“…The test-and-treat strategy has been compared with prompt endoscopy in eight RCTs (Table 1), 17, 18, 19, 20, 21, 22, 23, 24 which differ in important ways: in three trials 17, 19, 20 patients were recruited and randomized at the endoscopy unit after their general practitioner had referred them for investigation; in five trials 18, 21, 22, 23, 24 patients were randomized in primary care. It is noteworthy that the studies by Jones et al 18 and Duggan et al 24 used near-patient serology, which has very poor accuracy for the diagnosis of H. pylori infection; this important drawback markedly reduces the reliability of results.…”