Brief Report

Point-of-care testing for coagulation studies in a stroke protocol: a time-saving innovation^?,??

Michael J. Drescher MD ^a,? , Andrea Spence BS ^b, Darryl Rockwell RT ^c,

Ilene Staff PhD ^d, Alan J. Smally MD ^a

^aDivision of Emergency Medicine, Hartford Hospital and University of Connecticut, Hartford, CT 06102, USA ^bDepartment of Pathology and Laboratory Medicine, Hartford Hospital Hartford, CT 06102, USA ^cDepartment of Respiratory Care, Hartford Hospital, Hartford, CT 06102, USA

^dResearch Program Hartford Hospital, Hartford, CT 06102, USA

Received 24 June 2009; revised 11 September 2009; accepted 17 September 2009

Abstract

Study objectives: Time counts in thrombolytic therapy for stroke. An International normalized ratio greater than 1.7 may preclude its use. We studied whether the use of point-of-care testing for INR in the emergency department (ED) may substitute for the same test done in the central hospital laboratory, thereby reducing time to treatment.

Methods: We performed a prospective observational study comparing a POCT analysis of INR (i-STAT-1; Abbott Inc, Abbott Park, Ill) with a simultaneously drawn sample sent to the central laboratory. We tested a convenience sample of adult patients taking warfarin who presented to the ED of a tertiary teaching hospital.

Results: Thirty-two patients were enrolled. A receiver operator curve analysis was performed. Sensitivity and specificity were calculated for laboratory INR cutoff of 1.7. The area under the curve was 0.979 (95% confidence interval [CI], 0.843-0.991). When POCT INR was 2.1, the sensitivity for laboratory INR being higher than 1.7 was 100% (CI, 62.9%-100.0%), and the specificity was 90.5 (CI, 69.6-98.5). When POCT INR was 1.8, the specificity for laboratory INR being lower than 1.7 was 100% (CI, 83.7%-100%), and the sensitivity was 62.5% (CI, 24.7%-91.0%). The regression coefficient (r) value was 0.9648.

Conclusion: Correlation of POCT INR with that of the central laboratory and receiver operator curve characteristics are excellent. In general, POCT INR is about 0.3 higher than the laboratory INR. This is not generally of clinical importance, but when using cutoff of 1.7 (central laboratory), it may be. We describe a 3-tiered system for use of POCT INR in determining use of tissue-type plasminogen activator.

(C) 2011

^? An i-STAT point of care analyzer was provided by Abbott Inc for the purposes of this study.

^?? Presented at ACEP Research Forum, Scientific Assembly, Seattle, Wash, Oct 2007 (Ann Emerg Med 50(3):S25 Supplement Sept 2007).

* Corresponding author.

E-mail address: [email protected] (M.J. Drescher).

0735-6757/$ - see front matter (C) 2011 doi:10.1016/j.ajem.2009.09.020

POCT for coagulation studies in a stroke protocol

Introduction

Treatment with a thrombolytic has been shown to be effective treatment of acute thrombotic stroke if certain criteria are met. Among those are an International normalized ratio of less than or equal to 1.7 [1]. An even lower cutoff of INR of 1.5 has recently been suggested by one consensus of expert opinion [2]. There is a time window of 3 hours within which the therapy must be started to ensure a favorable risk/benefit profile. Further- more, it has been shown that the earlier thrombolysis is achieved within the 3 hours, the higher the chance of success [3]. In patients who are anticoagulated and present with acute stroke, verification of INR in the central hospital laboratory may become the rate-limiting step in initiating treatment. Indeed, it may prevent certain patients from achieving the 3-hour therapeutic window who would otherwise be candidates for thrombolysis.

We undertook this study to examine the correlation of INR done at the bedside using the i-STAT system (Abbott Inc, Abbott Park, Ill) and INR done at the centralized hospital laboratory.

Methods

We performed a prospective observational study com- paring a point-of-care testing (POCT) analysis of INR performed in the emergency department (ED) (i-STAT-1, Abbott Inc) with testing of a simultaneously drawn sample sent to the central laboratory on the Dade Behring BCS analyzer. We drew venous blood from a convenience sample of adult patients who were taking warfarin and who presented to the ED of a tertiary teaching hospital with a census of 85 000 annual visits. Blood was collected from each patient in a non-anticoagulated syringe. Approximately 20 uL of blood was dispensed into each of 2 i-STAT PT INR cartridges, and a whole-blood INR was performed in duplicate, side by side, on 2 bedside i-STAT-1 Analyzers [4]. The i-STAT PT/INR measurement was performed with a recombinant tissue factor reagent with an ISI of 1.05. The remainder of the sample was dispensed into a citrated blue top tube and sent to the central laboratory where a plasma INR was performed in duplicate on the laboratory’s Dade Behring BCS analyzer [5]. The central laboratory INR was performed with a rabbit brain thromboplastin with an ISI of 1.93 at the time of this study [6]. A precision study was also performed using 2 lyophilized control samples (known INR values of 1.2 and 2.7). Twenty of each control sample were run for a 10-day period [5]. The study was approved by the hospital’s institutional review board.

