The prevalence of clinically significant endoscopic findings in primary care patients with uninvestigated dyspepsia: the Canadian Adult Dyspepsia Empiric Treatment – Prompt Endoscopy (CADET–PE) study

Authors


Dr Alan B. R. Thomson, 519 Robert Newton Research Building, 11315–87 Avenue, Edmonton, Alberta, Canada T6G 2C2.
E-mail: alan.thomson@ualberta.ca

Summary

Background: Uninvestigated dyspepsia is common in family practice. The prevalence of clinically significant upper gastrointestinal findings (CSFs) in adult uninvestigated dyspepsia patients, and their predictability based on history, is unknown.

Methods: Prompt endoscopy was performed within 10 days of referral, in 1040 adult patients presenting with uninvestigated dyspepsia at 49 Canadian family practitioner centres. Subsequent management strategies during a 6-month follow-up period were determined by the individual family practitioners.

Results: CSFs were identified in 58% (603/1040) of patients. Erosive oesophagitis was most common (43%; N = 451); peptic ulcer was uncommon (5.3%; N = 55). Alarm symptoms were uncommon (2.8%; N = 29). Most patients had at least three dyspepsia symptoms, more than 80% had at least six, and approximately half had eight or more. Based on the dominant symptom, 463 (45%) patients had ulcer-like, 393 (38%) had reflux-like and 184 (18%) had dysmotility-like dyspepsia. The patients' dominant symptom was not predictive of endoscopic findings. Oesophagitis was more common in those with dominant reflux-like symptoms and was the most common finding in all subgroups. The prevalence of gastroduodenal findings was similar in all symptom subgroups. Helicobacter pylori (H. pylori) infection (30%; 301/1013) was associated with gastroduodenal findings.

Conclusions: Dyspepsia subclassifications, based on dominant symptom, are of limited value in predicting the presence and nature of CSFs. Oesophagitis was by far the most common diagnosis (43% of patients). CSFs were common in uninvestigated dyspepsia patients and their nature suggests patients could be initially treated effectively, without endoscopy, using empirical acid suppressive therapy.

Introduction

Dyspepsia is a common condition that is reported by up to 40% of the general population.1, 2 Patients with dyspepsia account for approximately 7% of all family physician visits in Canada.3 Dyspepsia defines an upper gastrointestinal symptom complex characterized by epigastric pain or discomfort and may include heartburn, acid regurgitation, excessive burping/belching, abdominal bloating, feeling of abnormal or slow digestion, early satiety or nausea.4, 5 These symptoms are often clustered into subgroups: dysmotility-like, ulcer-like, reflux-like or unspecified dyspepsia, although the clinical significance of doing so is uncertain.3, 6–8

Management requires the family physician to decide whether it is appropriate to arrange initial investigation of the upper gastrointestinal tract by endoscopy or barium X-ray before starting therapy, to ‘test and eradicate’Helicobacter pylori (H. pylori) infection or to start empirical symptomatic therapy. In practice, empirical anti-secretory treatment is commonly the first step unless the patient has new onset symptoms and is older or has alarm features, despite concerns that this approach may miss clinically significant upper gastrointestinal tract lesions.4, 8

The Canadian Adult Dyspepsia Empiric Treatment (CADET) series of studies was developed to define effective treatment strategies for uninvestigated dyspepsia at the primary care level. The CADET program included studies focusing on H. pylori-positive (CADET–H. pylori) and H. pylori-negative (CADET–HN) patients who all had epigastric pain and/or discomfort, as well as the CADET–HR study that focused on patients with heartburn as the dominant symptom in the dyspepsia symptom complex.9–11 The present study, the Prompt Endoscopy (CADET-PE) study was designed to estimate the frequency of clinically significant findings at endoscopy in a similar population of patients presenting to their family physicians with uninvestigated dyspepsia, and to evaluate whether specific symptoms were predictive of endoscopic findings.

