Diagnosis of gastroesophageal reflux disease (GERD) based on subjective evaluation of symptoms alone is challenging.
To test the validity and reliability of a graded response questionnaire to identify patients with GERD.
Patients presenting with upper abdominal symptoms for more than a month, and without alarm features were consecutively recruited. They completed a 12-item questionnaire encompassing 6 symptoms presented either in Chinese or English language. Scores were analyzed for validity in diagnosis of GERD, using gastroenterologist's diagnosis as reference standard. Receiver operating characteristic (ROC) analysis, predictive values, likelihood ratios, and diagnostic odds ratio were used to evaluate diagnostic accuracy.
Of 209 patients studied, 163 (78%) answered the English whereas 45 (21.5%) answered the Chinese version. At a cut-off of > or =28.5 in total symptom scores, the Chinese version had a sensitivity of 87.5% and specificity of 75.7% with an area under the ROC curve (AUCROC) of 0.824 for diagnosis of GERD. Corresponding sensitivity and specificity for the English version were 76.9% and 50.8%, respectively, with an AUCROC of 0.760. Positive predictive value, negative predictive value, positive likelihood ratio (+LR), negative likelihood ratio (-LR), and diagnostic odds ratio were 0.438, 0.966, 3.597, 0.165, and 21.778 and 0.330, 0.875, 1.564, 0.454, and 3.443 for the Chinese and English versions, respectively.
The questionnaires demonstrated good construct reliability, which were highly sensitive and reasonably specific for the detection of GERD. They are useful tools to help primary care physicians in identifying GERD in patients presenting with uninvestigated upper abdominal symptoms.