Patients with left ventricular assist device (LVAD) require warfarin to maintain adequate anticoagulation. The two major components required for a clot to occur are fibrin and thrombin. Warfarin has its effect by inhibiting synthesis of clotting factors 2 (thrombin), 7, 9, and 10 and regulatory factor proteins C, S, and Z. International normalized ratio (INR) blood draws measure effects of warfarin and are maintained within tight goal ranges to reduce risks of bleeding and clotting sequelae. For this reason, INR measurements are routinely completed for LVAD patients throughout the continuum of their support with a LVAD.
The Clinical Laboratory Improvement Amendments in 1988 allowed the clinical laboratory industry to develop point-of-care (POC) testing. Since this time, POC-INR measurements have been validated in a variety of patient populations. An early study comparing plasma and POC-INR measurements by the University of Texas Health Science Center reported that POC results were less variable and more repeatable.1 A later study at the University of Maryland demonstrated significant correlation between 41 plasma and POC-INR readings in hypertensive VAD patients.2 Mayo Clinic reviewed hemodialysis patients with anemia and found good correlation between POC-INR and plasma INR (P-INR) measures without anemia or dialysis influencing this result.3
Cork University Hospital in Ireland specifically addressed concordance of INR readings within particular goal ranges. Their results suggest that the best correlation of plasma and POC INR measurements are for INR values between 2.0 and 3.5 with increased discordance as INR rose above 3.5.4 Patients with LVAD typically maintain an INR goal range of 2.0–3.0 with some exceptions for demonstrated histories of bleeding or thrombosis. Cork Hospital further demonstrated that POC and P-INR measurements were within 0.5 INR units of each other 87% of the time.4
A recent Johns Hopkins publication studied POC and P-INRs in a pharmacist-managed single-center cohort. They found a statistically significant difference for time in therapeutic range in the POC-INR group, which was higher compared with the P-INR group. Importantly, there was no difference identified in the bleeding rate or thrombotic events when comparing both groups.5
These and many other studies suggest that POC-INR does not significantly differ from plasma INR values in other patient populations, although this has not been validated in a multicenter LVAD study. We sought to determine whether POC-INR and P-INR values differ significantly in the population with LVAD in a larger international study.
Materials and Methods
This retrospective, multicenter, nonrandomized study sought to evaluate correlation of plasma and POC INR measurements in an international LVAD population. The primary end-point was to assess concordance or discordance in POC and P-INR values. The secondary end-point was to determine whether comorbid conditions influenced validity of this correlation if a correlation was identified.
Data were abstracted for basic demographics, coagulopathy history, low-molecular-weight heparin use at the time of blood collection, gastrointestinal bleeding history, thrombosis or hemolysis history, specific INR machine utilized with reagent if available, and dates and times of INR collections. Two INR collections were obtained for every patient consisting of a P-INR measurement and a POC-INR measurement. Each individual center reported data to the primary center where data was compiled and subsequently analyzed for statistical significance of the findings. The institutional review board (IRB) at the primary center approved the study, and each participating center completed the necessary IRB requirements as required.
Patient self-testing of INR with a home monitoring device was performed with Food and Drug Administration–approved devices: CoaguCheck (Roche, Roche Diagnostics, Mannheim, Germany) and INRatio 2 (Alere Inc., South Waltham, MA).
Plasma International Normalized Ratio Testing
Venous blood samples were obtained during follow-up visits using standard technique, in a 3.2% sodium citrate vacutainer and sent to the each centers core laboratory for analysis on a STA-R system (Diagnostica Stago S.A.S, Asnim, D sur Seine, France) or as per their center-specific protocol.
Statistical analysis was performed by using SPSS for Windows Release 23.0.0 (SPSS Inc., Chicago, IL). Characteristics of patients are reported as mean ± standard deviation, compared with the Fischer exact test for categorical and the Mann–Whitney U test for continuous variables, respectively. The relationship between POC and plasma INRs was assessed using the Pearson correlation coefficient and Bland–Altman plot. Normal distribution was assessed by the Shapiro–Wilk test. In all analyses, a value of P < 0.05 was considered as statistically significant.
We specifically reviewed patients with permanent LVAD implantation from October 2008 to November 2016 at seven institutions internationally. Our cohort comprised 279 paired POC-INR and P-INR checks obtained on average 630 ± 598 days postimplant. Patients were only included if the POC and plasma INR checks were completed on the same day (Figure 1). 22.4% of the patients were on additional low-molecular-weight heparin (enoxaparin) therapy when the INR checks were performed. Presently available durable LVADs were included comprising HeartWare HVAD (57.3%, n = 160), HeartMate II (34.8%, n = 97), HeartMate III (7.9%, n = 22), and unknown devices (3.1%, n = 9). Our population averaged 57.9 years of age and 86.7% were of male gender. Comorbidities included coagulopathy history in 3.6%, bleeding history in 31.0%, and thrombosis history in 20.1% of the included LVAD patients (Table 1).
