- Email: [email protected]
Accuracy and precision of point-of-care testing for glucose and prothrombin time at the critical care units
Clinica Chimica Acta 307 Ž2001. 119–123 www.elsevier.comrlocaterclinchim
Accuracy and precision of point-of-care testing for glucose and prothrombin time at the critical care units Cecilia Yuoh, M. Tarek Elghetany, John R. Petersen, Amin Mohammad, Anthony O. Okorodudu) Clinical Chemistry DiÕision, Department of Pathology, UniÕersity of Texas Medical Branch, GalÕeston, TX 77555-0551, USA
Abstract The use of point-of-care testing ŽPOCT. in critical care patient units has continued to increase since the 1980s. This increase is due to the need for prompt therapeutic interventions that may impact mortality and morbidity, and reduce the overall cost of healthcare for critically ill patients. The diagnostic manufacturing industry has risen to this challenge by introducing portable andror handheld analyzers for use at the point-of-care. In order to ensure the public safety in the USA, the Food and Drug Administration ŽFDA. must approve the use of each POCT analyzer. The FDA approval is based on established performance criteria that includes relative accuracy and precision documentation. This study evaluated the precision and accuracy of the POCT prothrombin time and glucose analyzers relative to the manufacturers’ specifications, to the internal QC in the main laboratory, and to the results of the external proficiency-testing program. The QC for the prothrombin time had a precision that ranged from 2.84% to 3.45% ŽPOCT. and from 1.27–1.66% Žmain laboratory.. The precision for the glucose QC ranged from 5% to 5.2% ŽPOCT. and 0.9–2.7% Žmain laboratory.. Using the results of the external proficiency testing, the inter-laboratory CV% for the POCT prothrombin time ranged from 3.5% to 5.0% and the main laboratory had a range of 2.5–2.9%. The inter-laboratory CV% ranges for glucose POCT and the main laboratory were 4.9–10.6% and 1.8–3.5%, respectively. The main laboratory analyzers proved to be more accurate than the POCT analyzers as indicated by comparison to the mean prothrombin time and glucose results of all participating laboratories in the proficiency testing program. q 2001 Published by Elsevier Science B.V. Keywords: Accuracy; Precision; Point-of-care testing; Critical care; CoaguChek Plus System; AccuChek Advantage Monitor
1. Introduction The introduction of point-of-care testing ŽPOCT. has resulted in the enhancement of patient care in the hospital critical care units and various clinical settings. This is exemplified by the use of POCT ana-
) Corresponding author. Tel.: q1-409-772-3309; fax: q1-409772-9231. E-mail address: [email protected] ŽA.O. Okorodudu..
lyzers for measuring blood glucose levels in the diabetic patients and monitoring of prothrombin time for patients on oral anticoagulant therapy. For the diabetic patient, frequent measurement of blood glucose is required for optimum patient care. In the routine clinical laboratory, plasma and serum samples are often used for the measurement of glucose concentrations. However, at the POCT setting, glucose analysis is usually performed on whole blood obtained from the capillaries in the finger. The sample type is a critical factor because the glucose concentrations in capillary blood are higher than in
0009-8981r01r$ - see front matter q 2001 Published by Elsevier Science B.V. PII: S 0 0 0 9 - 8 9 8 1 Ž 0 1 . 0 0 4 4 2 - 9
120
C. Yuoh et al.r Clinica Chimica Acta 307 (2001) 119–123
venous blood. The difference between capillary and venous samples in the fasting state typically ranges from 2 to 5 mgrdl. In the post-prandial state, differences as high as 20–70 mgrdl have been observed w1x. Another important factor in the interpretation of glucose results is that plasma and whole blood concentrations are not identical due to the difference in water content. The water content of plasma is higher than that of red blood cells, thus the fasting whole blood glucose concentration is approximately 12– 15% lower than plasma glucose concentration w2x. To allow direct comparison between POCT and the central laboratory in the measurement of glucose, some POCT glucose monitoring devices convert the results to plasma values by multiplying the whole blood reading by 1.12. The use of this 1.12 factor is based on the assumption that the hematocrit in the sample is approximately 45%. Other factors may also influence the results of POCT glucose, such as