Intended for healthcare professionals

General Practice

Challenges in managing dyspepsia in general practice

BMJ 1997; 315 doi: (Published 15 November 1997) Cite this as: BMJ 1997;315:1284
  1. Lars Agréus (lars.agreus{at}, research associatea,
  2. Nicholas Talley, professor of medicineb
  1. a Department of Family Medicine, Uppsala University, Uppsala, Sweden
  2. b Department of Medicine, University of Sydney, Nepean Hospital, Penrith, NSW 2751, Australia
  1. Correspondence to: Dr L Agréus Vårdcentralen, Råghusgatan 6, 740 71 Öregrund, Sweden
  • Accepted 11 August 1997


Although dyspepsia has many possible causes, the primary causes are peptic ulcer disease, gastro-oesophageal reflux disease, and functional (non-ulcer dyspepsia). General practitioners managing dyspepsia face several dilemmas.

  • Is a diagnosis of organic dyspepsia reliable without an endoscopy?

  • Do patients with dyspepsia need to be referred for endoscopy when nothing of pathological importance will be found in most cases?

  • Are the new rapid blood tests for Helicobacter pylori helpful?

Summary points

Subgrouping of symptoms should not guide empirical treatment of dyspepsia

Symptoms of underlying serious disease and use of non-steroidal anti-inflammatory drugs are indications for prompt endoscopy

A breath test is superior to a blood test in documenting H pylori infection and guiding the management of dyspepsia in general practice

Empirical treatment with acid reducing or prokinetic drugs for all patients with dyspepsia is inappropriate as peptic ulcer disease will be inadequately treated

Empirical eradication treatment of H pylori in patients with dyspepsia who are positive for the organism is preferable to other empirical treatment if prompt endoscopy is not an option

Dyspepsia subgroups: an irrelevant concept

Broad (and vague) definitions of dyspepsia have been proposed,1 but a current internationally accepted definition of dyspepsia is chronic or recurrent pain or discomfort centred on the upper abdomen.2 People with heartburn or acid regurgitation alone and those in whom reflux symptoms are dominant and epigastric pain or discomfort is minor should probably not be classed as having dyspepsia but as having symptomatic gastro-oesophageal reflux disease. The same applies to patients who have a disturbed bowel habit and dyspeptic symptoms, which should be classed as the irritable bowel syndrome.2

Dyspepsia subgroups

During the past decade dyspepsia has been divided into clinical subgroups based on medical history: ulcer-like (typical ulcer symptoms), dysmotility-like (symptoms suggestive of gastric stasis), reflux-like (retrosternal and concomitant upper abdominal symptoms), and unspecified (symptoms cannot be classified) (box). Findings in the history or examination that are suggestive of serious underlying disease (alarming features) should result in prompt referral for an upper endoscopy or other investigations (box). Various management plans for the rational care of patients who have new dyspepsia but without alarming features have been suggested by clinicians and pharmaceutical companies.1 2 3 4 These plans have been based on the dyspepsia subgroups or on individual (predominant) dyspeptic symptoms.

Box 1: Dyspepsia subgroups12

  • Reflux-like dyspepsia

Heartburn plus dyspepsia

Acid regurgitation plus dyspepsia

  • Ulcer-like dyspepsia

Localised epigastric pain

Pain when hungry

Pain relieved by food

Pain relieved by antacids or acid reducing drugs

Pain that wakens the patient from sleep

Pain with remission and relapses

  • Dysmotility-like dyspepsia

Upper abdominal discomfort (pain is not a dominant symptom)

Early satiety

Postprandial fullness


Retching or vomiting

Bloating in the upper abdomen (without visible distension)

Upper abdominal discomfort often aggravated by food

  • Unspecified dyspepsia

Box 2: Alarming features in dyspepsia

Onset at age >45 years

Weight loss


Gastrointestinal bleeding



Persistent vomiting

Epigastric mass


Previous peptic ulcer

Use of non-steroidal anti-inflammatory drug

Management controversies

Dyspepsia is treated empirically or by early referral for upper endoscopy, with or without a non-invasive test for H pylori infection to guide the management decision. General practitioners are the main target for current recommendations because most patients with dyspepsia are managed in primary care, where the need for a simple, cost effective approach is obvious.

