Non-disease variables

Many factors other than disease influence the results of chemistry tests. These factors may be preanalytical, analytical and post-analytical. Extra-analytic errors is applied to errors in which the cause is not identified or due to more than one factor (Whipple et al 2020).

Preanalytical variables are variables associated with the patient, sample collection and sample handling. These generally affect the composition of the body fluid before analysis. Preanalytical variables account for 50-75% of the errors with laboratory testing (Whipple et al 2020) and include:
- Biological variables: Those inherent to the patient (e.g. age, breed) and those that can be controlled (avoided, e.g. post-prandial lipemia).
- Non-biological variables: Sample collection, handling and storage, including anticoagulants. These are typically controllable or avoidable. For more information on this, refer to the hematology, chemistry, hemostasis, urinalysis and cytology sample collection pages.
Analytical variables are factors which influence the analytical procedure.
- Interferences: Lipemia, hemolysis, icterus, hyperproteinemia, drugs. Also see, interference indexes measured on our chemistry results.
- Quality assurance: Ensures accurate testing performance and results
Post-analytical variables involve the different ways data from the laboratory is presented, stored and transferred to the clinician.

Whenever possible, these variables should be controlled in order to minimize their effect on test outcome. For reviews on this topic, see Braun et al 2015 and Camus 2016.

Preanalytical variables

Preanalytical variables affect the composition of the sample (and hence sample quality) before the sample is analyzed in the laboratory. These variables are the most important and common sources of error in laboratory analysis and are often overlooked when interpreting test data. These variables should be controlled, wherever possible, to minimize their effect on test results. Control of preanalytical variability involves proper sample collection, preparation and handling by the veterinarian or technician. When such factors cannot be controlled, e.g inherent biologic differences in species, they must be recognized and considered when evaluating results for disease.

Preanalytical variables include biological (patient) and non-biological (sample) factors.

Biological variables

Biological variables are associated with the patient. There are factors that are inherent to the patient, such as breed, age and sex, which cannot be controlled, but must be considered when evaluating test results. Remember, that most reference intervals provided by laboratories apply to healthy adult animals. Other biological variables involve those factors which can be controlled by you, the veterinarian, when drawing the blood sample, such as ensuring the animal is fasted for 12 hours before sample collection. These can be inherent or controlled.

Inherent biological variables

These variables are inherent to the patient and cannot be controlled. They must be considered when interpreting test results.

Species: There are species-specific differences in the source of analytes and the significance of analyte changes , e.g. alanine aminotransferase is a liver specific enzyme in small but not large animal species. Therefore, it is essential to compare chemistry results of an animal to a reference interval determined for the same species, e.g. chemisty results from a donkey cannot be interpreted using reference intervals for a horse.
Breed: There are breed-related differences in chemistry analytes, but these are usually minor, e.g. healthy Draught horses typically have higher creatine kinase values than Thoroughbreds.
Age: Many chemistry analytes change with age, i.e. young animals often have alterations in chemisty analytes that are not due to underlying disease, but are an expected age-related finding, e.g. elevated alkaline phosphatase in all species. In general, most reference intervals provided by clinical pathology laboratories are determined from adult animals and cannot be applied to young animals.
Gender: Physiological states associated with pregnancy or lactation in female animals can affect analyte results independently of disease. In some species, there may be separate reference intervals for the different sexes, e.g. serum iron values in humans. In general, there are few gender-related differences in chemistry results for domestic animals.

Controllable biological variables

Standardization of collection techniques can minimize the effect of these variables on chemistry results. If these variables are unavoidable or desirable (e.g. sample collection post-exercise to detect rhabdomyolysis in horses), they must be considered during test interpretation.