altitude, environmental temperatures, humidity, oxygen content, hypotension, and triglyceride concentration w3x. The methodology and analyzer design can also cause variation in the glucose results. The pre-analytical sample manipulation constitutes another source for variation, i.e., some methods require lysis of red blood cells while others separate the cells from the serum. Due to these inherent problems, professional groups that make decisions based on the results for many of the analytes have recommended practical guidelines for acceptable accuracy and precision. For example, the guideline for evaluation of the technical performance of glucose meters was proposed by the American Diabetes Association ŽADA.. In 1987, the ADA published a consensus statement, which recommended that POCT glucose concentration should fall within "15% of the laboratory values. The ADA also recommended that the goal for all future glucose meters should be to reduce this variability to within "10% for glucose concentrations between 30 and 400 mgrdl w4x. This was further modified by the ADA in consensus statements of 1994 and 1996, which set, as a goal for future glucose meters to be within "5% of laboratory values w5x. In the studies that have been published to date on glucose meters, the limits defined by the guidelines have not been met w3,5x. In one study that evaluated five glucose meters ŽAdvantage, Hemocue, One Touch II, Sure-
Step Pro and Precision G., 90% met the "15% and only 75–87% met the "10% goals, respectively w6x. The POCT instrument manufacturers have continued to strive at producing analyzers that meet the limits for accuracy and precision recommended by the consensus groups. The errors introduced by the end-user of the POCT analyzers, however, have represented in the past and continue to represent, a significant portion of the inaccuracy and imprecision associated with point-of-care testing w7–9x. In this study, we evaluated the accuracy and precision of POCT prothrombin time and glucose testing analyzers relative to the manufacturer’s specifications, proficiency testing and in-house QC at a 900 bed teaching hospital of the University of Texas Medical Branch at Galveston ŽUTMB..
2. Materials and methods 2.1. Analyzers The POCT analyzers evaluated in the retrospective study are CoaguChek Plus System ŽRoche Diagnostic, Indianapolis, IN. for prothrombin time and the AccuChek Advantage Monitor ŽRoche Diagnostic. for glucose. The comparative analyzers in the main laboratory are the MDA ŽOrganon Teknika, RaleighrDurham, NC. and Vitros 950 ŽJohnson and Johnson, Rochester, NY. for prothrombin time and glucose, respectively. In addition, the Chiron 865 ŽBayer, Medfield, MA. was used in the evaluation of POCT glucose relative to a satellite laboratory. 2.2. Procedures The daily quality control for the POCT prothrombin time was assessed using two levels of the CoaguChek Plus QC material. For glucose, Accu-Chek Advantage Monitors was done using the manufacturer’s QC material. The QC results for the POCT analyzers were compared to the quality control performance of the main laboratory’s MDA and the Vitros 950. For glucose POCT, the quality control was also evaluated relative to the Chiron 865 analyzer in the satellite laboratory. Additionally, the data from the 1999 voluntary participation in the College of American Pathologists
C. Yuoh et al.r Clinica Chimica Acta 307 (2001) 119–123 Table 1 Evaluation of manufacturer’s specifications relative to clinically useful limits ŽCUL. for POCT prothrombin time ŽPT. and glucose
PT-I PT-II Glucose-I Glucose-II
Value
Specification
CUL
12.2 s 20.1 s F100 mgrdl G100 mgrdl
3.5% 2.2% 5 mgrdl 5%
3.75% 4.5% 15 mgrdl 15%
ŽCAP. for Surveys WBP-A, CG2-A, WBGA and C-A were also evaluated. We assessed the inter-
121
laboratory performance for prothrombin time and glucose testing as an indicator of relative accuracy at the POCT vs. the main laboratory.
3. Results and discussion 3.1. Precision specifications and QC results relatiÕe to clinically useful limits The clinically useful limits ŽCUL. used for our in-house evaluations of the prothrombin time and
Fig. 1. Ža. The CV% of the quality control summary for prothrombin time at POCT relative to main laboratory. The clinically useful limits ŽCUL. are indicated as the upper allowable limit. N s 60. Žb. The CV% of the quality control summary for glucose at POCT relative to main laboratory. The clinically useful limits ŽCUL. are indicated as the upper allowable limit. N s 60.
