Lack of correlation between a self-administered subjective GERD questionnaire and pathologic GERD diagnosed by 24-h esophageal pH monitoring.

INTRODUCTION: Self-reported reflux symptoms do not always correspond to pathologic gastroesophageal reflux disease (GERD). We evaluated whether GERD-related symptoms in the self-reported Mayo-GERD questionnaire (GERDQ) were correlated with current gold standard definitions of pathologic GERD.
METHODS: Three hundred thirty-six consecutive consenting individuals with GERD symptoms referred for 24-h esophageal pH monitoring completed a baseline GERDQ. Univariate and multivariate analyses identified questions that were most associated with percent total time pH<4 at distal probe (DT) >4% or DeMeester score (DS) >or=14.7, two accepted definitions of pathologic GERD. A risk score was created from these analyses, followed by generation of receiver operating characteristic curves and determination of C-statistics, sensitivity, and specificities at various cut points, with prespecified minimal values of each that would be required to meet the definition of "potential clinical utility."
RESULTS: Forty-nine percent of patients were found to have pathologic GERD; half the patients (not necessarily those with pathologic GERD) described suffering from severe or very severe heartburn or acid regurgitation in the past year. Univariate logistic regression analysis identified six of 22 key GERD questions that were significantly related to DT or DS, in addition to age and gender. Three questions (duration of symptoms, nocturnal heartburn, hiatal hernia) along with age and gender remained significant in multivariate analyses. A risk score (RS) was created from these five questions separately for DT and DS. For DT, the C-statistic for RS was 0.75, and at the optimal cut point of >or=6 that maximizes sensitivity (SS) and specificity (SP), SS was 68% and SP was 72%. For DS, the C-statistic was 0.73, and at the optimal cut point, SS was 82%and SP 60%. When considering other cut points, the rare extreme case of very low RS (<or=2) was strongly predictive of lack of pathologic GERD: for DT, SS 100%/SP 18%, negative predictive value (NPV) 100%; and for DS, SS 97%, SP 25%, NPV 88%. However, only 10-15% of patients referred for pH testing had RS scores of <or=2.
CONCLUSION: Self-reported prolonged history of GERD-like symptoms, nocturnal heartburn, history of a hiatus hernia, and male gender were associated with abnormal 24-h esophageal pH monitoring. However, these factors lack clinical utility to predict pathologic GERD in patients referred for pH testing. We found that 51% of patients with severe GERD symptoms do not have true pathological GERD on objective testing. The clinical implications of this study are significant in that treatment with acid-suppressing medication in such patients would be inappropriate.