Andrea M Cook1, Andreas Moritz1, Kathleen P Freeman2, Natali Bauer3. 1. Department of Clinical Pathophysiology and Clinical Pathology, Faculty of Veterinary Medicine, Justus-Liebig-University, Gießen, Germany. 2. IDEXX Laboratories, Ltd, Grange House, Wetherby, UK. 3. Department of Clinical Pathophysiology and Clinical Pathology, Faculty of Veterinary Medicine, Justus-Liebig-University, Gießen, Germany. natali.bauer@vetmed.uni-giessen.de.
Abstract
BACKGROUND: Scarce information exists about quality requirements and objective evaluation of performance of large veterinary bench top hematology analyzers. OBJECTIVE: The study was aimed at comparing the observed total error (TEobs ) derived from meta-analysis of published method validation data to the total allowable error (TEa ) for veterinary hematology variables in small animals based on experts' opinions. Ideally, TEobs should be < TEa . METHODS: An online survey was sent to veterinary experts in clinical pathology and small animal internal medicine for providing the maximal allowable deviation from a given result for each variable. Percent of TEa = (allowable median deviation/clinical threshold) * 100%. Second, TEobs for 3 laser-based bench top hematology analyzers (ADVIA 2120; Sysmex XT2000iV, and CellDyn 3500) was calculated based on method validation studies published between 2005 and 2013 (n = 4). The percent TEobs = 2 * CV (%) + bias (%). The CV was derived from published studies except for the ADVIA 2120 (internal data), and bias was estimated from the regression equation. RESULTS: A total of 41 veterinary experts (19 diplomates, 8 residents, 10 postgraduate students, 4 anonymous specialists) responded. The proposed range of TEa was wide, but generally ≤ 20%. The TEobs was < TEa for all variables and analyzers except for canine and feline HGB (high bias, low CV) and platelet counts (high bias, high CV). CONCLUSIONS: Overall, veterinary bench top analyzers fulfilled experts' requirements except for HGB due to method-related bias, and platelet counts due to known preanalytic/analytic issues.
BACKGROUND: Scarce information exists about quality requirements and objective evaluation of performance of large veterinary bench top hematology analyzers. OBJECTIVE: The study was aimed at comparing the observed total error (TEobs ) derived from meta-analysis of published method validation data to the total allowable error (TEa ) for veterinary hematology variables in small animals based on experts' opinions. Ideally, TEobs should be < TEa . METHODS: An online survey was sent to veterinary experts in clinical pathology and small animal internal medicine for providing the maximal allowable deviation from a given result for each variable. Percent of TEa = (allowable median deviation/clinical threshold) * 100%. Second, TEobs for 3 laser-based bench top hematology analyzers (ADVIA 2120; Sysmex XT2000iV, and CellDyn 3500) was calculated based on method validation studies published between 2005 and 2013 (n = 4). The percent TEobs = 2 * CV (%) + bias (%). The CV was derived from published studies except for the ADVIA 2120 (internal data), and bias was estimated from the regression equation. RESULTS: A total of 41 veterinary experts (19 diplomates, 8 residents, 10 postgraduate students, 4 anonymous specialists) responded. The proposed range of TEa was wide, but generally ≤ 20%. The TEobs was < TEa for all variables and analyzers except for canine and feline HGB (high bias, low CV) and platelet counts (high bias, high CV). CONCLUSIONS: Overall, veterinary bench top analyzers fulfilled experts' requirements except for HGB due to method-related bias, and platelet counts due to known preanalytic/analytic issues.
Authors: Sarah Hindenberg; Stefanie Klenner-Gastreich; Nicole Kneier; Sabine Zielinsky; Kris Gommeren; Natali Bauer; Andreas Moritz Journal: BMC Vet Res Date: 2017-05-30 Impact factor: 2.741