Low sensitivity of anion gap to detect clinically significant lactic acidosis in the emergency department.

INTRODUCTION: Lactic acidosis represents the pathologic accumulation of lactate and hydrogen ions. It is important to efficiently diagnose lactic acidosis as delayed treatment will lead to poor patient outcomes. As plasma lactate levels may not be rapidly available, some physicians may use elevated anion gaps to test for the need to measure lactate. All Edmonton metropolitan hospitals have Radiometer blood gas/electrolyte instruments in the ED or close by. As lactate is measured for each set of electrolytes, we were able to determine the effectiveness of a screening anion gap for lactic acidosis. METHODS: Two years of emergency department lactates and electrolytes from Edmonton's 5 metropolitan hospitals were analyzed. We determined the sensitivity, specificity and positive predictive value of detecting an elevated lactate, defined as ≥2.5mmol/L or ≥4mmol/L. RESULTS: Depending on the elevated anion gap cut-off and the definition of elevated lactate, between 40-80% of elevated lactates are missed. In general, the positive predictive value approaches 40% for AGs ≥12mmol/L and 60% for AGs ≥16mmol/L. CONCLUSIONS: Anion gap is an inadequate marker of lactic acidosis. We recommend that lactate be done with each set of electrolytes and/or blood gases. In this way lactic acidosis will not be missed.

Diagnostic efficacy of six plasma proteins in evaluating consumptive coagulopathies. Use of receiver operating characteristic curves to compare antithrombin III, plasminogen, alpha 2-plasmin inhibitor, fibronectin, prothrombin, and protein C.

Six coagulation proteins were measured in 79 consecutive patients referred to the coagulation service for suspected disseminated intravascular coagulation. Antithrombin III, plasminogen, and alpha 2-plasmin inhibitor were measured with fluorescent substrate assays. Fibronectin, prothrombin, and protein C were measured with electroimmunoassays. Using history and physical findings and the results of a coagulation screen (prothrombin time, partial thromboplastin time, fibrinogen, fibrin[ogen] degradation products, platelet count, and peripheral smear), the 79 patients were classified into five categories: no disseminated intravascular coagulation (n = 21), elevated fibrin(ogen) degradation products without other evidence of coagulopathy (n = 44), defibrination syndrome (n = 9), microangiopathic thrombocytopenic purpura (n = 4), and primary fibrinolysis (n = 1). Because the sensitivity and specificity of each of the proteins could not easily be compared, receiver operating characteristic (ROC) curves and areas under the ROC curves were calculated for each of the six proteins as well as for the tests of the coagulation screen. The ROC curves indicated that, apart from plasminogen, the other coagulation proteins provided little additional information about the classification of the coagulopathy.

Prototype quantitative assay for fibrinogen/fibrin degradation products. Clinical evaluation.

A new quantitative assay for fibrinogen/fibrin degradation products (FDPs) was clinically evaluated in 123 tertiary-care patients for whom the standard semiquantitative FDP assay had previously been ordered. On the basis of a comprehensive chart review, 24 patients were categorized as having disseminated intravascular coagulation (DIC), 84 were considered not to have had DIC, ten had fibrinolysis (nine of ten streptokinase induced), and five had a complicated coagulopathy whose exact nature could not be determined. The quantitative and semiquantitative FDP values were significantly correlated. However, the FDP level indicative of DIC was lower by the quantitative assay than by the semiquantitative assay, approximately 18 mg/L vs 40 mg/L, respectively. The advantages of the quantitative over the semiquantitative assay included improved precision and ability to closely monitor changes in the severity of the coagulopathy.

Evaluation of capillary collection system for HbA1c specimens.

OBJECTIVE: We evaluated the accuracy and stability of a capillary HbA1c collection system for use with a high-performance liquid chromatography analyzer. RESEARCH DESIGN AND METHODS: The collection system requires that 5 ul blood is drawn into a calibrated capillary tube, which is then placed into a vial of stabilizing solution and sent for analysis. The study was conducted on simultaneously drawn capillary and venous blood specimens from 47 pediatric diabetes patients. Accuracy was determined by comparing the capillary to the venous HbA1c values. Stability was measured by analyzing 17 capillary specimens over 3 wk. RESULTS: There was excellent agreement between the capillary and venous HbA1c values (capillary 0.959, venous +0.494, R2 = 98.7%). The capillary HbA1c values were 0.2% higher than the venous HbA1c values and decreased gradually over time (0.1% HbA1c/week) when stored at room temperature. CONCLUSIONS: The Bio-Rad (Richmond, CA) collection system is accurate, stable, and simple to use.

Need for improved instrument and kit evaluations.

