Correspondence

Variation in the Prothrombin-Time Ratio during Oral Anticoagulation

N Engl J Med 1994; 330:509-510February 17, 1994

Article

To the Editor:

The international normalized ratio (INR) is a value calculated to help determine the intensity of anticoagulation therapy based on the results of prothrombin-time tests performed by laboratories using different reagents with different normal ranges. Eckman et al. (Sept. 2 issue)1 suggest that adoption of the INR will address the economic and clinical effect of insufficient or excessive anticoagulation. Where INRs are not available, they advocate providing the clinician with the international sensitivity index (ISI) of the thromboplastin reagent used, so that clinicians can independently calculate the INR for each prothrombin-time result (INR = [prothrombin time divided by mean of normal prothrombin-time range]^ISI).

Recent reports suggest that INRs are not all comparable2-5. Using the categories “low,” “low therapeutic,” “high therapeutic,” and “high,” one study reported a lack of concordance in INR results in 30 percent of paired samples measured at different laboratories3. A variety of factors contribute to this discordance, including inaccuracies in the ISI calibration by reagent manufacturers and the variable effects of different coagulation instruments on the ISI2-5. Indeed, it is now recommended that ISI values be instrument-specific3-5.

It appears that INRs have introduced confusion in some cases and a false sense of security in others. Until the limitations of the INR are more fully addressed and understood, clinicians may find that monitoring a patient's prothrombin time with measurements obtained from a single laboratory with which they are familiar is still a reasonable way to proceed.

Justine Meehan Carr, M.D.
Gary Horowitz, M.D.
Beth Israel Hospital, Boston, MA 02215

5 References

1
Eckman MH, Levine HJ, Pauker SG. Effect of laboratory variation in the prothrombin-time ratio on the results of oral anticoagulant therapy. N Engl J Med 1993;329:696-702
Full Text | Web of Science | Medline

2
Holland LJ, Lawrie AS, Hunt BJ. Evidence of discrepant commercial ISI assignment. Blood Coagul Fibrinolysis 1992;3:493-494
Web of Science | Medline

3
Ng VL, Levin J, Corash L, Gottfried EL. Failure of the International Normalized Ratio to generate consistent results within the local medical community. Am J Clin Pathol 1993;99:689-694
Web of Science | Medline

4
Hirsh J. Inadequate monitoring of warfarin dosage. Blood 1992;80:562-563
Web of Science | Medline

5
Swaim WR. Prothrombin time reporting and the International Normalized Ratio system: improvements are needed. Am J Clin Pathol 1993;99:653-655
Web of Science | Medline

To the Editor:

Eckman et al. suggest that the INR range for the prothrombin time in patients with prosthetic heart valves should be 2.5 to 3.5, with a goal of 3.0 for the “optimal intensity” of effectiveness. Furthermore, they note that the benefit of anticoagulation represents a trade-off between bleeding and thromboembolism. The safest INR for patients should be one at which the sum of the bleeding and the thromboembolic rates is at its nadir for a range of INR values. Accordingly, we solved the authors' equations for the total bleeding rate and the thromboembolic rate for INR values from 1.0 to 4.0. As they suggest, we calculated major bleeding to be 20 percent of the total bleeding rate. Since the authors state that “the consequences of thromboembolism are more severe than those of bleeding,” we have considered only major bleeding in our calculations. Thus, we summed the major bleeding rate and the thromboembolic rate for each INR (Table 1Table 1The INR and the Combined Risk of Major Bleeding and Thromboembolism.).

The nadir of the combined rate of major bleeding plus thromboembolism lies between the INR values of 2.3 and 2.4, not at a value of 3.0. Similarly, the quality of life and monetary costs for anticoagulation should be more favorable at an INR of 2.3 to 2.4. The combined risks of thromboembolic and hemorrhagic events as a function of the intensity of anticoagulant therapy should be lower at an INR of 2.35 than at an INR of 3.0. The recommended INR range for patients with prosthetic heart valves should be 1.85 to 2.85, or a value lower than that suggested by the authors.

Robert A. O'Reilly, M.D.
Stanford University School of Medicine, Stanford, CA 94305

Patrick J. Kearns, M.D.
Santa Clara Valley Medical Center, San Jose, CA 95128

To the Editor:

Eckman and his coauthors recommend that the INR be reported instead of the prothrombin-time ratio. However, both the INR and the prothrombin-time ratio are unsuitable if the thromboplastin reagents are sensitive to heparin1-3.