C. Yuoh et al.r Clinica Chimica Acta 307 (2001) 119–123
122
glucose is shown in Table 1. For the prothrombin time, the manufacturer’s specified precision at a clotting time of 12.2 s is 95% of the CUL while at 20.1 s it is 48% of the CUL. Similar evaluation of the glucose shows that the manufacturer’s specified precision is 33% of the CUL at the levels evaluated. Fig. 1 shows the result for the evaluation of the daily QC relative to the CUL for prothrombin time ŽFig. 1a. and glucose ŽFig. 1b.. The daily precision for prothrombin time satisfied the manufacturer’s specification of F 3.5% at clotting times of 10.9 s, and was well within the manufactures limits of 4.5% at 22.7 s. As expected, the QC for the prothrombin time on the MDA analyzer in the main laboratory proved to be more precise at levels I Ž1.27%. and II Ž1.66%.. The precision of "5 and "15.6 mgrdl were obtained for the POCT glucose at means of 100 and 300 mgrdl, respectively ŽFig. 1b.. The corresponding summary precision for the glucose QC at the main laboratory is "2.8 and "2.7 mgrdl. Thus, the main laboratory analyzers for prothrombin time and glucose are more precise relative to the POCT analyzers. These results also show that if Westgard’s 2 standard deviation Ž2SD. rule is used for evaluation of the QC data relative to the CUL, the ability of POCT analyzers to pass QC may be problematic.
Table 2 Proficiency testing for prothrombin time at POCT ŽA. and main laboratory ŽB. Mean, s CV% UTMB: PT meany Žall laboratories. Žall laboratories. mean Žs. UTMB mean (A) POCT 1 17.47 2 22.24 3 12.04 4 17.39 5 21.98
4.3 5.0 4.2 4.8 2.3
20.5 20.6 11.5 16.9 21.1
q3.03 y1.64 y0.54 y0.49 y0.88
(B) Main laboratory 1 12.53 2 13.26 3 12.40 4 12.55 5 12.55
2.6 2.9 2.6 2.5 2.7
12.90 13.50 13.00 13.00 13.10
q0.37 q0.24 q0.60 q0.45 q0.55
Evaluation of UTMB performance relative to the means and CV% of all participating laboratory.
Table 3 Proficiency testing for glucose at POCT ŽA. and main laboratory ŽB. Mean, mgrdl CV% UTMB: PT meany Žall laboratories. Žall laboratories. mean UTMB mean Žmgrdl. (A) POCT 1 248.7 2 98.7 3 43.9 4 199.2 5 149.6 (B) Main laboratory 1 49.5 2 185.7 3 77.2 4 329.4 5 241.8
4.9 6.5 10.6 5.1 5.5
3.5 1.9 2.7 1.8 2.1
250 112 65 198 153
49.5 189 77 334 246
q1.3 q13.3 q21.1 y1.2 q3.4
0 3.3 y0.2 4.6 4.2
Evaluation of UTMB performance relative to the means and CV% of all participating laboratory.
3.2. In-house proficiency test results relatiÕe to mean of all participating laboratories The comparison of the proficiency test results at UTMB for prothrombin relative to mean results for all the laboratories participating in the 1999 CAP survey is shown in Table 2A ŽPOCT. and B Žmain laboratory.. The inter-laboratory CV% ranged from 2.3% to 5.0% ŽPOCT. and 2.5–2.9% Žmain laboratory.. Table 2A and B also show that the results from the main laboratory are relatively more accurate as defined by being closer to the mean of all laboratories at all levels. Similar results are also seen when evaluating the proficiency testing survey for glucose as shown in Table 3A ŽPOCT. and B Žmain laboratory.. The inter-laboratory imprecision ŽCV%. ranged from 4.9% to 10.6% for the POCT ŽTable 3A. and 1.8–3.5% for the main laboratory ŽTable 3B.. This comparison of proficiency test results indicates that the testing in the main laboratory is more accurate relative to the mean of all participating laboratories. 4. Conclusion The manufacturer’s specifications for precision are within the UTMB CUL as shown in Table 1. Evaluation of UTMB’s daily QC performance showed that the POCT analyzers for prothrombin
C. Yuoh et al.r Clinica Chimica Acta 307 (2001) 119–123
time and glucose Žlevel I. would have failed the 2SD rule relative to the CUL. On the other hand, the QC for analyzers in the main laboratory were well within the 2SD established per the CUL criteria ŽFig. 1a and b.. The CAP proficiency testing indicates that the prothrombin time and glucose testing done using the main laboratory analyzers are relatively more accurate compared to the POCT. The fact that the POCT analyzers are not as accurate as indicated by the proficiency test results may be attributed to the calibration and the higher degree of imprecision.
