Heartburn-dominant, uninvestigated dyspepsia: a comparison of ‘PPI-start’ and ‘H2-RA-start’ management strategies in primary care – the CADET-HR Study


Dr D. Armstrong, Division of Gastroenterology, HSC-4W8, McMaster University Medical Centre, 1200 Main Street West, Hamilton, Ontario L8N 3Z5, Canada. E-mail: armstro@mcmaster.ca


Background : There are few data on empiric, stepped therapy for heartburn relief or subsequent relapse in primary care.

Aims : To compare heartburn relief produced by a proton pump inhibitor-start or an H2-receptor antagonist-start with step-up therapy, as needed, followed by a treatment-free period to assess relapse.

Methods : Heartburn-dominant uninvestigated dyspepsia patients from 46 primary care centres were randomized to one of two active treatment strategies: omeprazole 20 mg daily (proton pump inhibitor-start) or ranitidine 150 mg bid (H2-receptor antagonist-start) for the first 4–8 weeks, stepping up to omeprazole 40 or 20 mg daily, respectively, for 4–8 weeks for persistent symptoms. Daily diaries documented heartburn relief (score ≤3/7 on ≤1 of 7 prior days) and relapse (score ≥4 on ≥2 of 7 prior days).

Results : For ‘proton pump inhibitor-start’ (n = 196) vs. ‘H2-receptor antagonist-start’ (n = 194), respectively, heartburn relief occurred in 55.1% vs. 27.3% (P < 0.001) at 4 weeks and in 88.3% vs. 87.1% at 16 weeks. After therapy, 308 patients were heartburn-free (159 vs. 149); median times to relapse were 8 vs. 9 days and cumulative relapse rates were 78.6% vs. 75.8%, respectively.

Conclusions : An empiric ‘proton pump inhibitor-start’ strategy relieves heartburn more effectively than an ‘H2-receptor antagonist-start’ strategy up to 12 weeks but has no effect on subsequent relapse, which is rapid in most patients.


Heartburn, retrosternal burning and regurgitation, the cardinal symptoms of gastro-oesophageal reflux disease (GERD)1–3 are the primary inclusion criteria for studies in patients with erosive oesophagitis or endoscopy negative reflux disease (ENRD)4–12 However, patients with erosive oesophagitis have generally been studied separately from those with ENRD despite evidence that they have symptoms which are comparable in severity, frequency and duration.5 As a result, there are few data on which to base management strategies for patients with uninvestigated GERD13–16 despite the fact that most patients with heartburn or regurgitation are managed in primary care practice as a subset of patients with uninvestigated dyspeptic symptoms.17, 18

Patients with heartburn-dominant symptoms comprised 38% of a Canadian, uninvestigated dyspepsia study population17 and erosive oesophagitis was identified in 55% of these patients at endoscopy; the Canadian Dyspepsia Working Group recommended that such patients be treated empirically for reflux disease, provided that they have no alarm features.18 Initial acid suppression therapy with a proton pump inhibitor (PPI) for 4 weeks was recommended,18 based on evidence that PPIs are superior to H2-receptor antagonists (H2-RAs) for relief of symptoms in erosive oesophagitis and ENRD and for healing of erosive oesophagitis.7, 19–21 However, there is continuing debate as to the merit of both ‘H2-RA-start’22, 23 and ‘PPI-start’ strategies despite accumulating evidence to suggest that PPI therapy is cost effective in patients with erosive oesophagitis or ENRD.24, 25

Patients enrolled in this study were classed as having heartburn-dominant dyspepsia because, although the main enrolment criterion – heartburn or retrosternal burning – was similar to that of most GERD studies, primary care patients with dominant heartburn may have endoscopic lesions other than oesophageal erosions.17 The current study was designed to compare the efficacies of ‘H2-RA-start’ and ‘PPI-start’ strategies for the management of heartburn-dominant uninvestigated dyspepsia (HDUD) in primary care practice, to compare the proportions of patients requiring ‘step-up’ therapy in each strategy because of insufficient initial symptom control and to document the recurrence of symptoms after cessation of a successful therapeutic regimen.



Patients, aged 18 years or older, were eligible if they had at least 3 months of heartburn and moderate-to-severe heartburn (score ≥4 on a 7-point Likert scale) on at least three of the 7 days before randomization. Heartburn was defined as ‘a burning feeling rising from the stomach or lower part of the chest up towards the neck’.1 Patients were excluded if they had alarm features (unintentional weight loss, persistent vomiting, dysphagia, haematemesis, melaena, fever, jaundice, or anaemia),18 irritable bowel syndrome (IBS: ≥3 Manning criteria),26 serious concomitant disease, and investigations related to the upper gastrointestinal tract on one occasion (i.e. one or more investigations arising from presentation to a physician for dyspepsia symptoms) during the prior 6 months or more than two occasions during the previous 10 years.17, 27, 28

Exclusion criteria included a history of gastrointestinal disease (including peptic ulcers, malignancy, oesophageal dysmotility and a previous endoscopic or radiological diagnosis of GERD and Barrett's oesophagus), gastrointestinal surgery (except appendectomy, colonic resection and cholecystectomy). Hiatus hernia and non-specific endoscopic or radiological findings were not exclusion criteria. NSAIDs, ASA (>325 mg/day), H2-RAs, PPIs, prokinetic agents, misoprostol or sucralfate were not permitted for 15 days prior to randomization, or during the study whilst warfarin, theophylline, phenytoin, bisphosphonates, methotrexate, ketoconazole, fluconazole, itraconazole, diazepam, aminopyrine and antipyrine were not permitted at any time during the study. Antacids were not permitted, other than alginate/antacid tablets (Gaviscon®; SmithKline Beecham Consumer Healthcare, Oakville, Ontario, Canada), provided during study visits as rescue medication. Women of childbearing potential were eligible if they used effective contraception and had a negative urine hCG pregnancy test at randomization.

