FREE NEJM E-TOC    HOME   |   SUBSCRIBE   |   CURRENT ISSUE   |   PAST ISSUES   |   COLLECTIONS   |   Search Term  Advanced Search

Sign in | Get NEJM's E-Mail Table of Contents — Free | Subscribe

Previous Volume 329:696-702 September 2, 1993 Number 10 Next

Abstract Letters Add to Personal Archive Add to Citation Manager Notify a Friend E-mail When Cited PubMed Citation

ABSTRACT

Background Patients receiving long-term anticoagulant therapy may be subject to unnecessary risks of bleeding or thromboembolism because of variability in the commercial thromboplastins used to determine prothrombin time and consequent uncertainty about the actual intensity of anticoagulation.

Methods We explored the effect of this uncertainty on the benefits and risks of anticoagulation in patients with prosthetic heart valves, using models of thromboembolic and hemorrhagic complications as a function of the intensity of anticoagulation, with quality-adjusted life expectancy and average variable costs used to describe outcomes.

Results Anticoagulation provides a striking benefit for patients whose treatment is conducted within the recommended range of the international normalized ratio (INR) -- i.e., 2.5 to 3.5 -- but if uncertainty about the laboratory results causes the intensity of anticoagulation to fall outside this range, the gain becomes smaller. Uncertainty about the true intensity of anticoagulation may reduce the potential gain in life expectancy, adjusted for quality of life, by more than half and may increase the ratio of costs to effectiveness to almost five times the optimal value. Variability in the intensity of anticoagulation is even greater if older recommendations advocating a higher level of anticoagulation are followed.

Conclusions Uncertainty about the sensitivities of the commercially available thromboplastins used in the United States can have important clinical and economic effects. This problem could be eliminated if clinical laboratories uniformly reported the intensity of anticoagulation as the INR, by adjusting prothrombin-time ratios for variability in thromboplastins.

In many Western countries, laboratories report international normalized ratios (INRs), as if the prothrombin time had been determined with the World Health Organization reference thromboplastin². The INR is calculated as PTR^ISI, where PTR is the observed prothrombin-time ratio (the patient's prothrombin time divided by the mean normal prothrombin time in the laboratory) and ISI the international sensitivity index. A higher ISI corresponds to a thromboplastin that yields prothrombin-time ratios that are less responsive to anticoagulant effects (e.g., a decrease in the vitamin K-dependent factors II, VII, and X) and implies more intense anticoagulation for a given prothrombin-time ratio (see the Appendix). If the laboratory does not report the INR, clinicians can calculate it if they know the ISI of the thromboplastin used. Some guidelines suggest a target prothrombin-time ratio, assuming an average ISI of 2.4 that is to be used if the INR is not reported and the ISI is unknown³. Given the variability of thromboplastins, however, using the prothrombin-time ratio to guide decisions about the intensity of anticoagulation may prove risky. Because ISI values for the commercial rabbit-brain thromboplastins used in the United States range from 1.2 to 2.8,^1,2 treatment decisions based on the prothrombin-time ratio may result in a substantially lower or higher degree of anticoagulation. For example, a patient whose prothrombin time is 1.5 times the control value could have an INR as low as 1.6 (if the ISI is 1.2) or as high as 3.1 (if the ISI is 2.8). Furthermore, even if the laboratory uses thromboplastin from a single manufacturer, the problem is not eliminated, because the ISI can vary substantially from lot to lot of thromboplastin.

Methods

It is unlikely that a clinical trial will (or could, ethically) compare the outcomes of anticoagulant treatment based on INRs with the outcomes of treatment based on prothrombin-time ratios. Therefore, we used decision analysis to explore the effect of uncertainty about the actual INR (due to lack of information about the ISI) on the benefits, risks, and cost effectiveness of anticoagulation in patients with prosthetic heart valves. We chose this group because these patients need lifelong anticoagulant therapy and are at high risk for thromboembolism without it. We reviewed the medical literature for data describing changes in the risks of thromboembolic and hemorrhagic events as a function of the intensity of anticoagulant therapy. We extended to this setting a previously described Markov model⁴ depicting thromboembolic and hemorrhagic complications of disease and treatment. Further details of the model are available elsewhere.^*

Intensity of Anticoagulation and Risk of Bleeding

Only four prospective, randomized studies have explicitly examined the risk of bleeding with different intensities of oral anticoagulant therapy (Table 1)^5,6,7,8,9. In none of these four studies was there a statistically significant difference between the two groups tested in the incidence of thromboembolic events.

View this table: [in this window] [in a new window]   Table 1. Intensity of Anticoagulation and the Risk of Bleeding.

 
We used the rates of bleeding episodes for each group in the four trials as dependent data points with which to perform semilogarithmic weighted least-squares regression analyses^10,11 of the mean INR (Figure 1). We excluded the results from the high-intensity-anticoagulation group in the Saudi Arabian study⁸ because it was an extreme outlier, with a target INR of 9.1, far higher than that found in present clinical practice. In a sensitivity analysis, however, we included this group. The formula that describes the average annual bleeding rate as a function of the INR is as follows:

View larger version (38K): [in this window] [in a new window]   Figure 1. Annual Rate of All Bleeding Complications as a Function of INR, Shown on a Semilogarithmic Axis. The symbols represent the mean annual rates of bleeding events in each group in four studies examining different intensities of anticoagulation. Numbers in parentheses are the numbers of patients in each group (see Table 1). The solid line is a weighted least-squares regression to the data. Because Altman et al.9 reported only major bleeding episodes, the data from their study were adjusted to be equivalent to the rate of bleeding of all types. The high-intensity group in the study by Saour et al.8 was excluded because its target INR was 9.1, far higher than is found in clinical practice.