Recent food ingestion: This will produce a post-prandial lipemia.Lipemia may affect the analytical methods used to measure certain plasma constituents, thus producing invalid results. Post-prandial lipemia can be avoided by appropriate sample collection, namely after a 12-hour fast. Indeed, fasting lipemia is good evidence for altered lipid metabolism secondary to disease.What are some examples of diseases that produce a fasting lipemia?
Recent food ingestion may affect the concentration of certain plasma constituents independently of lipemia, e.g. glucose. This effect is distinctly species dependent due to differences in food metabolism (e.g. non-ruminant versus ruminant).
Stress: Stress (secondary to animal handling or underlying disease) may have profound effects on laboratory results, due to endogenous corticosteroid and/or adrenaline release. Handling stress is more prominent in young animals or un-acclimatized animals. Every effort should be taken to minimize stress during blood sample collection.
Exercise: The effect of exercise on plasma constituents is dependent on both the species and the intensity of the exercise. In general, blood samples should be collected from animals prior to exercise.
Drugs: The technique of drug administration, e.g. intramuscular, can directly affect analyte results. Which test results may be affected by intramuscular drug administration (hint: think muscle-related tests)? Drugs can also interfere with measurement of the analyte. Drug interference can be grouped into two general categories:
- Physiological (in vivo) effects of the drug or metabolites on the analyte to be measured
- In vitro effects due to some physical or chemical property of the drug or its metabolites, which interfere with the actual assay procedure. An example of a therapeutic agent that produce marked assay interference were hemoglobin-based oxygen carriers (HBOC), such as Oxyglobin. Although this product is no longer available, oxyglobin was polymerized hemoglobin of bovine origin and was, consequently, dark red.
  Whenever possible, the laboratory should be informed of any drug medication so steps can be taken to minimize the effect of the drug on assay performance. Unfortunately, little is known in veterinary medicine about the effect of drugs on both the animal or analytical techniques, with some notable exceptions (e.g. corticosteroids, bromide). Therefore, any abnormal chemistry results should be interpreted with full knowledge of the animal’s medication history.

Non-biological variables

Non-biological variables involve sample collection, handling and transport to the laboratory. All of these variables can be controlled to minimize the effect they have on laboratory results.

Sample collection

Venipuncture: Clean venipuncture is essential to minimize artifactual changes in the results. Poor venipuncture technique may cause hemolysis, which alters chemistry results in a variety of ways (falsely increase potassium in species with high potassium red blood cells, falsely increase AST in species with AST in red blood cells, e.g. cats, falsely increase CK due to release of constituents which participate in the chemical reaction for measurement). Difficult venipuncture during hemostasis testing could contaminate the sample with tissue factor, precipitating clotting in the sample collection tube, which could prolong or shorten coagulation screening assay times in unpredictable ways. This can be minimized by collecting blood directly into citrate anticoagulant (pre-filled syringe or directly into a vacutainer).
Anticoagulant: The anticoagulant can affect results of testing.
- Hematology testing: EDTA anticoagulant is preferred. Citrate will dilute the sample, decreasing counts and heparin causes clumping of cells, decreasing counts.
- Chemistry testing: The preferred samples for chemistry tests are heparinized plasma or serum (no anticoagulant). The same sample should be used if monitoring a patient sequentially because results are different between plasma and serum in the same patient. Plasma contains fibrinogen, which is lacking in serum (removed during clotting), so total protein (and globulin, since fibrinogen is a β globulin) results are higher in plasma versus serum. Potassium is released from platelets during clotting so serum potassium is higher than plasma potassium (the latter result is more accurate for potassium measurement).
  - Fluoride-oxalate anticoagulant can be used for glucose measurements, because it inhibits cell metabolism, however prompt separation of serum/plasma from cells is essential as this anticoagulant induces severe hemolysis.
  - Citrate and EDTA anticoagulants should not be used for chemistry tests, because they chelate divalent cations (calcium, magnesium) and contain high concentrations of sodium (citrate) or potassium (EDTA). Very high potassium and very low calcium (both incompatible with life) results are seen with even mild EDTA contamination of blood samples for chemistry testing. Divalent cations are also needed for activity of some enzymes. Therefore falsely low results may be seen for these enzymes in EDTA- or citrate-anticoagulated plasma (e.g. alkaline phosphatase).
  - The following tests cannot be run on EDTA plasma or serum/heparinized plasma contaminated with EDTA: Calcium, magnesium, iron, TIBC, % saturation, amylase, lipase, alkaline phosphatase, bicarbonate, bile acids, GLDH, electrolytes (sodium, potassium, chloride), SDH, direct bilirubin.
- Coagulation testing: Citrate anticoagulant is optimal (it gently chelates calcium so calcium can be added back to run coagulation screening assays). Heparin cannot be used (samples will not clot) and EDTA can be used (calcium can be added back into the sample), however the latter is not reliable and not recommended.

Sample handling and storage

After sample collection, strict attention should be paid to sample handling to minimize the effects of these variables on results.

