Establishing hemolysis, icterus, and lipemia interference limits for body fluid chemistry analytes measured on the Roche cobas instrument.

OBJECTIVES: The aims of this study were to (1) establish the maximum allowable interference limits for hemolysis, lipemia, and icterus for chemistry analytes tested in body fluid samples and (2) assess the effectiveness of serial dilution to mitigate spectral interferences. METHODS: Residual body fluids from clinically ordered testing were mixed (<10% by volume) with stock solutions of interferent (spiked) and compared with a control spiked with an equal volume of 0.9% saline. The analytes were measured on the Roche cobas c501 instrument. Difference and percentage difference were calculated and compared with allowable total error limits. A subset of samples were serially diluted with 0.9% saline. Mean (SD) difference and percentage difference were calculated. RESULTS: The interference thresholds were lower than the package insert for lactate dehydrogenase, cholesterol, triglycerides, and total protein for hemolysis; amylase, cholesterol, and total protein for icterus; and albumin for lipemia. Only cholesterol and triglyceride results returned to baseline upon dilution of icteric samples. CONCLUSIONS: Interference thresholds in body fluids were lower than blood for 6 analytes. Diluting interferences that surpass these limits does not produce reliable results that are comparable to the baseline results before spiking in the interferent.

Lack of observed interference by therapeutic monoclonal antibodies in select commonly utilized immunoassays.

BACKGROUND: Therapeutic monoclonal antibodies (tmabs) have been hypothesized to interfere with immunoassay measurements, although studies investigating this potential new class of interference are lacking. This study evaluated the effects of tmabs used in cancers ipilimumab (Bristol Myers Squibb), nivolumab (Bristol Myers Squibb), pembrolizumab (Merck) and autoimmune disorders adalimumab (AbbVie), infliximab (Janssen) and vedolizumab (Takeda) in common immunoassays used in the clinical laboratory. METHODS: Residual sera from 10 randomly chosen patients were split into two tubes and spiked with same volume (approximately 5 % final volume) of either saline (control) or 6 tmabs (final concentration of 100 µg/mL each). Concentrations from sixteen analytes in 19 different assays were assessed: TSH (Roche and Beckman), free thyroxine (Roche and Siemens), cortisol (Beckman), Cancer Antigens (CA): CA19-9 (Beckman), CA15-3 (Roche), CA125 (Roche), and CA27.29 (Siemens), carcinoembryonic antigen (Beckman), alpha-fetoprotein (Beckman), thyroglobulin (Beckman) and thyroglobulin antibodies (Beckman), thyroid peroxidase antibody (Beckman), beta-human chorionic gonadotropin (Roche and Beckman), total prostate-specific antigen (Roche), parathyroid hormone (Roche) and antinuclear antibodies IgG (Werfen). The tmab spiked residual sera were compared with matched saline spiked sera and percent error was assessed against allowable total error defined from biological variation or CLIA limits. RESULTS: None of the tested immunoassays were affected by the presence of the tmabs, in samples within or outside assay reference intervals. The median % error among all immunoassays ranged between -2.0% (for TSH) to 2.7% (for TPO Ab assay). CONCLUSION: These findings demonstrate no detectable tmab interference for the assessed immunoassays using spiked preparations of the tmabs in residual human sera. The findings are limited to the tmabs and immunoassays studied here.

A Review of Current Practices and Future Trends in Body Fluid Testing.

BACKGROUND: Body fluid testing in the clinical chemistry laboratory is a cornerstone in the diagnostic workup of pathological effusions. Laboratorians may not be aware of the preanalytical workflows used in the collection of body fluids though the value is evident whenever processes change or issues arise. The analytical validation requirements can vary depending on the regulations dictated by the laboratories' jurisdiction and accreditor requirements. Much of analytical validation hinges on how useful testing is to clinical care. Usefulness of testing varies with how well established and incorporated the tests and interpretation are in practice guidelines. CONTENT: Body fluid collections are depicted and described so clinical laboratorians have a basic appreciation of what specimens are submitted to the laboratory for testing. A review of validation requirements by major laboratory accreditation entities is presented. A review of the usefulness and proposed decision limits for common body fluid chemistry analytes is presented. Body fluid tests that show promise and those that are losing (or lost long ago) value are also reviewed. SUMMARY: The total testing process from collection to result interpretation can be complicated and easily overlooked by the clinical laboratory. This review aims to improve the understanding and awareness of collections, validation, result interpretation, and provide an update on recent trends.