A Linear regression analysis was performed on the average of the duplicate INR results paired with their respective laboratory INR results (Excel 2000 9.0.3821 SR-1; Micro- soft Corp). A receiver operator curve (ROC) analysis was

83

performed using a central laboratory cut point of INR of

1.7 on the first of the POCT INR results (MedCalc Version 8.2, Schoonjians, F., 1993-2006, www.medcalc.be).

Results

Thirty-two ED patients with various diagnoses currently being treated with warfarin were enrolled in a prospective PT/INR correlation study. Ages ranged from 19 to 92 years. Duplicate results for the POCT INR and central laboratory INR were each averaged on 28 of the 32 patients enrolled in the study; data from 4 patients could not be included, as their POCT INR was outside the analyzer’s reportable range (N8.0). The POCT INR ranged from 1.2 to 6.8. The linear regression analysis comparing the POCT with central laboratory results revealed that the regression coefficient (r) value was 0.9648 (Fig. 1). The y intercept was 0.276. The slope test was 0.07%, which was less than

2 times the overall coefficient of variation (CV) (3.5%) of the assay.

The paired data points were then analyzed to determine if the treatment decision might have been different had the POCT INR been used in lieu of the central laboratory INR using cutoff for treatment of 1.7 (Table 1). Five of the patients would have been denied tissue plasmi- nogen activator (tPA) based on the POCT INR result when the respective central laboratory INR indicated treatment was appropriate. No patient with a laboratory INR of higher than 1.7 would have received tPA based on the POCT result.

The (ROC) analysis revealed an area under the curve of 0.979. When the POCT INR was 2.1, the sensitivity for laboratory INR being higher than 1.7 was 100% (CI, 62.9%-100.0%), and the specificity was 90.5 (CI, 69.6-98.5). When POCT INR was 1.8, the specificity for laboratory INR being lower than 1.7 was 100% (CI, 83.7%-100%), and the sensitivity was 62.5% (CI, 24.7%-91.0%).

The mean, SD and CV of the level 1 control (known value, 1.2) were 1.2, 0.1, and 3.9%, respectively. The mean, SD, and CV for the level 2 control (known value, 2.65) were 2.4, 0.1, and 3.1%, respectively.

Fig. 1 Regression line of i-STAT whole-blood and laboratory plasma INR.

84 M.J. Drescher et al.

Table 1 Clinical differences in treatment based on i-STAT vs laboratory INR

Average i-STAT and laboratory INRs sorted ascendingly

i-STAT Laboratory i-STAT-

laboratory

Clinical difference

INR target <=1.7

Discussion

1.2	1.36	-0.16
1.5	1.18	0.32
1.7	1.24	0.46
1.8	1.59	0.21	Y
1.8	1.70	0.10	Y
2.0	1.62	0.38	Y
2.0	1.97	0.03
2.0	1.70	0.30	Y
2.1	2.28	-0.18
2.2	1.51	0.69	Y
2.2	2.28	-0.08
2.2	2.15	0.05
2.3	2.09	0.21
2.3	1.94	0.36
2.7	1.80	0.90
2.9	2.53	0.37
2.9	3.00	-0.10
3.1	2.70	0.40
3.1	3.03	0.07
3.2	2.95	0.25
3.3	2.67	0.63
3.3	3.22	0.08
3.7	3.35	0.35
3.9	3.59	0.31
4.2	4.26	-0.06
4.3	3.14	1.16
4.5	4.29	0.21
6.8	6.38	0.42
Average	difference	0.27
Y indicates yes.

• 1. Limitations

Our study evaluated INR values in patients taking warfarin but not actually having a stroke. Although this is not the true target population of the study, our study group served as a proxy. Our study sample was small, leading to wide confidence intervals around the levels of specificity and sensitivity. We take this into account in our clinical application by creating a 3-tiered rule, which allows for retesting of POCT values lying close to the treatment cutoff, at the central laboratory. Our criterion standard is the central laboratory INR value. As will be discussed, it is not certain that the central laboratory process is indeed superior to the POCT, although it is currently the standard on which Clinical decisions are made. In addition, without actually documenting the use of POCT in a stroke protocol, it is not clear what the real effect of this tool will be on increasing the number of appropriately treated patients or reducing time to treatment in this same group.

The correlation data of the POCT INR with that of the central laboratory were compelling in the usual therapeutic ranges for warfarin; there was, however, more scatter for subtherapeutic values. The ROC characteristics support the use of the i-STAT system for PT INR testing in the ED. The average difference between the POCT INR and the laboratory INR is 0.27, the POCT INR usually running higher. This is not generally of clinical importance for long- term INR monitoring, but when using cutoff of 1.7 (central laboratory) for decision making in patients with acute stroke, it may be. We developed a 3-tiered system for use of POCT INR in determining use of tPA when otherwise indicated for stroke in a patient taking warfarin:

• • • POCT INR <= 1.7--tPA may be given.
    • 1.7 b POCT INR b 2.4-send to central laboratory and await results.
    • POCT INR >= 2.4-withhold tPA.