Methods

Patients

Forty-nine family physician practices across Canada enrolled patients. Each centre was linked to a gastroenterology unit where endoscopies were performed. The study protocol was approved by local research ethics boards. Included patients were at least 18 years of age, provided written informed consent, and had a primary complaint of at least 3 months of either continuous or intermittent dyspepsia. This study used the Canadian Dyspepsia Working Group definition of dyspepsia, which describes dyspepsia as a symptom complex of epigastric pain or discomfort in association with other upper gastrointestinal symptoms, including heartburn and acid regurgitation.4 Importantly, patients reporting heartburn or acid regurgitation as their sole symptom were not included in this study.

Exclusion criteria were: a documented history of upper gastrointestinal pathology or upper gastrointestinal surgery, clinical investigation of dyspepsia by endoscopy or radiology in the previous 6 months or more than twice in the past 10 years, H. pylori eradication treatment in the previous 6 months, irritable bowel syndrome as assessed by the presence of three or more manning criteria,12 or severe concurrent disease. Patients could not have used proton pump inhibitors within 30 days or prokinetics or prescription H2-receptor antagonists (H2RAS) within 14 days of enrolment.

Study procedures

Baseline assessments included a medical history, physical examination, medication use and blood tests for clinical chemistry and haematology. Dyspepsia symptoms and alarm features (vomiting, gastrointestinal bleeding, abdominal mass, dysphagia, unexplained weight loss and anaemia) were recorded. The family physician did not prescribe proton pump inhibitors, H2RAs or prokinetics at this point, but patients were allowed to take antacids.

Within 10 days of initial consultation, a gastroenterologist performed an upper gastrointestinal endoscopy. Patients who were willing to participate, but who refused an endoscopy, underwent a 13C-urea breath test. The severity of reflux oesophagitis was graded using the Los Angeles classification.13 Gastroduodenal erosions were defined as superficial mucosal defects (< 5 mm in diameter) with a flat edge and could be either white or yellow. A gastric or duodenal ulcer was defined as a mucosal break > 5 mm in diameter. If patients had both an ulcer and erosions, they were counted only as having an ulcer. Gastroenterologists took biopsies from all endoscoped patients for the histological diagnosis of H. pylori. Also, when indicated, biopsies were taken to confirm diagnoses suspected at the time of endoscopy (e.g. Barrett's oesophagus). Barrett's oesophagus was diagnosed based on the endoscopic appearance of the distal oesophagus, suggesting gastric metaplasia. The number of biopsies taken was at the discretion of the endoscopist, as this depended on the length of the Barrett's epithelium. In keeping with recent guidelines, a final diagnosis of Barrett's epithelium was only made if intestinal metaplasia was histologically confirmed.

All endoscopic findings were reviewed by the study steering committee to determine whether they were clinically significant. Clinically significant findings were defined a priori in the protocol as lesions that could be responsible for the presenting dyspepsia symptoms or for a change in the patients' management. Any number of erosions was considered to be clinically significant, but the number of erosions was not recorded. Of note, gastritis (observed upon endoscopy or histology) was not considered to be a clinically significant finding. Similarly, as the clinical significance of hiatus hernia is uncertain, it was not considered to be a clinically significant finding. Patients with more than one clinically significant finding were counted only once in determining the overall prevalence of clinically significant findings.

Following endoscopy, the gastroenterologist could initiate treatment based on findings, or make a recommendation to the family practitioner regarding an appropriate management strategy. Patients were followed for 6 months, during which time the family physicians managed each patient according to their standard practice.

In parallel to the study reported here, we also examined the inter-observer agreement between endoscopists for upper gastrointestinal lesions by sending them a series of endoscopy video clips for interpretation. Inter-observer agreement was high (κ ∼ 0.70) for clinically significant lesions such as peptic ulcer and erosive oesophagitis, and for normal findings.14 The full data from this study will be reported separately.

Assessments

Dyspepsia Symptom Sub-group classification.

At baseline, the patient completed a 14-item symptom checklist and ranked, in order, the three symptoms that they found most bothersome.15 Patients were classified as having ‘ulcer-like’ dyspepsia if the most bothersome symptom was one of upper abdominal pain, pain often relieved by food, pain often relieved by antacids, pain often before meals or when hungry, periodic pain or night pain. Patients were classified as having ‘reflux-like’ dyspepsia if the most bothersome symptom was either heartburn or acid regurgitation. Patients were classified as having ‘dysmotility-like’ dyspepsia if the most bothersome symptom was one of nausea or vomiting, abdominal bloating and visible distension, anorexia or weight loss (≥ 7 lb), pain often aggravated by food or milk, pain often after meals or pain often relieved by belching.7, 15

Safety.