Population-Based Evaluation of Total Protein in Cerebrospinal Fluid.

OBJECTIVES: To present a normal range of cerebrospinal fluid (CSF) protein levels in a community-based population and to evaluate factors that contribute to CSF protein level variability. PATIENTS AND METHODS: Samples of CSF protein were obtained from participants aged 32 to 95 years who underwent lumbar puncture (LP) between November 1, 2007, and October 1, 2017, as part of the Mayo Clinic Study of Aging, a longitudinal, population-based study of residents of Olmsted County, Minnesota. RESULTS: A total of 633 participants (58.1% male; 99.1% White; mean ± SD age, 70.9±11.6 years) underwent LP with recorded CSF protein level. Mean ± SD CSF protein level was 52.2±18.4 mg/dL (to convert to mg/L, multiply by 10), with a 95% reference interval of 24.0 to 93.4 mg/dL (range, 14.0-148.0 mg/dL). Spinal stenosis and arterial hypertension were associated with higher CSF protein levels on univariable analysis (P<.001). Increasing age, male sex, and diabetes were all independently associated with higher CSF protein levels on multivariable analysis (P<.001). In the 66 participants with repeated LPs within 2.5 years, the coefficient of repeatability was 26.1 mg/dL. Eleven participants (16.7%) had a CSF protein level difference of 20 mg/dL or more between serial LPs, and 4 (6.1%) had a difference of 25 mg/dL or more. There was a trend toward greater CSF protein level variability in patients with spinal stenosis (P=.054). CONCLUSION: This large population-based study showed that CSF protein level can vary significantly among individuals. Elevated CSF protein level was independently associated with older age, male sex, and diabetes and is higher than listed in many laboratories. These findings emphasize the necessity of evidence-based reevaluation and standardization of CSF protein metrics.

Short- and Long-Term Biological Variability of Small Dense LDL, HDL3, and Triglyceride-Rich Lipoprotein Cholesterol.

BACKGROUND: Measurement of cholesterol within lipoprotein subfractions may aid in cardiovascular disease prediction. Simple, homogenous enzymatic assays for the direct measurement of lipoprotein subfractions have been developed to measure small dense low-density lipoprotein cholesterol (sdLDL-C), high-density lipoprotein-3 cholesterol (HDL3-C), and triglyceride-rich lipoprotein (TRL-C) cholesterol. The objective of this study was to determine biological variability for sdLDL-C, HDL3-C, and TRL-C in a healthy reference population to facilitate interpretation of these analytes. METHODS: Serum samples were collected from 24 healthy subjects (n = 14 female/10 male) daily for 3 days while non-fasting, and daily for 5 days, weekly for 4 weeks, and monthly for 6 months after overnight fasting. sdLDL-C, HDL3-C, and TRL-C cholesterol were measured by homogenous enzymatic assays. Sources of variability (between-subject, within-subject, and analytical) were calculated using random-effects regression models. Reference change value (RCV) and index of individuality (II) for each time period were determined from the variance components. RESULTS: Analytic variability (daily, weekly, and monthly CVA) was <3% for each analyte. Monthly within-subject variability (CVI) was 17.1% for sdLDL-C, 7.4% for HDL3-C, and 25.7% for TRL-C. Most of the monthly variation was attributed to between-subject variation for all 3 analytes. Overall RCVs for monthly measurements were 18.1 mg/dL for sdLDL-C, 6.1 mg/dL for HDL3-C, and 16.0 mg/dL for TRL-C. IIs were <0.6 for sdLDL-C and HDL3-C, and 0.81 for TRL-C. CONCLUSIONS: sdLDL-C, HDL3-C, and TRL-C showed moderate within-subject variability, but high between-subject variability, in a healthy reference population. Given the high individuality of each analyte, population-based reference intervals may be inadequate to detect clinically significant changes.

Comparison of Five Common Analyzers in the Measurement of Chemistry Analytes in an Authentic Cohort of Body Fluid Specimens.