 

total bleeding rate = 10^{(0.55 x INR - 2.10)} (bleeding events/yr)

with a correlation coefficient (R²) of 0.43. This formula suggests that each unit increase in the INR raises the risk of bleeding by 3.5 times. In the three studies that presented such data, an average of 20 percent of all bleeding episodes were major^5,7,8. In the study by Altman et al.,⁹ only major bleeding episodes were reported, and the rate was proportionately lower. Most major bleeding episodes do not have long-term sequelae, but they often necessitate transfusion or a short hospitalization. The chief long-term complication of anticoagulant therapy is neurologic impairment. From data in the Sixty Plus Reinfarction Study,¹² we calculated that approximately 5 percent of major bleeding events (or 1 percent of all bleeding episodes) result in permanent neurologic impairment. Most fatal bleeding events are intracerebral^{13,14,15,16,17}. In patients with prosthetic heart valves, roughly 7 percent of major bleeding episodes are fatal, according to our calculation of a weighted average based on the studies by Turpie et al.,⁷ Saour et al.,⁸ and Altman et al⁹.

Risk of Systemic Embolism

The few studies of patients with prosthetic heart valves that have compared the rate of embolic events in patients without anticoagulation with the rate in those who have received differing intensities of anticoagulation describe patients with older ball or Bjork-Shiley valves^{18,19,20,21,22}. To define the relation between the intensity of anticoagulation and the rate of thromboembolic events, we pooled data from numerous studies, excluding those in which patients were treated with antiplatelet agents, alone or in combination with warfarin. We divided the studies into those in which the patients did not receive anticoagulant therapy^{20,22,23,24,25,26} and those in which the patients did receive such therapy (with an INR 2.2 in each case)^{27,28,29,30,31,32,33,34,35,36,37,38,39,40}. We then calculated a weighted average rate of thromboembolic events for each group, using the inverse variance of each series as the weight. The rates of thromboembolic events were 18 percent for those not receiving therapy and 1.4 percent for those treated with anticoagulants. Among the patients receiving anticoagulant therapy there was no significant correlation between the intensity of anticoagulation and the rate of thromboembolic events (R² = 0.12), even when we analyzed subgroups of patients with valves of the same type and anatomical position. There were no studies with mean INRs below 2.2. Because recent data in other settings suggest there can be efficacy or at least partial efficacy at lower INRs,⁴¹ we assumed a declining exponential shape for the curve, with a flat tail beyond the mean INR of 5.9 for the 40 series of patients receiving anticoagulant therapy. The following equation resulted:

thromboembolic rate = 0.42 x 10^{-0.4 x INR} + 0.014 (thromboembolic events/year).

This equation yields a rate of 18 percent per year without anticoagulation (i.e., with an INR of 1); the curve has an asymptote of 1.4 percent per year for high INRs. We also explored this assumption in sensitivity analyses.

Quality-of-Life Adjustments and Costs

The probabilities and quality-of-life factors studied in the analysis are shown in Table 2. We used variable costs, not charges, and measured effectiveness in quality-adjusted years of life.

View this table: [in this window] [in a new window]   Table 2. Other Variables Used in the Model.

 
Quality-of-life Adjustments

We adjusted life expectancy to account for the quality of life associated with various states of health. For long-term states of health, such as permanent morbidity after a bleeding episode or an embolic event, we multiplied the life expectancy by a quality-adjustment factor, with full quality assigned a value of 1.0 and death a value of 0. For short-term morbidity resulting from thromboembolic or bleeding events, we subtracted an amount of time equal to the length of hospitalization, considering those days to have a value of 0. As the lower half of Table 2 shows, we report the quality of life for the month in which the short-term morbidity occurred (more complete information is available elsewhere^*). We also considered the minimal inconvenience of taking anticoagulant drugs in terms of the need to determine prothrombin times and the patients' concern about possible bleeding and the avoidance of other drugs and trauma^42,43.

Costs

Each outcome also had a monetary cost. Costs for the inpatient hospitalization were obtained from the cost-accounting system at New England Medical Center (Clinical Cost Manager, Transition Systems), whereas physicians' costs incurred by inpatients were estimated on the basis of length of stay and a fixed, average collection-to-charge ratio. Costs for ambulatory care were obtained from a large health maintenance organization in eastern Massachusetts (Tufts Associated Health Plans). For example, we calculated the average annual cost of routine follow-up for a patient with a prosthetic aortic valve who received anticoagulant therapy; the cost was $1,100 in 1992 dollars. For a complication involving long-term morbidity, such as systemic embolism, costs included the costs of hospitalization ($16,000) and outpatient care for the remainder of the patient's life ($4,200 per year). Table 3 shows the hospitalization-related costs for hemorrhagic and embolic complications of both anticoagulation strategies, as well as the costs of outpatient visits and anticoagulation therapy.