Endoscopy-related adverse events and all serious adverse events occurring throughout the study were recorded and assessed by the investigators on the basis of intensity and causality in relation to study procedures.

Data analysis

The estimated sample size of 1024 patients was calculated based on an estimated prevalence of clinically significant findings of 40%,16 with a margin for error of ± 3% in the estimation. Summary statistics, including 95% confidence intervals, were used to describe baseline characteristics, clinically significant endoscopic findings, adverse events and laboratory findings.

Results

Patients

Between October 1999 and January 2001, a total of 1100 patients (93.9% of 1171 patients enrolled into the study) consented to endoscopy. Sixty patients (5.1% of all patients) who consented to endoscopy did not undergo the procedure for the following reasons: eligibility criteria not fulfilled (N = 27), lost to follow-up (N = 3), withdrew consent (N = 9), non-compliant with the protocol (N = 1), endoscopy-intolerable (N = 2) and other (N = 18); the most common ‘other’ reason was that the endoscopy could not be performed within the specified time period (N = 7). Seventy-one patients (6.1% of all patients) did not consent to endoscopy, but underwent a 13C-urea breath test (UBT). Table 1 shows the similar demographic characteristics and history of dyspepsia symptoms of the 1040 patients who had an endoscopy and the 71 patients who underwent a UBT. Nearly two-thirds of the patient population was under the age of 50.

Table 1.  Demographics and baseline dyspeptic symptoms for 1040 patients who underwent endoscopy and 71 patients who declined endoscopy but underwent a urea breath test (UBT)
  Endoscopy1 (N = 1040)UBT1 (N = 71)
  • 1

     Percentage does not add to 100% because of rounding.

Sex, n (%)Male520 (50)32 (45)
Race, n (%)Caucasian991 (95)70 (99)
Black24 (2)1 (1)
Oriental15 (1)0 (0)
Aboriginal/Métis10 (1)0 (0)
Age (years)< 50 n (%)661 (64)39 (55)
Mean45.648.6
Range18–8419–82
Nicotine use, n (%)Current smoker296 (29)21 (30)
Province, n (%)Alberta (6 sites)87 (8)0 (0)
Newfoundland (3 sites)70 (7)2 (3)
Nova Scotia (3 sites)102 (10)17 (24)
Ontario (22 sites)467 (45)18 (25)
Quebec (12 sites)260 (25)33 (46)
Saskatchewan (3 sites)54 (5)1 (1.4)
Other current nicotine use8 (< 1)1 (1)
Alcohol use, n (%)Yes592 (57)37 (58)
Symptom duration (years)Median3.24.6
Mean6.56.5
Range0.2–44.70.3–29.8
Symptom duration, n (%)< 1 years244 (24)17 (24)
1 to < 2 years166 (16)7 (10)
2 to < 5 years230 (22)16 (23)
5 to < 10 years173 (17)11 (16)
10 to < 20 years149 (14)18 (25)
20 to < 50 years78 (8)2 (3)
Education, n (%)Partial elementary22 (2)5 (7)
Completed elementary52 (5)5 (7)
Partial secondary201 (19)16 (23)
Completed secondary261 (25)16 (23)
Partial college/university161 (16)9 (13)
Completed college/university294 (28)18 (25)
Completed postgraduate47 (5)2 (3)

Prevalence of clinically significant findings at endoscopy

Clinically significant findings were observed in 603 of 1040 patients (58.0%; 95% CI 55.1–61.1). The proportions (95% CI) of patients with findings in the oesophagus, stomach and duodenum were 47.1% (44.1–50.1), 18.3% (15.9–20.6) and 9.7% (7.9–11.5), respectively. Clinically significant findings were more prevalent in older patients (≥ 50 years of age) than in younger patients, particularly in the stomach (Figure 1). Oesophagitis was the most common endoscopic finding (43.4%, 451/1040, 95% CI 40.4–46.4), with the majority of patients being classified as Los Angeles grades A or B (51.4%, 232/451 and 35.7%, 161/451, respectively). More severe, Los Angeles grades C and D oesophagitis was reported in 9.8% (44/451) and 3.1% (14/451) of oesophagitis patients, respectively. Peptic ulcer disease was observed in 55 patients (5.3%; 95% CI 3.9–6.6; 31 patients had gastric ulcer, 29 had duodenal ulcer, and five patients had both). Eleven patients were found to have both oesophagitis and either a gastric or duodenal ulcer. Table 2 presents by age group the individual clinically significant findings observed in at least 1% of patients.

