Errors and Pitfalls Associated with Clinical Analyzers
Sally Lester Canada
The last decade has seen an increased utilization of in-house (in-clinic) chemistry and hematology analyzers as well as an increase in the POC (point-of-care) type analyzers used in emergency medicine. With this increased utilization comes additional responsibility. It is up to the practitioner providing this laboratory service to assure that the service is going to benefit the patient. The practitioner has the sole responsibility of quality assurance. This is especially important in North America where no regulations exist to govern the manufacturer of instruments for veterinary laboratory testing (or laboratories testing animal specimens).
The topic today revolves primarily around routine chemistry analyzers. Chemistry analyzers do not have “fail safe” mechanisms that are inherent in the hematology analyzers. A practitioner can (and should) always evaluate a blood film; and this will provide a rapid and accurate check on the validity of machine generated hematology results. Many of you are familiar with the sudden “outbreak “ of thrombocytopenia that developed when in-clinic hematology analyzers were used without evaluation of the blood film. This particular problem reflects the difficulty in measuring platelets rather than specific problems with any analyzer. The practitioners’ responsibility is to evaluate the film and make sure that the blood film matches the results.
Other than a refractometer to measure plasma protein as a check of the reliability of a machine generated total protein, there are no “quick” simple ways to access the reliability of chemistry results. Instead the practitioner must follow standard “quality assurance” guidelines. Unfortunately, there is no quicker method of putting an audience to sleep than to begin to discuss Quality Control. Therefore, for the purpose of this discussion we will refer to this subject as “critical thinking.” As part of the scientific community, we have all been taught the “scientific method” approach to problems—and that is the method I wish to employ in evaluating instrumentation. This requires us to be skeptical and to evaluate the machine as we would the patient. We must first establish what we want to achieve (goal). What is the purpose of the test(s)? Are we going to be evaluating well animals (pre-surgical screen)? Are we going to be accessing response to treatment in sick animals? Are we going to use this machine to make a diagnosis?
Once we have decided on the purpose then we need to consider the instrumentation and the test.
What specimen is required? How is the sample handled?
How does the machine deliver the specimen to the reaction chamber?
What is the basis of the test—is this an electrical procedure/photometric etc?
What is the range that the machine can measure?
How is temperature controlled?
What parts of the machine need replacing? What technical support is available?
What is the maintenance required to assure that the machine is working at optimum?
What can induce errors—both analytical and physiologic—in the test results?
What is necessary to assure that 1 + 1 = 2? Are there checks inherent within the machine and if so, how do I know they are performed and what are they?
How do we interpret the results that are generated?
FACTORS THAT INFLUENCE THE MACHINE ITSELF
The automated chemistry machine consists of an instrument that provides both reagent and sample delivery, a vessel in which the reaction occurs, a mixing of the two items, a controlled temperature, a timing device and then the actual measurement of the reaction. Many dry chemistry machines are based on reflectance photometry while electrolytes are measured with electrodes.
Factors that can influence the reactions include:
Speed of migration of the liquid phase through or along the slide, i.e., viscosity.
Most reactions are enzymatic and thus highly temperature dependent.
The accuracy of the dispensed volume both sample and reagent.
The timing of the reaction.
The individual performing the test
It is very important to have a way of assessing these factors either visually or as a part of the general machine maintenance so that we can confirm performance.
How is the sample collected?
Application of a tourniquet to induce venous stasis and allow blood collection can cause significant changes in analytes that include:
Increased Total Protein, AST/CK ( as high as 10%).
Decreased Potassium (as high as 8%).
Increased phosphorus and ionized calcium.
Increased bilirubin (even without visible hemolysis).
Anesthesia used for collection.
Decrease in total protein, pmns/wbcs.
Variability in PCV.
Increased muscle enzymes, glucose, decreased calcium.
Contamination of draw tubes.
Always fill tubes that have no anticoagulant first.
What changes occur with Na F/oxalate tubes?
What changes occur with EDTA?
What Changes Occur With Exercise?
(Some cats particularly undergo strenuous exercise during blood collection)
Increased CK, AST, creatinine and phosphorus.
Decreased iron, albumin and potassium.
The proficiency of the individual performing the test.
