Transformation of Sequential Hospital and Outpatient Laboratory Data into Between-Day Reference Change Values.

BACKGROUND: Serial differences between intrapatient consecutive measurements can be transformed into Taylor series of variation vs time with the intersection at time = 0 (y0) equal to the total variation (analytical + biological + preanalytical). With small preanalytical variation, y0, expressed as a percentage of the mean, is equal to the variable component of the reference change value (RCV) calculation: (CVA2 + CVI2)1/2. METHODS: We determined the between-day RCV of patient data for 17 analytes and compared them to healthy participants' RCVs. We analyzed 653 consecutive days of Dartmouth-Hitchcock Roche Modular general chemistry data (4.2 million results: 60% inpatient, 40% outpatient). The serial patient values of 17 analytes were transformed into 95% 2-sided RCV (RCVAlternate), and 3 sets of RCVhealthy were calculated from 3 Roche Modular analyzers' quality control summaries and CVI derived from biological variation (BV) studies using healthy participants. RESULTS: The RCVAlternate values are similar to RCVhealthy derived from known components of variation. For sodium, chloride, bicarbonate calcium, magnesium, phosphate, alanine aminotransferase, albumin, and total protein, the RCVs are equivalent. As expected, increased variation was found for glucose, aspartate aminotransferase, creatinine, and potassium. Direct bilirubin and urea demonstrated lower variation. CONCLUSIONS: Our RCVAlternate values integrate known and unknown components of analytic, biologic, and preanalytic variation, and depict the variations observed by clinical teams that make medical decisions based on the test values. The RCVAlternate values are similar to the RCVhealthy values derived from known components of variation and suggest further studies to better understand the results being generated on actual patients tested in typical laboratory environments.

Average of Patient Deltas: Patient-Based Quality Control Utilizing the Mean Within-Patient Analyte Variation.

BACKGROUND: Because traditional QC is discontinuous, laboratories use additional strategies to detect systematic error. One strategy, the delta check, is best suited to detect large systematic error. The moving average (MA) monitors the mean patient analyte value but cannot equitably detect systematic error in skewed distributions. Our study combines delta check and MA to develop an average of deltas (AoD) strategy that monitors the mean delta of consecutive, intrapatient results. METHODS: Arrays of the differences (delta) between paired patient results collected within 20-28 h of each other were generated from historical data. AoD protocols were developed using a simulated annealing algorithm in MatLab (Mathworks) to select the number of patient delta values to average and truncation limits to eliminate large deltas. We simulated systematic error by adding bias to arrays for plasma albumin, alanine aminotransferase, alkaline phosphatase, amylase, aspartate aminotransferase, bicarbonate, bilirubin (total and direct), calcium, chloride, creatinine, lipase, sodium, phosphorus, potassium, total protein, and magnesium. The average number of deltas to detection (ANDED) was then calculated in response to induced systematic error. RESULTS: ANDED varied by combination of assay and AoD protocol. Errors in albumin, lipase, and total protein were detected with a mean of 6 delta pairs. The highest ANDED was calcium, with a positive 0.6-mg/dL shift detected with an ANDED of 75. However, a negative 0.6-mg/dL calcium shift was detected with an ANDED of 25. CONCLUSIONS: AoD detects systematic error with relatively few paired patient samples and is a patient-based QC technique that will enhance error detection.

Barricor blood collection tubes are equivalent to PST for a variety of chemistry and immunoassay analytes except for lactate dehydrogenase.

