Clinical Impact of Point-of-Care vs Laboratory Measurement of Anticoagulation

Coagulation and Transfusion Medicine / CLINICAL IMPACT OF POC VS LABORATORY INR Clinical Impact of Point-of-Care vs Laboratory Measurement of Anticoagulation Rubina Sunderji, PharmD, FCSHP, 1,7 Kenneth Gin, MD, FRCPC, 3,8,9 Karen Shalansky, PharmD, FCSHP, 2,7 Cedric Carter, MB, FRCPC, 4,10 Keith Chambers, MD, 5 Cheryl Davies, RN, 5 Linda Schwartz, RN, 11 and Anthony Fung, MB, FRCPC 6,9 Key Words: Point-of-care; Self-testing; International normalized ratio; Self-management; Warfarin; Anticoagulation Abstract Patients using anticoagulation point-of-care (POC) monitors are advised to periodically test these systems against laboratory methods to monitor performance. The international normalized ratio (INR), however, can vary between test systems owing to different instrumentreagent combinations. In a randomized study evaluating warfarin self-management, we compared INR measured by patients on a POC monitor (ProTime, International Technidyne Corporation, Edison, NJ) with those obtained at a hospital laboratory within 1 hour. Ninety-one paired INR determinations from 55 patients met inclusion criteria. Clinical agreement in which POC and laboratory INR were within or outside the target INR range occurred in 56 (62%) of 91 cases (κ = 0.35). The mean (SD) difference between POC and laboratory INR was 0.44 (0.61). Six pairs differed by 1 or more INR units, 3 at study initiation resulting in POC monitor replacement. The accuracy of INR selftesting with ProTime was acceptable. The small failure rate of INR agreement might be clinically important, suggesting the need for external quality control systems. Current models of anticoagulation management include laboratory testing of the international normalized ratio (INR) coupled with warfarin dosage adjustment by a physician or through an anticoagulation clinic. Self-management of warfarin by the patient is an evolving model whereby trained patients can test their INR using point-of-care (POC) systems and adjust their warfarin dosages. The POC systems offer an alternative simplified method to traditional laboratory testing. The INR is determined from whole blood obtained by finger puncture. A key feature of these POC systems is the ability to produce an INR result within minutes to enable timely drug dosage adjustment and prompt attention to critical values. The ability to self-test the INR at home increases patient convenience and is particularly useful for those without ready access to laboratories or who experience difficulties with venous blood collection. Owing to their portability, POC systems are attractive for patients who frequently travel away from their home laboratory. Clinical trials that have compared self-management of warfarin with conventional methods have demonstrated improved anticoagulation control and enhanced patient satisfaction with the self-management strategy. 1-4 The success of self-testing depends on the accuracy and reliability of POC systems. Although these devices have internal quality control systems to ensure proper functioning of the instrument and reagent systems, formal external quality control systems are lacking. As a result, periodic checking of the POC INR concurrently at a laboratory is advised by the British Society of Haematology Task Force for Haemostasis and Thrombosis. 5 It is important to recognize, however, that variations in INR results can occur owing to differences in reagent-equipment combinations. 6 Disagreement between POC and standard laboratory testing could impose confusion 184 Am J Clin Pathol 2005;123:184-188 Downloaded 184 from https://academic.oup.com/ajcp/article-abstract/123/2/184/1759382

Coagulation and Transfusion Medicine / ORIGINAL ARTICLE in warfarin dosage decision making and anxiety for patients and health care providers. In addition, inappropriate therapeutic decisions could result in adverse outcomes owing to the narrow therapeutic range of warfarin. Recently, POC systems have been approved for patient use in Canada, making self-management of warfarin feasible. 7 We conducted a randomized single-center trial to compare anticoagulation control by self-management using the ProTime Microcoagulation System (International Technidyne Corporation, Edison, NJ) with the traditional physician method. 8 As part of our study protocol, patients in the selfmanagement arm tested their INR with this monitor and were required to have concurrent INR determinations at our hospital laboratory to assess their technique and the performance of the monitor. This is a report of our experience with INR agreement between the ProTime and the hospital laboratory and its impact on therapeutic decision making for patients enrolled in the self-management arm of the main study. The information also will determine the need for a quality assurance program for users of the POC system. Materials and Methods Patients Patients 18 years or older were recruited by physicians and clinical pharmacists in the main, randomized, open-label, 8-month study. 