Evaluation of a New POCT Bedside Glucose ... - Nova …

1 Evaluation of a New POCT Bedside Glucose Meterand Strip With Hematocrit and Interference CorrectionsCathy Holtzinger, MLT (ASCP),* Edwina Szelag, MHM, BSMT (ASCP),* Jeffrey A. DuBois, PhD, Terry L. Shirey, PhD, and Steven Presti, (ASCP), MT Introduction:Based on the expanding role of point of care testingglucose meters and the need to improve accuracy and precision, thenew nova Biomedical statstrip was evaluated and compared with theLifeScan SureStepFlexx (current point of care testing meter).Methods:Specimen volume variation, within-run imprecision,lot-to-lot bias, bias relative to a plasma hexokinase assay, andanalytical interferences likely to be encountered in hospitalizedpatients were :Strip dosing did not affect the statstrip meter but did affectthe SureStepFlexx at 5- and 50-KL specimen volumes.

2 Within-runprecision for each Glucose meter was less than 5% at 39 to 47 mg/dLof Glucose , less than at 215 to 265 mg/dL, and less than 370 to 470 mg/dL. Improper coding resulted in erroneousmeasurements on the SureStepFlexx. Each meter was comparedwith the Dade RxL hexokinase plasma reference method, givingthe following correlation equations: statstrip = (hexokinase) (r2= ); SureStepFlexx = (hexokinase) + (r2= ). At [ Glucose ] of 55 mg/dL, ascorbic acid interferedwith the SureStepFlexx but did not affect statstrip . Hematocrit alsoaffected the correlation of whole blood Glucose on the SureStep-Flexx to the plasma hexokinase reference Glucose but did not affectthe statstrip :These studies suggest that the new statstrip meter maybe more accurate and precise (elimination of hematocrit effect andelectrochemical interferences with no error because of strip dosing orcalibration) than the SureStepFlexx meter.

3 This reduction in totalerror may help achieve better glycemic control in Words: Glucose meter, point of care testing, accuracy(Point of Care2008;7:1Y6)Much attention is paid to the monitoring of Glucose inhospitalized patients to achieve tight glycemiccontrol and to minimize complications from hypoglycemiaand hyperglycemia of hospi-talized patients is common among patients with diabetesmellitus, it is not restricted solely to patients with thisdisease. In addition, Glucose levels within hospitalizedpatients, particularly critically ill patients, can changerapidly depending on stress and medications.

4 To maintaintight glycemic control for these patients, rapid turnaroundtime for Glucose analysis is required. Point of care testing(POCT) reduces turnaround time dramatically from thatobtained by traditional central laboratory testing. Glucosemeters are generally the instruments of choice for POCT Glucose major current concern is the accuracy of POCT Glucose target Glucose range, used to guideinsulin dosage in hospitalized patients, varies depending onthe institution and is typically narrower than that in theambulatory (home) setting. However, abnormal hematocritlevels and various electrochemical or chemical interferants,known to affect results from Glucose meters, are frequentlypresent in the specimens of hospitalized patients.

5 Medicationsand hematocrit in these patients are found to affect theperformance of almost all Glucose meter acid interfered with all Glucose metertechnologies available in the year 2000. Several studiesdemonstrated that low hematocrit gives a high bias to Glucose ,and conversely, high hematocrit causes a low bias in glucoselevels for virtually all meters addition, the degree to which Glucose meters correlatewith plasma hexokinase Glucose measurements (used fre-quently as a reference technology) varies greatly betweenglucose meter particular, correlation in thehypoglycemic and hyperglycemic ranges is highly variablewith most currently available study was designed to compare our current glucosemeter with a new Glucose meter that uses both interference andmeasured hematocrit corrections.

6 The study compared theaccuracy of both POCT hospital-based Glucose meters withthe reference plasma hexokinase method and also evaluatedthe effect of drug/chemical interferences and hematocrit oneach meter at 3 Glucose AND METHODSI nstrumentationThis study compared 3 Glucose assays. The referenceassay hexokinase method on the Dade-Behring DimensionRxL analyzer (Dade-Behring, Deerfield, Ill) was used tomeasure plasma Glucose . Hexokinase methods are suitable foruse as reference methods for Glucose determination becausethey correlate closely to definitive mass 2glucose meter technologies studied were the SureStepFlexx(LifeScan, Malpitas, Calif), which uses a photometric glucoseoxidase detection system and a new meter technology,ORIGINALARTICLEP oint of Care&Volume 7, Number 1, March 20081 From the *Holmes Regional Medical Center, Melbourne, FL.

7 And NovaBiomedical Corporation, Waltham, : Cathy Holtzinger, MLT (ASCP), Department of Pathology, HolmesRegional Medical Center, 1350 South Hickory St, Melbourne, FL 32901(e-mail: 2008 by Lippincott Williams & WilkinsStatStrip ( nova Biomedical, Waltham, Mass), which uses amodified Glucose oxidaseYbased amperometric test systemwith hematocrit and other interference Dosing StudyWhole blood droplets of varying volumes (5Y30KL)from blood specimens containing 4 different Glucose levels(as determined by the reference Dade RxL instrument) wereplaced on Parafilm on a flat surface.)

8 The statstrip meter,loaded with a strip, was touched to a droplet, as if it were on apatient s fingertip, and the specimen was drawn into the teststrip by capillary action to measure the Glucose . This processwas repeated 6 times each droplet volume (5Y30KL) at the 4different Glucose levels. The SureStepFlexx was evaluatedusing whole blood droplets of varying volumes (5Y50KL)that were pipetted directly onto the test pads of the process was repeated in replicates of 6 at each of the 4glucose Precision StudyFor within-run precision, venous heparinized wholeblood was drawn 12 to 24 hours in advance of performing thestudy.

9 Aerated blood was divided into three 2-mL aliquots,which received different volumes of a concentrated glucosesolution, such that the aliquots had 20 to 60, 200 to 300, and450 to 550 mg/dL of Glucose . Each aliquot was then tested 20times on each Code StudyA limited study of calibration bias from strip lot tostrip lot included looking at Glucose values across 4 striplots of statstrip (which all use the same company-designated calibration with no lot-to-lot variation) and 2strip lots of SureStepFlexx (each having been assigned itsown calibration code numbers). Six replicate analyses foreach statstrip at 3 Glucose levels were performed using thesame calibration code across strip lots.

10 Six replicate analysesfor each of 2 strip lots of SureStepFlexx using 2 calibrationcodes (ie, strip A with calibration code A, strip B withcalibration code B, strip A with calibration code B, and stripB with calibration code A) were also run for each strip lot at3 Glucose Correlations Using Patient SpecimensTwo hundred fresh, venous, whole blood newly dis-carded specimens in green top tubes were used in this study. Adrop from each well-mixed green top tube (spiked orunspiked) was removed and wicked onto 1 statstrip stripand 1 SureStepFlexx strip for immediate analyses.