View this table: [in this window] [in a new window]   Table 3. Average Costs and Long-Term Mortality Rates for Complications and Follow-up.

 
Analysis

The annual overall mortality rate was calculated as the sum of the various mortality rates associated with each health outcome. The average demographic-related instantaneous mortality rates were taken from life tables published by the Department of Health and Human Services⁴⁴. Additional annual excess mortality was associated with having a prosthetic valve, and it was highly dependent on the patient's clinical status at the time of valve replacement^45,46.

In the base-line analysis, we considered a 35-year-old woman with a prosthetic aortic valve. We assumed an excess valve-associated mortality rate of 0.01 per year (excluding thromboembolic and anticoagulation-related complications), as would be the case among patients who have such valves and have good preoperative cardiac function (i.e., that associated with New York Heart Association functional class II)⁴⁷. All cost-effectiveness ratios reflect 5 percent annual discounting of both costs and effectiveness, unless otherwise stated. We used a standard computer program (Decision Maker^48,49,50) that can analyze decision trees and perform sensitivity analyses.

Results

Base-Line Case

As the base-line case shows, the magnitude of the gain is related to the intensity of anticoagulation. If the target INR is within the recommended range of 2.5 to 3.5,^51,52 anticoagulant therapy for a 35-year-old woman with good cardiac function and a prosthetic aortic valve yields a striking benefit (9.5 quality-adjusted years of life, before discounting). The cost-effectiveness ratio of anticoagulant therapy in such patients is only $650 per quality-adjusted year of life gained when discounting is used for both costs and health benefits. If the intensity of anticoagulation falls outside the recommended range, the gain becomes smaller as the rate of thromboembolic events or bleeding events increases (in patients receiving inadequate or excessive anticoagulation, respectively). For patients receiving excessive anticoagulation there may even be a net loss.

Uncertainty about the Actual Value of the ISI

If the laboratory reports neither the ISI nor the INR, current guidelines suggest attempting to achieve a prothrombin-time ratio of 1.6, which would correspond to an INR of 3.0 for an ISI assumed to be 2.4 (the average for a "typical North American thromboplastin")⁵³. For the range of ISI values reported by the 53 SPAF sites,¹ we calculated the loss in quality-adjusted years of life that would result if the unadjusted prothrombin-time ratio were used rather than the INR (Figure 2). The loss in nondiscounted years is shown on the y-axis as a function of the laboratory's actual ISI (ranging from 1.2 to 2.8) on the x-axis. The second (lower) x-axis shows the INRs corresponding to a prothrombin-time ratio of 1.6. The curves represent four different assumptions with regard to the relative risk of bleeding for each unit increase in the INR. The solid curve represents the base-line assumption, whereas the topmost curve shows the results obtained with a model of bleeding that includes the high-intensity-anticoagulation group of Saour et al.⁸ (for which the relative risk of bleeding is only 2.0, rather than 3.5, as in the base-line case).

View larger version (46K): [in this window] [in a new window]   Figure 2. Loss of Quality-Adjusted Life Expectancy as a Result of Uncertainty about True Anticoagulation Intensity Due to the Use of the Prothrombin-Time Ratio Rather Than the INR. With the "typical" North American thromboplastin as a guide, clinicians would assume an ISI of 2.4. A target prothrombin-time ratio of 1.6 would correspond to an INR of 3.0. The upper x-axis shows the range of actual ISIs possible: from 1.2 to 2.8. In the base-line case (solid curve), there will be no loss in life expectancy if the actual ISI used is 2.4. If the actual ISI is not 2.4, then the effective target INR is no longer 3.0. INRs corresponding to various actual ISIs are shown on the lower x-axis. The uppermost curve (dotted line) summarizes the results in a more conservative model that includes outlying data from Saour et al.,8 in which the relative risk of bleeding was lower (only 2.0 for each unit increase in INR). The middle two curves (dashed lines) represent intermediate risks of bleeding.

 
In the base-line case, if the actual ISI at the laboratory was assumed to be 2.4, then the INR would be 3.0 (as expected), and monitoring anticoagulant therapy with the unadjusted prothrombin-time ratio would result in no loss of quality-adjusted life expectancy. However, if the laboratory used a thromboplastin with an ISI of 1.2, then a prothrombin-time ratio of 1.6 would correspond to an actual INR of only 1.7, resulting in a loss of 4.8 quality-adjusted years of life as compared with treatment based on the INR. Similarly, if the laboratory used a thromboplastin with an ISI of 2.8, then the actual INR would be 3.6 (rather than 3.0), resulting in a loss of 1.3 quality-adjusted years of life. If the physician used older recommendations with a target prothrombin-time ratio higher than 1.8 (corresponding to INRs higher than 4.0), uncertainty about the laboratory's actual ISI could result in a perverse effect, by which patients who received anticoagulant therapy would have poorer prognoses than those who did not, if the thromboplastin had an actual ISI higher than 2.4.