OBJECTIVES: Interpretation of body fluid (BF) results is based on published studies and clinical guidelines. The aim of this study is to determine whether the assays from five common commercial vendors produce similar results in BFs for 12 analytes in a BF cohort. METHODS: BFs (n = 25) and serum (n = 5) were analyzed on five instruments (Roche cobas c501, Ortho 5600, Beckman AU5800 and DXI800, Siemens Vista 1500, and Abbott Architect c8000) to measure albumin, amylase, total bilirubin, cholesterol, creatinine, glucose, lactate dehydrogenase (LDH), lipase, total protein, triglycerides, urea nitrogen, and carcinoembryonic antigen. Deming regression and Bland-Altman analysis were used for method comparison to Roche. RESULTS: Results were significantly different from Roche for LDH and lipase on Ortho and lipase on Siemens but similar for both BFs and serum. BF differences were larger than serum differences when measuring creatinine, glucose, and urea nitrogen on Ortho and glucose on Siemens. CONCLUSIONS: Five instruments used to perform BF testing produce results that are not significantly different except for lipase and LDH measurements. Bias of similar magnitude observed in both BF and serum should not affect interpretation. Further investigations into Ortho and Siemens measuring glucose and Ortho measuring creatinine and urea nitrogen are warranted.

The Biological Variability of Plasma Ceramides in Healthy Subjects.

BACKGROUND: Ceramides are bioactive lipid species that mediate numerous cell-signaling events. Elevated plasma ceramides concentration constitutes a risk factor for several pathologies. Multiple studies have affirmed the plasma concentrations of 4 specific ceramides (Cer16:0, Cer18:0, Cer24:0, and Cer24:1) can predict cardiovascular disease risk. Furthermore, these ceramides can be altered by many lipid-lowering therapies. Understanding the biological variability within an individual, and within a population, will further inform the clinical use of plasma ceramides as a biomarker. In this study, we aimed to define the intra- and interbiological variability of ceramides in a healthy reference population in a weekly and monthly manner. METHODS: Fasting plasma from 24 healthy adults was collected daily (5 days), weekly (4 weeks), and monthly (7 months). Ceramide concentrations were measured with liquid chromatography-mass spectrometry (LC-MS). For analysis, we used random-effects regression models to estimate variance components. RESULTS: The analytical variability was smaller compared to the biological variability overall. The greatest variation reported was between-subject variation for all ceramide species. The critical difference-reference change value (RCV) for within-subject variations monthly were 0.07 mcmol/L (Cer16:0), 0.04 mcmol/L (Cer18:0), 1.09 mcmol/L (Cer24:0), and 0.27 mcmol/L (Cer24:1). The index of individuality (IOI) of ceramides were 0.82 (Cer16:0), 0.96 (Cer18:0), 1.06 (Cer24:0), and 0.89 (Cer24:1). The most consistent ceramide species was Cer18:0 with the lowest within- and between-subject critical differences in weekly and monthly measurements. CONCLUSIONS: Overall, this study demonstrates that the variability of ceramide concentrations at different time points is minimal within individuals, allowing a single draw to be sufficient at least in a yearly time frame.

COVID-19 Mortality Prediction From Deep Learning in a Large Multistate Electronic Health Record and Laboratory Information System Data Set: Algorithm Development and Validation.

BACKGROUND: COVID-19 is caused by the SARS-CoV-2 virus and has strikingly heterogeneous clinical manifestations, with most individuals contracting mild disease but a substantial minority experiencing fulminant cardiopulmonary symptoms or death. The clinical covariates and the laboratory tests performed on a patient provide robust statistics to guide clinical treatment. Deep learning approaches on a data set of this nature enable patient stratification and provide methods to guide clinical treatment. OBJECTIVE: Here, we report on the development and prospective validation of a state-of-the-art machine learning model to provide mortality prediction shortly after confirmation of SARS-CoV-2 infection in the Mayo Clinic patient population. METHODS: We retrospectively constructed one of the largest reported and most geographically diverse laboratory information system and electronic health record of COVID-19 data sets in the published literature, which included 11,807 patients residing in 41 states of the United States of America and treated at medical sites across 5 states in 3 time zones. Traditional machine learning models were evaluated independently as well as in a stacked learner approach by using AutoGluon, and various recurrent neural network architectures were considered. The traditional machine learning models were implemented using the AutoGluon-Tabular framework, whereas the recurrent neural networks utilized the TensorFlow Keras framework. We trained these models to operate solely using routine laboratory measurements and clinical covariates available within 72 hours of a patient's first positive COVID-19 nucleic acid test result. RESULTS: The GRU-D recurrent neural network achieved peak cross-validation performance with 0.938 (SE 0.004) as the area under the receiver operating characteristic (AUROC) curve. This model retained strong performance by reducing the follow-up time to 12 hours (0.916 [SE 0.005] AUROC), and the leave-one-out feature importance analysis indicated that the most independently valuable features were age, Charlson comorbidity index, minimum oxygen saturation, fibrinogen level, and serum iron level. In the prospective testing cohort, this model provided an AUROC of 0.901 and a statistically significant difference in survival (P<.001, hazard ratio for those predicted to survive, 95% CI 0.043-0.106). CONCLUSIONS: Our deep learning approach using GRU-D provides an alert system to flag mortality for COVID-19-positive patients by using clinical covariates and laboratory values within a 72-hour window after the first positive nucleic acid test result.

