Can clinical features stratify use of endoscopy for dyspeptic patients with high background prevalence of upper gastrointestinal cancer?

Background: Whether clinical features can stratify priority of endoscopy remains controversial for dyspeptic patients with high background prevalence of upper gastrointestinal cancer. Aim: To examine the predictive performance of clinical features for cancerous lesions in dyspeptic patients in Taiwan. Methods: Between April 2008 and July 2009, 2530 consecutive dyspeptic outpatients underwent prospective evaluation with standardized questionnaire and then upper gastrointestinal endoscopy. Performance of using age thresholds and alarm symptoms to predict malignancy was determined. Independent risk factors associated with malignancy and those with negative endoscopic findings were identified. Results: Malignant lesions were found in 31 patients (1.2%) and were independently associated with age, male gender, gastrointestinal bleeding, weight loss, and alcohol consumption. Any symptom of weight loss, bleeding and dysphagia, or simply age >45 years predicted 97% of cancer cases, with the sensitivity, specificity, positive and negative predictive values being 96.8%, 29.3%, 1.7%, and 99.9%, respectively. This strategy achieved a low negative likelihood ratio (0.11) and a high diagnostic odds ratio (12.45). Negative endoscopic finding (n=1377, 54.4%) was independently associated with younger age, female gender, no use of non-steroidal anti-inflammatory drug, and no tobacco or alcohol consumption. Conclusions: Absence of weight loss, dysphagia, and gastrointestinal bleeding predicts low likelihood of malignancy in dyspeptic Taiwanese patients aged