In a sensitivity analysis, we did not exclude the high-intensity-anticoagulation group in the study of Saour et al.⁸. In that regression, the relative risk of bleeding for each unit increase in INR was 2.0, as compared with 3.5 when the data of Saour et al. were excluded. As shown in the uppermost curve of Figure 2, under this alternative assumption about the relative risk of bleeding, uncertainty about the actual ISI at a laboratory produces an even greater loss in quality-adjusted life expectancy. Because the benefit of anticoagulation represents a trade-off between bleeding and thromboembolism, a lower risk of bleeding allows the INR to have a greater influence on the risk of thromboembolic events, an influence that is magnified even more because the consequences of thromboembolism are more severe than those of bleeding. In other words, including the data of Saour et al.⁸ has two effects, decreasing the rate of bleeding and decreasing the slope of the relation between INR and the bleeding rate, and the first effect turns out to be more important. Including these data in the analysis also shifts the optimal intensity of anticoagulation toward a higher INR.

Although Figure 2 summarizes the loss in quality-adjusted years of life as a function of the actual ISI of each thromboplastin, it does not address the average loss that results from using the prothrombin-time ratio without adjustment for the sensitivity of thromboplastin, because in a typical patient's lifetime many different thromboplastins will be used. To examine this question, we obtained information on the frequency distribution of ISI values from three sources (Table 4): the laboratories involved in the SPAF study,¹ a recent survey of acute care facilities in Massachusetts⁵⁴ (and Ansell JE: personal communication), and a recent national survey of coagulation by the College of American Pathologists⁵⁵ (and Triplett DA: personal communication). Using these three distributions to calculate the weighted average loss, we found non-discounted losses of 0.75, 0.81, and 0.77 quality-adjusted year of life, respectively, when prothrombin-time ratios were used without correction for thromboplastin sensitivity.

View this table: [in this window] [in a new window]   Table 4. Frequency Distributions of ISI Values Observed in Three Studies.

 
Cost Effectiveness

Figure 3 examines the effect of uncertainty about the intensity of anticoagulation on the cost-effectiveness ratio. Discounted cost effectiveness, in dollars per quality-adjusted year of life, is shown on the vertical axis as a function of the actual ISI of the thromboplastin used. If the assumed ISI of 2.4 is correct, the target prothrombin-time ratio of 1.6 has a cost-effectiveness ratio of roughly $650 per quality-adjusted year of life. If the ISI is below 1.9 or above 2.4, however, the cost-effectiveness ratio of anticoagulation increases, becoming less favorable. On the other hand, if the actual ISI lies between 1.9 and 2.4, the cost-effectiveness ratio for the somewhat less intense anticoagulation is actually lower, because the savings from fewer bleeding complications more than offsets the costs of additional thromboembolic events. There is no reason to assume that the optimal intensity of anticoagulation from a perspective of effectiveness (an INR of 3.0) has the lowest cost-effectiveness ratio.

View larger version (46K): [in this window] [in a new window]   Figure 3. Cost Effectiveness as a Function of the Sensitivity of the Thromboplastin Used (Actual ISI). Data presuppose that a prescribed prothrombin-time ratio (rather than an INR) is used and that the ISI is assumed to be 2.4. The horizontal axes are similar to those shown in Figure 2. Cost effectiveness is expressed in dollars per quality-adjusted year of life. If the actual ISI is above 2.4 (corresponding to an INR above 3.0) or below 1.9 (corresponding to an INR below 2.4), the cost-effectiveness ratio increases.

 
Discussion

Anticoagulant drugs have a narrow therapeutic range, but many clinicians use them without precise information on the true intensity of therapy because of variability in commercially available thromboplastins^1,56. A patient with a prothrombin-time ratio of 1.5 may have an INR as low as 1.6 or as high as 3.1, with rates of bleeding and of thromboembolism that differ by factors of 6.5 and 4.0, respectively. We have used decision and cost-effectiveness analyses to highlight the clinical and economic effect of this variability. Variability in the ISI has greater impact with higher prothrombin-time ratios. Older recommendations that advocated more intense anticoagulation (INR, 3.0 to 4.5)⁵⁷ can produce far more bleeding if anticoagulant therapy is based on prothrombin-time ratios determined for thromboplastins with higher-than-average ISIs. In such circumstances, patients with relatively nonthrombogenic prosthetic valves who receive anticoagulant therapy may be less well served, on average, than those receiving no anticoagulant therapy.

Using a mean value for an unknown ISI is misleading, because the distribution of ISI values is bimodal (Table 4). The preponderance of values at the extremes (above and below 2.4) increases the weighted average loss incurred when clinicians use the prothrombin-time ratio without knowing the actual ISI.

Only four randomized, prospective trials have examined different intensities of anticoagulation in matched populations. We derived our relation between the INR and the bleeding rate from these four trials. We thought that these data would produce a more realistic model than would be achieved by pooling the bleeding rates from multiple series of patients with prosthetic valves, in whom the intensity of anticoagulation was not being studied and in whom episodes of bleeding were less rigorously defined.