w3x w4x w5x w6x
w7x
w8x
References w1x College of American Pathologists, Surveys 2000 Whole Blood Glucose, Participant Summary Report. 2000. p. 7. w2x Sacks DB. Carbohydrates. In: Burtis C, Ashwood E, editors.
w9x
123
Tietz Textbook of Clinical Chemistry. 3rd edn. Philadelphia, PA: Saunders 1999:750–808. American Diabetes Association, Consensus statement on selfmonitoring of blood glucose. Diabetes Care 1994;17:81–6. American Diabetes Association, Consensus statement on selfmonitoring of blood glucose. Diabetes Care 1987;10:95–9. American Diabetes Association, Self-monitoring of blood glucose. Diabetes Care 1996;19:S62–6. Chance JJ, Li DJ. Technical evaluation of five glucose meters with data management capabilities. Am J Clin Pathol 1999; 111:547–56. Nichols JH, Poe SS. Quality assurance, practical management, and outcomes of point-of-care testing: laboratory perspectives. CLM-R 1999;13:341–50. Kilgore ML, Steindel SJ, Smith JA. Continuous quality improvement for point-of-care testing using background monitoring of duplicate specimens. Arch Pathol Lab Med 1999;123: 824–8. Boldt J, Walz G, Triem J, Suttner S, Kumle B. Point-of-care ŽPOC. measurement of coagulation after cardiac surgery. Intensive Care Med 1998;24:1187–93.
Accuracy and precision of point-of-care testing for glucose and prothrombin time at the critical care units Cecilia Yuoh, M. Tarek Elghetany, John R. Petersen, Amin Mohammad, Anthony O. Okorodudu) Clinical Chemistry DiÕision, Department of Pathology, UniÕersity of Texas Medical Branch, GalÕeston, TX 77555-0551, USA
Abstract The use of point-of-care testing ŽPOCT. in critical care patient units has continued to increase since the 1980s. This increase is due to the need for prompt therapeutic interventions that may impact mortality and morbidity, and reduce the overall cost of healthcare for critically ill patients. The diagnostic manufacturing industry has risen to this challenge by introducing portable andror handheld analyzers for use at the point-of-care. In order to ensure the public safety in the USA, the Food and Drug Administration ŽFDA. must approve the use of each POCT analyzer. The FDA approval is based on established performance criteria that includes relative accuracy and precision documentation. This study evaluated the precision and accuracy of the POCT prothrombin time and glucose analyzers relative to the manufacturers’ specifications, to the internal QC in the main laboratory, and to the results of the external proficiency-testing program. The QC for the prothrombin time had a precision that ranged from 2.84% to 3.45% ŽPOCT. and from 1.27–1.66% Žmain laboratory.. The precision for the glucose QC ranged from 5% to 5.2% ŽPOCT. and 0.9–2.7% Žmain laboratory.. Using the results of the external proficiency testing, the inter-laboratory CV% for the POCT prothrombin time ranged from 3.5% to 5.0% and the main laboratory had a range of 2.5–2.9%. The inter-laboratory CV% ranges for glucose POCT and the main laboratory were 4.9–10.6% and 1.8–3.5%, respectively. The main laboratory analyzers proved to be more accurate than the POCT analyzers as indicated by comparison to the mean prothrombin time and glucose results of all participating laboratories in the proficiency testing program. q 2001 Published by Elsevier Science B.V. Keywords: Accuracy; Precision; Point-of-care testing; Critical care; CoaguChek Plus System; AccuChek Advantage Monitor
1. Introduction The introduction of point-of-care testing ŽPOCT. has resulted in the enhancement of patient care in the hospital critical care units and various clinical settings. This is exemplified by the use of POCT ana-
) Corresponding author. Tel.: q1-409-772-3309; fax: q1-409772-9231. E-mail address: [email protected] ŽA.O. Okorodudu..