Figure 1.

Prevalence of clinically significant findings by age.

Table 2.  Prevalence and 95% confidence intervals of clinically significant findings by region (only findings noted in ≥ 1% of patients are presented)
 Age: < 50 (N = 661)Age: ≥ 50  (N = 379)Total (N = 1040)
  • 1

     Patients with both gastric ulcer and duodenal ulcer were counted only once for the determination of peptic ulcer prevalence (e.g. for patients < 50 years of age, 15 had gastric ulcer and 17 had duodenal ulcer. Three patients in this age group had both types of ulcers. So, the number of patients with peptic ulcer in this age group was 15 + 17–3 = 29).

Oesophagus, n (%)
Reflux oesophagitis287 (43.4)164 (43.3)451 (43.4)
[39.6–47.2] [38.3–48.3] [40.4–46.4]
Benign stricture17 (2.6)19 (3.0)36 (3.5)
[1.4–3.8] [2.8–7.2] [2.4–4.6]
Barrett's oesophagus10 (1.5)15 (4.0)25 (2.4)
 (confirmed by pathology) [0.6–2.4] [2.0–5.9] [1.5–3.3]
Stomach/duodenum, n (%)
Peptic ulcer129 (4.4)26 (6.9)55 (5.3)
[2.8–5.9] [4.3–9.4] [3.9–6.6]
Gastric ulcer15 (2.3)16 (4.2)31 (3.0)
[1.1–3.4] [2.2–6.2] [1.9–4.0]
Duodenal ulcer17 (2.6)12 (3.2)29 (2.8)
[1.4–3.8] [1.4–4.9] [1.8–3.8]
Gastric erosions59 (8.9)43 (11.4)102 (9.8)
[6.8–11.1] [8.2–14.5] [8.0–11.6]
Duodenal erosions39 (5.9)15 (4.0)54 (5.2)
[4.1–7.7] [2.0–5.9] [3.8–6.5]

Clinically significant findings by dyspepsia symptom sub-group

Almost all patients (99.4%) had at least three dyspepsia symptoms at baseline, and more than 80% had at least six symptoms. Less than 0.1% of patients reported only one dyspepsia symptom. Upper abdominal pain (34%, N = 357), heartburn (25%, N = 255) and acid regurgitation (13%, N = 138) accounted for 72.1% of the symptoms ranked most bothersome (Figure 2).

Figure 2.

Most severe dyspeptic symptom at baseline.

There was considerable overlap among dyspepsia symptom subgroups as described by Talley et al.;7 only 10%, 5% and 3% of patients had symptoms that could allow categorization into reflux-like, ulcer-like and dysmotility-like subgroups, respectively. When classified according to most bothersome symptom reported, ulcer-like, reflux-like and dysmotility-like symptoms were dominant in 45% (N = 463), 38% (N = 393) and 18% (N = 184) of patients, respectively. There were no age differences in different subgroups (data not shown).

Patients with dominant reflux-like dyspepsia had a higher prevalence of clinically significant oesophageal findings than did those with dominant ulcer-like symptoms, who in turn had a higher prevalence of clinically significant oesophageal findings than those with dominant dysmotility-like symptoms (Figure 3). However, 36.5% (236/647) of patients with endoscopic oesophagitis did not have dominant heartburn or acid regurgitation (Table 3). Also, regardless of dominant symptom sub-group, the prevalences of gastric and duodenal findings were similar (Figure 3, Table 3).

Figure 3.

Prevalence (95% CI) of clinically significant findings by dyspepsia sub-group (based on most bothersome symptom reported).