Study design and medications

This was a prospective, randomized, controlled, double-blind, double-dummy active treatment trial, conducted at 46 primary care centres in Canada, with a 4–16-week treatment phase, followed by a treatment-free, follow-up phase up to a maximum of 6 months. At the initial screening visit, medical history and demographic features (including alcohol and tobacco use, age, gender and race) were recorded and a physical examination was performed; eligible patients received a supply of antacid/alginate tablets and a daily diary in which to record heartburn severity. On reassessment, 2 weeks later, patients with a heartburn score ≥4 on a 7-point Likert scale on at least three of the preceding 7 days were eligible for randomization to the ‘PPI-start’ or ‘H2-RA-start’ treatment strategies. Randomization, using a computer-generated table of random numbers, was concealed from patients, investigators and study personnel. Patient compliance with treatment was checked by counting the number of dispensed study tablets returned at each clinic visit. Patients were considered compliant if <25% of the dispensed study medication was returned, indicating that at least 75% had been consumed.

Treatment phase.  All patients received therapy, dispensed as four tablets daily (three tablets in the morning and one tablet in the evening) for the entire treatment phase of up to 16 weeks. For the ‘PPI-start’ strategy, patients received omeprazole 20 mg once daily (Losec MUPSTM AstraZeneca R&D, Mölndal, Sweden) and for the ‘H2-RA-start’ strategy, they received ranitidine 150 mg twice daily (Glaxo, Italy), each for 4 weeks. Double-blind dosing was maintained with an appropriate combination of active and placebo tablets (e.g. initially, in the ‘PPI-start’ strategy, patients received one active omeprazole tablet and one identical-looking placebo omeprazole tablet in the morning in addition to two placebo ranitidine tablets, one in the morning and one in the evening). Patients with intolerable heartburn at 4 weeks were stepped-up (‘PPI-start’: to omeprazole 40 mg once daily; ‘H2-RA-start’: to omeprazole 20 mg once daily) and reassessed at 8 and 12 weeks; all remaining patients, who did not report heartburn relief at 4 weeks, continued their initial therapy for another 4 weeks. At 8 weeks, all patients who did not report heartburn relief on their initial therapy were stepped up (‘PPI-start’: to omeprazole 40 mg once daily; ‘H2-RA-start’: to omeprazole 20 mg once daily) and reassessed at 12 and 16 weeks. Double-blind dosing was again maintained (e.g. ‘PPI-start’ patients who stepped up to omeprazole 40 mg once-daily were given two active omeprazole tablets in addition to two placebo ranitidine tablets, one in the morning and one in the evening). Patients who did not report heartburn relief at 12 weeks (after step up at 4 weeks) or at 16 weeks (after step up at 8 weeks), were withdrawn from the study and offered further investigation (endoscopy) in the ‘PPI-start’ strategy or open-label treatment with omeprazole 40 mg once daily in the ‘H2-RA-start’ strategy. Alginate/antacid tablets were supplied as rescue medication at a dose of two to four tablets, as required, up to a maximum of 16 tablets a day.

At all follow-up visits (4, 8, 12 and 16 weeks), patients with heartburn relief were withdrawn and entered into the treatment-free, follow-up phase.

Follow-up phase.  Patients’ symptom status in the 6-month follow-up phase was evaluated in a telephone interview after 4, 12 and 20 weeks and by direct, face-to-face interview after 8, 16 and 24 weeks by reference to the daily diary. The follow-up phase was terminated if heartburn had recurred on more than one of the previous 7 days with a severity score >3 on a 7-point Likert scale. Alginate/antacid tablets were supplied as rescue medication at a dose of two to four tablets, as required, up to a maximum of 16 tablets a day.

Measurement of outcomes

Primary outcome: heartburn relief.  The primary outcome of the study –‘treatment success’– was the proportion of patients with ‘heartburn relief’, defined as heartburn on no more than one of the previous 7 days, with a maximal allowable score of 3 on the 7-point Likert scale at 4 weeks (1 – no symptoms, 2 – minimal, 3 – mild, 4 – moderate, 5 – moderately severe, 6 – severe, 7 – very severe problem) (Table 1), based on the Global Overall Symptom (GOS) scale.29‘Sustained heartburn resolution’, defined as a maximal allowable score of 1 on all of the previous 7 days (i.e. no heartburn in the previous week) (Table 1), was a secondary outcome.

