprazole (n = 279)Placebo (n = 139)P-value
  1. * Wilcoxon rank sum test.

  2. ** Chi-squared test.

Mean change in symptom severity score from baseline0.71.0< 0.05*
Sufficient heartburn control (n, %)241 (86.4)94 (67.6)0.00002**
Maximum duration of symptoms (days)*
Maximum symptom episode duration   ≤2 days (%)30180.0106**
Maximum symptom episode duration   ≤4 days (%)59450.0096**
Mean weekly antacid use (n)*

Quality of life

Acute phase.

Treatment with rabeprazole once daily for 4 weeks increased mean PGWBI scores (transformed) from 65.0 to 72.8 (P < 0.0001) – a value within the normative range of the general population.14Post hoc analysis revealed a significant correlation between the severity of heartburn upon enrolment and greater improvement in PGWBI score (transformed) (P < 0.001).

On-demand phase.

For patients who completed the 6-month on-demand phase, irrespective of treatment group, PGWBI scores (transformed) remained at the level seen in normal populations (72.8, rabeprazole 10 mg; 73.2, placebo). The PGWBI scores of patients who discontinued the trial decreased to levels similar to those observed prior to entry into the acute phase, indicating a worsening of quality of life. In patients in the placebo group who discontinued the trial, the decrease in overall PGWBI (−12.9) was statistically significant (P < 0.0001), which was not the case in patients in the rabeprazole group who discontinued the trial (−7.4; P > 0.25).

Adverse events

Acute phase.

A total of 21% (n = 112) of all patients experienced at least one adverse event. The most commonly reported events (≥2%) were abdominal pain, diarrhoea, and headache. Fourteen patients discontinued because of adverse events, and one patient experienced a serious adverse event (neoplasm) considered unrelated to study drug.

On-demand phase.

In the rabeprazole group, 41% of patients (n = 113) reported adverse events vs. 34% (n = 47) in the placebo group. Less than 1% of these events were considered probably related to study medication. There was no significant difference between treatment groups.


This study demonstrated that rabeprazole 10 mg administered on-demand over 6 months was significantly superior to placebo in controlling NERD. These results are important for two reasons. First, they show that this rapidly acting proton-pump inhibitor is effective when administered in the manner in which most reflux patients are actually likely to use it.2 This aspect of the study design greatly facilitates generalization of results to physicians’ practices. Secondly, study results support the view that NERD can be effectively managed in most patients with on-demand therapy, a treatment strategy that has the potential to substantially lower the cost of treatment.1 Results for rabeprazole in this trial compare favourably with similar trials investigating the use of other proton-pump inhibitors in on-demand therapy. Here, the rabeprazole treatment group had lower rates of discontinuation because of insufficient heartburn control than seen in previous on-demand trials.15 The number of days of rabeprazole treatment required to achieve complete heartburn relief was very short, with 59% of patients requiring a maximum of 4 consecutive days of therapy, and 30% requiring only 1–2 days of rabeprazole treatment. In addition, use of rabeprazole was only 26%, indicating an average intake of only one tablet in 4 days, whereas two esomeprazole on-demand trials indicated average use of one tablet in every 3 days was required to maintain adequate heartburn control.15, 16 The lower usage seen with rabeprazole in this trial indicates that rabeprazole may provide both a cost and an efficacy advantage in terms of speed of onset of action. Several head-to-head pharmacodynamic studies8–11 have demonstrated that rabeprazole treatment produces greater acid suppression on day 1 than the other proton-pump inhibitors. Together, these results suggest that rabeprazole is ideally suited for on-demand treatment, which requires fast acid and, by association, symptom control.

Symptoms of GERD, both erosive and non-erosive, have been shown to have a significant and similar impact on quality of life.17 However, in the absence of erosions, the focus of treatment in the NERD patient is solely on the relief of symptoms. Here, 4 weeks of continuous rabeprazole treatment raised PGWBI scores to within the range of the general population. On-demand treatment with rabeprazole maintained this level of quality of life in patients over the 6-month treatment period.

