Performance of the Roche Accu-Chek Active Glucose Meter to Screen for Gestational Diabetes Mellitus Using Fasting Capillary Blood

Abstract

Background:

Point-of-care (POC) blood glucose measurement using glucose meters is used by diabetes patients to mange their disease. POC glucose testing also is also used in tight glycemic control protocols and as a screening tool for diabetes. We report the performance and effectiveness of the Accu-Chek^® Active (Roche Diagnostics GmbH, Mannheim, Germany) glucose meter to screen for gestational diabetes mellitus (GDM) using blood fasting capillary glucose (FCG).

Methods:

To screen for GDM, 1,465 pregnant women underwent an oral glucose glucose tolerance test. Correlation between the FCG and fasting plasma glucose (FPG) levels was determined by Passing and Bablok regression analysis. Total error (TE) of the glucometer was ascertained using the Bland–Altman method with the DXC-800 analyzer (Beckman-Coulter Instruments, Brea, CA) as the reference method. The area under the receiver operator characteristic curve was used to analyze the performance of the FCG to predict GDM.

Results:

FPG and FCG identified 361 (24.6%) and 338 (23%) women as having GDM, respectively. The Bland–Altman TE at 95% limits of agreement was −11.1% to 10.8%. The area under the receiver operator characteristic curve was 0.953 (95% confidence interval 0.943 to 0.964).

Conclusions:

The Roche Accu-Chek Active glucometer meets analytical and clinical quality requirements. A TE of±15% is acceptable for glucose meters used in ambulatory care, including home self-monitoring of blood glucose, and different TE targets should be set for acute critical care settings.

Introduction

A mong the wide array of point-of-care (POC) tests available, blood glucose measurement using a hand-held glucose meter remains the most frequently performed test. This is because self-monitoring of blood glucose is integral to managing diabetes mellitus as strongly recommended by the American Diabetes Association (ADA)¹ and the International Diabetes Federation.² Furthermore, POC glucose is used in many hospitals applying tight glycemic control protocols.³ POC glucose has also been used as a screening tool for diabetes.^4,5 A survey to establish the trends in POC testing found that all 317 participant U.S. hospitals performed blood glucose at the POC.⁶ Thus, a glucose result as reported by a glucose meter remains a common and important test. A glucose meter has many obvious advantages over the laboratory: it has a rapid turnaround time, is easy to use, avoids a venipuncture, and needs a minimal sample volume.

The results generated by glucose meters are used to make important therapeutic decisions. Therefore, the analytical quality achieved by a glucose meter is critical and should match that of laboratory analyzers.^7,8 Furthermore, there are serious concerns regarding the inaccuracy of many glucose meters with potential clinical errors, which could be detrimental.^9,10 There are no universally accepted criteria for the analytical performance of glucose meters; current recommendations for the accuracy of glucose meters compared with laboratory methods range widely from±5% to±20%.^8,11

–15 Previously, the ADA advocated that that glucose meter results should fall within±15% of the reference system; this recommendation was later revised to±5%.¹¹ Based on patient-derived glucometer quality specifications for self-monitoring of blood glucose, Skeie et al.¹² recommended a total error (TE) of±13%. Two simulation studies have reported that a TE of±13–16% for glucometers is unlikely to produce large errors in insulin dosage.^13,15 The ISO 15197 guideline states that meter measurements should be within 0.83 mmol/L (15 mg/dL) of the reference measurement for glucose values <4.2 mmol/L (75 mg/dL) and within 20% for glucose values ≥4.2 mmol/L (75 mg/dL). The performance of the glucometer is satisfactory if 95% of glucometer results are within the specified limits compared with a reference method.¹⁴

There is no consensus on objective quality targets for analytical performance. A hierarchical approach, “The Stockholm Conference Hierarchy,” recommends that, in descending order, the hierarchy of quality specifications should be based as follows: (1) clinical outcome; (2) biological variation; (3) published professional recommendations; (4) regulatory bodies and external quality assessment schemes; and (5) current state of the art.^16
–18 More recently, Klee¹⁹ has reviewed methods for establishing assay performance specifications. The aim of the present study was to evaluate the performance of the Accu-Chek^® Active (Roche Diagnostics GmbH, Mannheim, Germany) glucometer against published recommendations for analytical quality specifications against a clinical outcome: the diagnosis of gestational diabetes mellitus (GDM) in pregnant women using fasting capillary glucose (FCG).