INTRODUCTION: The BD Barricor tube uses a novel mechanical separator designed to eliminate gel artifacts, decrease cellular contamination, and improve stability. Here, we evaluated the Barricor tube as a possible replacement for PST using Beckman Coulter analyzers under both optimal, alternative, and suboptimal centrifugation conditions based on BD recommendations. METHODS: Paired PST and Barricor samples were collected from 4 local hospitals and processed based on site-specific preanalytical systems involving automated or manual centrifugation. Centrifugation conditions ranged from 1912 ×g for 10 min (suboptimal), 2060 g for 10 min (alternative), and 4000 ×g for 3 or 10 min (optimal). Tube volume (4.5 vs. 5.5 ml) was also assessed. Forty-three chemistry and immunochemistry analytes were measured on Beckman Coulter DxC and DxI analyzers. RESULTS: Using an automated preanlaytical system with suboptimal spin conditions, no bias between PST and Barricor was observed for all analytes tested except lactate dehydrogenase (LD). Further investigation revealed significant increase in LD when Barricor was spun for 10 min at 1912, 2060 and 4000 ×g, ranging from +7.4-19.4% vs. PST across the entire measurement interval (87-493 U/l). Smaller tube volume was also associated with higher LD. Differences in LD occurred despite no change in other hemolysis markers such as potassium, phosphate, and AST. CONCLUSIONS: LD is most sensitive to varying centrifugation conditions (time and speed) in Barricor tubes. We recommend that BD centrifugation protocols should be closely evaluated to determine if Barricor is equivalent to PST under local preanalytical configurations.

The BD Barricor blood collection tube is an acceptable and robust alternative to the PST for use with the Beckman AccuTnI+3 assay.

OBJECTIVES: BD Canada recently released a blood collection tube with a novel mechanical separator called the Barricor. We evaluated this tube as an alternate sample type for cardiac troponin I (cTnI) testing using the Beckman Coulter AccuTnI+3 assay. DESIGN AND METHODS: 3014 paired patient specimens (Barricor, plasma separator tube or PST) were obtained from the emergency departments and cardiac care units of nine hospitals in and around Edmonton, Alberta. After centrifugation, each plasma sample was analyzed for cTnI using the Beckman Coulter AccuTnI+3 assay. In addition, selected samples were analyzed multiple times within a single run or over 4-5days to generate imprecision data for the assay. RESULTS: Repeatability and within-laboratory studies revealed an imprecision of <10% at concentrations above 0.025µg/L for the Barricor as well as BD's traditional PST. Paired patient sample comparisons over the full range of the assay yielded linear regression slopes ranging from 0.956 to 1.011 and Pearson correlation coefficients ranging from 0.993 to 0.999. At a lower range of results closer to the manufacturer's 99th percentile cutoffs correlation was slightly worse, but still acceptable, with linear regression slopes ranging from 0.967 to 1.211 and Pearson correlation coefficients ranging from 0.983 to 0.987. Notably, at these lower concentrations the agreement between individual PST and Barricor results worsened with decreasing cTnI concentration. Differences between pairs of results became particularly large (-50 to +400%) at PST cTnI concentrations ≤0.015µg/L. Closer inspection of the data around the 0.02 and 0.04µg/L 99th percentile cutoffs revealed a number of discordances between PST and Barricor results, with at least some of these attributable to false elevations in the PST results. CONCLUSIONS: Together, our results suggest that the Barricor blood collection tube is good alternative to the traditional PST for cTnI testing using the AccuTnI+3 assay. The Barricor appears to minimize spurious, nonreproducible, and false elevations in cTnI results for a subset of patients but additional studies are needed to determine if it reduces overall false elevations. cTnI results below 0.04µg/L may still be of questionable accuracy even with the use of this new tube.

Impaired clinical utility of sequential patient GEM blood gas measurements associated with calibration schedule.