This protocol allows us to safely administer Thrombolytic treatment of acute stroke in appropriate patients and withhold tPA in overly Anticoagulated patients without waiting for results from our central laboratory in most cases. In patients in whom the POCT is over, but close to, the cutoff of 1.7, we await central laboratory confirmation of INR. These patients are very likely to be ineligible for tPA by central laboratory as well, because INR is nearly always higher than by POCT. We do this to ensure that we are not withholding potentially beneficial treatment in the patient whose INR is near, but over, the cutoff by POCT. A recent study using a different point-of-care device came to a similar conclusion [7].

Thus, we have widened the middle tier of INR i-STAT values to be more confident that we will, at the margins, not act in conflict with our present criterion standard, the central laboratory INR. It is possible that the POCT INR reagent, a recombinant thromboplastin with a low ISI (1.05), vs the laboratory reagent, a rabbit thromboplastin with a higher ISI (1.93), may account for the variability in data. Studies have shown that there is improved inter- laboratory variation when both methodologies use recom- binant thromboplastin with a low ISI value [8]. Further studies are needed to clarify whether using a reagent with a higher ISI (in this case, the central laboratory) should remain the criterion standard. Future studies with larger numbers of participants, should they reproduce our findings, will allow for a narrowing of confidence intervals and may eliminate the need for a range of POCT values for which it is still necessary to await confirmation from the central laboratory.

The guidelines limiting the use of thrombolytics in patients who are anticoagulated are derived from the original study describing the usefulness of intravenous tPA in acute

POCT for coagulation studies in a stroke protocol

stroke. In this study, a prothrombin time greater than 15 seconds was used as an exclusion criterion. International normalized ratio was not reported [9]. Later guidelines referring to INR result from consensus of expert opinion [1,2]. Although in at least one case, this was done using a systematic methodology [2], emergency physicians and neurologists treating acute stroke should be aware that the clinical guidelines for treatment with tPA were extrapolated from an exclusion criterion for entry into the National Institute of Neurological Disorders and Stroke trial and not determined by clinical trials or risk benefit analysis. Furthermore, the National Institute of Neurological Dis- orders and Stroke exclusion criterion for anticoagulation was PT of 15 seconds and not any given INR. Because the trial was done at multiple sites with potentially different reagents, it is not clear that this exclusion criterion was equivalent from site to site. Clinicians should take into account this significant limitation of current guidelines. They will need to act in keeping with their best clinical judgment as to when to use tPA in the face of existing anticoagulation. Clinical logic would dictate, however, that the more a patient is anticoagulated, the higher the risk in using tPA. Using INR in our protocol, measurements, and calculations [10], we offer a method that will allow a reliable and quick method for assessing the degree of anticoagulation.

In conclusion, this study suggests that point-of-care testing of INR, when indicated, in patients who are candidates for thrombolytics, correlates well with central laboratory values. This has the potential to reduce the time to treatment in these patients.

85

References

1. Adams H, Adams R, DelZoppo G, Goldstein LB. Guidelines for the early management of patients with ischemic stroke: 2005 guidelines update, a scientific statement from the Stoke Council of American Heart Association/American Stroke Association. Stroke 2005;26(4): 916-23.
2. Dirks M, Niessen LW, Koudstaal PJ, Franke CL, van Oostenbrugge RJ, Dippel DW, Delphi panel on indications and contraindications for intravenous thrombolysis in acute ischaemic stroke. Intravenous thrombolysis in acute ischaemic stroke: from trial exclusion criteria to clinical contraindications. An international Delphi study. J Neurol Neurosurg Psychiatry 2007;78(7):685-9.
3. Marler JR, Tilley BC, Lu M, Brott TG, Lyden PC, Grotta JC. Early stroke treatment associated with better outcome: the NINDS rt-PA stroke study. Neurology 2000;55:1649-55.
4. i-STAT system manual, cartridge and test information, prothrombin time (PT/INR), Revision 2/12/2007
5. Clinical Laboratory Improvement Amendment, Subpart K. Quality systems for non-waived testing, Section 493.1253
6. Dade Thromboplastin C Plus package insert, Lot # 537014. Dade Behring; 2006.
7. Green TL, Mansoor A, Newcommon N, Stephenson C, Stewart E, Hill MD. Reliability of point-of-care testing of INR in acute stroke. Can J Neurol Sci 2008;35(3):348-51.
8. Schmitz LL, Olson SL, Shapiro RS, McCormick SR, Kubic VL. Failure to generate comparable international normalized ratio values using five different thromboplastin reagents in parallel studies of patients receiving warfarin. Clin Appl Thromb Hemost 1995;1(2):142-50.
9. The National Institute of Neurological Disorders and Stroke rt-PA Stroke Study Group. Tissue plasminogen activator for acute ischemic stroke. N Engl J Med 1995;333(24):1581-7.
10. Adcock DM, Duff S. Enhanced standardization of the international normalized ratio through the use of plasma calibrants: a concise review. Blood Coagul Fibrinolysis 2000;11:583-90.