Medicare請求データにおける血友病患者の特定方法

Medical claims are increasingly used to study haemophilia, a rare and costly X-linked congenital bleeding disorder. However, there is no standard method to identify people with haemophilia A (PwHA) in medical claims, and it is unknown how existing approaches influence the cohort selected. Thus, this study evaluates cohort characteristics across three existing methods to identify PwHA in Medicare claims data.

Medicare claims data (2000-2020) were used to identify PwHA using three approaches: Cohort A: A historic validated algorithm relying on eight diagnosis, treatment, and health services variables; Cohort B: receipt of haemophilia treatment, and Cohort C: generic health services definition of ≥2 haemophilia diagnosis codes 30 days apart.

Five thousand six hundred and two unique beneficiaries were identified as having haemophilia A across the three methods. Three-quarters of beneficiaries (76.7%) were in Cohort C (diagnosis) alone, 16.1% and 3.7% were in Cohorts A (algorithm) and B (treatment), respectively. Fewer than 5% were in two or more cohorts. Over 85% of those in Cohorts A (88.8%) and B (85.2%) were male and received haemophilia treatment. In Cohort C, less than half (42.9%) were male, and <10% had haemophilia treatments recorded.

In this study, we compared three methods to identify PwHA in claims. Generic diagnosis coding may overcount those tested for haemophilia but ultimately not diagnosed, whereas historical algorithms and treatment-based selection may underrepresent those with non-severe haemophilia, including females. Contemporary validated algorithms are needed to improve the utility of claims for haemophilia research.