Validation and proposal of case definitions for identifying patients with myasthenia gravis

Abstract

Background and purpose Public medical insurance data are increasingly used to study myasthenia gravis (MG); however, a validated case definition is lacking. We assessed the clinical characteristics of patients identified according to previously used case definitions. Methods Patients with diagnosis code G70.0 (myasthenia gravis) who visited Severance Hospital between January 2019 and December 2020 were retrospectively analyzed. Three case definitions used in previous epidemiologic studies were applied to the included patients, and their performance was evaluated based on sensitivity, specificity, and positive predictive value (PPV). Results A total of 843 patients with diagnosis code G70.0 were identified, and 796 (96.7%) met the clinical diagnostic criteria for MG. All three previously-used case definitions showed a PPV of >98% in identifying clinically diagnosed MG, with sensitivities ranging from 68.7 to 71.1%. Case definition #1 was unable to identify 230 patients with MG: 42.6% were in remission, 23.5% were lost to follow-up, and 25.2% were receiving prescriptions elsewhere. A similar pattern was observed for definitions #2 and #3: Patients with thymoma-associated MG who were followed-up by thoracic surgeons were frequently missed based on the previously used case definitions. A proposed case definition that additionally includes a history of thymoma enhanced the sensitivity to 77.9%, while maintaining a PPV of 99.4%. Conclusion The previously used case definitions showed a high PPV but were unable to identify patients in remission or followed-up elsewhere. These limitations can be overcome by utilizing a new case definition and expanding the study duration.