We found no statistically significant difference (0.066 ± 0.38) between POC-INR and P-INR values with a correlation coefficient of r = 0.89 (r2 = 0.79; p = 0.001), which was our primary end-point (Figure 2). Our secondary end-point sought to assess whether comorbidities or time after LVAD implant influenced INR correlation for plasma and POC-INR measurements. There was no statistically significant difference when comparing paired INR values and patients with (−0.06 ± 0.32) and without (0.005 ± 0.24) gastrointestinal bleeding (p = 0.22), and there was no significant difference in the time between samples, respectively (1:25 vs. 1:35 hours). Comparison of paired INR values and time after implant to INR check (p = 0.43), age (p = 0.12), known coagulopathy (p = 0.12), or thrombosis history (p = 0.34) also did not reveal any statistically significant difference.
Time Between International Normalized Ratio Measurements and Relationship to Outcomes
The mean time between POC-INR and P-INR was 2 hours and 56 minutes with a range of 0 minutes to 14 hours 27 minutes. International normalized ratio accuracy correlated weakly with the time between INR measurements, r = 0.25 (r2 = 0.063; p < 0.001; Figure 3). When the time difference was less than 4 hours, the difference between INR pairs was significantly lower than measurements greater than 8 hours (0.038 vs. 0.32; p = 0.006; Figure 4).
Therapeutic, Subtherapeutic, and Supratherapeutic Bias Analysis
Paired INR valued were divided into three groups: INR <2.0, INR 2.0–3.0, and INR >3.0 to determine whether our data corroborated more variable performance for supratherapeutic and subtherapeutic INR levels. We performed a Bland–Altman plot (Figure 5) to assess the direction of bias and found a higher bias in the supratherapeutic range (plasma INR >3.0) compared with the therapeutic range (INR 2.0–3.0) and subtherapeutic range. The mean INR difference was 0.066 ± 0.24 in the subtherapeutic range (<2.0), 0.016 ± 0.32 in the therapeutic range (2.0–3.0), and 0.366 ± 0.52 in the subtherapeutic range. Of the 279 P-INR values, 21.3% were subtherapeutic (INR <2.0), 59.4% within the target range of 2.0–3.0, and 19.3% were supratherapeutic (INR >3.0). Of the 279 POC-INR values, 18.5% were subtherapeutic (INR <2.0), 66.8% were within the target range of 2.0–3.0, and 14.7% were supratherapeutic (INR >3.0).
Potential Benefit of POC INR
Comorbidities commonly present in LVAD patients including known coagulopathy, bleeding history, or thrombosis history did not demonstrate any statistically significant difference between plasma and POC-INR measurements in our patient population. Furthermore, there was no statistically significant difference between plasma and POC-INR measurements and the time after initial LVAD implant to the time of INR measurement. Patients with LVAD require frequent INR measurements to ensure that they remain appropriately anticoagulated, but patients may live long distance from a local laboratory able to perform plasma INR measurements, which could limit compliance. The ability and ease of performing a POC-INR measurement in the home environment could aid in compliance, patient satisfaction, and improved quality of life. Some LVAD patients will require more frequent INR checks, which can negatively impact their quality of life. Transition to POC-INR measurements instead of P-INR measurements can alleviate some of the burden associated with frequent trips to a laboratory for blood draws. Home INR monitoring has the potential to improve the percentage of time within therapeutic INR range in patients who require long-term warfarin therapy.6 Whether home INR monitoring will improve the long-term outcomes in patients with permanent LVADs remains unclear.
International Normalized Ratio Accuracy
Patients in our cohort who were provided POC-INR machines were instructed by trained providers on the guidelines for proper technique to obtain a valid INR reading to reduce possibility of inaccurate INR values. Adherence to these guidelines is necessary to avoid variable techniques influencing INR measurements. Patients should be instructed to follow the instructions for their particular POC machine to obtain an accurate reading; otherwise, actions such as squeezing the finger can prematurely activate clotting and prompt an inaccurate INR reading. It is important to note that POC-INR values outside of the LVAD patient’s INR goal will decrease in accuracy as they fall outside this range.4 Specifically, POC-INR may be less accurate when readings fall below 2.0 or above 3.5.4 If POC-INR values are outside this range, validating the POC-INR reading with a P-INR check could be considered. This can be supported by our results showing the influence of subtherapeutic (<2.0), therapeutic (2.0–3.0), and supratherapeutic (>3.0) P-INRs toward a higher bias in the POC-INR (Figure 5) once the INR is outside the target range.
Concomitant Enoxaparin Support
In contrast to previous studies7 comparing POC and plasma INR, only patients with a stable INR were included in the study cohort; thus, patients on low-molecular-weight heparin (enoxaparin) were excluded. In our study, no statistically significant difference was identified with paired INR values and patients with (0.067 ± 0.21) and without (0.011 ± 0.27) enoxaparin support (p = 0.10), and there was not significant differences in the time between samples, respectively (2:04 vs. 1:52 hours). Enoxaparin is utilized for bridging purposes in the LVAD population; thus, the ability to consider a POC-INR in place of a P-INR measurement can be useful for patients requiring enoxaparin.