lyzers for measuring blood glucose levels in the diabetic patients and monitoring of prothrombin time for patients on oral anticoagulant therapy. For the diabetic patient, frequent measurement of blood glucose is required for optimum patient care. In the routine clinical laboratory, plasma and serum samples are often used for the measurement of glucose concentrations. However, at the POCT setting, glucose analysis is usually performed on whole blood obtained from the capillaries in the finger. The sample type is a critical factor because the glucose concentrations in capillary blood are higher than in
0009-8981r01r$ - see front matter q 2001 Published by Elsevier Science B.V. PII: S 0 0 0 9 - 8 9 8 1 Ž 0 1 . 0 0 4 4 2 - 9
120
C. Yuoh et al.r Clinica Chimica Acta 307 (2001) 119–123
venous blood. The difference between capillary and venous samples in the fasting state typically ranges from 2 to 5 mgrdl. In the post-prandial state, differences as high as 20–70 mgrdl have been observed w1x. Another important factor in the interpretation of glucose results is that plasma and whole blood concentrations are not identical due to the difference in water content. The water content of plasma is higher than that of red blood cells, thus the fasting whole blood glucose concentration is approximately 12– 15% lower than plasma glucose concentration w2x. To allow direct comparison between POCT and the central laboratory in the measurement of glucose, some POCT glucose monitoring devices convert the results to plasma values by multiplying the whole blood reading by 1.12. The use of this 1.12 factor is based on the assumption that the hematocrit in the sample is approximately 45%. Other factors may also influence the results of POCT glucose, such as altitude, environmental temperatures, humidity, oxygen content, hypotension, and triglyceride concentration w3x. The methodology and analyzer design can also cause variation in the glucose results. The pre-analytical sample manipulation constitutes another source for variation, i.e., some methods require lysis of red blood cells while others separate the cells from the serum. Due to these inherent problems, professional groups that make decisions based on the results for many of the analytes have recommended practical guidelines for acceptable accuracy and precision. For example, the guideline for evaluation of the technical performance of glucose meters was proposed by the American Diabetes Association ŽADA.. In 1987, the ADA published a consensus statement, which recommended that POCT glucose concentration should fall within "15% of the laboratory values. The ADA also recommended that the goal for all future glucose meters should be to reduce this variability to within "10% for glucose concentrations between 30 and 400 mgrdl w4x. This was further modified by the ADA in consensus statements of 1994 and 1996, which set, as a goal for future glucose meters to be within "5% of laboratory values w5x. In the studies that have been published to date on glucose meters, the limits defined by the guidelines have not been met w3,5x. In one study that evaluated five glucose meters ŽAdvantage, Hemocue, One Touch II, Sure-
Step Pro and Precision G., 90% met the "15% and only 75–87% met the "10% goals, respectively w6x. The POCT instrument manufacturers have continued to strive at producing analyzers that meet the limits for accuracy and precision recommended by the consensus groups. The errors introduced by the end-user of the POCT analyzers, however, have represented in the past and continue to represent, a significant portion of the inaccuracy and imprecision associated with point-of-care testing w7–9x. In this study, we evaluated the accuracy and precision of POCT prothrombin time and glucose testing analyzers relative to the manufacturer’s specifications, proficiency testing and in-house QC at a 900 bed teaching hospital of the University of Texas Medical Branch at Galveston ŽUTMB..