Table 1.  Definitions of symptom severity used to characterize patients’ responses during the study
Heartburn reliefNo more than mild heartburn (score ≤3/7) on no more than one day during the 7 days prior to the visit (patient diary)
Sustained heartburn resolutionNo heartburn (score 1/7) on all 7 days in the week prior to the visit (patient diary).
Overall relief of other reflux symptoms (heartburn, acid regurgitation, epigastric pain, nausea)No more than mild symptoms overall (score ≤3/7) over the 7 days prior to the visit (investigator query about ‘overall’ symptom response)
Complete overall relief of other reflux symptoms (heartburn, acid regurgitation, epigastric pain, nausea)No symptoms overall (score 1/7) over the 7 days prior to the visit (investigator query about ‘overall’ symptom response)
RelapseModerate to severe heartburn (score ≥4/7) on ≥2 days during a 7-day period in the treatment-free follow-up period (patient diary)

Relapse, during the treatment-free follow-up phase was defined as the occurrence of heartburn on ≥2 days of the previous 7 days with a severity score ≥4 on a 7-point Likert scale (Table 1).29

The assessment of heartburn symptoms and their severity during the 7 days before each clinic and telephone visit, was based on the patients’ record, in a daily diary, of the severity of any heartburn symptoms, during the previous day and night. The daily diary was completed during the 16-week treatment phase and the subsequent 6-month treatment-free follow-up phase, using a 7-point Likert scale.29

Secondary outcomes.  At each visit, patients used the 7-point Likert scale to rate the overall severity of any heartburn, regurgitation (‘flow of sour acidic or bitter fluid into the mouth’), epigastric pain (‘abdominal pain centred in the upper abdomen’) and nausea during the previous week. Patients reporting a score ≥4 (moderate) were defined as ‘symptomatic’. For each symptom, relief was defined as a score or 3 or less and resolution was defined as a score of 1 (Table 1). Vomiting was recorded as absent or present at each visit. At the screening and final visits in the treatment phase, patients reported the presence of IBS symptoms according to the Manning criteria.26

The number of antacids consumed each day was recorded in the daily symptom diary. Quality of life was assessed using validated, self-administered questionnaires at each study visit, after a training session at the screening visit. The Quality of Life in Reflux and Dyspepsia (QoLRAD)30 questionnaire assessed dimensions of emotional distress, sleep, vitality, food/drink habits, and physical/social functioning and the Gastrointestinal Symptom Rating Scale (GSRS)31 assessed dimensions of indigestion, diarrhoea, constipation, abdominal pain and reflux. At the final treatment visit of the study, the overall treatment evaluation (OTE)32 evaluated the magnitude of change in health status on a 15-point scale (−7 to −1 = worse; 0 = no change and 1 to 7 = better) and the Satisfaction Survey (SS) evaluated patient satisfaction with treatment using a 14-point scale ranging from completely satisfied (7) to completely dissatisfied (−7).

Helicobacter pylori. Helicobacter pylori status was determined by 13C-urea breath test at the randomization visit; both patient and investigator were blinded to the results until the study was completed.

Sample size calculation

The calculated sample size was adjusted to ensure adequate enrolment for a maintenance treatment study that followed the present acute treatment study.32 Assuming complete heartburn resolution rates of 61 and 40% (a treatment difference of 21%) for omeprazole and ranitidine,4 respectively, after 4 weeks of therapy, a two-sided significance level of 0.05 and a minimum power of 99.5% (Fisher's exact test), at least 125 completed patients were needed for each treatment arm. Allowing for a 15% withdrawal rate of evaluable patients due to study discontinuation or persistence of symptoms, 295 patients were needed for randomization to both treatment arms. In all, 390 patients were enrolled to ensure adequate numbers of patients for the subsequent maintenance treatment study.33

Data analysis

The primary analysis of efficacy was an intention-to-treat (ITT) analysis of treatment success (heartburn relief) at 4 weeks, using daily diary card responses, for all randomized patients. Based on a priori hypotheses, the proportions of patients with treatment success at 4 weeks were stratified by H. pylori status (positive or negative) and age (less than or greater than or equal to 50 years of age) and results were compared between treatment strategies using the Cochran–Mantel–Haenszel test, controlling for geographical region. In addition, a proportional hazards survival analysis was performed to identify independent predictors of symptom relief in the treatment phase and symptom relapse in the follow-up phase.

The cumulative proportions of patients with heartburn relief and sustained heartburn resolution over the 16-week treatment phase were compared between the treatment strategies using the Cochran–Mantel–Haenszel test. QoLRAD and GSRS results were analysed using a mixed model ancova method with change from baseline as response and treatment and geographical region as covariates. The SS and OTE results were summarized in frequency tables by treatment strategy. The cumulative proportions of patients with recurrent symptoms during the treatment-free observation period were compared using the Cochran–Mantel–Haenszel test.


Patient characteristics and disposition

In total, 613 patients were screened (Figure 1), of whom 390 fulfilled the inclusion criteria; 196 patients were randomized to the ‘PPI-start’ strategy and 194 to the ‘H2-RA-start’; strategy. There were no significant differences in clinical characteristics between the two groups (Table 2) and compliance with therapy (consumption of at least 75% of dispensed study medication) was excellent during the first 4 weeks (‘PPI-start’: 184/196; 93.9% [95% CI: 90.5–97.2%], ‘H2-RA-start’: 177/194; 91.2% [87.3–95.2%]) and for the entire treatment phase (‘PPI-start’: 185/196; 94.4% [91.2–97.6%], ‘H2-RA-start’: 180/194; 92.8% [89.1–96.4%]). Symptom frequency prior to therapy was comparable in the two treatment arms (Table 1); 195 of 390 (50.2%) patients reported heartburn and/or regurgitation and 180 of 390 (46.2%) reported heartburn and/or regurgitation in combination with epigastric pain and/or nausea (Figure 2). After randomization, 11 (2.8%) patients without symptoms were discontinued (5 –‘PPI-start’; 6 –‘H2-RA-start’) as they had not fulfilled the entry criteria [did not have moderate to severe heartburn symptoms for at least 3 days in the previous week before randomization (6), IBS (2), previous history of oesophagitis (2), prohibited medication (1)]; they were, however, included in the ITT analyses.

