Professional Documents
Culture Documents
Reference Values in
Veterinary Medicine
i
be included in the reference population. Too
States. (902) in this issue of
many geriatric animals (older than 10 years) may Laboratory Medicine.
'Reference values were established at Antech shift the reference values for some analytes. There Participants will earn
Diagnostics, Irvine, Calif. can be important differences between various 3 CMLE credit hours.
breeds of animals. The same "normal" values do
not apply to all dogs, nor do the same reference
intervals apply to all breeds of cats, cows, horses,
or birds, for example. If possible, 40 to 120 clini-
cally healthy animals of various ages and breeds,
and both sexes should be included in the general
reference population (Fig 2). Before samples are analyte. If the patient daily mean is near the high
obtained, specific written criteria should be given end of the reference value, more than half of the
the veterinarian to ensure that a standardized patients had elevated values. This should also be
"reference" population is selected for the study. obvious from the number of phone calls from
Without specific guidelines, too many variables practitioners about elevated results received.
will be introduced into the population, decreasing Calculation of reference values should include
sensitivity and specificity. the mean and standard deviation (SD)—and are
Reference values can also be determined by usually the mean value ±2 SDs, or 95% of the
using the laboratory patient database, on the population. However, most of the analytes in a
premise that the values for a large enough patient chemical profile (primarily enzymes) and CBC do
population (generally several thousand animals) not show Gaussian (normal or symmetric) distri-
yield a valid distribution if guidelines for accept- bution, and conventional procedures involving
able values are used. This assumes that most val- mean and standard distribution will yield skewed
ues from most patients are "normal" and that results. This would be obvious if the data were
values at both the high and low extremes do not plotted on a graph, or by lack of agreement
influence the calculations because of the large between the median, mode, and mean. Nonpara-
number of observations in the population. Values metric analyses or transformation analyses should
for the upper and lower limits depend on the be performed to obtain the most useful reference
mathematical method used and its assumptions.3 values. To make the data appear more Gaussian,
Even if the patient database is not used to percentile or square-root transformations of the
determine reference values, it may prove valuable data can be used.4 Most computer spreadsheet
for confirming that the reference values are rea- programs or statistical packages enable perfor-
sonable. Patient daily mean reports can be gener- mance of these statistical analyses.
ated by most laboratory computer systems. These Analysis of the reference population may result
reflect the average of the previous day's results for in narrower values than previously used or pub-
all patients for all tests. The mean value generally lished in the literature. Many of the references in
does not fluctuate significantly from day to day, the literature are outdated, and the analyses were
and is often helpful for quality assurance. The performed on older equipment and reagents. In
patient daily mean value for an analyte should be addition, some of the older values are broader. It
near the middle of the reference value for that
•
mal species have considerably smaller RBCs Please let us know your opinion of the Veterinary 0
Vledicine (902) series. Place an X in one box for each ft'ifi 9
(mean corpuscular volume 30-70 fL), often less •
question. Return this form (or a photocopy) by fax to: WiPi U
than the RBC counting thresholds set for human 312) 850-8817; or, mail to: ASCP Press Administration, E
3
RBCs, therefore giving falsely decreased RBC 2100 W Harrison St, Chicago, IL 60612-3798. Thank you
'or your input. £
counts and hematocrit calculations and interfer- £
o
ing with calculation of RBC indices. Deficient Excellent o
In addition, certain reagent systems are not 1. The series met the objectives stated in the abstract.
9905
•
Downloaded from https://academic.oup.com/labmed/article-abstract/30/3/194/2504134
by guest MARCH 1999 VOLUME 30, NUMBER 3 LABORATORY MEDICINE 197
on 28 May 2018