A review of method comparison studies published in the American Journal of Clinical Pathology indicates that hematology evaluations are less rigorous than their chemistry counterparts and rely heavily on the correlation coefficient. While clinical chemistry evaluations depend more on linear regression, they tend to omit relevant, lesser-known statistics, such as the standard error of the estimate. The authors reiterate guidelines for the collection and statistical analysis of method comparison data and recommend that both hematology and chemistry evaluations be improved.

Quality control of multichannel hematology analyzers: evaluation of Bull's algorithm.

Bull's algorithm has been evaluated by computer simulation studies. Varying amounts of systematic analytic error were simulated in either hemoglobin (Hgb), red blood cell count (RBC), or mean corpuscular volume (MCV) with the resulting red blood cell indices averaged in batches of 20 using Bull's algorithm. The number of average indices outside the limits of 0.97 means and 1.03 means (means = stable patient mean index) was tabulated and plotted against the size of the systematic shift, expressed in multiples of the long-term analytic standard deviation (SD). The resulting plots, called power functions, show that Bull's algorithm can detect large shifts effectively and that its power increases with increasing batch number. Shifts less than 2 SD rarely are detected. The minimum error that is detected 50% of the time after nine consecutive batches is shown below: (Formula: see text) The simulation of populations with outlying indices, e.g., neonates and oncology patients, resulted in both decreased and increased power, depending on the proportion of outliers averaged, the index averaged, and the direction of the shift.

Performance characteristics of Bull's multirule algorithm for the quality control of multichannel hematology analyzers.

The multirule quality control procedure for interpreting Bull's moving averages, proposed by Levy and colleagues (Am J Clin Pathol 1986; 85: 719-721.), has been evaluated by computer simulation with the use of the approach of Cembrowski and Westgard (Am J Clin Pathol 1985; 83: 337-345.). With this procedure, a batch of patient specimens is rejected if either of two criteria are satisfied: (1) the Bull's mean of one of the red blood cell indices is outside its 3% limits, or (2) the average of three consecutive Bull's means is outside its 2% limits. Power function curves were used to summarize the performance of the multirule approach and demonstrated error-detection capabilities that are superior to the more common implementation of Bull's algorithm using 3% limits for single Bull's means. The increased error detection achieved by the multirule procedure allows shifts in hemoglobin and mean corpuscular volume to be more readily detected but also results in the detection of small shifts in red blood cell count. A modified multirule procedure was also tested and was found to be ineffective. The authors recommend the multirule of Levy and colleagues but caution that its use may result in the detection of small shifts in the red blood cell count.

Patient and physician analytic goals for self-monitoring blood glucose instruments.

The study's objective was to determine the maximum analytical error that is allowable in portable whole blood glucose meters. Interviews were conducted to derive personal reference values and significant deviations from these values for the limit of hypoglycemia, the limit of hyperglycemia, and the upper and lower limits of acceptable blood glucose for physicians and patients with diabetes at the Park Nicollet Medical Center, Minneapolis, Minnesota. Fifty patients with diabetes (30 type I and 20 type II), and 43 physicians (14 endocrinologists, 14 family practitioners, and 15 general internists) were enrolled in the study. The results showed no significant differences between type I and type II diabetic patient responses. Nor were there significant differences among family practitioner, internist, and endocrinologist responses for any of the parameters (the limit of hypoglycemia, the limit of hyperglycemia, the upper and lower limits of acceptable blood glucose for the patient, and the corresponding allowable coefficients of variation at each of these glucose levels). There were significant differences when patients were compared to physicians. Physicians require the highest degree of precision at the limit of hyperglycemia (8.4 +/- 0.28 mmol/L [150.8 +/- 5.1 mg/dL]) with a maximum allowable coefficient of variation (CV) of 7%, a CV significantly lower than that of the patients (CV = 10%). Patients require the highest precision for glucose concentration around the lower acceptable limit (4.7 +/- .013 mmol/L [84.1 +/- 2.5 mg/dL]), with an allowable CV of 8%, a CV significantly lower than that of the physicians (CV = 14%). The authors conclude that the accuracy required by patients and physicians at normal and higher glucose concentrations is achievable by currently available meters. Manufacturers should ascertain that glucose measurements are optimally accurate at glucose levels of 4.7 mmol/L (84.1 mg/dL) and have CVs no higher than 7%.

The use of CIE for the detection of Clostridium difficile toxin in stool filtrates: laboratory and clinical correlation.