Figure 1.

Flow diagram showing numbers of patients enrolled and randomized to each treatment strategy, and reasons for discontinuation at baseline, 4 and 16-week time points during the 16-week treatment phase and the 24 weeks of treatment-free observation period.

Table 2.  Demographic characteristics of patients at baseline (intention-to-treat analysis)
 ‘PPI-start’n = 196‘H2-RA-start’n = 194
  1. PPI, proton pump inhibitor; H2-RA, H2-receptor antagonist; GI, gastrointestinal.

 Female, n (%)100 (51.0) 92 (47.4)
Age (years)
 Mean 44 (12) 44 (12)
 Minimum–maximum 20–83 18–76
 <50 years, n (%)138 (70.4)133 (68.6)
Prior upper GI investigations (>6 months previously)
 Endoscopy 13 (6.6%) 11 (5.7%)
 Radiology 52 (26.5%) 49 (25.3%)
 Endoscopy + radiology  7 (3.6%) 15 (7.7%)
Helicobacter pylori infection 30.0% 31.6%
 Current smoker, n (%) 52 (27%) 60 (31%)
 Alcohol use, n (%)127 (65%)117 (60%)
Reflux symptoms (score ≥4 at baseline)
 Heartburn190 (97.4%)180 (93.3%)
 Regurgitation 94 (48.2%)107 (55.4%)
 Epigastric pain 79 (40.5%) 84 (43.3%)
 Nausea 31 (15.9%) 29 (15.0%)
Heartburn duration
 <12 months 19 (9.7%) 14 (7.2%)
 >1 year ≤5 years 73 (37.2%) 52 (26.8%)
 >5 year ≤10 years 43 (21.9%) 43 (22.2%)
 >10 ≤ 20 years 34 (17.3%) 56 (28.9%)
 >20 years 27 (13.8%) 29 (14.9%)
 Mean (s.d.) years 9.4 (9.9) 11.1 (9.6)
Days (n) during previous week with heartburn score ≥3
 3 12 (6.1%) 18 (9.3%)
 4 15 (7.7%) 17 (8.8%)
 5 31 (16.3%) 22 (11.3%)
 6 40 (20.3%) 35 (18.0%)
 7 97 (49.5%)102 (52.6%)
Figure 2.

Venn diagram showing overall percentages of randomized patients (‘PPI-start’ and ‘H2-RA-start’, combined; n = 390) with each of the four major upper gastrointestinal symptoms; 11 patients had no recorded symptoms and were withdrawn from the study.

Treatment phase

Symptom response at 4 weeks.  From the daily diary, heartburn relief was reported by 55.1% (108/196; 95% CI: 48.1–62.1%; ‘PPI-start’) and 27.3% (53/194; 21.1–33.6%; ‘H2-RA-start’) of patients (Figure 3) with a difference in heartburn relief of 27.8% (18.4–37.2%; P < 0.001); sustained heartburn resolution was reported by 36.2% (71/196; 25.9–43.0%; ‘PPI-start’) and 12.9% (25/194; 8.2–17.6%; ‘H2-RA-start’) of patients (Figure 3) with a difference between groups of 23.3% (15.1–31.6%; P < 0.001). The median times to heartburn relief were 3 (2–5) and 8 (5–15) days and to sustained heartburn resolution were 5 (2–8) and 29 (15–35) days, respectively for ‘PPI-start’ and ‘H2-RA-start’, respectively. At 4 weeks, the ‘number needed to treat’ for the ‘PPI-start’ strategy to achieve an additional heartburn relief success was 4 (95% CI 3–5) and to achieve an additional sustained heartburn resolution success was 4 (95% CI 3–7) compared with the ‘H2-RA-start’ strategy.

Figure 3.

Percentages (95% CI) of patients, by treatment strategy, at 4, 8, 12 and 16 weeks of treatment, reporting heartburn relief (heartburn score <3 on no more than 1 day during the previous week; upper graph) and sustained heartburn resolution (heartburn score = 1 during the previous week; lower graph); P-values indicate differences between treatment strategies at each time point (Cochran–Mantel–Haenszel test – intention-to-treat analysis).

On univariate analysis, neither H. pylori status nor age was associated with a difference in heartburn relief (Table 3); however, proportional hazards survival analysis indicated that heartburn relief was less likely and slower if patients had (i) received ranitidine (‘H2-RA-start’) strategy: hazard ratio = 0.77 (95% CI 0.62–0.96, P = 0.02); (ii) more IBS symptoms, hazard ratio = 0.85 (0.74–0.98, P = 0.03); and (iii) a higher heartburn symptom score at the start, hazard ratio = 0.78 (0.69–0.89, P < 0.001).

Table 3.  Effect of Helicobacter pylori status and age on heartburn relief at 4 weeks
 ‘PPI-Start’ (% [95% CI] n)‘H2-RA-Start’ (% [95% CI] n)Total (% [95% CI] n)
  1. PPI, proton pump inhibitor; H2-RA, H2-receptor antagonist.

