Data mining of reference intervals for coagulation screening tests in adult patients

Appropriate reference intervals are essential when evaluating laboratory test results. However, establishment of reference intervals is challenging, especially for coagulation screening tests, and uncertainty exists regarding age- and sex-dependency of test results. Data mining of laboratory information systems is an emerging approach to reference interval determination, and we evaluated its applicability to coagulation tests.

We analyzed measurements of activated partial thromboplastin time (aPTT), prothrombin time (PT), international normalized ratio (INR), thrombin time (TT), and fibrinogen performed during clinical care in the University Hospital Erlangen, Germany (1,778,738 samples from 116,754 adult patients, 45,577-509,859 samples per analyte). We identified the proportion of samples from healthy individuals using an established statistical approach (Reference Limit Estimator), in which the distribution of physiological test results is approximated using a parametrical function, and used for the calculation of reference intervals.

We established age- and sex specific reference intervals for aPTT, PT, INR, TT, and fibrinogen, and created batch- and reagent-specific aPTT-reference intervals. Additionally, we evaluated the sensitivity of the established aPTT reference intervals for the detection of factor VIII, IX, XI, XII deficiencies.

Data mining of laboratory test results allows the creation of age- and sex-reference intervals for coagulation tests that are specific to the examined population, analytical framework, and reagent. This approach can complement conventional methods when establishing reference intervals and improve clinical decision-making based on coagulation tests. The reference intervals established in this study show only minor variation with sex and age, supporting the practice of providing a common reference interval for adult women and men.