Using a retrospective, case-control study design, Landefeld et al. considered the association between the prothrombin-time ratio and the occurrence of bleeding events in outpatients treated with warfarin⁵³. In patients who had been receiving anticoagulant drugs for two to six months, the odds ratios for major and minor bleeding associated with an increase of 1.0 in the prothrombin-time ratio were 7.3 (95 percent confidence interval, 2.8 to 19.5) and 1.8 (95 percent confidence interval, 0.3 to 12.4), respectively. In our analysis, we calculated the risk of bleeding as a function of change in the INR rather than in the prothrombin-time ratio. For each unit increase in the INR, the relative risk of any type of bleeding was 3.5. Although Landefeld et al. did not report INRs in their study, we believe that their odds ratio would be higher than the one we calculated on the basis of the INR because a unit increase in the prothrombin-time ratio corresponds to a larger increase in the INR.

We calculated cost-effectiveness ratios that were extraordinarily favorable (predicting a cost of less than $1,000 per quality-adjusted year of life) for strategies prescribing INRs within the range of 2.5 to 3.5. Strategies prescribing INRs higher than 4.7 were unsatisfactory, being both more costly and less effective than a strategy of no anticoagulation.

Varying the intensity of anticoagulation with the prothrombin-time ratio rather than the INR as a guide to therapy may reduce by more than half the potential gain in quality-adjusted life expectancy afforded by the use of such therapy in patients with mechanical heart valves and may result in a 4.5-fold increase in the cost-effectiveness ratio.

The straightforward solution to this problem is for laboratories either to report INRs or to inform clinicians of the ISI of the thromboplastin used to determine the prothrombin-time ratio. Either way, clinicians would know the true intensity of anticoagulation. Alternatively, with recombinant technology it may become feasible for all laboratories to use a standard thromboplastin, so that variation between laboratories does not affect the clinical care of patients requiring anticoagulant therapy.

Supported in part by a grant from the Agency for Health Care Policy Research (HS-06503) and a grant from the John A. Hartford Foundation (87269-3H).

We are indebted to Jack E. Ansell, M.D., of the Department of Medicine, University of Massachusetts Medical School, Worcester, for graciously providing data on ISI frequency distribution in acute care facilities in Massachusetts; and to Doug Triplett, M.D., of the Department of Clinical Pathology, Ball Memorial Hospital, Muncie, Ind., for his assistance in providing ISI data from the College of American Pathologists.

^* See NAPS document no. 05048 for seven pages of supplementary material. To order, contact NAPS c/o Microfiche Publications, 248 Hempstead Tpk., West Hempstead, NY 11552.

Source Information

From the Divisions of Clinical Decision Making (M.H.E., S.G.P.) and Cardiology (H.J.L., S.G.P.), Department of Medicine, New England Medical Center and Tufts University School of Medicine, Boston.

Address reprint requests to Dr. Eckman at the Division of Clinical Decision Making, Box 302, New England Medical Center, 750 Washington St., Boston, MA 02111.