Separation: Whenever possible, especially when there is to be a delay between sample collection and submission to a laboratory, serum or plasma should be separated from cells as soon as possible after sample collection. This involves allowing the sample to clot (for serum) and centrifuging the sample to separate out the blood components. The serum or plasma should be separated from cells and placed into a new clot (no anticoagulant) tube. The tube should be labeled as “plasma” or “serum”. Samples that are not separated from cells have lower concentrations of glucose (consumed by cells during storage) and can have higher concentrations of some constituents that can leak out of red blood cells, even if there is not obvious hemolysis, e.g. potassium in horse blood.
There are special tubes available, called serum separator (or Corvac) tubes. These contain a silicon gel, that allows separation of serum from the clot . However, if these tubes are used, the serum should be still withdrawn and placed into a new clot tube, particularly if the tube is old (and the silicon cracks).
Labeling: All body fluid samples taken from a patient should be correctly labeled with the patient name or identification and the type of specimen (e.g. serum, plasma, synovial fluid, peritoneal fluid). The laboratory is not responsible for correct labeling of the sample, you are. As you can imagine, an incorrectly labeled specimen can have serious and deleterious consequences for patient care.
Storage: There is usually at least a 24-hour delay between sample collection and delivery to the laboratory under the latter circumstances. Therefore, prompt separation of serum/plasma from cells and keeping the sample cool (with an icepack) is essential to maximize enzyme stability and prevent false changes in the results. If a long (> 24 hour) delay is anticipated, it is advisable to freeze the sample and submit the sample to the laboratory frozen, on dry ice. However, always call the laboratory, because the method of storage depends on what you are testing – some samples are not so stable frozen.

Analytical variables

Analytical variables affect the procedure by which the analyte is measured by the instrument. Naturally, the impact on a chemistry test result will differ between laboratories, depending on the type of instrumentation. Analytical variables are caused by features inherent to the sample, e.g. interferences such as lipemia, or by features inherent to the analyzer. The latter are minimized by using quality assurance procedures within the laboratory.

Interferences

The degree to which endogenous substances, such as lipemia, free hemoglobin (hemolysis), icterus (high bilirubin), high globulins and drugs interfere with hematology and chemistry analyzers depends on the type of analyzer, the methods used to detect the analytes and the amount of interfering substances in the sample (i.e. laboratory-specific). The chemistry analyzer used in the Clinical Pathology Laboratory at Cornell University measures the amount of lipemia, hemolysis and icterus in a sample and provides these values as indices. These indices can be used to gauge the effect of these three endogenous interfering substances on the chemistry results.

Lipemia, hemolysis, icterus: For more information on lipemia, hemolysis and icterus, refer to the interference page of this section.
Hyperproteinemia: Monoclonal immunoglobulins (paraproteins), particularly when present in high concentration, interfere with chemistry analysis by the following mechanisms:
- Hyperviscosity: This affects sample volume and is dependent on the class of immunoglobulin (which immunoglobulins produce hyperviscosity and why?).
- Binding to analytes: Immunoglobulin binding to some analytes producing increased or decreased analyte values, e.g. hyperphosphatemia has been reported in a dog with chronic lymphocytic leukemia and an IgM monoclonal gammopathy.
- Volume displacement: This has effects similar to lipemia. What diseases produce a monoclonal gammopathy?
Drugs: There are a number of different methodology-related drug interferences that may bias results. One of the most common drug interferences seen in veterinary medicine is the artifactual elevation of chloride values in samples from dogs on bromide therapy (an anticonvulsant). It is always wise to inform the laboratory of any medications in a particular animal, so that drug influences on chemistry tests can be minimized (but usually not eliminated).

Assay performance

Controlling and minimizing factors which affect assay performance involves quality assurance.
The goal of the clinical pathology laboratory is to provide accurate and precise results. Quality assurance is achieved by using clearly defined laboratory protocols and a quality control program. Quality control procedures in the clinical chemistry laboratory depend on measuring the concentration of analytes in control sera (with predetermined analyte values). Control values that fall outside predetermined (acceptable) limits usually indicate a procedural problem. In the Clinical Pathology Laboratory at Cornell University, controls are run daily on our chemistry analyzer and we participate in external quality assurance programs. Analysis is not performed on patient samples unless the control values are within acceptable ranges. In addition, the laboratory participates in external quality control programs, in which unknown samples are analyzed and the results compared to those of other laboratories that use the same instrumentation. Major deviations of our laboratory’s results from the mean indicates a procedural problem, that can be investigated and corrected.