8 Patients had to be receiving warfarin for at least 1 month before enrollment with planned anticoagulation for at least 1 year to a target INR of 2.0 to 3.0 or 2.5 to 3.5. Exclusion criteria were known hypercoagulable disorder, mental incompetence, language barrier, or inability to attend training sessions. Study Design The study was approved by the University of British Columbia Clinical Research Ethics Board (Vancouver), and written informed consent was obtained from all patients. By using a computer-generated randomization code, eligible patients were randomized in varied blocks of 10 to warfarin selfmanagement or to usual care by their primary care physician. Patients randomized to the self-management arm received training lasting 3 to 5 hours divided in at least 2 separate visits. On the first visit, they were taught how to use the ProTime monitor and shown an instructional video. The technical features of the ProTime system have been described. 9 In brief, patients incise the finger and collect a few drops of capillary blood using the Tenderlett Plus blood collection device (International Technidyne Corporation). The collected blood is analyzed in triplicate simultaneously with 2 levels of controls. The international sensitivity index of the thromboplastin reagent contained in the disposable cuvettes is 1.0. The ProTime instrument and reagent strip are precalibrated, and no additional calibration is required. The instrument has multiple built-in controls to ensure proper instrument function, reagent integrity, and user technique. The monitor reports numeric results for INR values of 0.8 to 10.0, and an error message is displayed for INR results exceeding 10.0. On the first and last days of the 8-month study period, patients were required to determine their INR with the ProTime monitor and concurrently at our hospital laboratory within 1 hour to verify concordance. Venous samples were analyzed at the laboratory with the BCS Coagulometer (Dade Behring, Marburg, Germany). The thromboplastin reagent (Dade Innovin, Dade Behring) used to measure the prothrombin time has an instrument-specific international sensitivity index of 0.94. The laboratory performs prothrombin time assays using thromboplastin standardized against a World Health Organization (WHO) reference standard. Paired INR measurements that differed by 1 unit or greater were repeated, if possible, and the ProTime monitor was replaced if the difference in INR persisted or if repeated testing was not done. Data Analysis Clinical agreement between the ProTime and laboratory INR values was defined as both INR measurements falling within or outside the target INR range and was determined using the κ statistic. These cutoffs were selected such that therapeutic decision making would be unaffected. Numeric agreement of the ProTime and laboratory INR values was evaluated using correlation coefficient analysis and the mean difference for all paired INR determinations. A mean difference of 0.5 INR units or less was used to establish concordance. 5,9 Bias was evaluated graphically by plotting the difference of ProTime and laboratory INR against laboratory INR values. Results There were 114 expected paired INR values from 57 patients. Two patients had their INR measurements done at a community laboratory and were excluded from analysis. Of the remaining 110 expected paired INR values (55 patients), 2 initial INR measurements were done at a community laboratory and 17 final INR values were not available owing to early study withdrawal (4 patients) or lack of hospital INR measurement (13 patients), leaving 91 evaluable paired INR measurements. The baseline demographics of the 55 patients are summarized in Table 1. Clinical agreement in which both the POC and laboratory INR values fell within or outside the patients therapeutic range occurred in 56 (62%) of 91 cases (κ = 0.35, fair strength of agreement). When INR mismatch occurred, the ProTime Downloaded from https://academic.oup.com/ajcp/article-abstract/123/2/184/1759382 Am J Clin Pathol 2005;123:184-188 185 185 185

Sunderji et al / CLINICAL IMPACT OF POC VS LABORATORY INR Table 1 Demographics of 55 Self-Management Patients * Characteristic Value 8.00 7.00 6.00 Male 39 (71) Mean age (range), y 55.8 (20-79) Indication for warfarin Mechanical valve 37 (67) Atrial fibrillation 14 (25) Venous thromboembolism 2 (4) Other 2 (4) Target INR range INR 2.5-3.5 38 (69) INR 2-3 17 (31) INR, international normalized ratio. * Data are given as number (percentage) unless otherwise indicated. ProTime INR 5.00 4.00 3.00 2.00 1.00 0.00 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 Lab INR Figure 1 Correlation of ProTime and laboratory international normalized ratio (INR) values. Equation for the regression is y = 0.62x + 0.92. For proprietary information, see the text. INR was in target range in 22 (63%) of 35 cases, whereas the laboratory INR was subtherapeutic. Of the 35 