H. pylori positive59.7% [46.9–72.4%] 5732.2% [20.3–44.1%] 5945.7% [36.6–54.8%] 116
H. pylori negative53.4% [44.9–61.9%] 13324.2% [16.8–31.6%] 12839.1% [33.2–45.0%] 261
Age <50 years52.2% [43.8–60.5%] 13827.8% [20.2–35.4%] 13340.2% [34.4–46.1%] 271
Age ≥50 years62.1% [49.6–74.6%] 5826.2% [15.2–37.3%] 6143.7% [34.8–52.6%] 119

A global assessment of upper gastrointestinal symptoms over the 7-day period before the week 4 visit indicated heartburn relief in 75.8% (144/190; 69.7–81.9%) of the ‘PPI-start’ group compared with 57.8% (104/180; 50.6–65.0%) of the ‘H2-RA-start’ group (P = 0.0002) but there were no differences between treatment strategies with respect to regurgitation, epigastric pain, nausea or vomiting (data not shown).

Symptom response from 4 to 16 weeks.  ‘Step-up’ therapy was required by 143 patients (Figure 4), 51 of 196 (26.0%; 19.9–32.2%) from the ‘PPI-start’ group and 92 of 194 (47.4%; 40.4–54.5%) from the ‘H2-RA-start’ group by 8 weeks; step up at 4 weeks because of intolerable heartburn occurred in 16 of 196 (8.2%; 4.3–12.0%) and 37 of 194 (19.1%; 13.5–24.6%) patients, respectively.

Figure 4.

Flow diagram showing number of patients who achieved heartburn relief at each visit, were withdrawn for other reasons or required step-up therapy for each treatment strategy.

The ‘PPI-start’ strategy produced heartburn relief in a significantly greater proportion of patients up to 8 weeks and sustained heartburn resolution in a significantly greater proportion up to 12 weeks than did the ‘H2-RA-start’ strategy (Figure 2). There were no significant differences between the treatment strategies with respect to heartburn relief or sustained heartburn resolution by the end of the 16-week treatment period. The mean (s.d.) and median percentages of heartburn-free days during the entire 16-week treatment periods were: ‘PPI-start’: 52.2% (38.6%) and 58.2% and ‘H2-RA-start’: 41.0% (31.4%), and 38.0% (P = 0.002).

Quality of life and assessment of treatment effect.  At 4 weeks, patients in the ‘PPI-start’ group had significantly greater improvements in GSRS scores for indigestion, abdominal pain, reflux and overall GSRS domains (Table 4) and in all QoLRAD domains (Figure 5); however, by the end of the treatment phase (from 4 to 16 weeks), there was no difference between treatment groups for either scale. At the end of the treatment phase, OTE scores did not differ between treatment groups (data not shown) but SS were significantly higher in the ‘PPI-start’ group compared with the ‘H2-RA-start’ group (P = 0.0063; Table 5).

Table 4.  Mean Gastrointestinal Symptom Rating Scale (GSRS) scores (s.d.) for each of the five domains in patients at baseline, 4 weeks and completion of treatment (4–16 weeks) and at relapse (intention- to-treat population); P-values indicate whether the two strategies produced significantly different changes in symptom scores from baseline to 4 weeks and to 4–16 weeks and from 4 to 16 weeks to the time of relapse
 ‘PPI-start’ mean score (s.d.) n‘H2-RA-start’ Mean score (s.d.) nP-value for difference between strategies
  1. NS, not significant (P > 0.05).

  2. PPI, proton pump inhibitor; H2-RA, H2-receptor antagonist.

 Baseline4.3 (1.2) 1964.1 (1.2) 192 
 4 weeks2.1 (1.3) 1882.8 (1.4) 186<0.001 vs. baseline
 4–16 weeks1.7 (1.1) 1881.8 (1.1) 186NS vs. baseline
 Relapse3.8 (1.5) 1503.6 (1.5) 143NS vs. 4–16 wks
 Baseline3.1 (1.2) 1963.3 (1.4) 193 
 4 weeks2.4 (1.2) 1882.8 (1.7) 1870.02 vs. baseline
 4–16 weeks2.1 (1.2) 1882.2 (1.6) 187NS vs. baseline
 Relapse2.4 (1.1) 1502.5 (1.4) 142NS vs. 4–16 weeks
Abdominal pain
 Baseline2.9 (1.2) 1962.7 (1.1) 193 
 4 weeks2.1 (1.2) 1882.3 (1.2) 1870.014 vs. baseline
 4–16 weeks1.9 (1.1) 1881.8 (1.0) 187NS vs. baseline
 Relapse2.4 (1.1) 1502.3 (1.3) 143NS vs. 4–16 weeks
 Baseline2.0 (1.3) 1952.2 (1.4) 193 
 4 weeks1.8 (1.1) 1872.0 (1.5) 187NS vs. baseline
 4–16 weeks1.7 (1.0) 1881.8 (1.3) 187NS vs. baseline
 Relapse1.7 (1.0) 1501.7 (1.3) 142NS vs. 4–16 weeks
 Baseline1.9 (1.0) 1951.9 (1.2) 193 
 4 weeks1.7 (1.0) 1881.8 (1.3) 187NS vs. baseline
 4–16 weeks1.6 (0.9) 1881.7 (1.1) 187NS vs. baseline
 Relapse1.5 (0.8) 1501.6 (1.0) 143NS vs. 4–16 weeks
Total GSRS score
 Baseline2.8 (0.8) 1962.8 (0.9) 193 
 4 weeks2.0 (0.9) 1882.4 (1.1) 187<0.001 vs. baseline
 4–16 weeks1.8 (0.8) 1881.9 (1.0) 187NS vs. baseline
 Relapse2.4 (0.8) 1502.3 (1.0) 143NS vs. 4–16 weeks
Figure 5.

