Estimation of the Prevalence of Amyotrophic Lateral Sclerosis in the United States Using National Administrative Healthcare Data from 2002 to 2004 and Capture-Recapture Methodology

Abstract

Background: National administrative healthcare data may be used as a case-finding method for prevalence studies of chronic disease in the United States, but the completeness of ascertainment likely varies depending on the disease under study. Methods: We used 3 case-finding sources (Medicare, Medicaid, and Veterans Administration data) to estimate the prevalence of amyotrophic lateral sclerosis (ALS) in the United States for 2002–2004, and applied the capture-recapture methodology to estimate the degree of under-ascertainment when relying solely on these sources for case identification. Results: Case-finding completeness was 76% overall and did not vary by race, but was lower for males (77%) than for females (88%), and lower for patients under age 65 (66%) than patients over age 65 (79%). The uncorrected ALS prevalence ratio was 2.8/100,000 in 2002, 3.3/100,000 in 2003, and 3.7/100,000 in 2004. After correcting for under-ascertainment, the annual prevalence increased by approximately 1 per 100,000 to 3.7/100,000 in 2002 (95% CI 3.66–3.80), 4.4/100,000 in 2003 (95% CI 4.34–4.50), and 4.8/100,000 in 2004 (95% CI 4.76–4.91). Conclusions: Federal healthcare claims databases ascertained are a very efficient method for identifying the majority of ALS-prevalent cases in the National ALS Registry, and may be enhanced by having patients self-register through the registry web portal.