The aim of grouping the symptoms of dyspepsia is to try to match symptoms to pathophysiological disturbances to enable rational prescribing. This has led to recommendations such as acid reduction in reflux-like and ulcer-like dyspepsia and the use of motility stimulants (prokinetic drugs) in dysmotility-like dyspepsia. Such a solution if appropriate would meet everybody's needs: general practitioners could provide prompt treatment, there would be fewer unnecessary endoscopies in the healthcare system, and the pharmaceutical industry could promote the right drug for the right patient.

However, a symptom based strategy may be applied only if most patients with peptic ulcer present with ulcer-like dyspepsia. This has became even more important now that ulcer disease can be cured by eradication of H pylori infection, which is present in almost all patients with a duodenal ulcer and in three quarters of those with a gastric ulcer. Antisecretory treatment alone for these patients is no longer routinely justified. Moreover, acid reducing treatment must be of value in those who have ulcer-like dyspepsia but do not have a peptic ulcer, which is the majority.5 Indeed, the risk of having a peptic ulcer must be negligible in the other dyspepsia subgroups if a recommendation to tailor empirical drug treatment to specific dyspepsia subgroups is to be ethically justified.

Ulcer-like and dysmotility-like dyspepsia

Unfortunately for all parties, management strategies based on the dyspepsia subgroups as listed in box 1 fails to stand up to scientific examination. In endoscopy based studies about the same number of peptic ulcers are found regardless of whether the patients present with ulcer-like or dysmotility-like dyspepsia.5 6 7 Moreover, only four out of 165 dyspeptic patients who did not have a peptic ulcer at initial endoscopy developed an ulcer over 10 years of follow up, although 62 (58%) of the 106 who still had dyspepsia 10 years later had ulcer-like symptoms.8

Population based epidemiological data also fail to support the concept. Most patients with dyspepsia report symptoms that fit more than one of the proposed subgroups, and the change in symptoms from one year to the next is considerable.9 Prokinetic drugs are also at least as efficacious as H2 receptor blockers in patients with ulcer-like dyspepsia,10 especially in those in whom endoscopy confirmed the absence of peptic ulcer disease.11

Division of dyspepsia into subgroups on the basis of the predominant complaint has also been proposed as a predictor of the pathophysiology.4 This is also not clinically useful: the symptoms identified as typical ulcer-like and typical dysmotility-like in box are common regardless of whether the endoscopic diagnosis is peptic ulcer disease or functional dyspepsia.12 13 14 For example, Johannessen et al compared patients with peptic ulcer disease and functional dyspepsia: early satiety was reported by 33% and 29%, nausea by 35% and 46%, a distinct location of pain by 29% and 22%, respectively, and upper abdominal discomfort often aggravated by food by 36% in each group.13 Kang et al reported that the sensitivity and positive predictive value of localised abdominal pain for duodenal ulcer were only 13% and 14%, respectively, in outpatients.15 Stanghellini et al found that postprandial fullness was a clinically significant predictor of delayed gastric emptying (a purported mechanism for the induction of dysmotility-like dyspepsia) among younger women, but this has yet to be confirmed in general practice.16 Bytzer et al have expressed the dilemma about empirical treatment for dyspepsia in another way: at least 50% of patients with a peptic ulcer will be given the wrong empirical treatment because their doctor is not aware of their ulcer when prescribing treatment.17

