Identifying patients with gastroesophageal reflux disease: validation of a practical screening tool.

A reliable, accurate noninvasive method for identifying patients with gastroesophageal reflux disease (GERD) in the primary care setting is needed. A population-based case-finding instrument may assist managed care organizations in identifying candidates for disease management or quality improvement programs. Our aim was to develop and validate a GERD case-finding instrument. A 12-item "GERD Screener" was developed based on literature review and expert opinion with questions about heartburn and regurgitation frequency and severity and medication use, as well demographics and exclusion criteria. Categorical and continuous scoring methods with and without medication use were developed a priori. Using a telephone interview in a medical group, we identified and enrolled 100 subjects with a history of GERD-like symptoms and 103 controls. Each subject completed the GERD Screener, a validated gastrointestinal symptom questionnaire, the Digestive Health Symptom Index (DHSI), and was evaluated independently by two gastroenterologists using a structured format. Agreement by the two physicians that symptoms were consistent with GERD and required an intervention was considered the gold standard. In all, 70 subjects were classified as having GERD and 106 as controls using the gold standard (K = 0.73; 95% CI, 0.63-0.82). The scoring method for the screening instrument using a continuous measure of GERD symptoms (frequency and severity) and weighted medication use resulted in an area under ROC of 0.89 (95% CI, 0.84-0.94). Using a cutoff of >9 points, this measure was 83% sensitive and 83% specific. Compared to the gold standard, the DHSI GERD subscale has an area under ROC of 0.89 (95% CI, 0.84-0.94). The GERD Screener was highly correlated with the DHSI GERD subscale, r = 0.78 (95% CI 0.72-0.83; P < 0.0001). In conclusion, the GERD Screener has demonstrated construct, convergent, and predictive validity. It is shorter than existing validated instruments, practical, and easily administered, which may reduce the response and administrative burden. This may serve as a valuable case-finding instrument in primary-care and managed-care organizations wishing to implement programs to improve the quality and efficiency of care.