Changes in mean Quality of Life in Reflux and Dyspepsia scores for each domain assessed at 4 weeks, in comparison with baseline (upper graph: *P < 0.01, **P < 0.001) and at the time of heartburn relapse, in comparison with the end of the initial treatment period (lower graph: all P > 0.05).

Table 5.  Frequency distribution of Satisfaction Survey responses at the end of the treatment phase (4–16 weeks: intention-to-treat population)
 ‘PPI-start’, n (%) (n = 196)‘H2-RA-start’, n (%) (n = 194)
  1. PPI, proton pump inhibitor; H2-RA, H2-receptor antagonist.

Completely satisfied93 (47.4)62 (32.0)
Very satisfied52 (26.5)61 (31.4)
Quite satisfied13 (6.6)27 (13.9)
Satisfied17 (8.7)16 (8.2)
Dissatisfied7 (3.6)7 (3.6)
Completely dissatisfied0 (0.0)1 (0.5)

Outcomes after 16-week, blinded treatment phase.  After completion of the treatment phase, 22 patients had not achieved heartburn relief (‘PPI-start’: 13; ‘H2-RA-start: 9). Seven of the nine patients in the ‘H2-RA-start’ arm achieved heartburn relief after an 8-week open label course of omeprazole 40 mg daily. Of the 13 patients who underwent endoscopy because of persistent symptoms, six had mild, ‘LA’ grade ‘A’ (‘PPI-start’: 1, ‘H2-RA-start’: 2) or ‘LA’ grade ‘B’ (‘PPI-start’: 1, ‘H2-RA-start’: 2) oesophagitis.34, 35

Follow-up phase

Symptom response.  Of the 390 patients randomized, 308 patients with heartburn relief at the end of the treatment phase (‘PPI-start’: 159, ‘H2-RA-start’: 149) were eligible to enter the treatment-free, follow-up phase; after withdrawal of ineligible patients [e.g. withdrew consent (11), lost to follow-up (10), moved (2), other (4)], 281 patients were eligible to start the treatment-free, follow-up phase (‘PPI-start’: 145, ‘H2-RA-start’: 136). There were no differences in relapse rates between patients assigned to the initial ‘PPI-start’ and ‘H2-RA-start’ treatment strategies (Figure 6). Based on the daily patient diary data, heartburn relapse occurred in 79% (114/145; 72–85%: ‘PPI-start’) and 76% (103/136; 69–83%: ‘H2-RA-start’) at 6 months. Median times to relapse were 8 (7–9) and 9 (8–11) days, respectively. The median number of days to the first occurrence of moderate-to-severe heartburn (score ≥4) was 6 (95% CI 5–7) days and was comparable for the two treatment strategies; the mean proportion of days spent heartburn free, from the time between heartburn relief and relapse, was also comparable between strategies (‘PPI-start’: 22% [18–26%], ‘H2-RA-start’: 23% [18–27%]; P = 0.33).

Figure 6.

Incremental percentages of patients, by treatment strategy, reporting heartburn relapse (upper graph) and a first episode of moderate to severe heartburn (lower graph) during the 6-month treatment-free observation period; P-values indicate differences between treatment groups (Cochran–Mantel–Haenszel test – intention-to-treat analysis).

Proportional hazards survival analysis indicated that symptom relapse was more likely and faster if patients had (i) a long history of symptoms, hazard ratio = 1.03 per year (1.02 to 1.04, P < 0.001); (ii) more severe regurgitation symptoms, hazard ratio = 1.15 (1.05–1.25, P = 0.002); and (iii) no H. pylori infection, hazard ratio = 1.46 (1.10–1.93, P = 0.009).

Quality of life and assessment of treatment effect.  At the time of relapse in the follow-up phase, GSRS scores had deteriorated significantly for all domains (P < 0.03) except diarrhoea and QoLRAD scores had deteriorated significantly for all domains (P < 0.0001) compared with the end of the treatment phase, but there were no differences for either GSRS or QoLRAD with respect to the initial treatment strategy and no differences between the initial treatment strategies with respect to the magnitude of the decline in GSRS and QOLRAD scores (Table 3).

Antacid consumption

Daily intake [mean (s.d.)] of antacid rescue medication [‘PPI-start’: 4.9 (3.8) tablets vs. ‘H2-RA-start’: 4.2 (3.0); P = 0.192; Wilcoxon rank sum test] prior to randomization was similar for the two treatment strategies. During the active treatment phase, daily antacid intake decreased and was lower in the ‘PPI-start’ group [1.0 (1.8) tablets than in the ‘H2-RA-start’ group] [1.2 (1.6); P = 0.003]. Antacid intake increased again in the treatment-free, follow-up phase but there was no difference between the two treatment groups [‘PPI-start’: 2.7 (2.8) tablets vs. ‘H2-RA-start’: 2.5 (2.4); P = 0.989].

Adverse events

Both treatment strategies were safe and well tolerated. The frequency and total number of serious adverse events (SAEs) and adverse events reported were similar in both treatment arms. Altogether, seven non-fatal SAEs were reported; there were four SAEs reported in the ‘PPI-start’ (myocardial infarction, incarcerated ventral hernia, deep venous thrombosis and miscarriage) and three SAEs in ‘H2-RA-start’ groups (cholelithiasis with post-operative pain, myocardial ischaemia and lower abdominal pain); no SAE was judged by the investigators to be causally related to study therapy. The most frequently reported adverse events overall were headache (7.5%), respiratory infection (8.2%), diarrhoea (7.0%), and abdominal pain (6.6%) and these were reported equally in both strategies.


