Direct bilirubin in clinical practice — Interpretation and haemolysis interference guidance reassessed

Abstract

ObjectiveDiagnosis of hyperbilirubinaemia may rely on direct bilirubin (DBIL) measurement. Specimen suitability for DBIL depends on the degree of haemolysis present and method specifications. Our study reassesses the empiric, syncretised haemolysis guidance for DBIL measurement. MethodsDiazo and vanadate-oxidation assays for DBIL are compared, using sera from adults and neonates. Subsequent data analyses and statistics guide our conclusions on the clinical utility of haemolysed sera. Results48% Patients' specimens had haemolysis exceeding the diazo-DBIL assay's specified haemolysis limit, thus generating ‘specimen unsuitable’ comment on clinical reports. Our linearity study shows a diazo-DBIL bias of up to 21μmol/L in diluted specimens, regardless of grade of haemolysis. Haemolysed patients' serum gives unpredictably biased diazo-DBIL results relative to results from vanadate-oxidation method. The pleiotropy of effect includes negative, negligible, and positive bias; a greater bias is apparent when albumin is <37g/L. Predominant unconjugated hyperbilirubinaemia is defined as the 90th percentile D/TBIL limit of <36%, mixed hyperbilirubinaemia by D/TBIL 36–60%, and mainly conjugated hyperbilirubinaemia when D/TBIL is >60%. ConclusionsHaemolysed diazo-DBIL result is unsuitable for clinical use, due to unpredictability of observed bias. Our data shows that current haemolysis guidance is inaccurate and in need of revision. There is no direct association between haemolysis and diazo-DBIL results. Notwithstanding haemolysis, type of assay, and low albumin concentration, our study show D/TBIL% facilitates detection of the nature of hyperbilirubinaemia present. In addition, laboratories have an option to select a more befitting assay, such as the vanadate-oxidation method as it is unaffected by haemolysis.