distribution s'averait equivalente a celle de Gauss, ainsi qu'aux valeurs trompeu- ses. On utilisa l'analyse parametrique lorsque la distribution des valeurs obte-.
Hematology and Biochemistry Reference Values for the Light Horse J. H. Lumsden, R. Rowe and K. Mullen*
ABSTRACT Reference hematology and biochemistry intervals are presented for a number of variables of clinical interest determined for blood samples obtained from 60 thoroughbred mares, 12 thoroughbred foals and 50 standardbred horses in training. The observations for each variable were examined for outliers and Gaussian distribution. Parametric analysis was used where the observations were Gaussian initially or after any of four transformations, otherwise nonparametric analysis was required for estimation of the 2.5 and 97.5 percentiles. Description of the sample collection procedures, laboratory methods and statistical analysis are available allowing comparison and judicious application of these reference intervals by interested researchers and clinicians.
distribution s'averait equivalente a celle de Gauss, ainsi qu'aux valeurs trompeuses. On utilisa l'analyse parametrique lorsque la distribution des valeurs obtenues correspondait a celle de Gauss, initialement ou apres l'une ou l'autre de quatre transformations appropriees; il fallut cependant recourir a l'analyse non parametrique pour determiner les centiles 2.5 et 97.5. La disponibilite de la description des m'thodes de prelevement des echantillons, ainsi que des techniques de laboratoire et d'analyse statistique permet une comparaison et une application judicieuses de ces valeurs de reference par les chercheurs et les cliniciens interesses.
INTRODUCTION
Reference values are the observations made on an individual or groups of individuals in defined states of health (9). If considered for application outside of RESUMC the laboratory in which the observations Cette experience visait a colliger des were generated the reference values valeurs de reference, relatives a certains must be accompanied by the following: parametres hematologiques et biochimi- description of the stated health of the ques d'interet clinique basees sur l'ana- individuals, the specimen collection and lyse d'echantillons de sang preleves chez handling procedures, the laboratory 60 juments et 12 poulains Thoroughbred, methods including precision and quality ainsi que chez 50 chevaux Standardbred, control and the necessary support for astous a l'entrainement. On analysa a cette sumptions or hypothesis made regarding fin les donnees qui se rapportaient a cha- data distribution if parametric analysis cun des parametres choisis, en accordant is used to compute reference intervals une attention speciale a celles dont la e.g. 2.5 and 97.5 percentiles (2, 34). Introduction of this nomenclature as prescribed has been stated to be "an important step toward establishing a scientific basis for *Department of Pathology (Lumsden), De- clinical interpretation of laboratory data" partment of Clinical Studies (Rowe) and Department of Mathematics and Statistics (Mullen), University of Guelph, Guelph, Ontario NlG 2W1. Submitted March 23, 1979.
32
(34).
Even when the need to generate within laboratory reference intervals is recognized, too few veterinary laboratories have the necessary resources to generate
Can. J. comp. Med. 44: 32-42 (January 1980)
the statistically desirable numbers of ob- nificantly altered within stated time inservations (18, 22) for each subpopula- tervals (18). The blood sample handling and laboration of interest especially when the additional species, breed and husbandry con- tory methods used have been described ditions are considered. previously (23, 24) except for the followHematology and biochemistry reference ing procedures. Haptoglobin was deterintervals for 60 thoroughbred mares, 12 mined by electrophoretic separation of thoroughbred foals and 50 standardbred hemoglobin-haptoglobin complex from an horses were developed according to de- excess but known amount of hemoglobin2, scribed procedures for use in a veterinary stained with o-dianisidine dihydrochloride teaching hospital. These reference in- and quantitated by transmission densitotervals are presented for appropriate use by metry3. clinicians and possibly researchers beBlood collected in