Utility of Capture-Recapture Methodology to Assess Completeness of Amyotrophic Lateral Sclerosis Case Ascertainment

Abstract

Background: With the establishment of a national amyotrophic lateral sclerosis (ALS) registry in the United States, methods are needed to ascertain the completeness of case ascertainment, especially in view of the proposal to rely largely on existing data sources. Methods: Data about ALS patients residing in the 5-county metropolitan Atlanta area (within the State of Georgia) from 2001 to 2005 were categorized according to their source – ALS Association, clinical (Emory Healthcare, community neurologist, Veterans Health Administration, Veterans Benefits Administration), Medicare and death certificates. ALS diagnoses were verified using chart review. Capture-recapture analyses were carried out using log-linear modeling, stratified by age and race. Results: The final model (based on 798 cases), which included the 4 main sources and 3 two-way interaction terms, yielded an estimated total population of 880 (95% CI 816–965), indicating that the combination of case-finding methods identified about 90.7% of cases. The estimated 5-year period prevalence is 38.5/100,000 (95% CI 35.66–42.19). Conclusion: This study highlights gaps in data based on existing data sources and illustrates a method for combining data from multiple sources to help facilitate the successful establishment of a US national ALS registry.

Epidemiology of Amyotrophic Lateral Sclerosis and Effect of Riluzole on Disease Course

“…The proportion of unobserved cases in epidemiological studies on ALS has previously been estimated to be as high as 28% [14,20]. Not accounting for such an important bias is therefore likely to lead to a substantial underestimation of the true IR and PR [10,21]. In our study 10.3% of the incident and 14.0% of the prevalent cases from 2009 to 2011 were determined by the use of the capture-recapture method.…”

“…To account for unobserved cases neither being detected by the hospital discharge database (HDD) or the prescription database (PD) the two-source capture-recapture method was performed [9,10]. This refers to outpatients never having been hospitalised or prescribed riluzole during the study period.…”

Prevalence of Amyotrophic Lateral Sclerosis — United States, 2012–2013

“…A statistical approach to determine the completeness of the Registry is capture-recapture analysis which is a method for estimating the maximum population with a disease based on the "capture" of specific individuals different sources (44)(45)(46). This technique has been utilized previously on a smaller and national geographic scale for the under reporting of ALS cases (47,48). Using capture-recapture on a national level for the administrative databases and portal data are currently underway by ATSDR to estimate the completeness of the Registry.…”

The Epidemiology of Amyotrophic Lateral Sclerosis in New Hampshire, USA, 2004-2007

“…There could also be a bias according to the accuracy of the MND diagnosis listed on the death certificates. False positives have been documented in other studies, and it should also be noted that the current ICD-10 code for ALS is less specific than previous codes (ICD-9 codes for ‘other paralytic syndromes' and ‘other specified idiopathic peripheral neuropathy' were assigned to the ICD-10 code for ALS) [28,29,30]. …”

“…The hospital and clinic encounters of these patients were then reviewed to determine whether the case represented probable ALS (from documentation of an encounter with a neurologist whose notes confirmed the diagnosis, by electromyography criteria or if the patient had received a prescription for riluzole). Regional MDA registration data were then examined [28,29]. Diagnoses were confirmed and additional information was obtained by review of the medical records [30] such as: the year of diagnosis, the age of the patient at diagnosis, their gender, occupation and residential address, the certainty of the ALS diagnosis, any family history of ALS and a history of tobacco use.…”