Effect of pH on the Quantification of Common Chemistry Analytes in Body Fluid Specimens Using the Roche cobas Analyzer for Clinical Diagnostic Testing.

OBJECTIVES: To determine the influence of pH on recovery of analytes in body fluids (BFs), investigate the mechanism of pH interference, measure the frequency of abnormal-pH BFs received, and compare pH measured by meter and paper. METHODS: We performed pH titration in residual BFs. A low-pH BF was spiked and neutralized to investigate pH interference. We measured analytes on a Roche cobas c501 analyzer (Roche Diagnostics) and calculated the percent recovery. Measurement of pH using a meter and paper was conducted on 122 BF samples received in the laboratory. RESULTS: Enzyme activity in BFs was unaffected when pH = 7.4-8.5 lactate dehydrogenase, pH = 7.3-10.2 amylase, pH = 6.0-9.9 lipase, and pH = 1.3-11.7 all other analytes. BFs had mean (range) pH of 8.0 (5.1-8.9), with a mean (range) difference (paper ‒ meter) of ‒0.4 (‒0.6 to 1.1). CONCLUSIONS: Irreversible loss of enzyme activity occurs in BFs at low pH. Few clinical BFs have pH < 7.0, but laboratories should incorporate pH measurement in BF workflows.

Synovial fluid α defensin has comparable accuracy to synovial fluid white blood cell count and polymorphonuclear percentage for periprosthetic joint infection diagnosis.

AIMS: The aim of this study was to determine the diagnostic accuracy of α defensin (AD) lateral flow assay (LFA) and enzyme-linked immunosorbent assay (ELISA) tests for periprosthetic joint infection (PJI) in comparison to conventional synovial white blood cell (WBC) count and polymorphonuclear neutrophil percentage (PMN%) analysis. METHODS: Patients undergoing joint aspiration for evaluation of pain after total knee arthroplasty (TKA) or total hip arthroplasty (THA) were considered for inclusion. Synovial fluids from 99 patients (25 THA and 74 TKA) were analyzed by WBC count and PMN% analysis, AD LFA, and AD ELISA. WBC and PMN% cutoffs of ≥ 1,700 cells/mm3 and ≥ 65% for TKA and ≥ 3,000 cells/mm3 and ≥ 80% for THA were used, respectively. A panel of three physicians, all with expertise in orthopaedic infections and who were blinded to the results of AD tests, independently reviewed patient data to diagnose subjects as with or without PJI. Consensus PJI classification was used as the reference standard to evaluate test performances. Results were compared using McNemar's test and area under the receiver operating characteristic curve (AUC) analysis. RESULTS: Expert consensus classified 18 arthroplasies as having failed due to PJI and 81 due to aseptic failure. Using these classifications, the calculated sensitivity and specificity of AD LFA was 83.3% (95% confidence interval (CI) 58.6 to 96.4) and 93.8% (95% CI 86.2 to 98.0), respectively. Sensitivity and specificity of AD ELISA was 83.3% (95% CI 58.6 to 96.4) and 96.3% (95% CI 89.6 to 99.2), respectively. There was no statistically significant difference between sensitivity (p = 1.000) or specificity (p = 0.157) of the two AD assays. AUC for AD LFA was 0.891. In comparison, AUC for synovial WBC count, PMN%, and the combination of the two values was 0.821 (sensitivity p = 1.000, specificity p < 0.001), 0.886 (sensitivity p = 0.317, specificity p = 0.011), and 0.926 (sensitivity p = 0.317, specificity p = 0.317), respectively. CONCLUSION: The diagnostic accuracy of synovial AD for PJI diagnosis is comparable and not statistically superior to that of synovial WBC count plus PMN% combined. Cite this article: Bone Joint J 2021;103-B(6):1119-1126.