BACKGROUND: Within- and/or between-instrument variation may falsely indicate patient trends or obscure real trends. We employ a methodology that transforms sequential intra-patient results into estimates of biologic and analytic variation. We previously derived realistic biologic variation (sb) of blood gas (BG) and hematology analytes. We extend this methodology to derive the imprecision of two GEM 4000 BG analyzers. METHODS: A laboratory data repository provided arterial BG, electrolyte and metabolite results generated by two GEM 4000s on ICU patients in 2012-2013. We tabulated consecutive pairs of intra-patient results separated by increasing time interval between consecutive tests. The average between pair variations were regressed against time with the y-intercept representing the sum of the biologic variation and short term analytic variation: yo2=sb2+sa2. Using an equivalent equation for the Radiometer ABL, the imprecision of the two GEMs was calculated: saGEM=(yoGEM2-yoABL2+saABL2)1/2. This analysis was performed for nearly all measurements, regardless of time as well for values obtained over two 12h mutually exclusive periods, starting either at 2am or 2pm. RESULTS: Regression graphs were derived from 1800 patients' blood gas results with least 10,000 data pairs grouped into 2h intervals. The calculated saGEM exceed the directly measured saABL with many GEM sigma ratios of biologic variation/analytic variation being close to unity. All of the afternoon saGEM exceeded their morning counterparts with pH, pCO2, K and bicarbonate being statistically significant. CONCLUSION: For many analytes, the average analytical variation of tandem GEMs approximates the biologic variation, indicating impaired clinical usefulness of tandem sequential measurements. A significant component of this variation is due to increased variation of the GEMs between 2pm and 2am.

Requirements for Successful Adoption of a Glucose Measurement System Into a Hospital POC Program.

Widespread and successful implementation of any glucose measurement system in a hospital point-of-care (POC) program requires a number of features in addition to accurate and reliable analytical performance. Such features include, but are not limited to, a system's glucose-hematocrit dependence, durability, information technology capabilities, and battery capacity and battery life. While the study of Ottiger et al in this issue supports the analytical accuracy and reliability of Bayer's CONTOUR XT® blood glucose monitoring system, the suitability of other features of this system for a hospital POC program remains to be established.

Limiting the testing of AST: a diagnostically nonspecific enzyme.

OBJECTIVES: Annually, millions of pairs of alanine aminotransferase (ALT) and aspartate aminotransferase (AST) tests are ordered. These enzymes are highly correlated, and ALT is far more specific diagnostically than AST. To reduce AST testing, we suggest measuring AST only when ALT exceeds a predetermined limit. METHODS: We derived the proportions of elevated ASTs that would not be measured based on 15 months of paired inpatient and outpatient ALT and AST data. RESULTS: For inpatients, a 35 U/L ALT limit for initiating AST testing would reduce AST testing by 51%, missing only 3% and 7.5% of ASTs exceeding 50 U/L and 35 U/L, respectively. In outpatients, AST testing can be reduced by more than 65%, with fewer missed elevated ASTs (0.5% and 2% of the ASTs exceeding 50 U/L and 35 U/L, respectively). CONCLUSIONS: Conservatively, $100 million could be saved annually in the US health care budget by selectively limiting AST testing in just the US outpatient environment.

Quality control of automated cell counters.

The hematology analyzers of today provide more reproducible analyses compared with those of a few decades ago, necessitating an evolution in hematology quality practices. The improved performance allows use of simple quality control rules. This improved performance also renders the repeat analysis of critical value specimens non value added. Some of the patient averaging techniques have become outmoded and need to be replaced by less complicated calculations but with truncation of predictable outlying populations. The ready availability of comparative peer quality control values helps to improve analyzer precision and accuracy, and simplifies instrument validation and calibration.

Urinary metal concentrations among female welders.

