Objective: To explore the optimal cut-off point of symptom association probability (SAP) in the diagnosis of gastroesophageal reflux-induced chronic cough (GERC) and therefore to improve the diagnostic accuracy.
Methods: Patients with suspected GERC consecutively referred to our respiratory clinic were enrolled into this prospective study between July 2011 and February 2013. After multi-channel intraluminal esophageal impedance and pH monitoring, SAP was calculated by associating the cough recordings on the patients' diary with the detected reflux. GERC was confirmed when there was a favorable response to the following anti-reflux therapy despite the laboratory findings. The optimal cutoff point of SAP was defined according to the highest Youden index. Then, the sensitivity, specificity, positive and negative predictive values, the area under the curve of ROC, and the Kappa value for the optimal cut-off point of SAP was calculated and compared to those of SAP standards currently used in China or generally accepted in the diagnosis of GERC.
Results: During the study period, 103 patients with suspected GERC were recruited. Among them, GERC was confirmed in 87 patients (84.5%), including 54 patients (62.1%) due to acid reflux and 33 patients (37.9%) due to non-acid reflux. The optimal cut-off point of SAP was defined at ≥ 80% based on the highest Youden index of 0.372. For the diagnosis of GERC, SAP ≥ 80% had the area under the curve of ROC of 0.686, the Kappa value of 0.264, the sensitivity of 74.7%, the specificity of 62.5%, positive predictive value of 91.5% and negative predictive value of 31.3% respectively, which were superior to those of SAP ≥ 75% currently used in China, and to those of SAP ≥ 95% ( the generally accepted cut-off) in that the balance between higher sensitivity and higher specificity was maintained. When combined with DeMeester score ≥ 12.7, the diagnostic accuracy of SAP ≥ 80% was further improved, with the area under the curve of ROC of 0.820, the Kappa value of 0.689, the sensitivity of 87.0%, the specificity of 76.0%, positive predictive value of 94.1% and negative predictive value of 80.0%.
Conclusion: SAP ≥ 80% may be a more suitable standard for the diagnosis of GERC.