Comparative analysis of 23 synovial fluid biomarkers for hip and knee periprosthetic joint infection detection.

There is interest in novel synovial fluid biomarkers for the detection of periprosthetic joint infection (PJI). Here, we assessed the diagnostic accuracy of 23 simple or sophisticated synovial fluid biomarkers for periprosthetic hip or knee infection detection. One hundred seven subjects were studied, 57 of whom had aseptic failure (AF) and 50 PJI. The following synovial fluid biomarkers were tested using spectrophotometric assays, immunoassays, lateral flow tests, or test strips: leukocyte count, monocyte percentage, lymphocyte percentage, neutrophil percentage, C-reactive protein (CRP), glucose, lactate, granulocyte-macrophage colony-stimulating factor, interferon-γ, interleukin-1ß (IL-1ß), IL-2, IL-4, IL-5, IL-6, IL-8, IL-10, IL-12p70, IL-13, IL-17A, IL-23, tumor necrosis factor-α, α-defensin, and leukocyte esterase. The best-performing synovial fluid biomarkers to differentiate PJI from AF-that is, those with highest area under the curve compared to all other biomarkers-were leukocyte count, percent neutrophils and percent monocytes, CRP, and α-defensin (P < .0001).

Proposed Plasma Ammonia Reference Intervals in a Reference Group of Hospitalized Term and Preterm Neonates.

BACKGROUND: Plasma ammonia is commonly measured in the diagnostic evaluation of hospitalized newborns, but reference values are not well defined. METHODS: We prospectively enrolled newborns admitted to the level III/IV neonatal intensive care unit and level II intermediate special care nursery from January 2017 to January 2018. Infants with inborn errors of metabolism or liver disease were excluded. Plasma ammonia concentrations were measured once within the first week of life and evaluated by sex, gestational age, timing of the draw, blood collection method, and type of nutrition. Reference intervals were calculated. RESULTS: 127 neonates were included; one third (34%) were term infants born at ≥37 weeks gestation, and two thirds (66%) were born preterm at <37 weeks gestation. Median plasma ammonia concentrations were 32 µmol/L (range <10 to 86 µmol/L). Median ammonia concentrations were higher among preterm compared to term infants (35 vs. 28 µmol/L, p = 0.0119), and term female compared to term male infants (34 vs. 26 µmol/L, p = 0.0228). There was no difference in median ammonia concentrations between female and male preterm infants, based on gestational age within the preterm group, timing of the blood draw, presence of hyperbilirubinemia, blood collection method, or type of nutritional intake. CONCLUSIONS: Plasma ammonia concentrations among newborns are higher than the expected adult concentrations and may vary by gestational age and sex. Blood collection method, type of nutrition, hyperbilirubinemia, and timing of the draw do not impact concentrations. We propose a reference limit of ≤82 µmol/L for newborns less than one week of age.

Prevalence of Clinically Significant Differences in Sodium Measurements Due to Abnormal Protein Concentrations Using an Indirect Ion-Selective Electrode Method.

BACKGROUND: Indirect ion-selective electrode (ISE) is the primary method used to measure sodium in automated clinical laboratories and is susceptible to the electrolyte exclusion effect. Pseudohyponatremia due to hyperproteinemia can affect patient management. The aims of this study were to (a) establish the relationship between serum total protein (TP) concentration and the magnitude of the electrolyte exclusion effect on indirect ISE-measured sodium values (b) estimate the frequency at which TP concentrations outside the reference interval may impact indirect-ISE measured sodium values, and (c) determine whether clinical decision support (middleware) rules in the laboratory would be effective for detecting cases of pseudohyponatremia. METHODS: Residual waste serum specimens from physician-ordered TP testing were collected (n = 112). Sodium concentration was measured using indirect ISE (Cobas 8000, Roche Diagnostics) and direct ISE (ABL 825, Radiometer) methods. The difference in sodium concentration (Δ[Na+]) was calculated as follows: ([Na+]indirect-ISE - [Na+]direct-ISE). Retrospective TP results reported from July 31, 2013, to September 24, 2014, were stratified by ordering location and sodium and TP co-ordering rates were quantified. RESULTS: Δ[Na+] was inversely proportional to TP concentration (y = -1.29x + 8.6, R = -0.883). The average difference (SD, range) was -6.1(3.4, -16-0) mmol/L when TP >7.9 g/dL (>79g/L), with 69% of samples demonstrating differences ≥4.0 mmol/L. A majority of intensive care unit patients (81%) were hypoproteinemic (<6.3 g/dL, <63g/L). Only 10.9% of sodium test orders include an order for TP on the same collection. CONCLUSIONS: Indirect sodium measurement is impacted when TP concentrations are increased. TP concentration outside the reference interval is prevalent and sodium is usually not ordered with TP. Health systems need to be aware of the limitations of their indirect-ISE method for sodium measurement.