Materials and Methods

Setting and subjects

The study, conducted at a tertiary-care hospital in the United Arab Emirates between July 2007 and June 2008, was approved by the institutional Research and Ethics Committee. To screen for GDM, 1,643 pregnant women underwent a 75-g oral glucose tolerance test (OGTT). Because of many reasons, like wrong time of collection, eating between the test, not fasting, and refusal of the OGTT, 188 women were excluded from the study. Data from the remaining 1,465 women were used for the study.⁴ For this study the data were analyzed using the latest International Association of Diabetes and Pregnancy Study Groups guidelines: GDM is diagnosed if one or more plasma venous glucose values were ≥5.1 mmol/L at 0 h, ≥10 mmol/L at 1 h, or ≥8.5 mmol/L at 2 h.²⁰ For purposes of convenience and practical considerations, only FCG and fasting plasma glucose (FPG) values were compared.

Test methods

FCG was determined by the Accu-Chek glucometer following the manufacturer's instructions. All measurements were performed by experienced trained laboratory staff. The venous plasma sample (fasting and 1 and 2 h following oral 75-g glucose) was centrifuged immediately, and the plasma glucose level was analyzed within 30 min of collection in the hospital laboratory. The plasma glucose was estimated by the glucose oxidase method (DXC-800 Synchron system, Beckman-Coulter Instruments, Brea, CA). The imprecision (CVa) of the glucometer and DXC-800 was determined by using two and three levels of commercial liquid control material, respectively. CVa of the DXC-800 was 2.0%, 1.8%, and 1.6% at 2.6 mmol/L, 13.3 mmol/L, and 23.6 mmol/L, respectively. The CVa of the Accu-Chek glucometer at mean glucose levels of 3.2 mmol/L and 9.0 mmol/L was 2.9% and 1.8%, respectively. The hospital laboratory subscribes to an overseas external quality assessment scheme from the United Kingdom (U.K. National External Quality Assessment Scheme). The average absolute deviation of glucose laboratory results from the target mean during the study period was 2.3%. Hence, the laboratory met the analytical standards for glucose.²¹

Statistical methods

Data were logged into an Excel (Microsoft, Redmond, WA) spreadsheet and analyzed using SPSS version 15 for Windows (SPSS Inc., Chicago, IL). For glucose results <4.2 mmol/L, the absolute number and percentage of glucometer values within±0.28, 0.56, and 0.83 mmol/L of laboratory results were reported. For glucose results ≥4.2 mmol/L, the absolute number and percentage of glucometer values within±5%, 10%, 15%, and 20% were reported.¹⁴ In addition, the Clarke Error Grid Analysis (EGA) was used to determine the clinical accuracy of the Accu-Chek Active. The EGA specifies five zones, A–E, signifying varying degrees of accuracy of glucose estimations: Zone A and B errors are clinically acceptable, whereas values in Zones C, D, and E are potentially dangerous and clinically significant.²² The area under the receiver operator characteristic curve was used to analyze the performance of the FCG to predict GDM. A diagnostic threshold of 5.1 mmol/L for FPG was used as the “gold standard.” Correlation between the FCG and FPG levels was determined by Passing and Bablok regression analysis. TE of the glucometer was determined using the Bland–Altman method with the DXC-800 as the reference method.

Results

The glucose results ranged from 3.3 to 10.7 mmol/L. Of the 1,465 glucometer results, 163 (11.1%) were <4.2 mmol/L. The Accu-Chek glucometer accuracy results as per the ISO 15197 guideline are shown at Table 1. Of the results, 99.4% and 0.6% were in Zones A and B, respectively, of the EGA with no results in Zones C–E. The linear relationship observed between the Accu-Chek glucometer and DXC-800 glucose results is described by the following formula: FCG=([1×DXC-800 results]−0.06). The 95% confidence intervals (CIs) for the slope and intercept are 0.95 to 1.0 and −0.06 to 0.18, respectively. The mean FPG and FCG levels were 4.83 and 4.82 mmol/L, respectively (P=0.1371). The mean bias between the FCG and FPG was −0.2% (95% CI −0.4% to 0.1%). The Bland–Altman TE at 95% limits of agreement is shown at Figure 1. The FCG and FPG results are highly correlated and show very good agreement.

FIG. 1.

Bland–Altman plot showing the percentage total error between between fasting capillary glucose (FCG) (by Roche Accu-Chek glucometer) and fasting plasma glucose (FPG) (by Beckman-Coulter DXC-800) at 95% limits of agreement. The solid line at 0% difference is the line of identity; the dashed black line close to the solid line is the bias. The two outer dashed lines are the 95% limits of agreement. Color images available online at www.liebertonline.com/dia

Table 1.