References

1. 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. [Free Full Text]
2. Poller L. Progress in standardization in anticoagulant control. Hematol Rev Commun 1987;1:225-41.
3. Hirsh J, Poller L, Deykin D, Levine M, Dalen JE. Optimal therapeutic range for oral anticoagulants. Chest 1989;95:Suppl:5S-11S. [Erratum, Chest 1989;96:962.] [Free Full Text]
4. Eckman MH, Levine HJ, Pauker SG. Decision analytic and cost-effectiveness issues concerning anticoagulant prophylaxis in heart disease. Chest 1992;102:Suppl:538S-549S.
5. Hull R, Hirsh J, Jay R, et al. Different intensities of oral anticoagulant therapy in the treatment of proximal-vein thrombosis. N Engl J Med 1982;307:1676-1681. [Abstract]
6. Hirsh J. Oral anticoagulant drugs. N Engl J Med 1991;324:1865-1875. [Medline]
7. Turpie AG, Gunstensen J, Hirsh J, Nelson H, Gent M. Randomised comparison of two intensities of oral anticoagulant therapy after tissue heart valve replacement. Lancet 1988;1:1242-1245. [Medline]
8. Saour JN, Sieck JO, Mamo LAR, Gallus AS. Trial of different intensities of anticoagulation in patients with prosthetic heart valves. N Engl J Med 1990;322:428-432. [Abstract]
9. Altman R, Rouvier J, Gurfinkel E, et al. Comparison of two levels of anticoagulant therapy in patients with substitute heart valves. J Thorac Cardiovasc Surg 1991;101:427-431. [Abstract]
10. Rothman KJ. Modern epidemiology. Boston: Little, Brown, 1986:338.
11. SAS procedures guide, version 6, 3rd ed. Cary, N.C.: SAS Institute, 1990:210.
12. Risks of long-term oral anticoagulant therapy in elderly patients after myocardial infarction: Second report of the Sixty Plus Reinfarction Study Research Group. Lancet 1982;1:64-68. [CrossRef][Medline]
13. Altman R, Boullon F, Rouvier J, Raca R, de la Fuente L. Aspirin and prophylaxis of thromboembolic complications in patients with substitute heart valves. J Thorac Cardiovasc Surg 1976;72:127-129. [Abstract]
14. Chesebro JH, Fuster V, Elveback LR, et al. Trial of combined warfarin plus dipyridamole or aspirin therapy in prosthetic heart valve replacement: danger of aspirin compared with dipyridamole. Am J Cardiol 1983;51:1537-1541. [CrossRef][Medline]
15. Dale J, Myhre E, Storstein O, Stormorken H, Efskind L. Prevention of arterial thromboembolism with acetylsalicylic acid: a controlled clinical study in patients with aortic ball valves. Am Heart J 1977;94:101-111. [Medline]
16. Sullivan JM, Harken DE, Gorlin R. Pharmacologic control of thromboembolic complications of cardiac-valve replacement. N Engl J Med 1971;284:1391-1394.
17. Turpie AGG, Gent M, Laupacis A, et al. Reduction in mortality by adding aspirin (100 mg) to oral anticoagulants in patients with heart valve replacement. J Am Coll Cardiol 1992;19:Suppl A:103A-103A.abstract 
18. Edmunds LH Jr. Thromboembolic complications of current cardiac valvular prostheses. Ann Thorac Surg 1982;34:96-106. [Abstract]
19. Vidne B, Levy MJ. Incidence of thromboembolic complications using totally cloth-covered Starr-Edwards prostheses. Isr J Med Sci 1974;10:586-589. [Medline]
20. Moggio RA, Hammond GL, Stansel HC Jr, Glenn WW. Incidence of emboli with cloth-covered Starr-Edwards valve without anticoagulation and with varying forms of anticoagulation: analysis of 183 patients followed for 3 1/2 years. J Thorac Cardiovasc Surg 1978;75:296-299. [Abstract]
21. Larsen GL, Alexander JA, Stanford W. Thromboembolic phenomena in patients with prosthetic aortic valves who did not receive anticoagulants. Ann Thorac Surg 1977;23:323-326. [Abstract]
22. Bjork VO, Henze A. Management of thrombo-embolism after aortic valve replacement with the Bjork-Shiley tilting disc valve: Medicamental prevention with dicumarol in comparison with dipyridamole-acetylsalicylic acid. Surgical treatment of prosthetic thrombosis. Scand J Thorac Cardiovasc Surg 1975;9:183-191. [Medline]
23. Baudet EM, Oca CC, Roques XF, et al. A 5 1/2 year experience with the St. Jude Medical cardiac valve prosthesis: early and late results of 737 valve replacements in 671 patients. J Thorac Cardiovasc Surg 1985;90:137-144. [Abstract]
24. Duvoisin GE, Brandenburg RO, McGoon DC. Factors affecting thromboembolism associated with prosthetic heart valves. Circulation 1967;35:Suppl 1:70-76. [Free Full Text]
25. Akbarian M, Austen G, Yurchak PM, Scannell JG. Thromboembolic complications of prosthetic cardiac valves. Circulation 1968;37:826-831. [Free Full Text]
26. Stein DW, Selden R, Starr A. Thrombotic phenomena with non-anticoagulated, composite-strut aortic prostheses. J Thorac Cardiovasc Surg 1976;71:680-684. [Abstract]
27. Kopf GS, Hammond GL, Geha AS, Elefteriades J, Hashim SW. Long-term performance of the St. Jude Medical valve: low incidence of thromboembolism and hemorrhagic complications with modest doses of warfarin. Circulation 1987;76:Suppl III:III-132. 
28. DeSesa VJ, Collins JJ Jr, Cohn LH. Hematological complications with the St. Jude valve and reduced-dose Coumadin. Ann Thorac Surg 1989;48:280-283. [Abstract]
29. Vogt S, Hoffmann A, Roth J, et al. Heart valve replacement with the Bjork-Shiley and St. Jude Medical prostheses: a randomized comparison in 178 patients. Eur Heart J 1990;11:583-591. [Free Full Text]
30. Horstkotte D, Korfer R, Seipel L, Bircks W, Loogen F. Late complications in patients with Bjork-Shiley and St. Jude Medical heart valve replacement. Circulation 1983;68:Suppl II:II-75. 
31. Czer LS, Chaux A, Matloff JM, et al. Ten-year experience with the St. Jude Medical valve for primary valve replacement. J Thorac Cardiovasc Surg 1990;100:44-55. [Abstract]
32. Eberlein U, von der Emde J, Rein J, Esperer HD. Thromboembolic and bleeding complications after mitral valve replacement. Eur J Cardiothorac Surg 1990;4:605-612. [Abstract]
33. Sethia B, Turner MA, Lewis S, Rodger RA, Bain WH. Fourteen years' experience with the Bjork-Shiley tilting disc prosthesis. J Thorac Cardiovasc Surg 1986;91:350-361. [Abstract]
34. Bjork VO, Henze A. Ten years' experience with the Bjork-Shiley tilting disc valve. J Thorac Cardiovasc Surg 1979;78:331-342. [Abstract]
35. Cortina JM, Martinell J, Artiz V, Fraile J, Rabago G. Comparative clinical results with Omniscience (STM1), Medtronic-Hall, and Bjork-Shiley convexo-concave (70 degrees) prostheses in mitral valve replacement. J Thorac Cardiovasc Surg 1986;91:174-183. [Abstract]
36. Vallejo JL, Gonzalez-Santos JM, Albertos J, et al. Eight years' experience with the Medtronic-Hall valve prosthesis. Ann Thorac Surg 1990;50:429-436. [Abstract]
37. Beaudet RL, Poirier NL, Doyle D, Nakhle G, Gauvin C. The Medtronic-Hall cardiac valve: 7 1/2 years' clinical experience. Ann Thorac Surg 1986;42:644-650. [Abstract]
38. Sullivan JM, Harken DE, Gorlin R. Effect of dipyridamole on the incidence of arterial emboli after cardiac valve replacement. Circulation 1969;39:Suppl I:I-149. 
39. Dale J. Arterial thromboembolic complications in patients with Starr-Edwards aortic ball valve prostheses. Am Heart J 1976;91:653-659. [Medline]
40. Dale J. Arterial thromboembolic complications in patients with Bjork-Shiley and Lillehei-Kaster aortic disc valve prostheses. Am Heart J 1977;93:715-722. [Medline]
41. Ezekowitz MD, Bridgers SL, James KE, et al. Warfarin in the prevention of stroke associated with nonrheumatic atrial fibrillation. N Engl J Med 1992;327:1406-1412. [Abstract]
42. Lancaster TR, Singer DE, Sheehan MA, et al. The impact of long-term warfarin therapy on quality of life: evidence from a randomized trial. Arch Intern Med 1991;151:1944-1949. [Erratum, Arch Intern Med 1992;152:825.] [Free Full Text]
43. Hirsh J. Influence of low-intensity warfarin treatment on patients' perceptions of quality of life. Arch Intern Med 1991;151:1921-1922. [Free Full Text]
44. National Center for Health Statistics. Vital statistics of the United States, 1987. Vol. 2. Mortality. Part A. Section 6, Life tables. Washington, D.C.: Government Printing Office, 1990:10-1. (DHSS publication no. (PHS) 90-1101.)
45. Rackley CE, Edwards JE, Karp RB. Mitral valve disease. In: Hurst JW, ed. The heart. 6th ed. New York: McGraw-Hill, 1986:754-84.
46. Singer RB, Levinson L. Medical risks: patterns of mortality and survival. Lexington, Mass.: D.C. Heath, 1976:3-122.
47. Copeland JG, Griepp RB, Stinson EB, Shumway NE. Long-term follow-up after isolated aortic valve replacement. J Thorac Cardiovasc Surg 1977;74:875-889. [Abstract]
48. Pauker SG, Kassirer JP. Clinical decision analysis by personal computer. Arch Intern Med 1981;141:1831-1837. [CrossRef][Medline]
49. Lau J, Kassirer JP, Pauker SG. Decision Maker 3.0: improved decision analysis by personal computer. Med Decis Making 1983;3:39-43.
50. Sonnenberg FA, Pauker SG. Decision Maker 6.0. In: Salamon R, Blum B, Jorgensen M, eds. Proceedings of Medinfo 86. Amsterdam: Elsevier North-Holland, 1986:1152.
51. Stein PD, Alpert JS, Copeland J, Dalen JE, Goldman S, Turpie AG. Antithrombotic therapy in patients with mechanical and biological prosthetic heart valves. Chest 1992;102:Suppl:445S-455S.
52. Hirsh J, Dalen JE, Deykin D, Poller L. Oral anticoagulants: mechanism of action, clinical effectiveness, and optimal therapeutic range. Chest 1992;102:Suppl:312S-326S. [Free Full Text]
53. Landefeld CS, Rosenblatt MW, Goldman L. Bleeding in outpatients treated with warfarin: relation to the prothrombin time and important remediable lesions. Am J Med 1989;87:153-159. [Medline]
54. Ansell JE. Imprecision of prothrombin time monitoring of oral anticoagulation: a survey of hospital laboratories. Am J Clin Pathol 1992;98:237-239. [Medline]
55. College of American Pathologists Comprehensive Coagulation Survey. Skokie, Ill.: CGD-2, 1991 (survey).
56. Hirsh J. Substandard monitoring of warfarin in North America: time for change. Arch Intern Med 1992;152:257-258. [Erratum, Arch Intern Med 1992;152:959.] [Free Full Text]
57. Quick AJ, Stanley-Brown M, Bancroft FW. A study of the coagulation defect in hemophilia and in jaundice. Am J Med Sci 1935;190:501-511. [CrossRef]

