Misleading Biochemical Laboratorytest Results
Total Page:16
File Type:pdf, Size:1020Kb
Misleading biochemical laboratory test results will examine some of the conditions bility increases with each additional AMIN A. NANJI, MB, CH B, FRCP[C] under which commonly performed test, as shown in Table I. Interpre- laboratory tests may give misleading tation of the table is subject to the results. caveat that the probability of all This article reviews the general and A few general principles may be tests giving results within the nor- specific factors that interfere with helpful. The great progress that has mal limits is slightly increased by the performance of common bio- been made in recent years in the the fact that low values for some chemical laboratory tests and the quality of laboratory tests means tests do not indicate an abnormality. interpretation of their results. The that most errors are now due to Indeed, several authors have warned clinical status of the patient, drug nonanalytic factors;' primarily these against pursuing a marginally ab- interactions, and in-vivo and in-vitro include the steps involved in request- normal test result.4-6 biochemical interactions and changes ing the test, in preparing the patient, Other factors can also alter test may alter the results obtained from and in storing and processing the results. These include the patient's biochemical analysis of blood con- specimens. Drugs can also alter lab- sex, diet, physical activity and smok- stituents. Failure to recognize invalid oratory test results, either by affect- ing habits; these factors have been laboratory test results may lead to ing a physiologic process or by inter- discussed in detail elsewhere.78 injudicious and dangerous manage- fering with the laboratory analysis.2 Table II presents a concept out- ment of patients. All laboratories publish the line of the major sources of poten- ranges of normal values for the tially misleading laboratory test re- Cet article passe en revue les fac- results of the tests they perform. sults. In the following discussion, teurs generaux et specifiques qui Limits of normal are set to include however, it will be easier for you to genent la realisation de tests de labo- the central 95% of the population, review each laboratory test com- ratoire biochimiques courants ainsi excluding the 2.5% in each tail of pletely while referring to the general que l'interpretation de leurs resul- the normal curve. In other words, by principles set out in the table. The tats. L'etat clinique du malade, les definition the probability of an ab- three categories of headings in the interactions medicamenteuses et les normal result of any given test in a following discussion represent the interactions ou changements biochi- normal individual is 0.05; the proba- main area of testing, the specific miques in vivo et in vitro peuvent fausser les resultats des analyses biochimiques des constituants san- Table I-Predicted probability that the results of at least one test will exceed guins. Le fait de ne savoir identifier laboratory limits des resultats non valables de tests de Probability* laboratoire peut conduire a un traite- ment inapproprie et dangereux pour That results of all That results of one le malade. tests will be or more tests within limits* will exceed limits Laboratory results that do not corre- No. of tests (p) (1 - p) late with the patient's clinical status 1 0.95 0.05 or with the disease process may lead 6 0.74 0.26 to an incorrect diagnosis and injudi- 8 0.66 0.34 cious management. In this paper I 12 0.54 0.46 14 0.49 0.51 From the division of clinical chemistry, Van- 17 0.42 0.58 couver General Hospital and the department 20 0.36 0.64 of pathology, University of British Columbia, 100 0.006 0.99 Vancouver = = = Reprint requests to: Dr. Amin A. Nanji, *p sw, where p the statistical probability, s the probability (usually 95%) Director of clinical biochemistry, Ottawa that a normal person will have values within the normal range for each test, and General Hospital, 501 Smyth Rd., Ottawa, n = the number of tests performed. Ont. KIG 8L5 CAN MED ASSOC J, VOL. 130, JUNE 1, 1984 1435 component being tested and the type instrument and hyperproteinemia. monoclonal protein in the plasma of error or the condition leading to In a study of 11 patients with vari- causing a decrease of 0.7 mmol/L in error. ous degrees of hyperviscosity Blank the sodium level.'8 and Ill found that the measurements Presence of cationic paraproteins Electrolytes obtained with direct potentiometry in the blood: Bloth and coworkers" were accurate, whereas those ob- described three asymptomatic pa- Serum sodium tained with flame photometry and tients with multiple myeloma whose indirect potentiometry were artifi- serum sodium levels were reported Sampling: One of the commonest cially low. No correction factor can to be around 110 mmol/L. Cationic causes of a falsely low value for the be applied to measurements ob- paraproteins may