sodium fluoridecause of the particular subpopulations oxalate and stored at -70°C was used for and the range of variables included. lactate determination4 at 30°C with absorbance readings at 340 nm. Serum sorbitol dehydrogenase was determined according to the method of Gerlach and Schurmeyer (14)5 with reaction at 30°C MATERIALS AND METHODS and monitored at 340 nm. The hemoglobin, hematocrit, erythrocyte The thoroughbred mares and foals and leukocyte counts were determined were stabled under the supervision of a with the Coulter Model S' and also the resident veterinarian. Blood samples Coulter hemoglobinometerl, microhemawere obtained from ten animals each tocrit7 and the Coulter Model ZBIP respecweek during May, June and July. The tively. The reference intervals reported mares were in late pregnancy (n = 9), in the tables were determined from obopen (n = 18) or in early pregnancy (n servations made using the Model S instru= 33). The foals were between two and ment. eight weeks of age. The standardbred Daily known within-lot quality control horses, in light to heavy training pro- sera at one or more levels were examined grams, were from several stables and in- with each group of samples, daily charts cluded various ages and sex. Blood sam- maintained and on a monthly basis statisples were obtained from six horses each tically compared with the previous three week of collection during September months observations for significant through January. Each animal was con- changes in the variable mean value or sidered to be clinically normal based upon standard deviation. Bimonthly, unknown careful assessment of the health history sera were examined to provide external and physical examination prior to collec- quality control. As an indication of anation of the blood samples. lytical bias inherent in the methods and Jugular blood samples were collected to aid clinical interpretation, the standdirectly into vacuum tubes', by separate ard deviation at one level of known conindividuals for each group of animals, trols, generally that closest to the high between 07:30 h and 09:30 h. The blood normal range, is included in the tables. samples were brought to the laboratory within 11/2 h. Plasma and serum were immediately separated from the cells by 2Corning ACI Agarose Film, Corning ACI, double centrifugation (23). Many test Palo Alto, California. procedures were begun immediately, e.g. platelet counts, and the majority, a stand- 3Clifford Electrophoresis Densitometer, Clifard number for each group of animals, ford Instruments Inc., Natick, Massachusetts. were completed within the working day. 4Calbiochem Rapid Lactate kit, Calbiochem, Plasma and serum were stored at 4°C or La Jolla, California. -200C for determination of the additional 5Calbiochem SDH kit, Calbiochem, La Jolla, variables previously shown to not be sig- California. Coulter Electronics of Canada Ltd., Missis'Becton, Dickinson and Company, Clarkson, Ontario.
sauga, Ontario. 7Adams Micro-hematocrit Centrifuge, ClayAdams Inc., New York, New York.
33
The within group observations for each variable were examined for outliers and Gaussian distribution. The reference intervals presented were calculated using parametric analysis when the observations were Gaussian, or able to be made Gaussian by any of four transformations or calculated using nonparametric analysis if nonGaussian as previously described (22). The SI units expected to be initially adopted by Canadian medical laboratories have been used for these reference intervals.
RESULTS The hematology and biochemistry reference intervals obtained for the 60 thoroughbred mares, 12 foals and 50 standardbred horses are presented in Tables I-IV, V-VII and VIII-X respectively. Where the numbers of observations (n) for a variable are less than the maximum number of animals within the group, there may have been inadequate technical time to complete the test(s) within the
defined time limits, inadequate sample volume or the observation (s) was discarded as an outlier according to previously described criteria (22). The observations from the thoroughbred mares were examined for the effects of the stage of pregnancy. Mean values for the variables significantly affected by the stage of pregnancy are presented in Table IV. Because of the small thoroughbred foal population sample, the minimum and maximum observations are the best estimates of the lower and upper limits of 95% of the population and are presented in Tables V-VII. The 2.5 and 97.5 percentiles calculated as described correlate well with the minimum and maximum observations. Much poorer correlation was found when nonGaussian data was analysed parametrically.
