Incidence of the myelodysplastic syndromes using a novel claims-based algorithm: high number of uncaptured cases by cancer registries

Abstract

The myelodysplastic syndromes (MDSs) are hematologically diverse hematopoietic stem cell malignancies primarily affecting older individuals. The incidence of MDS in the United States is estimated at 3.3 per 100 000; however, evidence suggests underreporting of MDS to centralized cancer registries. Contrary to clinical recommendations, registry guidelines from 2001-2010 required the capture of only one malignancy in the myeloid lineage and did not require blood count (BC) or bone marrow (BM) biopsy for MDS confirmation. To address these potential limitations, we constructed 4 claims-based algorithms to assess MDS incidence, applied the algorithms to the 2000-2008 Surveillance Epidemiology and End Results (SEER)–Medicare database, and assessed algorithm validity using SEER-registered MDS cases. Each algorithm required one or more MDS claims and accounted for recommended diagnostic services during the year before the first claim: 1+, 2+, 2 + BC, and 2 + BCBM (ordered by sensitivity). Each had moderate sensitivities (78.05%-92.90%) and high specificities (98.49%-99.84%), with the 2 + BCBM algorithm demonstrating the highest specificity. Based on the 2 + BCBM algorithm, the annual incidence of MDS is 75 per 100 000 persons 65 years or older—much higher than the 20 per 100 000 reported by SEER using the same sample.