This is illustrated by the following experiment. Plasma, which was obtained from a patient two days after the first dose of coumarin, was supplemented with increasing amounts of heparin (Figure 1Figure 1Sensitivity to Heparin Shown by Two Thromboplastin Reagents.). The INR was then determined with the use of two commercially available thromboplastin reagents. Reagent B was insensitive to heparin up to a concentration of 1 U per milliliter. Reagent A, however, showed a heparin-dependent increase in the INR. In the absence of heparin, the INR indicated that anticoagulation with coumarin was inadequate. In the case of reagent A, a heparin concentration of less than 0.5 U per milliliter in the blood sample would have shown an INR value within the therapeutic range.

After an acute myocardial infarction or implantation of artificial prosthetic material (for example, stents or heart valves), patients often receive anticoagulation therapy with heparin followed by the long-term use of coumarin, which is initiated in an overlapping fashion. It is important that anticoagulation be monitored with a thromboplastin that is insensitive to heparin, since false values of the prothrombin-time ratio and the INR may suggest sufficient anticoagulation and heparin therapy may therefore be stopped too early4.

Peter Schuff-Werner, M.D.
Ekkehard Schutz, M.D.
B.-D. Gonska, M.D.
University Clinics, 37075 Gottingen, Germany

4 References

1
Moser KM, Hajjar GC. Effect of heparin on the one-stage prothrombin time: source of artifactual “resistance” to prothrombinopenic therapy. Ann Intern Med 1967;66:1207-1213
Web of Science | Medline

2
Hellstern P, Anders CU, Oberfrank K, Faller B, Speatt A. Heparin sensitivity of thromboplastins for the determination of the prothrombin time. Klin Lab 1992;38:183-8.

3
Salzman EW, Deykin D, Shapiro RM, Rosenberg R. Management of heparin therapy: controlled prospective trial. N Engl J Med 1975;292:1046-1050
Full Text | Web of Science | Medline

4
Bussey HI, Force RW, Bianco TM, Leonard AD. Reliance on prothrombin time ratios causes significant errors in anticoagulation therapy. Arch Intern Med 1992;152:278-282
CrossRef | Web of Science | Medline

Author/Editor Response

The authors reply:

To the Editor: Long-term anticoagulation in patients with cardiovascular disease requires that the clinician balance the risks of bleeding and thromboembolism by adjusting the intensity of anticoagulation. The optimal intensity depends on the specific risks of thromboembolism and bleeding in an individual patient. Recent recommendations support less intense anticoagulation in patients with prosthetic heart valves than that recommended only a few years ago1,2. As suggested by O'Reilly and Kearns, the optimal INR range for such patients may well lie somewhat below the currently recommended range of 2.5 to 3.5, although the quality and quantity of data concerning the efficacy of anticoagulation at lower INRs are limited. However, the simple summation of the two risks does not account for the differential impact of thromboembolism and bleeding. Among patients with a prosthetic aortic valve, the case fatality rate for thromboembolic events is more than twice that for bleeding events (0.18 vs. 0.07), and the risk of long-term sequelae after a thromboembolic event is six times that after a bleeding event (0.30 vs. 0.05). If the event rates are weighted according to their relative clinical impact, the nadir of risk occurs at a higher INR than these correspondents suggest.

Laboratory variability in the measurement of anticoagulation intensity makes an already complex decision too prone to error. Although some thromboplastins have poorly calibrated ISIs, Carr and Horowitz propose an unwieldy solution in suggesting that clinicians monitor their patients' prothrombin times by obtaining measurements from a single laboratory. Patients are mobile and change laboratories; laboratories themselves sometimes change reagents, often without notifying clinicians. The imperfection of INRs is hardly an argument to fall back on even more variable methods of reporting and interpreting results; instead, it should be a call to standardize thromboplastins. As discussed by Schuff-Werner et al., clinicians should note that additional variability in the response of some thromboplastins may be introduced by concomitant therapy with heparin.

As the indications for anticoagulation expand (for example, to include some patients with nonvalvular atrial fibrillation), the number of patients put at unnecessary risk will increase if the prothrombin-time ratio continues to be used to monitor the intensity of anticoagulation. To maximize the effectiveness of anticoagulation therapy and conserve our limited health care resources, we must not only consider the specific risks of thromboembolism and bleeding in individual patients but also minimize the uncertainty surrounding the measurement of anticoagulation intensity.

Mark H. Eckman, M.D.
Herbert J. Levine, M.D.
Stephen G. Pauker, M.D.
New England Medical Center, Boston, MA 02111

2 References

1
Hirsh J, Dalen JE, Deykin D, Poller L. Oral anticoagulants: mechanisms of action, clinical effectiveness, and optimal therapeutic range. Chest 1992;102:Suppl:312s-326s
Web of Science | Medline