Table 3.  Prevalence and 95% confidence intervals of reflux oesophagitis, gastric and duodenal ulcer, and gastric and duodenal erosions by dyspepsia symptom sub-group (classified according to the reported dominant symptom)
n (%) [95%CI]Reflux
oesophagitis
Gastric
ulcer
Duodenal
ulcer
Gastric
erosions
Duodenal
erosions
All patients (N = 1040)451 (43.4)31 (3.0)29 (2.8)102 (9.8)54 (5.2)
[40.4–46.4] [1.9–4.0] [1.8–3.8] [8.0–11.6] [3.8–6.5]
Reflux-like (N = 393)215 (54.7)7 (1.8)10 (2.5)40 (10.2)25 (6.4)
[49.8–59.6] [0.5–3.1] [1.0–4.1] [7.2–13.2] [3.9–8.8]
Non-reflux-like (N = 647)236 (36.5)24 (3.7)19 (2.9)62 (9.6)29 (4.5)
[32.8–40.2] [2.3–5.2] [1.6–4.2] [7.3–11.9] [2.9–6.1]
Ulcer-like (N = 463)175 (37.8)22 (4.8)17 (3.7)44 (9.5)22 (4.8)
  [33.4–42.2] [2.8–6.7] [2.0–5.4] [6.8–12.2] [2.8–6.7]
Dysmotility-like (N = 184)61 (33.2)2 (1.1)2 (1.1)18 (9.8)7 (3.8)
[26.4–40.0] [-,-] [-,-] [5.5–14.1] [1.0–6.6]

Clinically significant findings by ASA/NSAID use

Use of acetylsalicylic acid or non-steroidal anti-inflammatory drugs (ASA/NSAIDs) alone or in combination was reported by 20% (131/662) of patients under the age of 50 years, and by 28% (105/379) of older patients. Clinically significant findings were more prevalent with use of these medications. Gastric lesions were more prevalent both in patients under 50 (ASA/NSAID: 21.4% (95% CI 14.4–28.4) vs. no ASA/NSAID: 14.3% (11.4–17.3) and in patients over 50 years of age [ASA/NSAID: 33.3% (24.3–42.4) vs. no ASA/NSAID: 18.6% (14.0–23.2)]. Gastric ulcers, but not duodenal ulcers, were more common if patients took ASA/NSAIDs (Figure 4), and were more prevalent in older patients [over 50 years: 8.6% (3.2–13.9) vs. ≤ 50 years: 3.8% (0.0–7.1)].

Figure 4.

Most prevalent (95% CI) clinically significant findings by ASA/NSAID use. Gastric ulcer: ASA: 7/93 patients (7.5%; 95% CI 2.2–5.4); NSAID: 8/161 patients (5.0%; 95% CI 1.6–8.3). Duodenal ulcer: ASA: 4/93 patients (4.3%; 95% CI not calculated); NSAID: 5/161 patients (3.1%; 95% CI not calculated).

Clinically significant findings by H. pylori status

H. pylori infection was confirmed in 30% of patients (301/1013) for whom gastric biopsy data were available. The prevalence of overall clinically significant findings was lower in H. pylori-infected patients. Gastric and duodenal ulcers were more common in H. pylori-positive patients (5.7%; 95% CI 3.0–8.3 vs. 2.0%; 95% CI 0.9–3.0 for gastric ulcer and 6.6%; 95% CI 3.8–9.5 vs. 1.3%; 95% CI 0.4–2.1 for duodenal ulcer) (Figure 5). Taking into account the fact that five patients had both a gastric ulcer and a duodenal ulcer (three infected and two uninfected patients), the prevalence of peptic ulcer disease was 11.3% (34/301) in H. pylori-positive patients and 2.9% (21/712) in H. pylori-negative patients. Oesophagitis was less prevalent in H. pylori-positive patients (35.6%; 95% CI 30.1–41.0) than H. pylori-negative patients (46.2%; 42.5–49.9). If patients with peptic ulcer are excluded, oesophagitis was found in 34.8% (95% CI 29.1–40.5) of H. pylori-positive patients, compared to 46.2% (95% CI 42.4–49.9) of H. pylori-negative patients (Figure 5).

Figure 5.