Although to date, this is the largest and first multicenter international study in a population with LVAD comparing POC and plasma INR, there are some limitations in our study. This retrospective study was performed nonrandomized and assessed the correlation of data pairs only at single time points. In addition, because of the multicenter design of this study, different phlebotomists obtained venous blood samples and blood was analyzed in each centers’ respective core laboratory per their respective protocols. A prospective study measuring baseline data on antiplatelet use and correlation with INR measurements could be valuable. Each center in our study followed center-specific guidelines regarding antiplatelet therapy but because of the retrospective nature of our study, this information was not included in our analysis. The lack of observed difference in POC and P-INR values may be attributed to the behavior in the therapeutic range with only 0.016 bias as nearly 60% of the INR measurements were from this area, but it is also possible that the lack of power based on the relatively modest sample size could have limited the observed differences.
The further apart timewise that P-INR and PC-INR checks are, the greater the variability. In our cohort of 279, there were only seven measurements with an INR difference between P-INR and POC-INR that was higher than 1.0. All seven measurements were INR pairs where the time difference between POC and P-INR samples was greater than 8 hours. Five of the measured INR pairs were supratherapeutic plasma INR values showing underestimation by POC and two of the measured INR pairs reflected an overestimation by the POC. Our data support previous studies4 suggesting interindividual differences in INR values are related not to the P-INR or POC-INR test but to the time elapsing between these two measurements.
In 145 of the 279 data pairs, the POC device utilized was known. There was a statistically significant difference (p < 0.001) in the INR difference between CoaguCheck (0.035 ± 0.27) and INRatio 2 (−0.147 ± 0.26) POC machines in our cohort. However, of the two POC devices used, there were 107 CoaguCheck devices compared with 38 INRatio 2 patients; thus, the different sample sizes could have led to this finding. Additionally, all INRatio 2 measurements were performed at the same time as the P-INR, whereas there was an average time difference of 2:30 hours between CoaguCheck and P-INR, which may have also led to the difference between devices.
Main Findings of the Study
Despite the known tendency to overestimate the INR as reported by Dionizovik-Dimanovski et al.7 and Hur et al.,8 this is the first large multicenter international study of most commonly used durable LVADs, comparing POC-INR and P-INR measurements and identifies no statistically significant difference between either method, particularly when measured within less than 4 hours of each other. For this reason, it is unlikely that obtaining a P-INR instead of a POC-INR measurement would influence the warfarin dosage or patient management. Rather, it is plausible that patients who struggle to obtain plasma INR checks because of distance from a laboratory may have earlier identification of potential bleeding or clotting concerns attributable to the ease of a POC-INR check obtained at home.
We found no statistically significant difference in comparison of paired INR values and time after implant to time of INR check, coagulopathy history, bleeding history, or thrombosis history. Despite the limitations of the study, we found an reasonable correlation (r = 0.89) and small bias (0.066) between POC and plasma INR; thus, POC testing offers a reasonable option to consider in LVAD patients and may increase compliance with INR checks while offering a more convenient patient option than traveling to a lab.
The authors would like to acknowledge The International Consortium for Circulatory Assist Clinicians (ICCAC) for the multicenter collaborative platform provided for research endeavors.
1. Bussey HI, Chiquette E, Bianco TM, et al. A statistical and clinical evaluation of fingerstick and routine laboratory prothrombin time measurements. Pharmacotherapy 1997.17: 861866.
2. Joshi A, Smith D, Arora M, Poston R. Anticoagulant monitoring in ventricular assist device patients: a feasibility study. Interact Cardiovasc Thorac Surg 2008.7: 10351038.
3. Hoel RW, Albright RC, Beyer LK, et al. Correlation of point-of-care international normalized ratio to laboratory international normalized ratio in hemodialysis patients taking warfarin. Clin J Am Soc Nephrol 2009.4: 99104.
4. Ryan F, O’Shea S, Byrne S. The reliability of point-of-care prothrombin time testing. A comparison of CoaguChek S and XS INR measurements with hospital laboratory monitoring. Int J Lab Hematol 2010.32(1 pt 1): e26e33.
5. Bishop MA, Streiff MB, Ensor CR, Tedford RJ, Russell SD, Ross PA. Pharmacist-managed international normalized ratio patient self-testing is associated with increased time in therapeutic range in patients with left ventricular assist devices at an academic medical center. ASAIO J 2014.60: 193198.
6. Völler H, Glatz J, Taborski U, Bernardo A, Dovifat C, Heidinger K. Self-management of oral anticoagulation in nonvalvular atrial fibrillation (SMAAF study). Z Kardiol 2005.94: 182186.
7. Dionizovik-Dimanovski M, Levin AP, Fried J, et al. Correlation between home INR
and core laboratory INR in patients supported with continuous-flow left ventricular assist devices. ASAIO J 2015.61: 386390.
8. Hur M, Kim H, Park CM, et al. Comparison of international normalized ratio measurement between CoaguCheck XS Plus and STA-R coagulation analyzers. Biomed Res Int 2013.2013: 213109.