As part of a Canada-wide study of women entering non-traditional trades [Women's Health in Apprenticeship Trades-Metalworkers and Electricians (WHAT-ME)], we examined spot urine samples from women welders in Alberta to determine whether urinary metal concentrations exceeded those of the general population, to compare levels to previously published urinary concentrations in male welders and to examine the relationship with welding tasks. Women mailed-in urine samples collected close to the time of completing a detailed exposure questionnaire, including welding tasks on their most recent day welding at work. Of 53 welders working in their trade, 45 had urinary creatinine >0.3-≤3.0g l(-1) and were included in analyses. Seven metals were examined for which both population and male welder urinary concentrations were available: cadmium, chromium, cobalt, copper, manganese, nickel, and zinc. Principal component analysis was used to extract three components from natural log transformed creatinine-corrected metal concentrations. Of the 45 women, 17 reported more than one main task. Overall two thirds worked in fabrication, a third on pipe welding, and smaller numbers on repair, in construction or other tasks: manual metal arc welding was reported by 62%, semi-automatic arc welding by 47%, and arc welding with a tungsten electrode by 15%. In multiple regression analyses, little relation was found between urinary metals and task or type of welding, except for cadmium where lower levels were seen in those reporting semi-automatic manual welding (after adjustment for age and smoking). The proportion of women welders exceeding the selected general population 95th percentile was high for manganese (96%) and chromium (29%). Urinary metal concentrations were similar to those reported for male welders with only manganese, with a geometric mean in women of 1.91 µg g(-1) creatinine, and perhaps copper (11.8 µg g(-1) creatinine), consistently lower in male welders. Although not evident from the task analysis reported here, differences in exposure by sex may be explained by type of welding or by other work practices. A closely comparable cohort of male welders would be necessary to examine this hypothesis more fully.

The chronic injury glucose error grid: a tool to reduce diabetes complications.

Traditional glucose error grids provide error limits for glucose meters. These criteria help to assess the meter's suitability to prevent acute injury. We present a rationale for an error grid that provides a different set of error limits to help prevent chronic injury in diabetes. For example, glucose values in the no treatment zone of a traditional error grid could be harmful in diabetic retinopathy. The same method comparison data informs both the acute and chronic injury error grids. All of the data are used in an acute injury error grid, whereas only long-term biases populate a chronic injury error grid. These biases can be due to reagent lots and patient specific interferences. An example of a chronic injury glucose error grid is provided using simulated data.

Tighter precision target required for lactate testing in patients with lactic acidosis.

BACKGROUND: Allowable analytic errors are generally based on biologic variation in normal, healthy subjects. Some analytes like blood lactate have low concentrations in healthy individuals and resultant allowable variation is large when expressed as a coefficient of variation (CV). In Ricós' compendium of biologic variation, the relative pooled intra-individual lactate variation (si) averages 27% and the desirable imprecision becomes 13.5%. We derived biologic variability (sb) from consecutive patient data and demonstrate that sb of lactate is significantly lower. METHODS: A data repository provided lactate results measured over 18 months in the General Systems intensive care unit (ICU) at the University of Alberta Hospital in Edmonton, Canada. In total 54,000 lactate measurements were made on two point-of-care Radiometer 800 blood gas systems operated by Respiratory Therapy. The standard deviations of duplicates (SDD) were tabulated for the intra-patient lactates that were separated by 0-1, 1-2...up to 16 h. The graphs of SDD vs. time interval were approximately linear; the y-intercept provided by the linear regression represents the sum of sb and short-term analytic variation (sa):y0=(sa²+sb²)½. The short-term sa was determined from imprecisions provided by Radiometer and confirmed with onsite controls. The derivation of sb was performed for multiple patient ranges of lactate. RESULTS: The relative desirable lactate imprecision for patients with lactic acidosis is about half that of normal individuals. CONCLUSIONS: As such, evaluations of lactate measurements must use tighter allowable error limits.

Within-individual hematocrit variations and self-monitoring of blood glucose.

Many self-monitoring of blood glucose (SMBG) systems have generated artefactually increased glucose results in low-hematocrit patients (e.g., intensive care unit and renal failure patients); conversely, these devices could produce artefactually decreased glucose results in high-hematocrit patients (e.g., neonates). The introduction of hematocrit-independent SMBG systems permits more accurate testing in anemic or polycythemic individuals. In this issue of Journal of Diabetes Science and Technology, Ramljak and coauthors have created glucose bias graphs for 19 common SMBG devices and declared certain systems to be optimally accurate because of insensitivity to hematocrit variation over a broad hematocrit range. Luckily, the average within-individual variation of hematocrit is low (between 2.9 and 3.3%). As such, a larger spectrum of SMBG devices can be regarded as optimally hematocrit independent.