Whereas the present findings support on-demand therapy with rabeprazole for NERD, further long-term studies are needed to validate the efficacy of this treatment approach and to determine whether or not it results in any negative effects (e.g. development of erosions) on the oesophageal mucosa. This progression to erosive GERD is considered unlikely, as NERD appears to be a condition distinct from erosive GERD and has not been shown to progress to erosive GERD in the majority of patients.18, 19 Furthermore, the issue of which proton-pump inhibitor is more suitable for this treatment regimen is best addressed through head-to-head clinical comparisons.

Safety results for rabeprazole in this long-term trial are consistent with those of previous studies.20–23 These results indicate that long-term therapy with this proton-pump inhibitor presents no special concerns.

In summary, results from this 6-month trial showed that rabeprazole is effective as on-demand therapy for NERD, a condition generally considered more difficult to treat than erosive reflux disease.13

Study Investigators

Patients were enrolled by the following investigators: Greece: D. Karamanolis, Piraeus; E. Tsianos, Ioannina; N. Skandalis, Athens; N. Evgenidis, Thessaloniki; T. Rokkas, Athens; Italy: D. Spotti, Milano; E. Benedetti, Gorizia; G. Frosini, Siena; the Netherlands: C. Mulder, Arhnem; R. Ouwendijk, Rotterdam; W. Dekker, Haarlem; Spain: M. Diaz-Rubio, Madrid; J. Ponce Garcia, Valencia; M. Manrique, Barcelona; R. Sainz Samitier, Zaragoza; France: M.-A. Bigard, Vandoeuvre; F. Spilthooren, Evreux; A. Campagne, Tours; A. Boye, Nantes; L. Boucher, Murs Erigne; D. Lejay, Vieux Condé; C. Magnani, L'Aigle; J. Marty, Murs Erigne; T. Latté, St Julien des Landes; E. De Sainte Lorette, Paris; Portugal: C. Pinho, Oporto; H. Gouveia, Coimbra; H. Barros, Santa Maria de Feira; Sweden: T. Franzén, Linköping; S. Vallgren, Helsingborg; G. Lindberg, Huddinge; T. Kjellin, Växjö; H.-O. Hakansson, Kalmar; S. Nilson, Kristianstad; J. Ronkainen, Haparanda; J. Bark, Stockholm; E. Toth, Malmö; H. Högstr&#x006f;̆m, Halmstad; Denmark: P. Bytzer, Glostrup; L. Hendel, Birkerod; E. Skoubo Kristensen, Holstebro; A. Bak-Christensen, Allerod; Ireland: A. Byrne, Dublin; B. O'Doherty, Gorey; F. Bradbury, Loughboy; Belgium: D. Urbain, Brussels; J. Vandervoort, Aalst; J. Deviere, Brussels; J.-P. Henry, Charleroi; United Kingdom: I. Allwood, Stratford-Upon-Avon; S. Blagden, Chesterfield; A. Connolly, Bradford; J. Evans, Stevenage; B. King, Sheffield; P. Lane, Barnsley; N.H. Patel, Sheffield; S.K. Patel, Spennymoor; N. Sinclair, Doncaster; Russia: O. Minushkin, Moscow; S. Bourkov, Moscow; Poland: E. Butruk, Warsaw; K. Marlicz, Szczecin; E. Malecka-Panas, Lodz; K. Linke, Poznań; A. Nowak, Katowice; W. Laszewicz, Bialystok; Lithuania: A. Irnius, Vilnius; L. Kupčinskas, Kaunas.


This study was supported by Janssen-Cilag EMEA, a division of Janssen Pharmaceutica N.V., Beerse, Belgium. PB has been a study investigator for AstraZeneca, Janssen-Cilag, and Eisai Ltd; has spoken for AstraZeneca, Wyeth, Janssen-Cilag, Eisai Ltd, and Byk Gulden; and has been a consultant for AstraZeneca, Janssen-Cilag, Eisai Ltd, and Wyeth. AB has been a study investigator for Janssen-Cilag and Eisai Ltd; has been a consultant for Novartis, Eisai Ltd, and Nestec; holds shares in Roche and Merck & Co., Inc.; and has received honoraria from AstraZeneca. DdH and DD hold shares in Johnson & Johnson.