2. Materials and methods 2.1. Analyzers The POCT analyzers evaluated in the retrospective study are CoaguChek Plus System ŽRoche Diagnostic, Indianapolis, IN. for prothrombin time and the AccuChek Advantage Monitor ŽRoche Diagnostic. for glucose. The comparative analyzers in the main laboratory are the MDA ŽOrganon Teknika, RaleighrDurham, NC. and Vitros 950 ŽJohnson and Johnson, Rochester, NY. for prothrombin time and glucose, respectively. In addition, the Chiron 865 ŽBayer, Medfield, MA. was used in the evaluation of POCT glucose relative to a satellite laboratory. 2.2. Procedures The daily quality control for the POCT prothrombin time was assessed using two levels of the CoaguChek Plus QC material. For glucose, Accu-Chek Advantage Monitors was done using the manufacturer’s QC material. The QC results for the POCT analyzers were compared to the quality control performance of the main laboratory’s MDA and the Vitros 950. For glucose POCT, the quality control was also evaluated relative to the Chiron 865 analyzer in the satellite laboratory. Additionally, the data from the 1999 voluntary participation in the College of American Pathologists
C. Yuoh et al.r Clinica Chimica Acta 307 (2001) 119–123 Table 1 Evaluation of manufacturer’s specifications relative to clinically useful limits ŽCUL. for POCT prothrombin time ŽPT. and glucose
PT-I PT-II Glucose-I Glucose-II
Value
Specification
CUL
12.2 s 20.1 s F100 mgrdl G100 mgrdl
3.5% 2.2% 5 mgrdl 5%
3.75% 4.5% 15 mgrdl 15%
ŽCAP. for Surveys WBP-A, CG2-A, WBGA and C-A were also evaluated. We assessed the inter-
121
laboratory performance for prothrombin time and glucose testing as an indicator of relative accuracy at the POCT vs. the main laboratory.
3. Results and discussion 3.1. Precision specifications and QC results relatiÕe to clinically useful limits The clinically useful limits ŽCUL. used for our in-house evaluations of the prothrombin time and
Fig. 1. Ža. The CV% of the quality control summary for prothrombin time at POCT relative to main laboratory. The clinically useful limits ŽCUL. are indicated as the upper allowable limit. N s 60. Žb. The CV% of the quality control summary for glucose at POCT relative to main laboratory. The clinically useful limits ŽCUL. are indicated as the upper allowable limit. N s 60.
C. Yuoh et al.r Clinica Chimica Acta 307 (2001) 119–123
122
glucose is shown in Table 1. For the prothrombin time, the manufacturer’s specified precision at a clotting time of 12.2 s is 95% of the CUL while at 20.1 s it is 48% of the CUL. Similar evaluation of the glucose shows that the manufacturer’s specified precision is 33% of the CUL at the levels evaluated. Fig. 1 shows the result for the evaluation of the daily QC relative to the CUL for prothrombin time ŽFig. 1a. and glucose ŽFig. 1b.. The daily precision for prothrombin time satisfied the manufacturer’s specification of F 3.5% at clotting times of 10.9 s, and was well within the manufactures limits of 4.5% at 22.7 s. As expected, the QC for the prothrombin time on the MDA analyzer in the main laboratory proved to be more precise at levels I Ž1.27%. and II Ž1.66%.. The precision of "5 and "15.6 mgrdl were obtained for the POCT glucose at means of 100 and 300 mgrdl, respectively ŽFig. 1b.. The corresponding summary precision for the glucose QC at the main laboratory is "2.8 and "2.7 mgrdl. Thus, the main laboratory analyzers for prothrombin time and glucose are more precise relative to the POCT analyzers. These results also show that if Westgard’s 2 standard deviation Ž2SD. rule is used for evaluation of the QC data relative to the CUL, the ability of POCT analyzers to pass QC may be problematic.
Table 2 Proficiency testing for prothrombin time at POCT ŽA. and main laboratory ŽB. Mean, s CV% UTMB: PT meany Žall laboratories. Žall laboratories. mean Žs. UTMB mean (A) POCT 1 17.47 2 22.24 3 12.04 4 17.39 5 21.98
4.3 5.0 4.2 4.8 2.3
20.5 20.6 11.5 16.9 21.1
q3.03 y1.64 y0.54 y0.49 y0.88
(B) Main laboratory 1 12.53 2 13.26 3 12.40 4 12.55 5 12.55
2.6 2.9 2.6 2.5 2.7
12.90 13.50 13.00 13.00 13.10
q0.37 q0.24 q0.60 q0.45 q0.55
Evaluation of UTMB performance relative to the means and CV% of all participating laboratory.
