Evaluation of the YSI 2300 glucose analyzer: algorithm-corrected results are accurate and specific.

OBJECTIVES: To insure commutability (equivalence) between whole blood and plasma glucose values and to assess potential interferences in glucose oxidase methods. DESIGN AND METHODS: We compared plasma glucose results by the Ektachem 700 analyzer to glucose results on whole blood by the Nova Stat Profile 6, and by the YSI 2300, both without hematocrit correction (YSIunc) and with two different hematocrit (Hct) corrections. The two correction methods were: (a) whole blood results multiplied by 1.11 (YSI1.11), and (b) whole blood results corrected for Hct with the programmed YSI algorithm (YSIcor). Several compounds were tested for interference. RESULTS: 88% of YSIcor results agreed with plasma results within plus/minus 5%, and 78% of Nova and 74% of YSI1.11 results were within this limit. YSIcor results were unaffected by Hct and were valid even when erythrocyte size was abnormal. At low Hct, Nova results were falsely decreased, while YSI1.11 results were falsely elevated. The Ektachem was most affected by ascorbic acid, and the Nova was especially affected by acetaminophen. CONCLUSIONS: The YSI algorithm-corrected whole blood glucose results were commutable with plasma results and the YSI was unaffected by the compounds tested, including acetaminophen and acetylsalicylic acid.

Estimating the linear analytic range.

Although many methods have been proposed to verify assay linearity, or to test for nonlinearity, there are few, if any, statistically sound methods to establish an assay's linear range. In this article, we propose a simple and statistically sound method for initially establishing an assay's linear range as a paradigm for standardization of manufacturers' claims of assay linearity. Simulations verify that the method outperforms using (Pearson's) coefficient of determination (r2) to estimate the linear range, consistently yielding estimates for the linear range which are more accurate. In addition, the method permits the addition of tolerance limits when a certain amount of nonlinearity is acceptable clinically.

Viral load test reports: a description of content from a sample of US laboratories.

CONTEXT: Human immunodeficiency virus (HIV) RNA testing (viral load testing) is increasingly important in the care of patients infected with HIV-1 to determine when to initiate, monitor, and change antiretroviral therapy. Patient viral load testing information is communicated to the clinician through the laboratory test report. OBJECTIVES: To examine the format and information used in reporting viral load testing results and determine the clarity of the information provided in these reports. DESIGN: Patient test reports with all personal identifiers removed were requested of viral load testing laboratories participating in a telephone survey of laboratory practices. Hospital, independent, health department, and "other type" laboratories identified as university-associated laboratories participated in the telephone survey. RESULTS: Thirty-seven unique test reports were collected. All laboratories reported results in copies/mL, while 14% also reported results as "log(10) copies/mL." The test kit was identified by only 24% of the laboratories. Reportable ranges were specified by 70% of the laboratories, but there was considerable variation in terminology. One laboratory reported a viral load copy number below the manufacturer's test kit lower limit of sensitivity. The layout and format differed among reports. Some results were expressed in log(10), others contained nonsignificant integers, while others contained exponential numbers. Supplemental information in some reports included previous patient test results and significance of changes from baseline. The format of some reports made it difficult to read the report information and interpret the testing results. CONCLUSION: This study emphasizes the importance of standardizing the reporting of HIV-1 viral load test results to minimize result misinterpretation and incorrect treatment.

Artifactual hypoglycemia associated with hematopoietic cytokines.

We observed apparent hypoglycemia in seven patients manifesting granulocytosis associated with hematopoietic cytokine treatment or with a leukemoid reaction. All seven patients had confounding preanalytic conditions of specimen transport delay and lack of antiglycolytic agents. Cytokine-stimulated leukocytes may cause artifactual hypoglycemia by consuming glucose in vitro, possibly leading to unnecessary diagnostic evaluation. The glucose depletion was faster in blood drawn from patients receiving granulocyte colony-stimulating factor (0.29 mmol/L/h) than in blood from a control group (0.17 mmol/L/h) or from a group with leukemia (0.23 mmol/L/h). Stabilization with sodium fluoride (60 mmol/L) slowed the glucose depletion in both the cytokine group (0.13 mmol/L/h) and the leukemic group (0.09 mmol/L/h), which were then statistically indistinguishable from the control rate (0.10 mmol/L/h). In blood obtained from patients being treated with hematopoietic cytokines or who have leukemoid reactions, an antiglycolytic agent should be used whenever separation of plasma might be delayed more than 1 hour.

Pneumatic transport exacerbates interference of room air contamination in blood gas samples.

