M E Numans1, N J de Wit. 1. Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht, The Netherlands. m.e.neumans@med.uu.nl
Abstract
BACKGROUND: Amongst primary care patients with dyspeptic symptoms, those with reflux-like symptoms or gastro-oesophageal reflux disease are expected to benefit most from empirical proton pump inhibitor therapy. Recognition of this patient group, however, is difficult. The Carlsson-Dent gastro-oesophageal reflux disease questionnaire was developed to justify the selection of primary care patients for empirical proton pump inhibitor treatment. AIM: To evaluate the diagnostic test characteristics of the Carlsson-Dent questionnaire in a primary care population. METHODS: A prospective, open-label, multi-centre diagnostic prevalence study was conducted amongst primary care adults with reflux symptoms. All patients completed the questionnaire and underwent gastroscopy. If no abnormalities were found, the patients were prescribed 2 weeks of treatment with omeprazole (20 mg daily). Receiver operating characteristic curve analysis was used to determine the overall discriminative value of the questionnaire. Diagnostic test characteristics of both the cut-off score (total diagnostic score) in the questionnaire and the physician's provisional classification were measured using oesophagitis and omeprazole treatment success as the reference tests. RESULTS: Of the 536 patients included, 515 underwent endoscopy. No abnormalities were found in 286 (55%). Omeprazole treatment success occurred more frequently in patients with a total diagnostic score of > 7 and oesophagitis II-IV in patients with a total diagnostic score of > 9. The diagnostic test characteristics of the questionnaire were comparable with those of the physician's provisional classification. The area under the receiver operating characteristic curve did not exceed 0.65. Kappa values for observer agreement with the gold standard were far below 0.4. CONCLUSION: The diagnostic performance of this version of the Carlsson-Dent questionnaire was poor. It equalled the physician's clinical judgement and therefore its value for clinical use is limited.
BACKGROUND: Amongst primary care patients with dyspeptic symptoms, those with reflux-like symptoms or gastro-oesophageal reflux disease are expected to benefit most from empirical proton pump inhibitor therapy. Recognition of this patient group, however, is difficult. The Carlsson-Dent gastro-oesophageal reflux disease questionnaire was developed to justify the selection of primary care patients for empirical proton pump inhibitor treatment. AIM: To evaluate the diagnostic test characteristics of the Carlsson-Dent questionnaire in a primary care population. METHODS: A prospective, open-label, multi-centre diagnostic prevalence study was conducted amongst primary care adults with reflux symptoms. All patients completed the questionnaire and underwent gastroscopy. If no abnormalities were found, the patients were prescribed 2 weeks of treatment with omeprazole (20 mg daily). Receiver operating characteristic curve analysis was used to determine the overall discriminative value of the questionnaire. Diagnostic test characteristics of both the cut-off score (total diagnostic score) in the questionnaire and the physician's provisional classification were measured using oesophagitis and omeprazole treatment success as the reference tests. RESULTS: Of the 536 patients included, 515 underwent endoscopy. No abnormalities were found in 286 (55%). Omeprazole treatment success occurred more frequently in patients with a total diagnostic score of > 7 and oesophagitis II-IV in patients with a total diagnostic score of > 9. The diagnostic test characteristics of the questionnaire were comparable with those of the physician's provisional classification. The area under the receiver operating characteristic curve did not exceed 0.65. Kappa values for observer agreement with the gold standard were far below 0.4. CONCLUSION: The diagnostic performance of this version of the Carlsson-Dent questionnaire was poor. It equalled the physician's clinical judgement and therefore its value for clinical use is limited.