Abstract
PURPOSE: To examine the validity of diagnosis code algorithms for identifying amyloidosis and its most common subtypes in the Danish National Patient Registry (DNPR). DESIGN AND SETTING: Validation study of routine amyloidosis registrations in the DNPR. As the reference standard, we used medical record review of discharge summaries and diagnostic examinations. STUDY POPULATION: We included all patients identified through an algorithm with a first-time primary or secondary, inpatient or outpatient amyloidosis diagnosis at Aarhus University Hospital between December 1, 2018, and November 30, 2023. Subtype classification was performed using predefined algorithms to identify wild-type transthyretin amyloidosis (ATTRwt), amyloid light-chain (AL) amyloidosis, and variant ATTR (ATTRv). MAIN OUTCOME: We calculated positive predictive values (PPVs) as the number of patients with a confirmed diagnosis divided by the total number of registered diagnoses in the DNPR. Subgroup analyses were performed for ATTRwt. RESULTS: Among 334 patients registered with a first-time amyloidosis diagnosis, 313 had their diagnosis confirmed, corresponding to an overall PPV of 94% (95% CI: 91-96). The PPVs for the individual subtypes were 85% (95% CI: 80-90) for ATTRwt, 90% (95% CI: 79-96) for AL amyloidosis, and 57% (95% CI: 33-79) for ATTRv. For ATTRwt, the PPV increased to 94% (95% CI: 90-97) among patients aged ≥60 years registered in cardiology department. CONCLUSION: Routine amyloidosis registration in the DNPR demonstrated high validity overall and for the ATTRwt and AL amyloidosis subtypes.