displace other cat- concentration of sodium in the tained with flame photometry; how- ions, especially sodium. serum occurs when the venous sam- ever, for the results determined by Presence of osmotically active ple is taken downstream from an indirect potentiometry each centi- particles in the blood: Glucose is ah intravenous infusion of hypotonic stoke increase in the serum viscosity osmotically active substance that fluid, such as 5% dextrose in water. accounted for a 2-mmol/L decrease can cause water to move from the I attempted to determine the cause in the serum sodium level. Serum intracellular to the extracellular of hyponatremia in 87 patients with viscosity. is temperature-dependent.'6 fluid space and thus dilute the sodi- reported serum sodium levels of less Blank and I'` found that the differ- um ions in the extracellular fluid. In than 125 mmol/L and found that in ences in the viscosity of hyperviscous patients with diabetic ketoacidosis 10 patients the blood sample had serum at 20°C and 37°C caused or with hyperglycemic, hyperosmo- been diluted by the intravenous flu- differences in the measured sodium lar, nonketotic diabetic coma the id. This type of error can usually be levels. high concentration of plasma glu- identified by the high glucose level Monoclonal gammopathy: The cose causes a dilution of sodium ions in the sample. reduced water content in the plasma in the plasma. In this case the low Hyperlipemia: With some analy- of patients with monoclonal gam- values for the serum sodium level tic techniques the reported sodium mopathy without hyperviscosity can result from renal loss of sodium and levels in serum samples from pa- result in falsely low reported serum the shift of water into the extracel- tients with hyperlipemia may be sodium levels, each gram per litre of lular fluid space. Katz20 suggested, falsely low because of the decrease in the proportion of water in the Table II-Major sources of misleading results of biochemical laboratory tests plasma secondary to the large amount of lipid present.9'0 Ladenson Use of a specimen that is not representative as a result of one of the following: and colleagues" showed that errors Biochemical change or interaction in the measurement of serum sodi- In vitro um concentrations in hyperlipemic Hemolysis/lysis of cells, with leakage of cellular constituents (e.g., lactate patients were related to the method dehydrogenase or potassium) Storage, leading to alkalinization of serum, with subsequent increase in of measurement. If the serum sam- alkaline phosphatase activity ple is diluted, as it is when analysed Evaporation-induced errors (e.g., concentration of electrolytes) by indirect potentiometry, the re- Artefactual hypoglycemia in the presence of large numbers of cells (i.e., sulting values will be falsely low; erythrocytes, leukocytes or platelets) however, with direct potentiometry In vivo the results will be valid because the Interference by endogenous metabolites or drugs (e.g., cephalosporins or concentration of sodium ions is acetoacetate, which falsely elevate the serum creatinine level) being measured in the aqueous Cell rupture, with release of cellular constituents (e.g., platelet acid phospha- phase. Frier and associates'2 des- tase, which spuriously increases the prostatic acid phosphatase level) cribed a patient with hyperlipemia, Sampling error Contamination of specimen (e.g., by residual detergent, which spuriously hemiplegia and convulsions who was increases the phosphate level) treated with normal saline because Venous stasis, which leads to increases in the lactate, potassium and calcium the serum sodium level was reported levels to be 116 mmol/L. When the level Upright posture, which leads to hemoconcentration and an increase in the total was corrected for the hyperlipemia, serum protein level however, it became 222 mmol/L. As Blood drawn downstream from an intravenous line, which leads to spurious a rule, every increase of 11.3 hyponatremia and hyperglycemia mmol/L in the serum triglyceride The patient's not being in the basal state level results in an apparent decrease After exercise, which increases the serum creatine kinase level of 1.5 mmol/L in the serum sodium During diet, which increases the serum triglyceride levels After ethanol consumption, which increases the y-glutamyl transpeptidase level level.'" Laboratory error in one of the following areas: Hyperviscosity: Serum samples Analysis from patients with hyperviscosity Interference by bilirubin, turbidity or chemicals yield falsely low values for the sodi- Poor quality control um level when analysed by flame Reporting of results with identification error photometry'4 because of problems in Normal variation sample aspiration, dilution by the 1436 CAN MED ASSOC J, VOL. 130, JUNE 1, 1984 and I have verified clinically,2' that as a cause of reported hyperkalemia decrease by an average of 4 for every 5.6-mmol/L increase in the in patients with myeloproliferative mmol/L.