Identifying patients with gastroesophageal reflux disease in a managed care organization

The ability of various strategies to identify patients with gastroesophageal reflux disease (GERD) and the relative economic impact on disease management programs for GERD were studied. A telephone interview was conducted of a random sample of patients enrolled in any of three health plans in a 100,000-member managed care organization who had either a pharmacy claim or an encounter claim during 1997. The telephone interview identified patients with GERD and served as the standard by which the sensitivity, specificity, and predictive values of the following patient-identification strategies were compared: (1) telephone interview, (2) chart review, (3) use of encounter claims, (4) use of pharmacy claims, (5) use of both encounter claims, and pharmacy claims, and (6) use of encounter claims or pharmacy claims. Conservative estimates of costs and projected savings were then used to model the potential return on investment of the strategies. A total of 1186 patients completed the telephone interview, of whom 390 (33%) met the case definition of GERD. The most sensitive method for identifying patients with GERD was using either pharmacy or encounter claims (26%). The most specific strategy with the highest positive predictive value (PPV) (87%) was using both pharmacy and encounter claims, but this approach had a case-detection rate of only 3%. Encounter claims were significantly more sensitive than pharmacy claims and yielded a higher estimate of prevalence. The telephone interview identified the most subjects who could have benefited from a disease management program and cost 84% less than chart review. While use of administrative data (pharmacy and encounter claims) was the least costly strategy, it identified 74% fewer patients expected to benefit from disease management. The efficiency of disease management programs for GERD may depend on the method of patient identification, which in turn may depend on whether PPV or negative predictive value (NPV) should be maximized. If there is a need to identify all cases (i.e., sensitivity and NPV are most important), then telephone interview may provide the greatest opportunity for disease management with the greatest return on investment, but at the expense of enrolling many patients who may not benefit.