Gastro-oesophageal reflux symptoms

Reflux-like symptoms (box), whether considered as a part of the dyspepsia concept or not, are presumed to have a higher sensitivity for reflux oesophagitis,2 but the symptoms are probably not sensitive enough now that ulcer treatment (brief treatment aimed at cure of H pylori infection) has became so different from the treatment of oesophagitis (acid suppression rather than curative treatment).5 14 In one study Talley et al reported that the prevalence of peptic ulcer was similar among patients with reflux-like (11%), ulcer-like (9%), and dysmotility-like dyspepsia (7%).5 If endoscopic signs of chronic peptic ulcer disease among those with dysmotility-like dyspepsia were included every 10th patient would have been treated inadequately if endoscopy had been omitted. Hungin similarly reported that 7% of patients presenting with reflux symptoms also had a peptic ulcer,18 while Carlsson et al found that 20% of patients with a gastric ulcer presented with predominant reflux symptoms.19 Although proton pump inhibitors may be used as a diagnostic tool to identify and treat gastro-oesophageal reflux disease,20 such an empirical strategy will fail to treat concomitant peptic ulcer disease in a substantial proportion of patients. There is also a risk of prescribing unneeded and expensive acid reducing drugs long term. The placebo response with acid reducing drugs may be as high as 60%.21 What many patients in general practice want rather than drug treatment is an explanation for their symptoms.22 Reassurance and dispelling concerns about serious disease are the most important considerations when planning management.2 23

Endoscopy versus empiric treatment

Two well designed studies have shown that prompt endoscopy with treatment based on the results was preferable to blind empirical treatment, although it was initially more expensive.24 25 They also found a noticeable decrease in drug consumption, visits to doctors, and sick leave in the year after the endoscopy compared with the year before. Patients were more satisfied with their management, although the symptomatic outcome was similar with the two strategies.24 Concerns about the inadequate correlation between dyspeptic symptoms and pathophysiology is shared by the British Society of Gastroenterology, which has avoided identifying dyspepsia subgroups in its new guidelines on management of dyspepsia.26

Non-invasive tests for H pylori: problems and opportunities

Accuracy of blood tests

The British Society of Gastroenterology's guidelines recommend that blood tests for H pylori should be used to help decide when to perform an endoscopy in younger patients who do not have sinister symptoms and are not taking non-steroidal anti-inflammatory drugs (box).26 Such an approach has also been proposed by others.27 Blood tests reduce the need for endoscopy by about a quarter because almost all peptic ulcer disease in patients who are not taking non-steroidal anti-inflammatory drugs is due to the infection. Stomach cancer in younger patients is rare but is often associated with the infection.28 An important question is whether patients will be as reassured by normal results in these indirect peptic ulcer and cancer tests as they are by normal results on endoscopy.24 A study by Patel et al suggests that patients are reassured by blood rest results,27 while preliminary results from Lassen et al contradict this conclusion.29

The strategy of testing for H pylori infection is also dependent on the diagnostic accuracy of the serology kit used. Both rapid whole blood tests that can be done on the spot and enzyme linked immunosorbent assay (ELISA) kits that need laboratory facilities for the analysis are available. The accuracy of the commercial ELISA kits has recently been systematically reviewed.30 The test cut off points gave an average sensitivity and specificity of 85% and 79%, respectively, with the low and high extremes being 49% and 85% and 99% and 63%, respectively (1). The rapid whole blood tests in preliminary reports seem to have similar sensitivity and specificity. For example, Enroth et al report some of the best results to date, with a sensitivity of 96% and a specificity of 85% (1).31

Diagnostic value of blood and breath tests forHelicobacter pyloriinfection in population with prevalence of infection of 25

View this table:

The clinical picture is, however, complex when blood test results are relied on to determine the next step in management. Some authors have recommended empirical eradication of H pylori in all cases with positive results on serology,32 but most recommend that further management should be based on the results of upper endoscopy in those with a positive blood test result.3 26 Empirical treatment of H pylori infection is probably preferable when access to prompt upper endoscopy is limited or when endoscopy is costly. If studies confirm that substantial relief of symptoms occurs after eradication of H pylori infection in infected patients with functional dyspepsia,33 then an empirical eradication strategy will be of even more interest. In addition, the demands on the non-invasive screening test will be greater.