Prevalence (95% CI) of erosive oesophagitis, gastric ulcer and duodenal ulcer by H. pylori status.

Gastric ulcer was more than twice as prevalent in H. pylori-positive ASA/NSAID users than in H. pylori-uninfected users or infected patients who did not use either medication. Likewise, duodenal ulcer was almost twice as prevalent in H. pylori positive users of ASA/NSAIDs than in infected non-users, and almost 10 times more prevalent than in uninfected users (Table 4).

Table 4.  Prevalence and 95% confidence interval of clinically significant findings with respect to H. pylori status and ASA/NSAID use in 1013 patients with available H. pylori status
n (%) [95%CI]H. pylori –ve
non-user
(N = 550)
H. pylori –ve
user
(N = 162)
H. pylori  + ve
non-user
(N = 232)
H. pylori  + ve
user
(N = 69)
Reflux oesophagitis (N = 451)250 (45.5)79 (48.8)85 (36.6)22 (31.9)
[41.3–49.6] [41.1–56.5] [30.4–42.8] [20.9–42.9]
Gastric ulcer (N = 31)7 (1.3)7 (4.3)10 (4.3)7 (10.1)
[0.3–2.2] [1.2–7.5] [1.7–6.9] [3.0–17.3]
Duodenal ulcer (N = 29)7 (1.3)2 (1.2)13 (5.6)7 (10.1)
[0.3–2.2] [-,-] [2.6–8.6] [3.0–17.3]
Gastric erosions (N = 102)51 (9.2)19 (11.7)22 (9.5)10 (14.5)
[6.8–11.7] [6.8–16.7] [5.7–13.3] [6.2–22.8]
Duodenal erosions (N = 54)26 (4.7)10 (6.2)14 (6.0)3 (4.3)
[3.0–6.5] [2.5–9.9] [3.0–9.1] [-,-]

Alarm symptoms

Alarm symptoms were reported in 2.8% (29/1040) of study patients and were more common in patients under 50 years of age (3.5%; 95% CI 2.1–4.9) than in patients 50 years of age and over (1.6%; 95% CI 0.3–2.8); however, there was no correlation between age, the presence of alarm symptoms and the presence of clinically significant findings (data not shown). Dysphagia was the most commonly reported alarm symptom (N = 13) with gastrointestinal bleeding, unexplained weight loss, anaemia and vomiting being reported in seven, six, three and two patients, respectively.

Barrett's oesophagus

In total, 2.4% (25/1040) of patients had Barrett's oesophagus on biopsy of oesophageal mucosa. Of these 25 patients, 16 (64%) had dominant reflux-like symptoms and 17 (68%) had reflux oesophagitis. Seventeen (3.8%) of the 451 patients with reflux oesophagitis had Barrett's oesophagus. Forty percent (10/25) of patients with Barrett's oesophagus had dyspepsia symptoms for less than 5 years, while four (16%) reported symptoms for less than 1 year. In patients with dominant symptoms of heartburn or acid regurgitation, the prevalence of Barrett's oesophagus was 4.1% (16/393), compared to 1.4% (9/647) in the remaining patients.

Malignancies

Two cases of malignancy were found on biopsy. In one case, a biopsy of ‘nodular distal esophagitis’ in a 56-year-old Caucasian female revealed unexpected in situ squamous cell carcinoma. In the other case, gastric biopsies from a 52-year-old Caucasian male with no alarm symptoms and a normal endoscopy revealed H. pylori-negative mucosa associated lymphoid tissue lymphoma (MALToma).

Safety

A total of 16 endoscopy-related adverse events (sore throat, nausea, gagging or breathing difficulty) were reported by 13 patients (1.2%), and all resolved spontaneously. In addition, of 38 serious adverse events in 34 endoscopy patients, none was considered causally related to the study procedures. Two patients discontinued participation in the study, one due to depression, and the other patient died due to a non-gastrointestinal malignancy.