Accuracy of the Roche Accu-Chek Glucometer Results Compared Against the Beckman Coulter DXC-800 Using ISO 15197 Criteria ¹⁴

Glucose results	Accuracy
<4.2 mmol/L
Within±0.28 mmol/L	150/163 (92%)
Within±0.56 mmol/L	157/163 (96.3%)
Within±0.83 mmol/L	161/163 (98.8%)
≥4.2 mmol/L
Within±5%	1,150/1,302 (88%)
Within±10%	1,272/1,302 (97.7%)
Within±15%	1,300/1,302 (99.8%)
Within±20%	1,301/1,302 (99.9%)

The numbers of women identified as having GDM by FPG and FCG were 361 (24.6%) and 338 (23%), respectively. The performance of the FCG to predict GDM is shown in Figure 2 as a receiver operator characteristic curve; the area under the curve was 0.953 (95% CI 0.943 to 0.964).

FIG. 2.

The receiver operator characteristic curve showing the performance of fasting capillary glucose (FCG) using the fasting plasma venous glucose as the gold standard. The area under the curve is 0.953 (95% confidence interval 0.943 to 0.964).

Discussion

The FCG and FPG results were highly correlated with very good agreement by both the analytical and the diagnostic criteria. Although not meeting the very stringent ADA criteria of±5% TE, the Roche Accu-Chek Active glucometer meets the quality specifications of DIN EN ISO 15197 and other published recommendations.^{8,9,11

–15} The performance of the Accu-Chek Active against the ISO 15197 and EGA criteria is virtually identical to that reported recently.⁹ The results of both studies show the Accu-Chek Active to be very accurate.

This study reports the excellent performance of the Accu-Chek Active glucometer compared with a laboratory-based method using a defined clinical outcome (i.e., a diagnosis of GDM). Therefore, according to the Stockholm Hierarchy the highest level of quality has been achieved. The relatively large number of subjects (1,465 women) and results over a study period of 12 months carried out in a real-life outpatient setting would support the data presented as being valid and not attributed to chance occurrence.

We wish to recognize a few shortcomings. The full range of glucose levels was not studied (reported range is 3.3–10.7 mmol/L); ideally, the laboratory method should be performed in duplicates. Blood hematocrit was not estimated. The manufacturer's instructions provided with the test strips state that it is permissible to use blood with a hematocrit between 25% and 55%. As there was excellent analytical correlation and clinical agreement between the laboratory and glucometer results, it may be assumed that the hematocrit of the study subjects did not affect results. It is generally accepted that the hexokinase method is the reference method for glucose estimations. In a survey to determine the trueness and interlaboratory harmonization for 10 analytes, Miller et al.²³ reported that glucose met the desirable analytical goals for bias against the stringent biological variation analytical criteria. The Beckman hexokinase and glucose oxidase methods were among the methods evaluated against the reference isotope dilution mass spectrometry and hexokinase methods. Therefore, by extrapolation the glucose oxidase method is traceable to a reference system and may be designated as a reference method when comparing the performance of glucometers with laboratory methods.²³

The differences between glucometer performance and the laboratory methods may be affected by pre-analytical variables. In clinical situations with decreased peripheral blood flow use of capillary finger stick blood glucose measurements may not be appropriate. Examples include severe dehydration, hypotension, shock, or peripheral vascular disease. The reliability of glucose meters depends on the underlying clinical conditions and the analytical performance of the glucometer.²⁴ Therefore, recommendations for quality specifications may need to be refined based on the clinical setting in which glucose meters are used. In clinical practice, plasma and blood glucose results are used interchangeably. There is therefore a risk of exceeding the analytical allowable error and clinical misinterpretation. To avoid such errors a constant factor of 1.11 is recommended to convert blood glucose values to plasma equivalent glucose concentrations, thus allowing for consistent harmonization of results.²⁵ The Accu-Chek Active glucometer applies this factor to report plasma equivalent results. The owner's booklet states clearly that the Accu-Chek Active meter is not approved to diagnose diabetes. In light of the data reported further studies are required to ascertain the feasibility of using glucose meters for diagnostic purposes.

In conclusion, the Roche Accu-Chek Active glucometer meets analytical and clinical quality requirements. The authors recommend (1) a TE of±15% be acceptable for glucose meters used in ambulatory care, including home self-monitoring of blood glucose, and (2) different TE targets be set for acute critical care settings.

Footnotes

Author Disclosure Statement

No competing financial interests exist.

References

American Diabetes Association. Standards of medical care in diabetes—2009. Diabetes Care, 2009; 32,Suppl 1:S13–S61.

IDF Clinical Guidelines Task Force. Global Guidelines for Type 2 Diabetes. Brussels: International Diabetes Federation, 2005.

van den Berghe

, Woulters

, Weekers

, Verwaest

, Bruyninckx

, Schetz

, Vlasselaers

, Ferdinande

, Lauwers

, Bouillon

. Intensive insulin therapy in critically ill patients. N Engl J Med, 2001; 345:1359–1367.