Table 3 Proficiency testing for glucose at POCT ŽA. and main laboratory ŽB. Mean, mgrdl CV% UTMB: PT meany Žall laboratories. Žall laboratories. mean UTMB mean Žmgrdl. (A) POCT 1 248.7 2 98.7 3 43.9 4 199.2 5 149.6 (B) Main laboratory 1 49.5 2 185.7 3 77.2 4 329.4 5 241.8
4.9 6.5 10.6 5.1 5.5
3.5 1.9 2.7 1.8 2.1
250 112 65 198 153
49.5 189 77 334 246
q1.3 q13.3 q21.1 y1.2 q3.4
0 3.3 y0.2 4.6 4.2
Evaluation of UTMB performance relative to the means and CV% of all participating laboratory.
3.2. In-house proficiency test results relatiÕe to mean of all participating laboratories The comparison of the proficiency test results at UTMB for prothrombin relative to mean results for all the laboratories participating in the 1999 CAP survey is shown in Table 2A ŽPOCT. and B Žmain laboratory.. The inter-laboratory CV% ranged from 2.3% to 5.0% ŽPOCT. and 2.5–2.9% Žmain laboratory.. Table 2A and B also show that the results from the main laboratory are relatively more accurate as defined by being closer to the mean of all laboratories at all levels. Similar results are also seen when evaluating the proficiency testing survey for glucose as shown in Table 3A ŽPOCT. and B Žmain laboratory.. The inter-laboratory imprecision ŽCV%. ranged from 4.9% to 10.6% for the POCT ŽTable 3A. and 1.8–3.5% for the main laboratory ŽTable 3B.. This comparison of proficiency test results indicates that the testing in the main laboratory is more accurate relative to the mean of all participating laboratories. 4. Conclusion The manufacturer’s specifications for precision are within the UTMB CUL as shown in Table 1. Evaluation of UTMB’s daily QC performance showed that the POCT analyzers for prothrombin
C. Yuoh et al.r Clinica Chimica Acta 307 (2001) 119–123
time and glucose Žlevel I. would have failed the 2SD rule relative to the CUL. On the other hand, the QC for analyzers in the main laboratory were well within the 2SD established per the CUL criteria ŽFig. 1a and b.. The CAP proficiency testing indicates that the prothrombin time and glucose testing done using the main laboratory analyzers are relatively more accurate compared to the POCT. The fact that the POCT analyzers are not as accurate as indicated by the proficiency test results may be attributed to the calibration and the higher degree of imprecision.
w3x w4x w5x w6x
w7x
w8x
References w1x College of American Pathologists, Surveys 2000 Whole Blood Glucose, Participant Summary Report. 2000. p. 7. w2x Sacks DB. Carbohydrates. In: Burtis C, Ashwood E, editors.
w9x
123
Tietz Textbook of Clinical Chemistry. 3rd edn. Philadelphia, PA: Saunders 1999:750–808. American Diabetes Association, Consensus statement on selfmonitoring of blood glucose. Diabetes Care 1994;17:81–6. American Diabetes Association, Consensus statement on selfmonitoring of blood glucose. Diabetes Care 1987;10:95–9. American Diabetes Association, Self-monitoring of blood glucose. Diabetes Care 1996;19:S62–6. Chance JJ, Li DJ. Technical evaluation of five glucose meters with data management capabilities. Am J Clin Pathol 1999; 111:547–56. Nichols JH, Poe SS. Quality assurance, practical management, and outcomes of point-of-care testing: laboratory perspectives. CLM-R 1999;13:341–50. Kilgore ML, Steindel SJ, Smith JA. Continuous quality improvement for point-of-care testing using background monitoring of duplicate specimens. Arch Pathol Lab Med 1999;123: 824–8. Boldt J, Walz G, Triem J, Suttner S, Kumle B. Point-of-care ŽPOC. measurement of coagulation after cardiac surgery. Intensive Care Med 1998;24:1187–93.