Discussion

The results of this multicentre study indicate that there is a high prevalence of clinically significant upper gastrointestinal findings in patients presenting to a family physician with uninvestigated dyspepsia. The main aim of the study was to document the prevalence of significant endoscopic findings in patients presenting with uninvestigated dyspepsia to their family physician. As patients with previous investigations for dyspepsia and those who recently received treatment for dyspepsia were excluded, we believe that our large sample is representative of such patients in Canada. It is important to note that this study was not intended to describe the prevalence of dyspepsia in the general population. The patients enrolled in the study were representative of the Canadian population with respect to age, gender, smoking and education, based on available demographic data for the year 2001.17 Importantly, the study patients were identified using essentially the same inclusion and exclusion criteria as those used to enrol patients in three complementary, randomised, controlled treatment studies for patients with uninvestigated dyspepsia: the CADET–H. pylori, CADET–HN and CADET–HR studies. Clinically significant endoscopic findings were observed in 58% of patients. Erosive oesophagitis was by far the most common finding (43%), while gastric or duodenal ulcers were found in 5.3%. Malignancies were found in only two patients, and although they had no alarm symptoms, both were over the age of 50 years and would have been investigated endoscopically had the current recommendations for investigation been pursued.16, 18 Thus, the overall prevalence of clinically significant findings, when limited to oesophagitis, peptic ulcer, oesophageal strictures, Barrett's oesophagus and malignancies, was 55.2% (N = 574).

This overall prevalence of endoscopic findings is higher than that reported previously.19–22 It is important to stress that this high prevalence may in part be explained by the fact that use of acid suppression or prokinetic therapy in the preceding 14–30 days was not permitted. Many studies reporting on the prevalence of endoscopic findings do not make it clear that recent use of acid suppression was a reason for exclusion from the study. The prevalence of individual endoscopic diagnoses differed from that summarized in the AGA technical review which reported a peptic ulcer prevalence of 15–25% of patients, reflux oesophagitis in 5–15%, and malignancy in less than 2% of these individuals.16 However, the findings of this review must be interpreted in the context of the study design, the timing of the endoscopy, and the definition of dyspepsia used in the included studies. Many of the referenced studies included in the American Gastroenterological Association (AGA) technical review were carried out in gastroenterology clinics rather than in the primary care setting.

Another explanation for the differences in prevalence may be due to the use of the Canadian Dyspepsia Working Group (CanDys) dyspepsia definition in this study.4 In contrast to the Rome definition of dyspepsia, which specifically excludes heartburn and acid regurgitation from the dyspepsia symptom complex,23 the CanDys definition allows for the inclusion of these symptoms along with epigastric pain and/or discomfort. One of our objectives was to determine whether symptoms were predictive of endoscopic findings. We excluded patients with isolated heartburn and/or regurgitation without epigastric pain. Although the prevalence of oesophagitis was higher (54.7%) in patients with dominant heartburn or regurgitation, oesophagitis was also by far the most common diagnosis in patients reporting dominant ulcer-like or dysmotility-like dyspepsia symptoms (36.5% of such patients). This indicates that upper gastrointestinal symptoms are poor predictors of endoscopic findings. It also supports the notion that the Rome definition of dyspepsia, although perhaps useful for research purposes, does not apply well to patients who present with upper gastrointestinal symptoms in primary care.

Therefore, a management strategy based on specific symptoms or groups of symptoms does not reliably predict a specific underlying pathology, a conclusion also reached by others.15, 16, 19 However, the higher prevalence of oesophagitis, especially in those with dominant reflux-like symptoms, supports the use of acid suppressive therapy as the initial management recommended.4, 24, 25 Given that reflux oesophagitis was the most common finding regardless of the presenting symptoms, empirical initial acid suppressive therapy may be considered for these patients.

We confirmed that clinically significant endoscopic findings were slightly more common in older patients with dyspepsia.16 Also, consistent with previously reported Canadian data, 30% of patients were H. pylori-positive.26 As expected, peptic ulcer disease was more common in patients infected with H. pylori: 27–29 11.3% of H. pylori infected patients had a peptic ulcer compared to 2.9% of H. pylori-negative patients. In our related CADET–H. pylori trial, H. pylori eradication therapy produced symptom relief in significantly more patients than did placebo, even in those with reflux-dominant dyspepsia.9 The findings in this patient population are also consistent with recent data that suggest that reflux oesophagitis is less common in individuals infected with H. pylori.30–32 However, this study was intended to be an observational study, and was not designed to make any inferences about the causality of such relationships.