In general practice, patients with a negative result on testing for H pylori have to be reassured as they will usually not be retested. The negative test result must consequently be as certain as possible—that is, the ability of the test to identify those not infected (the negative predictive value) is most important. Unlike the sensitivity and specificity, the predictive value is dependent on the prevalence. In the relevant age group (<45 years of age) the prevalence of H pylori is estimated to be about 25% in the Western world.34 A high, almost maximal, negative predictive value is possible with this prevalence and a serology kit that has a sensitivity and specificity of 99% and 63%, respectively (A in 1, a negative predictive value of 99%). However, over half (53%) of the patients with a positive test result will be falsely positive and consequently will undergo endoscopy or be treated unnecessarily. Use of a kit with an average or lower sensitivity (B and C in 1) does not help: four out of 10 patients with a positive result will still be given a wrong diagnosis, and almost two out of 10 (17%) will have false negative results. Even a better ratio of sensitivity to specificity, 96%:85% respectively (D in 1), will result in three patients with false positive results out of 10. A recent study confirms the high rate of false positive results when using blood tests.35 Moreover, good blood tests may perform poorly in different places in the same country because of variation in strains. Thus, application of a locally validated test is vitally important.36

Urea breath test

An alternative non-invasive test is the urea breath test. This test is easy to perform in primary health care. One disadvantage is that the answer is delayed because the breath sample has to be sent away for analysis. A breath test is more expensive (about £30 v £15 for a serological test) and to reduce false negative results should not be performed while a patient is taking potent acid reducing drugs.37 However, the average sensitivity and specificity is 95%,38 and even higher values have been reported.39 In patients under 45 this means an almost negligible risk of having infection if the breath test gives negative results (a negative predictive value of 99% in 1), giving one out of 10 unnecessary referrals for endoscopy because of a false positive test result (a positive predictive value of 89%). Consequently, because of the higher costs associated with an increased number of endoscopies with blood testing, breath testing may be more cost effective despite its higher initial costs.

Because of the clinical and economic consequences of applying suboptimal blood tests for H pylori we question the rationale of using them in general practice, as recommended by the British Society of Gastroenterology.26 The prompt use of the breath test seems a better option. Indeed, McColl et al found that half of those with dyspepsia and a positive breath test had a peptic ulcer on endoscopy.40

An alternative, but to our knowledge not studied, strategy would be to start with a blood test that has a high negative predictive value. A negative result would then be considered reliable but a positive result would be complemented by a breath test to confirm the presence of infection.

Management strategy for patients with new dyspepsia

We recommend that the management strategy for patients under 45 years of age with dyspepsia of new onset should comprise prompt endoscopy for patients with alarming features, for worried patients or patients demanding long term drug treatment, and for patients taking non-steroidal anti-inflammatory drugs or with a positive result on breath testing. This recommendation will be practical only if the patient is reassured by a negative breath test result and if the waiting lists at endoscopy units are short. Otherwise, a doctor will be obliged to prescribe empirically either acid reducing drugs or eradication treatment for H pylori, with the risk of overprescribing or providing inappropriate treatment.

The choice between empirical treatment with acid reducing drugs alone or prokinetic drugs in all patients with undiagnosed dyspepsia remains arbitrary, and all peptic ulcers will be inappropriately managed with such a strategy. Therefore this approach should be abandoned. If it is not applicable to arrange a prompt endoscopy for a patient we recommend empirical eradication treatment of H pylori instead, provided that the screening test for H pylori is reliable. Most patients with peptic ulcer disease will then be cured, and the total endoscopic load should be reduced since only patients in whom treatment fails or those with a symptom relapse will need to be investigated.

In patients who are negative for H pylori and are not satisfied with reassurance alone a trial of an acid reducing or prokinetic drug is reasonable. Our recommendations are summarised in the figure. The algorithm needs to be formally tested in well designed community trials that compare different management strategies.


Algorithm for management of patients with new dyspepsia in general practice


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
View Abstract