Monitoring Oral Anticoagulant Therapy with the Prothrombin-Time Ratio and the INR

The prothrombin-time assay, first described in 1935 by Quick et al., soon became the basis for monitoring oral anticoagulant therapy⁵⁷. The prothrombin time measures the effect of a reduction in the vitamin K-dependent coagulation factors II, VII, and X. It is usually reported as the ratio between the patient's prothrombin time and the mean control value used at the laboratory. Unfortunately, this ratio depends on the reagent used to initiate the extrinsic coagulation cascade. Thromboplastins derived from a number of different tissues have been used for this purpose, in particular brain tissue from humans (in the United Kingdom) and from rabbits (in North America). The sensitivity of the prothrombin time to the depletion of vitamin K-dependent clotting factors (due to anticoagulant therapy) varies substantially among rabbit-brain thromboplastins, from manufacturer to manufacturer and from lot to lot. Because the prothrombin-time ratio is markedly affected by the activity of thromboplastin, the World Health Organization (WHO) has developed an international reference standard, the ISI, to gauge the responsiveness of the prothrombin time to the reduction in coagulation factors, as measured with a given thromboplastin. Each lot of thromboplastin can be characterized by an ISI that calibrates the reagent with the first WHO reference human-brain thromboplastin, which was assigned an ISI of 1.0.