OBJECTIVE: To characterize and control the potential interference to P(O2) determinations when blood contaminated with air is sent via a pneumatic tube system (PTS). DESIGN: Both tonometered blood at P(O2)s of 65, 75, 142, and 339 mm Hg and arterial blood gas samples from patients with baseline P(O2)s from 70 to 400 mm Hg were analyzed for P(O2) to determine possible effects of air contamination from PTS transport. SETTING: A large teaching hospital in which a variety of personnel routinely send samples to the laboratory by PTS transport. PATIENTS: Twenty patients under anesthesia for elective surgery and 21 patients in an intensive care unit who had a wide range of P(O2)s. Several additional patients with a preexisting lung pathology likely to cause hypoxemia were selected to provide samples with low P(O2)s. MAIN OUTCOME MEASURES: Measurement of bias in P(O2) between samples sent via PTS and samples walked to the laboratory. RESULTS: Interference from air contamination was worse after PTS transport compared with manual transport of the specimen. Over a wide range, the P(O2) in specimens after PTS transport tended toward 160 mm Hg. Samples from hypoxemic patients were prone to errors in P(O2) that could have resulted in clinical misinterpretation; 5 of 10 samples with a baseline P(O2) less than 85 mm Hg had increases of 10 mm Hg or more when contaminated with air. Cooling samples with high P(O2)s minimized changes to P(O2), probably by increasing the solubility of oxygen. Mechanical buffering by various liners used in the carriers did little to alleviate the interference. Decreasing the speed of pneumatic transport by 50% lessened the effect on P(O2). CONCLUSION: Interference can be minimized by carefully purging samples of all air bubbles using the following protocol: invert syringe to check for air bubbles, then retap and reexpel bubbles if necessary. Personnel that collect and send blood gas samples via PTS should be educated about the problem of interference. Modifications both to pneumatic sample transport systems and to blood gas syringes should be investigated to minimize the effect.

Impact of quality control on accuracy in enzyme immunoassay testing for human immunodeficiency virus type 1 antibodies.

OBJECTIVE: To assess use of quality control (QC) material, supplemental to internal kit controls (calibrators), as protection against errors in enzyme immunoassay testing for human immunodeficiency virus type 1 antibodies. DESIGN: From August 1994 to January 1996, enzyme immunoassay testing accuracy was assessed for laboratories participating in the Centers for Disease Control and Prevention Model Performance Evaluation Program that provided information regarding their use of QC material. Error rates were examined for human immunodeficiency virus type 1 antibody-negative, strongly positive, and weakly positive samples. RESULTS: The overall error rate with QC (2.20%) was significantly (P = .0023) lower than the error rate without QC (2.90%). With QC use there was a significant reduction in the relative risk of error for negative (P = .014) and weakly positive (P = .0067) samples. After multivariate analysis, use of QC lowered overall error rate by 29% (P = .0009). Laboratories not using QC were at increased risk of systematic error. Following the Clinical Laboratory Improvement Amendments of 1988 guidelines for QC material was relatively more protective against error than lower frequencies/number of levels. CONCLUSIONS: Using QC protected against errors in enzyme immunoassay testing for human immunodeficiency virus type 1 antibodies. Two levels of QC should be used with each run as mandated by the Clinical Laboratory Improvement Amendments of 1988.

Myocardial ischemia correlates with reduced fibrinolytic activity following peripheral vascular surgery.

STUDY OBJECTIVES: To evaluate the relationship between perioperative ischemia and serial concentrations of D-dimer, which is a sensitive and specific marker of fibrinolytic activity. Myocardial ischemia and infarction are well-recognized complications of peripheral vascular surgery. We hypothesized that patients at increased risk of perioperative myocardial ischemia might be identified preoperatively by abnormal hemostatic indices. DESIGN: Prospective clinical outcomes study. SETTING: A 1,124-bed tertiary care medical center. PATIENTS: 42 ASA physical status II, III, and IV patients undergoing peripheral vascular surgery. INTERVENTIONS: Serial D-dimer concentrations were measured preoperatively, and at 24 and 72 hours postoperatively. Continuous 12-lead ST-segment monitoring (Mortara Instrument, Inc., Milwaukee, WI) was performed with the acquisition of a 12-lead ECG every 20 seconds for 72 hours. MEASUREMENTS AND MAIN RESULTS: D-dimer measurements were performed in duplicate using the Dimer Gold assay (American Diagnostica, Greenwich CT). Ischemic episodes, as defined by continuous 12-lead ST-segment monitoring, occurred in 49% of patients. There were no demographic differences between ischemic and nonischemic groups. Although baseline D-dimer concentrations were not statistically significantly different between groups, patients experiencing perioperative myocardial ischemia generated significantly less D-dimer during the perioperative period (p = 0. 014). CONCLUSIONS: PATIENTS with an impaired fibrinolytic response, as defined by reduced generation of D-dimer, experienced an increased incidence of perioperative myocardial ischemia.

