BACKGROUND: Results from hemolyzed, icteric, and lipemic samples may be inaccurate and can lead to medical errors. These preanalytical interferences may be detected using visual or automated assessment. Visual inspection is time consuming, highly subjective and not standardized. Our aim was to assess the comparability of automated spectrophotometric detection and visual inspection of lipemic, icteric and hemolyzed samples. METHODS: This study was performed on 1727 routine biochemistry serum samples. Automated detection was performed using the Olympus AU2700 analyzer. We assessed: 1) comparability of visual and automated detection of lipemic, icteric and hemolyzed samples, 2) precision of automated detection, and 3) inter-observer variability for visual inspection. RESULTS: Weighted kappa coefficients for comparability of visual and automated detection were: 0.555, 0.529 and 0.638, for lipemic, icteric and hemolyzed samples, respectively. The precision for automated detection was high for all interferences, with the exception of samples being only slightly lipemic. The best overall agreement between observers was present in assessing lipemia (mean weighted kappa=0.698), whereas the lowest degree of agreement was observed in assessing icterus (mean weighted kappa=0.476). CONCLUSIONS: Visual inspection of lipemic, icteric and hemolyzed samples is highly unreliable and should be replaced by automated systems that report serum indices.
BACKGROUND: Results from hemolyzed, icteric, and lipemic samples may be inaccurate and can lead to medical errors. These preanalytical interferences may be detected using visual or automated assessment. Visual inspection is time consuming, highly subjective and not standardized. Our aim was to assess the comparability of automated spectrophotometric detection and visual inspection of lipemic, icteric and hemolyzed samples. METHODS: This study was performed on 1727 routine biochemistry serum samples. Automated detection was performed using the Olympus AU2700 analyzer. We assessed: 1) comparability of visual and automated detection of lipemic, icteric and hemolyzed samples, 2) precision of automated detection, and 3) inter-observer variability for visual inspection. RESULTS: Weighted kappa coefficients for comparability of visual and automated detection were: 0.555, 0.529 and 0.638, for lipemic, icteric and hemolyzed samples, respectively. The precision for automated detection was high for all interferences, with the exception of samples being only slightly lipemic. The best overall agreement between observers was present in assessing lipemia (mean weighted kappa=0.698), whereas the lowest degree of agreement was observed in assessing icterus (mean weighted kappa=0.476). CONCLUSIONS: Visual inspection of lipemic, icteric and hemolyzed samples is highly unreliable and should be replaced by automated systems that report serum indices.
Authors: Euan J McCaughey; Elia Vecellio; Rebecca Lake; Ling Li; Leslie Burnett; Douglas Chesher; Stephen Braye; Mark Mackay; Stephanie Gay; Tony C Badrick; Johanna I Westbrook; Andrew Georgiou Journal: Clin Biochem Rev Date: 2016-12