The ISI is used as a correction factor in calculating the INR by the formula PTR^ISI, where PTR is the prothrombin-time ratio; this formula yields the prothrombin-time ratio that would have been obtained had the reference thromboplastin been used. If the prothrombin-time ratio indicated by a given thromboplastin is especially sensitive to the depletion of vitamin K-dependent factors II, VII, and X, the prothrombin-time ratio will need little correction and the ISI will be only slightly higher than 1.0. If the prothrombin-time ratio measured with another thromboplastin is less sensitive to the effect of anticoagulation, the prothrombin-time ratio will need a larger correction and the ISI will be higher. Thus, the use of a more responsive thromboplastin (or more correctly, a more responsive prothrombin time as measured by the thromboplastin) requires a lower ISI.

Abstract Letters Add to Personal Archive Add to Citation Manager Notify a Friend E-mail When Cited PubMed Citation

Related Letters:

Variation in the Prothrombin-Time Ratio during Oral Anticoagulation
Carr J. M., Horowitz G., O'Reilly R. A., Kearns P. J., Schuff-Werner P., Schutz E., Gonska B.-D., Eckman M. H., Levine H. J., Pauker S. G.
Extract | Full Text  
N Engl J Med 1994; 330:509-510, Feb 17, 1994. Correspondence

This article has been cited by other articles:

• Spiegel, B. M.R., Targownik, L., Dulai, G. S., Gralnek, I. M. (2003). The Cost-Effectiveness of Cyclooxygenase-2 Selective Inhibitors in the Management of Chronic Arthritis. ANN INTERN MED 138: 795-806 [Abstract] [Full Text]  
• Fitzmaurice, D A, Murray, E T, Allan, T F, Holder, R L, Rose, P E, Hobbs, F D R (2000). A comparison of international normalised ratio (INR) measurement in hospital and general practice settings: evidence for lack of standardisation. J. Clin. Pathol. 53: 803-804 [Full Text]  
• Taube, J., Halsall, D., Baglin, T. (2000). Influence of cytochrome P-450 CYP2C9 polymorphisms on warfarin sensitivity and risk of over-anticoagulation in patients on long-term treatment. Blood 96: 1816-1819 [Abstract] [Full Text]  
• COLWELL, C. W., COLLIS, D. K., PAULSON, R., McCUTCHEN, J. W., BIGLER, G. T., LUTZ, S., HARDWICK, M. E. (1999). Comparison of Enoxaparin and Warfarin for the Prevention of Venous Thromboembolic Disease After Total Hip Arthroplasty. Evaluation During Hospitalization and Three Months After Discharge. JBJS 81: 932-40 [Abstract] [Full Text]  
• Cohen, J. S., Insel, P. A. (1996). The Physicians' Desk Reference: Problems and Possible Improvements. Arch Intern Med 156: 1375-1380 [Abstract]  
• The Stroke Prevention in Atrial Fibrillation Inves, (1996). Bleeding During Antithrombotic Therapy in Patients With Atrial Fibrillation. Arch Intern Med 156: 409-416 [Abstract]  
• Gage, B. F., Cardinalli, A. B., Albers, G. W., Owens, D. K. (1995). Cost-effectiveness of Warfarin and Aspirin for Prophylaxis of Stroke in Patients With Nonvalvular Atrial Fibrillation. JAMA 274: 1839-1845 [Abstract]  
• Schmitz, L. L., Olson, S. L., Shapiro, R. S., McCormick, S. R., Kubic, V. L. (1995). Failure to Generate Comparable International Normalized Ratio Values Using Five Different Thromboplastin Reagents in Parallel Studies of Patients Receiving Warfarin. CLIN APPL THROMB HEMOST 1: 142-150 [Abstract]  
• Sakata, T., Kario, K., Matsuo, T., Katayama, Y., Matsuyama, T., Kato, H., Miyata, T. (1995). Suppression of Plasma-Activated Factor VII Levels by Warfarin Therapy. Arterioscler. Thromb. Vasc. Bio. 15: 241-246 [Abstract] [Full Text]  
• Eckman, M. H., Pauker, S. G. (1995). Clinical Impact of Inaccuracies in Anticoagulation Monitoring. Arch Intern Med 155: 115-116 [Abstract]  
• Carr, J. M., Horowitz, G., O'Reilly, R. A., Kearns, P. J., Schuff-Werner, P., Schutz, E., Gonska, B.-D., Eckman, M. H., Levine, H. J., Pauker, S. G. (1994). Variation in the Prothrombin-Time Ratio during Oral Anticoagulation. NEJM 330: 509-510 [Full Text]  
• Kirk, J. K., McGann, K. P. (1994). International Normalized Ratio: The New Standard for Monitoring Oral Anticoagulation. Arch Fam Med 3: 71-74 [Abstract]