Biological variation outweighs pre-analytical factors in coagulation reference intervals: a multicenter big data...

Defining reference intervals (RIs) for coagulation tests is challenging due to pre-analytical variability and the logistical difficulty of recruiting healthy donors. This study aimed to define RIs for Prothrombin Time (PT) and Activated Partial Thromboplastin Time (APTT) using a "Big Data" indirect approach and to evaluate the impact of pre-analytical and biological variables within a harmonized hospital network.

A multicenter retrospective study was conducted across nine Italian hospitals using shared analytical platforms (ACL TOP 50). Data from over 300,000 patients were analyzed using the indirect refineR algorithm, with local estimates aggregated via random-effects meta-analysis. Results were validated against direct methods performed on healthy blood donors.

Indirect estimates showed strong concordance with direct methods. The harmonized upper reference limit (URL) for APTT ratio was 1.19, unaffected by inpatient inclusion. For PT ratio, the pooled URL was 1.25, decreasing to 1.17 when excluding inpatients, closely matching direct estimates (1.15). While preanalytical variables (tube manufacturer, reagent type and sample transportation) showed negligible impact, biological variables were significant drivers of variation. PT URL displayed a marked J-shaped increase after age 50, up to 1.35 after age 80, with males consistently exhibiting higher values than females.

Harmonizing RIs across a hospital network is feasible, as technical differences do not justify local partitioning. However, the substantial influence of age and sex on PT suggests that introducing stratified RIs for the patients with age > 50 and sex partitioning may be necessary to improve diagnostic specificity.