paired INR values that failed to meet clinical agreement criteria, 19 (54%) were within 0.5 INR units of each other. In only 1 case was there contradiction in INR results such that the POC INR was above and the laboratory INR was below the therapeutic range. The correlation coefficient for the ProTime vs the laboratory INR was 0.62 Figure 1. The mean (SD) difference in INR between POC and laboratory measurements was 0.44 (0.61). Overall, 69 (76%) of 91 paired INR values were within 0.5 INR units of each other, and 78 (86%) of 91 were within 0.7 INR units. Of all 98 paired INR measurements obtained from hospital and community laboratories, 5 pairs differed by 1 or more INR units, 3 at study initiation resulting in replacement of the POC monitor and 2 at study end. A fourth POC monitor was replaced partway through the study when a patient observed unexpectedly high INR results by the ProTime on 2 separate occasions. Repeated INR testing at the patient s community laboratory after several hours confirmed that the earlier ProTime INR was higher by 1 unit or more on both occasions. Of the 4 POC monitors that were replaced, 3 were returned to the manufacturer for repair, recalibration, or both. The fourth instrument demonstrated acceptable concordance on repeated testing and was retained for future use. The remaining 2 POC monitors that showed disagreement at study end subsequently showed acceptable agreement and were not returned to the manufacturer. A description of all paired INR values that differed by 1 or more units and the course of management is given in Table 2. The distribution of the absolute differences between ProTime and laboratory INR values in relation to the laboratory Table 2 Description of Paired INR Values Differing by One or More Units Timing of INR/Paired Absolute INR Values Difference Repeated INR * Comments Study initiation 7.25 (POC); 2.28 (lab) 4.97 ND POC monitor replaced 4.09 (POC); 3.09 (lab) 1.00 3.13 (POC); 3.20 (lab) POC monitor replaced per patient preference 2.73 (POC); 4.20 (lab) 1.47 2.91 (POC); 5.00 (lab) Repeated INR third time with POC and hospital lab next day were 2.72 and 2.50, respectively; POC monitor replaced Study end 3.13 (POC); 2.08 (lab) 1.05 ND Patient resumed usual care by family physician and lab INR testing 2.61 (POC); 5.68 (lab) 3.07 2.37 (POC); 1.89 (lab) Erroneous lab INR of 5.68 owing to inadequate blood sample During study 6.03 (POC) (first episode) 3.38 (POC); 2.50 (lab) Both episodes reported by same patient; all repeated INR measured 5.53 (POC) (second episode) 3.75 (POC); 3.60 (lab) on same day; POC monitor replaced after second episode of INR mismatch INR, international normalized ratio; lab, laboratory; ND, not done; POC, point-of-care. * Paired INR tests were repeated the next day except in the last case. INR performed at a community lab. Unexpectedly high routine INR result partway through study. 186 Am J Clin Pathol 2005;123:184-188 Downloaded 186 from https://academic.oup.com/ajcp/article-abstract/123/2/184/1759382

Coagulation and Transfusion Medicine / ORIGINAL ARTICLE INR value is shown in Figure 2. The values seem to be centered near the horizontal line of agreement between POC and laboratory INR values, although the ProTime INR results more often were higher than laboratory values. Discussion Our results show that there is acceptable agreement between INR values obtained by patients using the ProTime monitor and the INR measurement repeated within 1 hour at our hospital laboratory. We observed some intrinsic noise in the system, emphasizing the need for a quality assurance program for optimal use of POC systems. Of 91 paired INR measurements, 56 (62%) were in agreement such that the therapeutic decision of warfarin dosage adjustment was unaffected by the test system used to measure the INR. Although we did not observe perfect clinical agreement, the overall differences in INR between paired values were small and within accepted clinical decision limits. 5,10,11 This is indicated by 76% (69/91) and 86% (78/91) of paired values measuring within 0.5 and 0.7 INR units of each other, respectively. Our results compare favorably with those of others who have evaluated the accuracy of INR self-testing. 11-13 In one study, blood was obtained by finger prick from 212 patients by health care professionals for analysis using the ProTime and compared with venous blood tested at the local laboratory. 12 Similar to our results, clinical agreement was observed in 66% of cases. In a multicenter study of hospital-based anticoagulation centers, 82 trained patients self-tested their INRs using ProTime and had repeated INR measurements within 3 hours at a central reference laboratory. 