EDTA blood samples from 40 birds with various clinical signs that were admitted between May 2014 and July 2015 to the Clinical Laboratory, Vetsuisse Faculty, University of Zurich, were included in the study. All samples were analyzed within 24 h after collection with the reference method and the test method. The study population consists of 22 different species: 7 Gy parrots, 2 gamefowl, 5 common buzzards, 1 capercaillie, 1 lovebird, 3 keas, 1 stork, 1 scarlet macaw, 1 bar-headed goose, 1 African pygmy goose, 1 toucan, 4 cockatoos, 1 blue-fronted amazon, 1 cockatiel, 1 domestic chicken, 1 crow, 1 Virginia quail, 1 peacock, 2 geese, 3 guinea fowl, 1 parrot, and 1 mute swan.
The within-assay precision of one EDTA-blood sample from a kea was analyzed 10 consecutive times by a skilled technician with the reference method and the Natt-Herrick-Tic® test method.
Reference method: Natt-Herrick method 
Natt-Herrick solution for the reptiles and birds was prepared in-house according to the methods described by Natt and Herrick (1952). The prepared staining solution was filtered and stored at room temperature protected from light. The storage life of the staining solution is unlimited. However, filtering procedures were performed from time to time. EDTA blood samples were mixed in a ratio of 1:100 with Natt-Herrick solution. For this purpose, 990 μL staining solution and 10 μL blood were pipetted and mixed with a micropipette (Calibra, Socorex, Ecublens, Switzerland) in a sample vial (tubes 3.5 mL, Sarsted AG & Co., Nümbrecht, Germany). The sample-stain mixture was incubated and mixed for 5–10 min on an automated mixer (Rock ‘n Roller 34,201, Snijders Scientific B.V., AR Tilburg, Nederland) at room temperature. Afterwards, the sample-stain mixture was filled into a Neubauer Improved hemocytometer (BLAUBRAND®-Zählkammern, Brand Gmbh, Wertheim, Germany). The hemocytometer was incubated in a petri dish containing wet filter paper for 2 min, allowing sedimentation of the cells. After sedimentation, instead of counting all leukocytes, only the granulocytes were enumerated using a Laborlux S microscope (Leitz, Lenzburg, Germany) by counting the cells in the four large outer squares containing sixteen small squares at 400 x magnification. Counting was performed in duplicate by counting both chamber grids of one chamber. The mean of both chamber grids was calculated and multiplied by a factor of 250 to obtain the noncorrected white blood cell count (NCLC) per μL. The NCLC only includes eosinophils, basophils and heterophils. To include lymphocytes and monocytes in the total white blood cell count (TLC), the NCLC was multiplied by 100 and divided by the sum of the percentages of heterophils, eosinophils and basophils determined by microscopic differentiation.
Test method: Natt-Herricks-tic® (bioanalytic GmbH, Freiburg, Germany)
A 5-μL end-to-end capillary tube was filled with EDTA blood and transferred into a test vial prefilled with Natt-Herricks-Tic® reagent, as provided by the manufacturer. The vial was mixed carefully until the blood was removed completely from the capillary. The sample-stain mixture was incubated and mixed for 2 to 5 min on an automated mixer at room temperature and then transferred into the Neubauer improved counting chamber using a capillary tube. The counting chamber was placed in a petri dish containing wet filter paper for 2 min, allowing sedimentation of the cells. Differently from the manufactures instruction, not all leukocytes but only granulocytes (heterophils, eosinophils, and basophils) were counted in one whole counting grid (9 large squares) of the counting chamber with the same microscope used for counting in the Natt-Herrick method at 100 x magnification (10x objective). To obtain the NCLC per μL, the number of counted cells was multiplied by a factor of 222.2. To include lymphocytes and monocytes in the TLC, the NCLC was multiplied by 100 and divided by the sum of the percentages of heterophils, eosinophils and basophils determined by microscopic differentiation.
Differentiation of white blood cells
From each EDTA blood sample, two blood smears were prepared and stained with an automated staining instrument (HEMA-TEK 2000 slide stainer, Bayer HealthCare AG, Berlin, Germany), using a modified Wright-Giemsa solution (Hematek® Stain Pak, Siemens Healthcare Inc., New York, USA). Two technicians with experienced in avian hematology differentiated one hundred white blood cells each. Out of the two differential counts, the mean was calculated to obtain a percentage for each cell type. If deviations between the two technicians exceeded 10% for heterophils and lymphocytes, a clinical pathologist was consulted for a final decision. Finally, the percentages were used to calculate the TLC as described above.
Statistical quality control is inadequate to control manual white blood cell counts in birds, and standardized control blood is not available for avian species. Nevertheless, to ensure the stability of both methods, the reference and test methods, non-statistical measures of quality control were applied to each single analysis. For each sample, chamber counting was performed in duplicate, and the mean was used to calculate the NCLC. If the deviation between the two counts exceeded 15%, the counting procedure was repeated. Furthermore, correlation with a blood smear evaluation was performed to ensure plausibility of the counting results.
To determine the clinical relevance of the observed differences between the reference and test methods for each sample, the results from the Natt-Herricks-Tic® method and the reference method (Natt-Herrick method) were compared with avian reference intervals from the current literature [14,15,16,17,18,19,20]. The results were classified as within, below (leukopenia) or above (leukocytosis) the reference interval. The resulting interpretations from the Natt-Herricks-Tic® and reference methods were compiled and compared to each other. If incongruous results were observed, i.e., a pathological finding in the test method (Natt-Herricks-Tic®) and not in the standard method (Natt-Herrick method), they were classified as false positive results. If no pathological finding was found in the test method (Natt-Herricks-Tic®) and a pathological finding was reported in the standard method, the result of the test method was classified as a false negative result. After completing the classification, the total number of false samples was identified, and the percentage of false samples in relation to the total number of samples was calculated.
All results were compiled in a table calculation program, and the Microsoft Excel add-in “Analyse-it” was used for statistical analyses (Analyse-it for Microsoft Excel 2010, Analyse-it-Software, Ltd. http://www.analyse-it.com). For the agreement study, TLC was used to calculate the Pearson coefficient of correlation (r), a Passing–Bablok linear regression analysis was performed providing an intercept and slope with a 95% confidence interval, and a Bland–Altman difference plot with biases and 95% limits of agreement was graphed. For precision analysis, the standard deviation (SD) and coefficient of variation (CV) were calculated.