Consistent with previous studies, the prevalence of gastric ulcer was higher in patients who took ASA and NSAIDS than in patients who did not.29 Recently, there has been considerable debate about a possible interaction between H. pylori infection and ASA/NSAID use.33–37 In this study, both gastric and duodenal ulcers were more common in infected patients who used ASA/NSAIDs than in infected non-users or uninfected users (Table 4).

Barrett's oesophagus was detected in 2.4% of patients in this uninvestigated dyspepsia study; 4.1% of patients with dominant heartburn or acid regurgitation were found to have Barrett's oesophagus. This prevalence is lower than reported in several reviews.38, 39 Most of the reviewed studies were carried out in gastroenterology centres, not in primary care. Neither the presence of reflux oesophagitis, the age of the patient, the prevalence of dominant reflux symptoms nor the duration of dyspepsia symptoms appeared to be associated with the presence of Barrett's oesophagus.

In conclusion, clinically significant endoscopic findings were observed in 58% of patients with uninvestigated dyspepsia. Reflux oesophagitis was by far the most common finding. Most patients presented with a complex of three or more dyspeptic symptoms, and the symptom profile was not predictive of the endoscopic findings. Our study did not specifically address management strategies. However, the high prevalence of oesophagitis, the 30% prevalence of H. pylori infection, and the observation that 62% (34/55) of patients with a peptic ulcer were infected with H. pylori, suggests that most patients presenting with uninvestigated dyspepsia in primary care can be safely managed initially with acid suppressive therapy or treatment of H. pylori, if the patient is infected.

Acknowledgements

We thank Dr Eileen Grace, AstraZeneca Canada, for her input into the development of this study. The CADET–PE Study group of principal investigators are R. Akhras, St-Laurent; J. Akitt, London; D. Armstrong, Hamilton; R. Bacchus, Windsor; A. Barkun, Montreal; P. Bentz, Thunder Bay; K. Berger, Toronto; B. Bernucci, Montreal; H. Bhamjee, London; M. Bradette, Quebec City; R. Brownoff, Edmonton; F. Bursey, St. John's; D. Callaghan, Hamilton; D. Candler, Edmonton; D. Caplan, Toronto; D. Chernoff, Saskatoon; B. Choi-Fung, Toronto; N. Chiba, Guelph; L. Cohen, North York; L. Dionne, Montreal; D. Doell, Westmount; J. Duckett, Montreal; R. Evanson, Montreal; C. Fallone, Montreal; D. Fay, Halifax; S. Finkelstein, Etobicoke; F. Fraser, Stoney Creek; A. Frechette, Quebec City; R. Goldsmith, Guelph; B. Gore, Montreal; M. Gould, Rexdale; W. Haver, Saskatoon; K. Hayes, Halifax; S. Hoddinott, London; J. Janes, Mount Pearl; B. Klar, Woodbridge; E. Krikke, Edmonton; W. Kutcher, Thunder Bay; J-R. Lachance, Montreal; M. Leclair, Anjou; W. Liang, Hamilton; B. Lumb, Hamilton; D. Makinen, Edmonton; J. McDonald, London; E Michalska, Edmonton; J Michiels, Hamilton; D. Morgan, Hamilton; G. Noad, Hamilton; M. Overington, Kingston; W. Paterson, Kingston; P. Piechota, Pointe-Claire; N. Pinsky, Halifax; J. Reddington, Edmonton; A. Rolfe, St. John's; J. Roy, Les Saules; D. Sadowski, Edmonton; L. Schmidt, London; P. Scullion, Guelph; C. Sosnowski, Napanee; K. Stakiw, Saskatoon; A. Teplinsky, Toronto; R. Therrien, Ste-Foy; A. Thomson, Edmonton; B. Trainor, Guelph; R. Tytus, Hamilton; S. Veldhuyzen van Zanten, Halifax; R. West, Kingston; K. Wong, Toronto; R. Woodland, St. John's; L. Worobetz, Saskatoon; and P. Ziter, Windsor.

This study was financially supported by AstraZeneca Canada Inc.

Ancillary