Measurement of free phenytoin in blood with a self-contained fiber-optic immunosensor.

We describe the measurement of the anticonvulsant drug phenytoin (5,5-diphenylhydantoin) in human blood and plasma using a self-contained fiber-optic immunosensor which gave a reversible response to changes in concentration. No regenerative treatment of the immunosensor was required between measurements. The analytical signal depended on the degree of energy transfer from B-phycoerythrin labeled with phenytoin to Texas Red labeled anti-phenytoin antibody. Dextran 70K was added to the reagent system to equalize the oncotic pressure across the encapsulation membrane. A gas chromatography reference method was used to measure free drug concentrations. Regression analysis for plasma samples gave the relationship (sensor) = 1.02 (reference) + 0.07 microM; Syx = 1.00 microM; r = 0.953; sensor mean 4.45 microM; reference mean 4.31 microM. Sensor performance in plasma and whole blood was essentially equivalent. We demonstrate the feasibility of measuring free phenytoin directly in blood and suggest that the sensor design is generally applicable for the measurement of other haptens in blood.

Quantifying the bias associated with use of discrepant analysis.

Discrepant analysis is a widely used technique for estimating the performance parameters of a laboratory test. In discrepant analysis, each specimen is initially tested with the candidate test and a comparison method, and when the results of the two tests disagree, a confirmatory test is used to resolve the discrepancy. Discrepant analysis usually produces biased estimates. This report quantifies this bias and shows that it is usually positive, leading to overestimation of the performance parameters of a laboratory test. The direction and magnitude of the bias are predictably influenced by the analytical sensitivity and specificity of the candidate test, comparison method, and confirmatory test. The proportion of abnormal specimens tested also affects the magnitude of the bias, particularly the estimates of analytical sensitivity and positive predictive value when this proportion is low. Alternative approaches are suggested.

Lipase activity measured in serum by a continuous-monitoring pH-Stat technique--an update.

Using recent knowledge regarding the roles of colipase, bile acids, Ca2+, and emulsifiers, we optimized a previously published pH-Stat method for lipase (EC 3.1.1.3) activity measurements. The recommended assay conditions are: olive oil/triolein, 100 mL/L; sodium glycocholate, 35 mmol/L; Ca2+, 8.5 mmol/L; and colipase, 6.0 mg/L. The sample volume is 0.10 mL, the reaction pH 9.0, the temperature 30 degrees C, and the concentration of titrant 15 mmol/L. Hydroxypropyl methylcellulose, 20 g/L, replaces acacia as emulsifier to avoid inhibition by excess Ca2+. The standard curve is linear to greater than 4566 U/L. The reference interval with olive oil as substrate is 30-235 U/L. Lipase activities with triolein substrate are 9.9% greater than with olive oil. Interference by pancreatic carboxylesterase (EC 3.1.1.1) activity is inhibited by incubating the sample with diisopropylfluorophosphate. Results correlate well with those by the optimized SingleVial method of Boehringer Mannheim Diagnostics (r = 0.997) and the immunochemical assay of Beckman Instruments, Inc. (r = 0.995). Correlation with the aca method (E.I. DuPont de Nemours & Company) is less satisfactory (r = 0.892), probably owing to lack of colipase in the latter method.

Turbidimetric measurement of lipase activity--problems and some solutions.

We optimized a commercial turbidimetric method for lipase (EC 3.1.1.3) activity (Boehringer Mannheim Diagnostics) and overcame some of its deficiencies. Increasing the bile salt concentration to 35 mmol/L and the colipase concentration to 6 mg/L and using a continuous recording of the reaction-rate curve greatly improved the reaction kinetics, eliminated false results from increases in absorbance, reduced the lag phase, and increased the analytical sensitivity and accuracy. Differentiation of pancreatitis from nonpancreatitis sera by adding NaCl, 140 mmol/L, to the assay mixture to observe the degree of enzyme activation has important limitations. Sera from patients with pancreatitis and only slight or modest increases in lipase behave like sera from healthy individuals or from patients with nonpancreatic disease. The assay shows no interference by lipoprotein lipase and carboxyl esterase. Results compare well by this optimized method and by an optimized "pH-Stat" titrimetric method.