Heartburn is common36–39 and, with regurgitation, it is a cardinal symptom of GERD; it is also common in association with other symptoms of dyspepsia.13–18 Despite management guidelines for GERD3, 22, 23 and dyspepsia,18, 40 and studies of empiric therapy in dyspepsia patients, many of whom may have heartburn,13–16 there are few data on the initial, symptom-based treatment of primary care dyspepsia patients with dominant heartburn, the effect of different empiric treatment strategies on heartburn relief over periods of up to 16 weeks or on relapse of heartburn after treatment cessation.41–44

In the present study, initial therapy with a standard-dose PPI (omeprazole 20 mg once daily) produced heartburn relief in twice as many, and sustained heartburn resolution in three times as many patients as a standard-dose H2-RA (ranitidine 150 mg b.d.) at 4 weeks and this was associated with a greater improvement in health-related quality of life and a reduced requirement for ‘step-up’ therapy after 4–8 weeks. The ‘PPI start’ strategy was dominant until 12 weeks for sustained heartburn resolution and, although heartburn relief was eventually achieved in about 80% of patients with either strategy, the rapidity of symptom resolution was significantly higher with the ‘PPI-start’ strategy and, by 8 weeks, it had produced comparable results for sustained heartburn resolution to those produced by the ‘H2-RA-start’ strategy at 16 weeks. Of patients in the H2RA-start group, 92 of 194 (47.4%; 37 at 4 weeks and 55 at 8 weeks) stepped up to PPI therapy. Heartburn control was subsequently achieved in 69 of these 92 patients (75.0%). After treatment discontinuation for patients who had achieved heartburn relief, relapse rates were comparable for the ‘PPI-start’ and ‘H2RA-start’ treatment groups.

A small proportion of patients remained symptom-free, off therapy, at the end of the 6-month follow-up phase but the majority had recurrent symptoms within a few weeks, sufficient to warrant resumption of treatment. Thus, initial therapy with a PPI rather than an H2-RA does not appear to predispose to more rapid relapse after cessation of acute therapy.

Enrolment in this study, as in previous GERD studies,4–10 was predicated on the symptom of ‘retrosternal burning, rising towards the neck’.1 In an endoscopic study in Canadian primary care practice,17 about 55% of patients with dominant heartburn had erosive oesophagitis, similar to the findings of other endoscopic studies.5, 7, 45 These data support the recommendation18 that patients with heartburn-dominant dyspepsia should be offered empiric acid suppressive therapy, before non-invasive testing for H. pylori infection.27, 28 This does not preclude endoscopy46 but, although heartburn may be associated with peptic ulcer disease,17 Barrett's oesophagus or malignancy,47 clinically-significant lesions are infrequent in the absence of alarm features17, 18 and severe oesophagitis (LA Grades ‘C’ and ‘D’) occurs in <5–10% of patients with heartburn-dominant symptoms. In patients who had persistent symptoms after the 16-week treatment phase, the only significant endoscopic findings were LA Grade A or B oesophagitis. There is evidence that, for patients receiving PPIs, heartburn relief is associated with healing of erosive oesophagitis.48 Thus, a treatment strategy that ‘starts high’, with the most effective acid suppression therapy,8–10 is likely to be safe and effective for the vast majority of patients with heartburn-dominant upper gastrointestinal tract symptoms. Symptoms will persist in some patients who will, therefore, need to be re-evaluated to determine whether alarm features are present or whether there is another underlying condition that requires further investigation18 but the empiric approach to therapy evaluated in this study does, at least, obviate the need for investigation in a high proportion of patients.

Proton pump inhibitors produce healing and symptom resolution in a higher proportion of erosive oesophagitis patients than do H2-RAs or prokinetics3, 19, 49 and more potent gastric acid suppression is associated with a further improvement in outcomes.8–10 Standard dose PPIs also produce symptom resolution in a higher proportion of ENRD patients than do low-dose PPIs,6 H2-RAs4, 50 and prokinetics.7, 21 In the present study, overall heartburn relief rates, with standard-dose PPI or H2-RA therapy, were similar to those observed after 4 weeks in a previous Canadian study11 of GERD patients with and without erosive oesophagitis, after endoscopy. Endoscopy was not performed at enrolment in the present study; although it would have allowed risk stratification with respect to oesophagitis severity 8, 34, 35 and H. pylori infection,11, 51 it would also have introduced the potential for bias52, 53 in a study intended to assess heartburn response to empiric therapy in a primary care environment.

The primary outcome measure –‘heartburn relief’ (Table 1) – was selected because it was considered to be a reasonable clinical outcome; the more stringent, secondary outcome –‘sustained heartburn resolution’– was associated with markedly lower response rates. The 7-point Likert scale used to rate symptom severity was identical in wording to the GOS scale29 used in the related CADET-Hp27 (H. pylori positive uninvestigated dyspepsia) and CADET-HN28 (H. pylori negative uninvestigated dyspepsia) studies, although the evaluation period (7 days) was different.28 Heartburn resolution was identified less frequently by patients in their daily diary assessments than when heartburn (as well as regurgitation, epigastric pain, nausea and vomiting) were assessed overall for the previous week. This suggests that the daily diaries were completed regularly; if there had been significant ‘hoarding’, one would have expected less discrepancy between the daily diary data and the retrospective assessment of symptoms. Both measures of symptom response were based on the patients’ recall, consistent with the increasing emphasis on patient reported outcomes.54–56 As in previous studies,30, 57 symptom relief was associated with an improvement in patients’ quality of life. In the present study, the disease-specific, quality of life tool (QOLRAD) demonstrated a significant difference between strategies after 4 weeks but not at the end of therapy; similarly, the OTE was comparable for the two strategies by the end of therapy as were the proportions of patients in each group with symptom resolution. On the contrary, the SS results suggested a persistent difference in outcomes for the two groups, perhaps because of the more rapid and complete symptom response in the ‘PPI-start’ group.

