OBJECTIVES: To develop an analytic method to measure uncoded disease outcomes in a retrospective claims database. METHODS: An analysis of multiple sclerosis (MS) patients in a large, vertically integrated healthcare system database was conducted with the primary outcome of interest being a relapse event. MS only has only one International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code, which does not specify whether the claim was a result of a relapse or other event. A review of the literature found no surrogate markers or composite scores that could be identifiable in claims data. Though Magnetic Resonance Imaging (MRI) is useful in the treatment of MS, typical claims only report whether or not a scan was performed and no results or reason for the scan are available. Utilizing information from published literature and a physician panel, a treatment pattern was identified that could be used in a claims database and would indicate a relapse in MS patients – two consecutive days of an IV steroid (either methylprednisolone or dexamethasone). The date of relapse was recorded as the first date of IV steroid treatment. RESULTS: Using a retrospective claims database, a multiple sclerosis (MS) population utilizing one of four major MS drugs (341 patients) was identified. Within the study period (1994-2005), 67 relapses occurred, defined as two consecutive dates of IV steroid therapy, which is a common clinical treatment for symptoms related to MS relapse. This method did not identify all relapses in the study population; it would be necessary to identify more clinical combinations to achieve comparable relapse rates as those found in the published literature. CONCLUSION: Uncoded events can be identified using clinical treatment measures, which are prominent in claims databases.