11 Of a total of 431 specimens, 66% of self-tested INR values matched the reference laboratory ProTime-Lab INR 6.00 5.00 4.00 3.00 2.00 1.00 0.00 1.00 2.00 3.00 4.00 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 Lab INR Figure 2 Difference between ProTime and laboratory international normalized ratio (INR) vs laboratory INR. The horizontal line at zero represents perfect agreement between the 2 test systems. For proprietary information, see the text. result. These authors also demonstrated that differences between paired INR values were small, with a systematic error of approximately 0.2 INR for an INR range of 2.0 to 5.0. Thus, although achievement of higher rates of clinical agreement is desirable, the overall small differences in INR from the different test systems suggest a minimal clinical impact of INR mismatch on anticoagulation therapy. Of note, there were no complications of major bleeding or thrombosis in the selfmanagement arm of our study. The accuracy of INR values tested from identical samples across test systems has been challenged owing to differences in reagent-equipment combinations. 6,14 Kaatz et al 6 compared INR determinations using 2 POC monitors (not ProTime) and 4 clinical laboratories against a criterion WHO standard. The authors reported that INR results from 2 laboratories using sensitive thromboplastin had good agreement with the criterion, whereas INR values determined from less sensitive reagents at the remaining 2 laboratories had poor agreement. Clinical agreement of both POC systems was between these extremes at 77% and 78%. The authors concluded that large interlaboratory variation in INR results could occur, indicating the need for cautious interpretation of INR tested with POC against conventional laboratory methods. This might explain the poor INR agreement between various POC systems and laboratory methods reported in some studies. 15,16 It has been recommended that for reliable measures of accuracy, POC instruments ideally should be tested against a criterion standard with a manual technique using the WHO international reference thromboplastin as described previously. 6,17 In the multicenter study, patient-generated INR values using ProTime were equivalent to INR values determined at a central reference laboratory that originally calibrated its assay against the WHO standard method. 11 The ProTime monitor has been shown to correlate well with laboratory testing of INR, with correlation coefficients on the order of 0.86 to 0.93. 11,15,16,18 In comparison, we calculated a lower correlation coefficient of 0.62 for the ProTime INR vs the INR determined by our hospital laboratory. Our analysis of 91 paired INR values included 5 pairs that differed by at least 1.0 unit, which could have skewed the overall correlation analysis. Of note, correlation analyses are useful in demonstrating similarity of 2 tests, but they do not provide meaningful assessment of agreement. Despite a suboptimal correlation coefficient, we observed acceptable agreement with three quarters of paired INRs measuring within 0.5 units of each other. Some studies have shown that ProTime overestimates the INR with biases exceeding 0.5 INR units. 15,16 Similar to the large, multicenter study discussed earlier, 11 we did not detect any significant bias. Our mean difference in INR values between test systems was 0.44. Recalculation of this parameter after removing 2 major outliers (4.97 and 3.07, Table 2) that were due to a faulty POC monitor or laboratory error resulted in a smaller mean difference of 0.36. Downloaded from https://academic.oup.com/ajcp/article-abstract/123/2/184/1759382 Am J Clin Pathol 2005;123:184-188 187 187 187

Sunderji et al / CLINICAL IMPACT OF POC VS LABORATORY INR In our study, 6 paired measurements differed by 1 or more INR units. In 1 case in which the POC monitor generated an INR of 7.25 with a corresponding laboratory value of 2.28, the concordance failure was attributed to ProTime malfunction. It is not clear why the instrument generated such an elevated INR result despite its internal quality control systems designed to display an error message in this situation. Of note, erroneous INR results also can occur as a result of laboratory error, as was demonstrated in one case and possibly in a second case. The small failure rate in INR concordance between test systems observed in our study could be clinically important and emphasizes the need for external quality control systems. Our criterion for retesting using an INR difference of 1 unit or more between test systems was arbitrary for safety reasons; clinicians may select different criteria for assessing instrument failure. Our study was limited because it was not powered to determine the impact of INR disagreement on clinical outcomes. As with laboratory INR testing, patients who plan to self-test should be trained to recognize unusual POC INR results so that timely action can be taken to prevent inappropriate dosage adjustments. In our study, patients were required to call the study investigator if their POC INR exceeded 4.5 or 5.0, depending on their target INR range, to allow assessment of the need for laboratory INR measurement to confirm the POC result. 