Clinical Question: Which patients with dyspepsia should undergo urgent endoscopic evaluation?

Setting: Outpatient (specialty)

Study Design: Decision rule (validation)

Synopsis: Physicians in this study, which was conducted in England, had the option of referring patients for an urgent endoscopy if they felt it was indicated. Patients eligible for urgent endoscopy were those at least 55 years of age, those with alarm symptoms (e.g., dysphagia, anorexia, vomiting, weight loss, anemia), and those in whom one or more high-risk features were present (e.g., family history, Barrett’s esophagus, pernicious anemia, peptic ulcer surgery, known dysplasia). A sample of 1,852 patients was studied to identify the accuracy of predictors and to develop a new set of criteria.

The alarm features most strongly associated with cancer were dysphagia, weight loss, and age of at least 55 years. For the prediction of any significant pathology (i.e., cancer, peptic ulcer, stricture, esophagitis), age, dysphagia, and high-risk features were the strongest predictors. These results were used to develop a new set of criteria for urgent endoscopy: dysphagia or weight loss at any age or onset of dyspepsia after 55 years of age associated with any alarm feature (e.g., anemia, anorexia, vomiting, dysphagia, weight loss).

The new decision rule was validated in a subsequent sample of 1,785 patients during the next 12 months. Of this group, 570 patients would not have required urgent endoscopy using the new criteria; 0.7 percent in this group had cancer compared with 4.0 percent of the 1,215 patients identified by the old rule as being at high risk.

Bottom Line: Patients with dysphagia or weight loss at any age, as well as patients older than 55 years with any alarm symptoms or high-risk factors, should have urgent upper endoscopy to identify those who have cancer. However, even in this large study, which detected 70 upper gastrointestinal cancers, 60 percent of the findings were tumor, lymph node, or metastasis stage T3 or above, and only 16 percent of patients were candidates for surgical resection. (Level of Evidence: 1b)