Are all individuals using the test following identical procedures (is there a uniform procedure manual)? Can two people make the same cookies?
Is it Time to Consider Medical Decision Levels?
A medical decision level is a critical value—this value can only be established if you know:
The “reference range,” how was this established, does it truly represent “normal” individuals in your practice area? Is age, sex, and breed important?
Need to know how precise your instrument is for each analyte. Is a calcium of 2.4 actually 2.35 to 2.45 or is it 2.41? If it is 2.45, what does that mean if your reference range upper limit is 2.43? Precision is really a test of reproducibility: if you measure an analyte 10 times does the value change and by how much?
Accuracy is a different concept and may not be a requirement of a test if it is reproducible, and can distinguish between “normal and abnormal.”
What is the effect of various drugs on the methodology used by the machine?
Cephalosporin, transaminases; Metronidazole decreases transaminases.
Hemoglobin substitutes will decrease ALP transaminases, for how long?
What other factors will influence the methodology?
Will storage of the sample affect the result? This is important if you want to compare prior samples with a sample taken three days later. If the results are disparate, how much of the disparity is the result of how the sample was stored?
Does the color of the sample interfere? Can anything be done about the sample color?
Will an alteration in protein present in the sample affect the result?
When using whole blood, does the hematocrit influence the result or can an abnormal WBC count influence the result?
The major post analytical variable that must be addressed is how you are going to interpret the results. The interpretation requires:
1. Correlation with the history and physical examination.
2. Understanding of the limits of the assay.
3. Understanding of the effects of disease and treatment on the various parameters; a sense of the probabilities of any particular disease creating a set of altered parameters.
4. Remembering the uniqueness of each individual animal, rather than the statistics.
Glucometers are routinely used to assess the response to insulin in diabetic animals and to determine routine blood glucose curves as an assessment of diabetic control:
The American Diabetes Association consensus statement recommends that the total error (analytical + user) be less than 10% of the glucose concentrations between 30 and 400 mg/dL (SI units – 1.7– 22.2 mol/L), 100% of the time. In addition, the measurements should be within 15% of the reference glucose value. In a study of 11 different meters; 19–48% of the analyses failed to achieve results within 15%, and the failure rate for accuracy (10%) was 36–68%.
In veterinary medicine, recent papers have analyzed the performance of various glucometers. The basic technology is either electrical or photometric and the statistical evaluations have been satisfactory. Over the range suggested by the ADA, the meters did not perform to specifications. Does this make a difference? If the meter reads on an individual single hyperglycemic sample either 4.0 mol/L above or below the reference range, is that important? If however, you are running a curve and can be either above or below the actual value, and if in your patients case the values were all in one direction, you would have a flat curve—no response to insulin and a hyperglycemic animal—would that make a difference to your treatment?
1. Teitz, NW; Clinical Guide to Laboratory Tests: Third edition. WB Saunders Co; 1995.
2. Teitz Fundamentals of Clinical Chemistry; 5th Edition. Editors: CA Burtis, ER Ashwood, WB Saunders Co; 2001.
3. Young, DS; Effects of drugs on clinical laboratory tests; 5th Edition. Volume 1; AACC Press, 2000.
4. Vap LM; Mitzner B. An Update on chemistry analyzers; the Veterinary Clinics of North America, small animal practice. 26:5:1996.
5. Lanevscki A; Kramer JW. Comparison of two dry chemistry analyzers and a wet chemistry analyzer using canine serum. Vet Clin Path 25:1:10, 1996
6. James KM, Polzin DJ. Effects of sample handling on total carbon dioxide concentrations in canine and feline serum and blood. AJVR 58:4:343, 1997.
7. Cohn LA, McCaw DL. Assessment of five portable blood glucose meters, a point of care analyzer and color test strips for measuring blood glucose concentration in dogs. JAVMA 216:2:198, 2000.
8. Wess G, Rausch, C. Assessment of five portable blood glucose meters for use in cats. AJVR 61:12:1587.
9. Gaynor JS, Wertz EM. Effect of intravenous administration of fluids on packed cell volume, blood pressure, and total protein and blood glucose concentrations in healthy and halothane anesthetized dogs, JAVMA 208:12:2013, 1996
WSAVA Contact Information