DISCUSSION The clinician must interpret laboratory observations in relation to the history, clinical findings and disease incidence
TABLE I. Hematology Reference Values for Thoroughbred Maresa
B-Hemoglobin B-Hematocrit BErythrocytes
Ery-MCV Ery-MCH Ery-MCHC B-ESR P-Protein B-Platelets B-Leukocytes B-Neutrophils Segmented Band B-Lymphocytes
g/L % x1012/L fl pg g/L mm/h g/L
x109/L x109/L x109/L % x109/L NO x1O9/L % x1O9/L
2.5 109 29 6.5 39 15.1 337 37 60 80 5.3
97.5b 188 53 11.6 49 18.2 386 62 75 397 11.0
x 148 40.8 9.0 45.3 16.6 361 49.3 68 239 8.1
SD 20 6.0 1.3 2.1 0.8 12 6.4 4 80 1.5 1.0 8.2 0.06 0.7 0.8 7.8 0.1 1.7 0.2 2.6 0.03
n
de
60 60 60 60 60 60 52 59 60 60
G G G NP G G G G G G G G NP NP G G
SDd 2 0.6 0.09 0.8 0.8 5 0.1
2.1 6.0 4.1 59 35 68 51.4 60 0 0.2 0.03 59 0 2 0.4 60 1.7 5.0 3.4 59 26 57 60 41.8 B-Monocytes 0 v/x 0.6 0.2 60 0 6 NP 2.9 60 5% B-Eosinophils x109/L 0 Vx 0.8 0.2 60 0 10 60 2.7 V/x No B-Basophils x109/L 0 0.1 NP 0.02 59 0 0.4 NP 1.0 0.2 59 % -Thoroughbred mares open, early and late pregnancy bDeternined 2.5 and 97.5 percentile intervals oDistribution of data or method of analysis: G = Gaussian, >/x = square root, log e - log base e transformation, NP =nonparametric analysis dAnalytical bias (1SD) observed for one level, generally high normal range, control sample (n > 30, < 50)
34
TABLE II. Biochemistry Reference Values for Thoroughbred Mares 2.5 2.67
97.5b 3.09 1.42 9.3 0.95 142 4.2 66 106 1.72 296 24.1 185 29 8.33 3.57
x
SD
n
dc G G G G G NP NP G G G G NP NP 1/x G
SDd 0.05
dc
SDd
S-Calcium mmol/L 2.89 0.10 60 S-Phosphorus 0.36 mmol/L 0.26 60 0.87 0.09 2.5 4.7 1.9 60 Ca/P 0.66 0.82 0.08 60 S-Magnesium mmol/L 0.02 P-Sodium mmol/L 134 2.2 138 59 1.2 P-Potassium mmol/L 2.1 0.1 3.5 60 0.5 Na/K 33 40.4 7.3 59 P-Chloride mmol/L 94 99.5 3.1 60 0.9 B-Lactate mmol/L 0.28 1.01 60 0.37 P-Osmolality mmol/kg 279 288 4.5 60 2.9 B-Urea Nitrogen mmol/L 11.3 60 17.7 3.3 1.0 S-Creatinine 80 ,umol/L 18 60 5 115 BUN/Creatinine 9 60 20.5 5.0 P-Glucose mmol/L 3.83 0.21 5.27 1.54 58 S-Cholesterol mmol/L 1.89 2.72 0.43 60 0.2 S-Bilirubin Total tmol/L 15.4 2.9 22.2 44.5 6.8 60 1/x Unconj. ,umol/L 5.1 29.1 15.4 6.8 60 V/x Conj. 3.4 8.6 ,mol/L 13.7 3.4 60 W Unconj/conj 0.7 4.5 2.2 1.1 60 NP S-Iron 15.2 wxmol/L 41.3 G 28.3 6.6 60 1.0 S-Iron Binding 88 Capacity 57 G ,umol/L 73 7.8 60 2.4 S-Iron Binding Saturation % 23 64 39 8.7 60 NP S-Thyroxine nmol/L 10 30 49 9 G 51 7 P-Cortisol nmol/L 50 637 342 22 G 149 36 &Thoroughbred mares open, early and late pregnancy bDeternined 2.5 and 97.5 percentile intervals "Distribution of data or method of analysis: G = Gaussian, Vx = square root, log e = log base e transformation, 1/x - inverse, NP = nonparametric analysis dAnalytical bias (1SD) observed for one level, generally high normal range control samples (n > 30, < 50) TABLE III. Biochemistry Reference Values for Thoroughbred Maresa
S-Amylase S-Alkaline phos-
Caraway units/L (37°C)
phatase U/L (37°C) S-Creatine phosphokinase U/L (30°C) S-Lactate dehydrogenase U/L (30°C) S-Sorbitol dehydrogenase U/L (30°C) S-Transaminases Aspartate amino (SGOT) U/L (30°C) Alanine amino U/L (30°C) (SGPT) S-Protein g/L S-Albumin g/L S-Globulin g/L Alb/globulin
2.5
97.5b
i
SD
n
120
590 59.0
350 16.7
60
26
1400 92
60
W G
1
88
44.5
22.3
59
G
81
225
36.7
60
1.8
1.2
59
G v7
165
33.6
60
0.2 99
4.7 231
153
G G G G G
144
2.1
18.8 8.8
2.2
0 5 2.8 1.4 60 2.3 4 60 57 74 65 16 29 32 2 60 36 41 33 60 26 4 12 1 V/x 7 10 50 S-czglobulin g/L 5 14 2 NP 10 60 9 18 13 2 g/L 60 V/x S-5 globulin S- globulin 6 14 10 2 60 G g/L P-Fibnnogen ,umol/L 4.7 8.4 G 6.6 1.0 59 P-Haptoglobin g/L 48 32 G 172 110 60 &Thoroughbred mares open, early and late pregnancy bDetermined 2.5 and 97.5 percentile intervals eDistribution of data or method of analysis: G = Gaussian, Vi/ = square root, log e = log base e transformation, NP = nonparametric analysis dAnalytical bias (1 SD) observed for one level, generally high normal range, control sample (n >30,