Two CIE procedures (CIE-1, CIE-2) for the detection of Clostridium difficile in diarrheal stools were evaluated by comparison to cytotoxin assay and culture results and by comparison to a clinical likelihood of C. difficile-induced diarrhea. Using a combination of toxin assay and culture results for reference, the CIE-1 and CIE-2 procedures had sensitivities of 33% and 47%, specificities of 89% and 91%, and positive predictive values of 42% and 54%, respectively. Using clinical likelihood for reference, the best results were obtained by the CIE-2 method, which yielded a positive predictive value of 77%. Neither CIE procedure provided an acceptable sensitivity for the detection of C. difficile in stools.

Computer-assisted differential diagnosis of laboratory abnormalities and follow-up testing. Evaluation of the accuracy of a computer program.

An evaluation is made of a computer program which generates a differential diagnostic list given a set of input data obtained from an admission chemistry screening profile. The program is tested by supplying input data on patients for whom diseases are known. The laboratory data from 367 patients are examined. Accurate computer diagnosis is obtained in many disease categories. The original computer program is modified to suggest the follow-up tests indicated based on the diagnoses it makes. By using this program and some additional clinical input from the physician, the clinical pathologist can select the most appropriate computer diagnosis and begin the follow-up testing. Accelerated diagnosis and patient care should result.

Use of anion gap for the quality control of electrolyte analyzers.

A simple model for the simulation of patient Na, CO2, Cl, and anion gap was formulated from patient electrolyte data. Analytical error, either random or systematic, was incorporated into the simulation of the electrolyte data and allowed study of the response of anion gap to error. Power functions, plots of probability of error detection vs. size of analytical error, were constructed and indicated a low probability of error detection when single patient specimens with abnormal anion gaps were reanalyzed. These power functions showed that pooling of the anion gap data by averaging consecutive anion gaps resulted in a high probability for detecting systematic error. We recommend, as a useful quality control procedure, averaging at least eight consecutive anion gaps and testing for a significant difference between the average and the established mean gap.

Assessment of "Average of Normals" quality control procedures and guidelines for implementation.

The capabilities of "Average of Normals" control procedures have been assessed by determining power functions, graphs of the probability of error detection versus the size of analytic error. The power functions indicate that the most important determinants of statistical power are the ratio of the standard deviation of the patient population (Sp) to the analytic standard deviation (Sa), Sp/Sa; the number of data points averaged (N); the control limits (probability for false rejection); truncation limits for selecting the population; and the magnitude of the population lying outside the truncation limits. General guidelines for the implementation of "Average of Normals" are provided, along with a nomogram for the selection of N as a function of Sp/Sa and the probability of error detection. Optimal performance of these procedures may require simulation studies on a per analyte basis.

Statistical analysis of method comparison data. Testing normality.

A Lilliefors test of normality has been applied to data from precision and accuracy studies. Most data sets tested as non-normal. Simulation studies showed that the test is extremely sensitive to the rounded, narrowly distributed data that are typical of method performance studies in clinical chemistry. The Lilliefors test can be modified to be applicable to rounded data so that it gives fewer indications of non-normality. The authors conclude that the selection of a test of normality requires careful study of the properties of the test. Otherwise, the subsequent choice between parametric and nonparametric statistics may not be meaningful.

An optimized quality control procedure for hematology analyzers with the use of retained patient specimens.

The authors propose guidelines for the use of retained patient specimens for the quality control of multichannel hematology analyzers. They demonstrate that control limits for patient specimen replicates may be derived from the long-term standard deviations (s) of commercial whole blood controls. They then use computer stimulation of the Coulter multichannel hematology instrument to determine power functions of various procedures using retained specimens. These power functions show that the use of three patient specimens and +/- 2 s limits are optimal for the detection of systematic error. They recommend that three different, previously analyzed normal range specimens be periodically analyzed, e.g., at eight-hour intervals. The differences between the current and original measurements should then be calculated and compared with their +/- 2 s limits. If at least two of the three differences for any directly measured parameter exceed the +/- 2 s limits, there will be a high probability of significant analytic error. Because the power functions of the derived red blood cell parameters, hematocrit, mean corpuscular hemoglobin, and mean corpuscular hemoglobin concentration demonstrate relatively low error detection capabilities, the authors recommend that these parameters not be monitored with the retained patient specimen procedure.

Evaluation of the accuracy of the diff3/50 with a cell by cell comparison.