Human chorionic gonadotropin suspected heterophile interference investigations in immunoassays: a recommended approach.

Background Heterophile antibody (HAb) interferences in immunoassays can cause falsely elevated hCG concentrations leading to incorrect diagnosis and treatments options. When results are not consistent with the clinical findings, hCG HAb interference investigation may be requested by the physician. A retrospective evaluation of the frequency of HAb interference was performed among cases of physician-requested investigations and the effectiveness of commercially available blocking reagents to detect HAb interference in two immunoassay systems was evaluated. Methods One hundred and thirteen physician requests for hCG HAb investigation from 2008 to 2017 were reviewed. The primary method used to measure hCG was the Beckman Coulter Access Total ßhCG (2008-2010) and the Roche Elecsys HCG+ß (2014-2017). HAb investigation included measurement by two immunoassays before and after treatment of samples with heterophile blocking reagents and serial dilution studies. Results Five cases of HAb and HAb-like interference were identified. The interference frequency was 6.7% for the Beckman assay and 2.9% for the Roche assay. The presence of HAb was detected using heterophile blocking reagents and an alternative method in three cases. The other two cases were detected due to discrepant results with an alternative method and non-linear serial dilutions (HAb-like). Conclusions HAb interference was observed in the Beckman and the Roche assays. The heterophile blocking reagents failed to detect 40% of interference cases. Blocking reagents should not solely be used for these investigations. Multiple strategies including the use of serial dilutions and using an alternative platform are critical when troubleshooting interferences in hCG immunoassays.

Impact of an electronic decision support rule on ESR/CRP co-ordering rates in a community health system and projected impact in the tertiary care setting and a commercially insured population.

INTRODUCTION: Erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) are common laboratory assays used as markers of inflammation. ESR suffers from higher false positive and false negative rates than CRP. To that end, the American Board of Internal Medicine's (ABIM's) Choosing Wisely campaign has recommended against ESR testing for those with undiagnosed conditions in favor of CRP testing. This study describes the impact of a computerized provider order entry (CPOE) decision support rule against ESR/CRP co-ordering within a community health system that predates the ABIM's Choosing Wisely national guidance. To demonstrate the potential impact of such a CPOE rule within other healthcare settings, ESR/CRP ordering data from a multi-site tertiary care practice and from the commercially insured population in the OptumLabs® Data Warehouse (OLDW) were analyzed and the relative reduction in ESR/CRP co-ordering achieved within the community health system was projected onto these populations. MATERIALS AND METHODS: ESR and/or CRP orders from a community health system were assessed from 2012 to 2016. Co-ordering and test concordance rates between ESR and CRP were compared before and after CPOE decision support rule launch. Similarly, ESR/CRP co-ordering across three tertiary care sites from 2015 to 2016 and the OLDW from 2009 to 2013 were assessed and the co-ordering rate reduction achieved in the community health system was mathematically projected onto these populations. Estimated payer savings from the rule's effect were calculated within each population using Medicare reimbursement rates. RESULTS: The CPOE decision support rule realized an unadjusted 42% relative rate reduction in ESR/CRP co-ordering within the community health system yielding an annual payer savings of $15,000 with a modest increase in ESR/CRP concordance rates. Projecting a 40% relative reduction in ESR/CRP co-ordering rates from a similarly effective CPOE rule, annual payer cost reductions exceeding $100,000 within a multi-site tertiary care setting and $1,000,000 within the OLDW would be expected. CONCLUSION: ESR/CRP co-ordering represents an opportunity to eliminate testing waste and reduce payer costs. A CPOE decision support rule stably reduces ESR/CRP co-ordering rates. Similar results may occur as one component of new commercially available decision support platforms.