Is HbA1c a good screening test for diabetes mellitus?

OBJECTIVES: HbA(1c) has been recently recommended as the primary diagnostic test for diabetes. This study evaluated the positive predictive value (PPV) and negative predictive value (NPV) of HbA(1c) against the oral glucose tolerance test (OGTT) in three locations. DESIGN AND METHODS: Three years of data with concurrent OGTT and HbA(1c) tests were extracted from Laboratory Information Systems (LIS) and receiver operator (ROC) curves and positive and negative predictive values calculated comparing the OGTT with the HbA(1c) values using a 10% prevalence of diabetes. RESULTS: The recommended threshold HbA(1c) value of 6.5% did not give the optimal combination of NPV (0.93 to 0.92) and PPV (0.40 to 0.61) compared to a threshold HbA(1c) value of 7.0% (NPV 0.91 to 0.92, PPV 0.61 to 0.73). CONCLUSION: The optimal HbA(1c) value for the diagnosis of diabetes is 7.0% but even at this HbA(1c) the PPV is suboptimal and may cause up to 12% of patients without diabetes, as defined by a normal OGTT, to be classified having diabetes mellitus.

Preanalytic and analytic accuracy: toward more realistic and meaningful self-monitoring of blood glucose submissions for regulatory approval.

Dr. Cembrowski provides an analysis of an article by Harrison and colleagues in this issue of Journal of Diabetes Science and Technology in which the authors describe the evaluation of a new device for self-monitoring of blood glucose, the Bayer CONTOUR® blood glucose monitoring system.

Towards more complete specifications for acceptable analytical performance - a plea for error grid analysis.

Abstract We examine limitations of common analytical performance specifications for quantitative assays. Specifications can be either clinical or regulatory. Problems with current specifications include specifying limits for only 95% of the results, having only one set of limits that demarcate no harm from minor harm, using incomplete models for total error, not accounting for the potential of user error, and not supplying sufficient protocol requirements. Error grids are recommended to address these problems as error grids account for 100% of the data and stratify errors into different severity categories. Total error estimation from a method comparison can be used to estimate the inner region of an error grid, but the outer region needs to be addressed using risk management techniques. The risk management steps, foreign to many in laboratory medicine, are outlined.

The use of serial patient blood gas, electrolyte and glucose results to derive biologic variation: a new tool to assess the acceptability of intensive care unit testing.

BACKGROUND: Most estimates of biologic variation (s(b)) are based on periodically acquiring and storing specimens, followed by analysis within a single analytic run. We demonstrate for many intensive care unit (ICU) tests, only patient results need be statistically analyzed to provide reliable estimates of s(b). METHODS: Over 11 months, approximately 28,000 blood gas measurements (including electrolyte panels and glucose) were performed on one of two Radiometer ABL800 FLEX analyzers (Radiometer, Copenhagen, Denmark) from 1676 ICU patients. We tabulated the measurements of paired intra-patient blood samples drawn within 24 h of each other. After removal of outliers, we calculated the standard deviations of duplicates (SDD) of the intra-patient pairs grouped in 2-h intervals: 0-2 h, 2-4 h, 4-6 h, … 20-22 h and 22-24 h. The SDDs were then regressed against the time intervals of 2-14 h; extrapolation to zero time represents the sum of s(b) and short-term analytic variation (s(a)). RESULTS: Substitution of experimentally derived analytic error permitted the calculation of coefficient of variation (biologic) (CV(b)) (100 s(b)/mean): pH, 0.3%; pCO(2), 5.7%; pO(2), 13%; Na(+), 0.6%; K(+), 4.8%; Cl(-), 0.8%; HCO(3)(-), 3.2%; iCa(++), 2.4%; and glucose, 10.3%. The CV(b) of the electrolytes very closely matches the lowest estimates obtained in the usual manner. CONCLUSIONS: Derivation of the ratio of biologic to analytic variation indicates that the ABL800 is extremely suitable for ICU testing. This analysis should be extended to other point of care instrument systems.