8 INRs tested using the ProTime POC monitor showed acceptable agreement with our hospital laboratory INR. Owing to the small failure rate in INR concordance, it is recommended that POC systems be tested periodically (eg, every 6 months 5 ) against a laboratory method to monitor performance and that patients be alert for unexpected INR values. Because errors in INR testing can occur with POC and laboratory systems, discrepancies in INR results should be investigated for optimal management of anticoagulation. From 1 Cardiology and 2 Nephrology, Pharmaceutical Sciences Clinical Service Unit, 3 Coronary Care Unit and Echocardiography Laboratory, 4 Department of Laboratory Medicine, 5 Center for Clinical Epidemiology and Evaluation, and 6 Cardiology, Vancouver General Hospital; 7 Faculty of Pharmaceutical Sciences, 8 Post-Graduate Cardiology Training Program, 9 Faculty of Medicine, and 10 Faculty of Pathology and Laboratory Medicine, University of British Columbia, Vancouver; and 11 Programs- Seniors, Evergreen House, North Vancouver, Canada. Supported by a Grant-in-Aid from the Heart and Stroke Foundation of British Columbia and Yukon, Vancouver, the Vancouver General Hospital Interdisciplinary Research Grant, and International Technidyne Corporation, Edison, NJ. Address reprint requests to Dr Sunderji: Pharmaceutical Sciences CSU, Vancouver General Hospital, 855 W 12th Ave, Vancouver, BC, Canada, V5Z 1M9. Acknowledgment: We thank Anar Dossa, BSc(Pharm), staff development coordinator, Pharmaceutical Sciences Clinical Service Unit, Vancouver General Hospital, for assistance with the conduct of this study and data management. References 1. Kortke H, Korfer R. International normalized ratio selfmanagement after mechanical heart valve replacement: is an early start advantageous? Ann Thorac Surg. 2001;72:44-48. 2. Sawicki PT. A structured teaching and self-management program for patients receiving oral anticoagulation: a randomized controlled trial. JAMA. 1999;281:145-150. 3. Cromheecke ME, Levi M, Colly LP, et al. Oral anticoagulation self-management and management by a specialist anticoagulation clinic: a randomised cross-over comparison. Lancet. 2000;356:97-102. 4. Horstkotte D, Piper C, Wiemer M, et al. Improvement of prognosis by home prothrombin estimation in patients with life long anticoagulation therapy [abstract]. Eur Heart J. 1996;17(suppl):S230. 5. Fitzmaurice DA, Machin SJ, on behalf of the British Society of Haematology Task Force for Haemostasis and Thrombosis. Recommendations for patients undertaking self-management of oral anticoagulation. BMJ. 2001;323:985-989. 6. Kaatz SS, White RH, Hill J, et al. Accuracy of laboratory and portable monitor international normalized ratio determinations. Arch Intern Med. 1995;155:1861-1867. 7. Sunderji R, Fung A, Gin K, et al. Patient self-management of oral anticoagulation: a review. Can J Cardiol. 2003;19:931-935. 8. Sunderji R, Gin K, Shalansky K, et al. A randomized trial of patient self-managed versus physician-managed oral anticoagulation. Can J Cardiol. 2004;20:1117-1123. 9. Sunderji R, Campbell L, Shalansky K, et al. Outpatient selfmanagement of warfarin therapy: a pilot study. Pharmacotherapy. 1999;19:787-793. 10. Lassen JF, Brandslund I, Antonsen S. International normalized ratio for prothrombin times in patients taking oral anticoagulants: critical difference and probability of significant change in consecutive measures. Clin Chem. 1995;42:444-447. 11. Oral Anticoagulation Monitoring Study Group. Prothrombin measurement using a patient self-testing system. Am J Clin Pathol. 2001;115:280-287. 12. Shermock KM, Bragg L, Connor JT, et al. Differences in warfarin dosing decisions based on international normalized ratio measurements with two point-of-care testing devices and a reference laboratory measurement. Pharmacotherapy. 2002;22:1397-1404. 13. Pierce MT, Crain L, Smith J, et al. Point-of-care versus laboratory measurement of the international normalized ratio. Am J Health Syst Pharm. 2000;57:2271-2274. 14. Lind SE, Pearce LA, Feinberg WM, et al, for the SPAF Investigators. Clinically significant differences in the international normalized ratio measured with reagents of different sensitivities. Blood Coagul Fibrinolysis. 1999;10:215-217. 15. Chapman DC, Stephens MA, Hamann GL, et al. Accuracy, clinical correlation, and patient acceptance of two handheld prothrombin time monitoring devices in the ambulatory setting. Ann Pharmacother. 1999;33:775-780. 16. Reed C, Rickman H. Accuracy of international normalized ratio determined by portable whole-blood coagulation monitor versus a central laboratory. Am J Health Syst Pharm. 1999;56:1619-1623. 17. Ansell J, Hirsh J, Dalen J, et al. Managing oral anticoagulant therapy. Chest. 2001;119(suppl):S22-S38. 18. Oral Anticoagulation Monitoring Study Group. Point-of-care prothrombin time measurement for professional and patient self-testing use: a multicenter clinical experience. Am J Clin Pathol. 2001;115:288-296. 188 Am J Clin Pathol 2005;123:184-188 Downloaded 188 from https://academic.oup.com/ajcp/article-abstract/123/2/184/1759382