The traditional approach to the evaluation of accuracy of discrete differential counters uses a graphic comparison of automated and manual counts. This type of comparison introduces so much sampling variation that significant differences often are obscured. To exclude this sampling variability, one of three technologists, a reference technologist and the Coulter diff3/50, each performed a differential count on exactly the same 100 cells of 47 normal and 56 abnormal blood smears. Linear regression analysis demonstrated excellent agreement between the technologists for all cell types with the exception of band neutrophils. Comparison of the instrument results with the technologists' results showed very good agreement for segmented neutrophils, lymphocytes, monocytes, and eosinophils. The identification of band neutrophils, basophils, immature granulocytes and atypical lymphocytes requires improvement. The overall performance is acceptable, with only 11 leukocytes, on the average normal smear, misclassified out of 100 normal leukocytes (four bands called segs, four segs called bands, and three non-neutrophils incorrectly classified). We recommend similar evaluations for all discrete differential counters.

Quality control of electrolyte analyzers. Evaluation of the anion gap average.

The authors have shown previously that averaging at least eight patient anion gaps provides a sensitive technic for the detection of systematic error in electrolyte analysis (Am J Clin Pathol 79:688-696, 1983). They conducted a retrospective and prospective evaluation of this technic on the ASTRA 4. One month of patient and control data were studied retrospectively and showed that 17/71 abnormally low patient anion gap averages were associated with violations in a multi-rule procedure, and 41/71 low averages were associated with violations in cusum, a more sensitive procedure. In the prospective study, a total of 36 runs of eight patient specimens with low anion gap averages (less than 7.5 mmol/L) were reanalyzed after appropriate recalibration and/or maintenance. Thirty-one of the 36 groups had significant changes in either Na (nine groups, delta Na = +1.5 mmol/L), Cl (14 groups; delta Cl = -1.8 mmol/L), or in both Na and Cl (eight groups; delta Na = +1.2 mmol/L; delta Cl = -0.9 mmol/L). Because the average error detected was small, the authors recommend that the average of anion gaps be used as an early indicator of drift. It must be used, however, in conjunction with standard quality control procedures such as the multi-rule approach.

Plasma fibronectin concentration in patients with acquired consumptive coagulopathies.

Plasma fibronectin was assayed in 179 hospitalized patients referred for workup of possible acquired coagulopathy. Based on laboratory results and chart review, these patients were classified as having no coagulopathy (N = 36), defibrination syndrome (N = 31), compensated defibrination syndrome (N = 100), microangiopathic thrombocytopenia (N = 7), and primary fibrinolysis (N = 5). Compared to patients with no coagulopathy, fibronectin concentration was reduced in patients with defibrination syndrome (p less than 0.005) and compensated defibrination syndrome (p less than 0.10). Fibronectin concentration was not reduced in patients with microangiopathic thrombocytopenia and primary fibrinolysis. In patients with defibrination syndrome, the reduction of fibronectin was correlated to the degree of liver impairment. This finding is consistent with the liver being the primary site of synthesis of plasma fibronectin. Fibronectin was significantly correlated to plasminogen and antithrombin III. The sensitivity of fibronectin for the diagnosis of coagulopathy is low except for patients with defibrination syndrome.

Evaluation of the average of patients: application to endocrine assays.

Recent work (Cembrowski et al, Am J Clin Pathol 1984;81:492-499) has shown that the average of selected patient data can theoretically be used to demonstrate systematic errors. We formerly demonstrate that the sensitivity or power of the 'average of patients' (AOP) is related to the index (NpSw2)/Sp2 where Np is the number of patient results averaged, Sw the within-run component of the standard deviation of the analytical method and Sp the standard deviation of the trimmed population. The utility of the AOP can therefore be assessed rapidly without power function curves or the calculation of the daily AOP itself. The index showed that, for 12 radioimmunoassays done in our laboratory, the power of the AOP was greatest for T3 uptake, then FTI, then T4. Daily AOP were followed for these tests with control results on parallel Shewhart plots. We also applied an exponential smoothing procedure on the AOP and control results to improve trend detection. These analyses demonstrated that the AOP can improve the detection of systematic errors and the differentiation of true errors from errors affecting only control specimens.

Laboratory restructuring in metropolitan Edmonton: a model for laboratory reorganization in Canada.

In 1994 the Alberta government acted to reduce to a decade-long deficit in the provincial budget with draconian reductions in the health, education and welfare expenditures. As a result, funding to Alberta clinical laboratories was to be reduced by approximately 40%. In response, the private and public laboratories in metropolitan Edmonton formed a unique alliance to provide laboratory testing in a more coordinated and efficient manner. Of the five metropolitan hospitals, only University of Alberta Hospital preserved its full service laboratory and its specialty reference testing. The other hospital laboratories were converted to rapid response laboratories with a merged private reference laboratory providing routine testing and support to the four hospitals, and far fewer outpatient collection facilities. This paper describes the steps in the laboratory restructuring from inception to execution.