The traditional ‘step-up’ approach,22, 23 progressing from ‘life-style modifications’ to over-the-counter medications and prescription H2-RAs and PPIs has been challenged3, 18 by data demonstrating the superiority of PPI therapy in randomized trials19, 20 and in a comparison of ‘step up’ and ‘step down’ strategies.58 The present study did not compare ‘step up’ and ‘step down’ GERD strategies directly as the latter addresses both initial therapy and a switch to maintenance therapy. Instead, the study evaluated two initial treatment strategies, with respect to treatment response and safety, for patients with heartburn-dominant symptoms; the strategies were intended to be relevant to primary care, clinical practice42 and consistent with published guidelines.18 although this paper has not addressed the economic consequences of these strategies. The subsequent treatment-free, follow-up phase, was designed to determine how many patients would develop recurrent symptoms during a 6-month period, requiring further medical therapy.33

Risk factors for delayed symptom resolution included the presence of IBS symptoms at baseline, consistent with previous observations.59 The relationship between H. pylori status and GERD remains controversial.60, 61 In this study, neither univariate nor multivariate analysis indicated a significant effect of H. pylori on symptom relief although H. pylori negative subjects were at greater risk of recurrent symptoms after cessation of therapy. An interaction between H. pylori and acid suppression therapy appears to be less marked in HDUD than in GERD patients.11, 51 Despite some evidence that H. pylori infection is less prevalent in GERD patients,17, 60 a possible association between H. pylori infection and delayed GERD symptom relapse is probably not clinically significant.

In summary, empiric, ‘PPI-start’ therapy produces relief of symptoms and improvement in health-related quality of life in a higher proportion of primary care patients with uninvestigated heartburn-dominant dyspepsia after 4 weeks than does an ‘H2-RA-start’ strategy. A ‘step up’ of therapy in non-responders, at 4–8 weeks, produces a further increase in symptom response in both groups and, although both groups achieve good response rates by 16 weeks, sustained symptom relief at 8 weeks with the ‘PPI-start’ regimen is similar to that achieved at 16 weeks with the ‘H2-RA-start’ regimen. This study confirms the appropriateness of a PPI-based, empiric treatment strategy for primary care practice patients with heartburn-dominant symptoms to provide effective acid suppression and rapid symptom relief. The study also confirms that GERD is a chronic condition; symptoms recur rapidly after cessation of therapy, suggesting that some form of maintenance therapy is required for the majority of patients with heartburn-dominant, uninvestigated dyspepsia. The role of ‘on demand’ therapy in patients with uninvestigated heartburn-dominant dyspepsia33 will be reported separately.


This study was financially supported by AstraZeneca Canada Inc.


The CADET HR Study group of principal investigators are: S. Arndt, Regina; M. D. Atin, Toronto; M. Audet, Ste-Foy; J. Bachand, Sherbrooke; G. Baran, Kingston; K. Bayly, Saskatoon; S. Behiels, Edmonton; B. Bellemare, Granby; W. Booth, Antigonish; R. Bouchard, Sherbrooke; M. Bradette, Quebec; L. Breger, Pointe-Claire; S. Brien, Peterborough; F. Bursey, St John's; A. Cockeram, Saint John; A. Cohen, Montreal; H. Conter, Halifax; VS. J. Coyle, Winnipeg; B. N. Craig, Saint John; D. Crowley, Strathroy; G. D'Ignazio, Hawkesbury; O. Doiron, Saint John; J. Dollin, Ottawa; J. Drexler, Winnipeg; B. Elfassy, St-Laurent; C. Ellis, Calgary; D. Fay, Granby; W. Fouad, Winnipeg; G. Fullerton, Woodstock; R. Hamilton, Wolfville; J. Hii, Vancouver; R. Hilsden, Calgary; B. Hogenbirk, Pointe-Claire; W. P. House, Vancouver; E. Howlett, Saskatoon; D. Johnson, Winnipeg; H. A. Johnston, Orleans; A. Kelly, Edmonton; J. Kennedy, Antigonish; E. N. L. Larkai, Saskatoon; Y. Lee, North York; J. Li, Moncton; J. Loutfi, Pierrefonds; J. MacKenzie, Ottawa; W. P. McKeough, London; J. McMorran, Vancouver; J. T. Momoh, Winnipeg; A. Munroe, Saint John; W. Olsheski, Toronto; P. O'Shea, St John's; G. Pannozzo, Waterloo; W. Paterson, Kingston; L. Pontikes, Regina; I. Prokopiw, London; G. Rideout, Goulds; L. Rochon, Ottawa; F. Schweiger, Saint John; V. Senikas, Montreal; D. Shu, Coquitlam; R. J. Smith, Mount Pearl; G. Tellier, St-Jerome; A. Thomson, Edmonton; F. Turcotte, Sherbrooke; S. J. O. Veldhuyzen van Zanten, Halifax; G. R. Webb, Grand Bay-Westfield; R. Whatley, Peterborough; D. P. Willoughby, Woodstock; B. Zidel, Mississauga.