Agarwal

, Dhatt

, Othman

, Gupta

. Gestational diabetes: fasting capillary glucose as a screening test in a multi-ethnic, high risk population. Diabet Med, 2009; 26:760–765.

Rush

, Crook

, Simmons

. Point-of-care testing as a tool for screening for diabetes and pre-diabetes. Diabet Med, 2008; 25:1070–1075.

Winter

. Market trends in point-of-care testing. Point of Care, 2010; 9:12–20.

Khan

, Vasquez

, Gray

, Wians

Jr , Kroll

. The variability of results between point-of-care testing glucose meters and the central laboratory analyzer. Arch Pathol Lab Med, 2006; 130:1527–1532.

Dhatt

, Agarwal

, Bishawi

. Evaluation of a glucose meter against analytical quality specifications for hospital use. Clin Chim Acta, 2004; 343:217–221.

Freckmann

, Baumstark

, Jendrike

, Zschornack

, Kocher

, Tshiananga

, Heister

, Haug

. System accuracy evaluation of 27 blood glucose monitoring systems according to DIN EN ISO 15197. Diabetes Technol Ther, 2010; 12:221–231.

10.

Pitkin

, Coursin

, Rice

. Review article: glucose measurement in the operating room: more complicated than it seems. Anesth Analg, 2010; 110:1056–1065.

11.

American Diabetes Association:. Self-monitoring of blood glucose. Diabetes Care, 1996; 19,Suppl 1:S62–S66.

12.

Skeie

, Thue

, Sandberg

. Patient-derived quality specifications for instruments used in self-monitoring of blood glucose. Clin Chem, 2001; 47:67–73.

13.

Boyd

, Bruns

. Quality specifications for glucose meters: assessment by simulation modelling of errors in insulin dose. Clin Chem, 2001; 47:209–214.

14.

DIN EN ISO 15197: In Vitro Diagnostic Test Systems—Requirements for Blood-Glucose Monitoring Systems for Self-Testing in Managing Diabetes Mellitus (ISO 15197:2003) Brussels: European Committee for Standardization, 2003.

15.

Karon

, Boyd

, Klee

. Glucose meter performance criteria for tight glycemic control estimated by simulation modeling. Clin Chem, 2010; 56:1091–1097.

16.

Kallner

, McQueen

, Heuck

. The Stockholm Consensus Conference on Quality Specifications in Laboratory Medicine, 25–26 April 1999. Scand J Clin Lab Invest, 1999; 59:475–476.

17.

Kenny

, Fraser

, Petersen

, Kallner

. Consensus agreement. Scand J Clin Lab Invest, 1999; 59:585.

18.

Dhatt

, Agarwal

, Bishawi

, Jill

. Implementing the Stockholm Conference hierarchy of objective criteria in a routine laboratory. Clin Chem Lab Med, 2007; 45:549–552.

19.

Klee

. Establishment of outcome-related analytical performance goals. Clin Chem, 2010; 56:714–722.

20.

International Association of Diabetes and Pregnancy Study Groups recommendations on the diagnosis and classification of hyperglycemia in pregnancy. Diabetes Care, 2010; 33:676–682.

21.

Sacks

, Bruns

, Goldstein

, Maclaren

, McDonald

, Parrott

. Guidelines and recommendations for laboratory analysis in the diagnosis and management of diabetes mellitus. Clin Chem, 2002; 48:436–472.

22.

Clarke

, Cox

, Gonder-Frederick

, Carter

, Pohl

. Evaluating clinical accuracy of systems for self-monitoring of blood glucose. Diabetes Care, 1987; 10:622–628.

23.

Miller

, Myers

, Ashwood

, Killeen

, Wang

, Ehlers

, Hassemer

, Lo

, Seccombe

, Siekmann

, Thienpont

, Toth

. State of the art in trueness and interlaboratory harmonization for 10 analytes in general clinical chemistry. Arch Pathol Lab Med, 2008; 132:838–846.

24.

Khan

, Vasquez

, Gray

, Wians

Jr , Kroll

. The variability of results between point-of-care testing glucose meters and the central laboratory analyzer. Arch Pathol Lab Med, 2006; 130:1527–1532.

25.

D'Orazio

, Burnett

, Fogh-Andersen

, Jacobs

, Kuwa

, Külpmann

, Larsson

, Lewenstam

, Maas

, Mager

, Naskalski

, Okorodudu AO;

IFCC-SD-WG-SEPOCT

. Approved IFCC recommendation on reporting results for blood glucose: International Federation of Clinical Chemistry and Laboratory Medicine Scientific Division, Working Group on Selective Electrodes and Point-of-Care Testing (IFCC-SD-WG-SEPOCT) Clin Chem Lab Med, 2006; 44:1486–1490.