The most common symptoms leading to referral for upper GI endoscopy were epigastric pain, dyspepsia, upper gastrointestinal bleeding (haematemesis and melaena) and chest pain respectively which is the same as that reported in earlier publications . The current study found the incidence of H pylori infection in Korle Bu Teaching Hospital to be 51.3 % based on the HUT. The prevalence is much lower than that quoted in other studies from West Africa [3, 4, 9] which have shown a high prevalence of H. pylori. Previous studies in Korle Bu Teaching Hospital reported an incidence of 75 % [3, 4]. However, a recent publication also reports an H. pylori prevalence of 69.7 % (1999) and 45.2 % (2012) in Accra . This suggests a change in H. pylori prevalence with time.
Gastritis (59.2 %) was the most common identifiable lesion at endoscopy in this study. One study done in Tanzania with H. pylori prevalence of 65 % also reported gastritis as the most common finding and gastritis and duodenal ulcer were statistically associated with H. pylori. 72 % of gastritis and 89.5 % of duodenal ulcers were H. pylori positive . Based on previous prevalence studies, it has been established that the finding of a non-bleeding duodenal ulcer has a positive predictive value of over 90 % for H. pylori diagnosis, making confirmatory tests not mandatory. However, the positive predictive values for gastric ulcers, bleeding duodenal ulcers and perforated duodenal ulcers are lower and confirmatory tests would be necessary .
Nearly 58 % of gastritis and 60 % of duodenal ulcers tested positive for H. pylori, while 33 % of gastric ulcers were also positive. Interestingly our study did not find any positive association between duodenal ulcers, gastric ulcers or gastritis and a positive HUT. This may be explained by the fact that, among the 10 patients with duodenal ulcer, 3 of the 4 patients who were HUT negative admitted to using PPI’s. While of the 6 patients diagnosed with gastric ulcer, 3 of the 4 who had negative HUT admitted to using PPI’s. Cohen et al.  have reported that the use of PPI’s may lead to unreliable H pylori diagnosis.
Moderate to severe symptoms were associated with HUT results such that the odds of a patient reporting with moderate to severe symptoms of having a positive HUT test result was 12.06 times (OR 12.06 [95 % CI 1.38–105.59], p = 0.024) compared with patients who reported with mild symptoms. This means endoscopy with H. pylori diagnosis and treatment is mandatory in these patients whereas those with mild symptoms may be managed without H. pylori testing.
The gold standard recommended by the World Gastroenterology Organisation Global Guidelines, 2010 is endoscopy and rapid urease test which may not be readily available or cost effective in developing countries. They recommend then that in resource limited settings with a high prevalence, the decision to treat may be based on the assumption that H. pylori is present. This has been the practice over the years, however, recent development has seen the introduction of several endoscopy set ups in both public and private facilities. This has brought with it increased opportunities for the accurate diagnosis and treatment of H. pylori. Accurate diagnosis and treatment is necessary if the prevalence is not as high as previously reported. The empirical treatment with antibiotics based on endoscopic diagnosis and assumption of high prevalence should be discouraged. Efforts should be made to make an accurate diagnosis of H. pylori before treatment. Where the HUT is not available other tests should be done such as histological diagnosis or stool for H. pylori antigen. The lower prevalence we report may be due to improving standards of living and sanitation in Ghana as compared to previous studies published several years ago also using HUT [3, 4]. In addition the vast majority of our study subjects were from urban areas where the standard of living is higher. However, the frequent use of antibiotics may also contribute to the relatively low prevalence being reported recently . Also the recent use of PPI’s may contribute to the low prevalence. Of all the patients who admitted to taking PPI’s only 18 % had positive HUT as compared to 51 % who tested negative. While for those who denied PPI use, 82 % had positive HUT and 49 % were negative. This supports the fact that recent use of PPI’s may result in falsely negative HUT results .
With regards to the endoscopic diagnosis this study reports a diagnostic yield of 55.3 % with a sensitivity and specificity of 71.8 % [95 % CI 55.1–85.0] and 37.8 % [95 % CI 22.5–55.2] respectively, indicating that the endoscopic diagnosis does not predict H. pylori status (as determined by HUT) better than chance. After excluding those on PPI, analysis resulted in a diagnostic yield of 58 %, sensitivity of 75.03 [95 % CI 56.6–88.5] and specificity of 38.8 % [95 % CI 13.9–68.4]. The value of endoscopic diagnosis of H. pylori is unclear. The value of the HUT in the patients who have recently been on PPIs and antibiotics maybe limited. In this case treatment based on the endoscopic diagnosis should be considered as recommended by WHO. On the other hand in treatment naive patients, the HUT maybe more reliable and treatment can be prescribed when indicated.
It is known that certain endoscopic mucosal features indicate H. pylori infection such as atrophic changes, rugal hyperplasia, oedema, spotty erythema, linear erythema and haemorrhage, amongst others. However, recognition and diagnosis depends on the experience of the endoscopist. Several studies including that by Khaloo et al. (41.8 %)  and Redeen et al. (43–53 %)  have reported a low diagnostic yield. Khazuhiro et al. however, published a relatively higher diagnostic yield in the H. pylori-uninfected (88.9 %) but lower in H. pylori-infected (62.1 %) and in the H. pylori-eradicated (55.8 %) patients . This current study, like others, does not support endoscopic diagnosis of H. pylori as a standard of care.
Magnifying endoscopy affords the opportunity to make an accurate endoscopic diagnosis of H. pylori based on the surface structure of the gastric mucosa and has sensitivity and specificity of 100 and 92.7 % respectively for H. pylori infected mucosa . This facility is not available at many endoscopy centers worldwide and not in Ghana and other developing countries. It may now be cost effective to make an accurate diagnosis before treatment so as to cut down cost on antibiotics, its side effects and reduce antibiotic resistance.
The obvious advantage of the rapid urease test in our environment is that it is inexpensive (less than $10/test), does not require any technical expertise and gives rapid results within an hour. It is also reliable with the sensitivity of Campylobacter-like organism (CLO) test 75–98 % and specificity 95–100 % . This will improve the accurate diagnosis especially where endoscopic expertise is not certain.
With histologic diagnosis the disadvantages are sampling errors, observer variations, high cost and long duration of processing. There is also an additional delay in our facility as the patient is responsible for getting the specimen to the pathologist and getting the report back to the physician.
One of the limitations of our study is taking antral biopsy samples. We also did not have the sufficient resources to recruit more patients, which may have improved the statistical power of our studies. This study was also undertaken in a single medical center, Korle Bu Teaching Hospital, which may not be a general representation of Ghana although the patients came from variety of ethnic groups and socioeconomic backgrounds.