Skip to main content
Device Global Header

Subscribe
or Renew

Advanced Search

Original Article

Trial of Different Intensities of Anticoagulation in Patients with Prosthetic Heart Valves

List of authors.
  • Jalal Najib Saour, M.R.C.P.,
  • Jens Otto Sieck, M.R.C.P.,
  • Layla Abdul Rahim Mamo, Ph.D.,
  • and Alexander Stephen Gallus, F.R.C.P.A.

Abstract

We compared the efficacy and complications of anticoagulation with warfarin in 258 patients with prosthetic heart valves treated with regimens of "moderate intensity" (prothrombin-time ratio, 1.5; international normalized ratio, 2.65) or "high intensity" (prothrombin-time ratio, 2.5; international normalized ratio, 9) in a prospective, randomized study. The two patient groups were followed up for 421 patient-years and 436 patient-years, respectively. Eleven patients were lost to follow-up.

Thromboembolism occurred with similar frequency in the two groups (4.0 and 3.7 episodes per 100 patient-years, respectively), but there was a total of 6.2 bleeding episodes per 100 patient-years in the moderate-intensity group, as compared with 12.1 episodes in the high-intensity group (P<0.002). There were 5.2 episodes of minor bleeding per 100 patient-years in the moderate-intensity group, as compared with 10.1 episodes in the high-intensity group (P<0.01). Major bleeding was also more common in the high-intensity group (2.1 episodes per 100 patient-years — including the only two fatal hemorrhages — as compared with 0.95 episode in the moderate-intensity group), but the difference was not statistically significant.

We conclude that a moderate anticoagulant effect (prothrombin-time ratio, about 1.5) in patients with a mechanical prosthetic heart valve offers protection equivalent to that of more intensive therapy, but at a significantly lower risk. (N Engl J Med 1990; 322:428–32.)

Introduction

THROMBOSIS affecting prosthetic heart valves and arterial thromboembolism remain major causes of late morbidity and mortality after the replacement of heart valves with mechanical prostheses, despite improvements in valve design.1

Factors that influence the risk of arterial thromboembolism include the type, site, and number of valves replaced,1 2 3 4 5 the presence or absence of atrial fibrillation,6 treatment with warfarin or other oral anticoagulants,1 the adequacy of warfarin therapy, and the addition of antiplatelet drugs such as dipyridamole and aspirin.1

Although a need for lifelong warfarin treatment after the mechanical replacement of valves, particularly in the mitral site (and probably in the aortic site), is now generally accepted, uncertainty remains about its optimal intensity.4 , 7 , 8 Because retrospective surveys strongly suggest that thromboembolism is much more likely when the effect of warfarin is low or absent,7 , 9 10 11 more rather than less intense therapy has usually been recommended. In Europe, therapy usually aims for an international normalized ratio (a system for standardizing the results of prothrombin-time tests performed with different reagents) of 3.0 to 5.0, and in North America it aims for even higher levels,12 although recently there appears to be some agreement on an international normalized ratio of between 3.0 and 4.5 in patients with a mechanical prosthetic valve.13 The question is both important and topical, since the risk of bleeding increases with higher levels of anticoagulation and since randomized studies have shown that secondary prevention after venous thrombosis requires a less intense anticoagulant effect (international normalized ratio, 2.0 to 2.5) than was previously thought.14

We conducted a prospective, open clinical trial in which patients with prosthetic heart valves were randomly assigned to therapy designed to achieve one of two levels of anticoagulant effect: a prothrombin-time ratio (the ratio of the patient's prothrombin time to that of pooled normal plasma) of 1.5 (international normalized ratio, 2.65) or a prothrombin-time ratio of 2.5 (international normalized ratio, 9). These values correspond to the extremes of the accepted therapeutic range in many centers.15 , 16

Methods

Patients

All patients who received mechanical prosthetic valves for rheumatic heart disease and attended the anticoagulation clinic between 1981 and 1985 were randomly assigned to one of two treatment groups, which differed only in the intended level of anticoagulant effect (moderate or high). There were no exclusions. All those enrolled in the study gave their informed consent. Each patient, including those who were later lost to follow-up, was given a consecutive, randomly distributed number at entry. Patients with even and odd numbers constituted the moderate- and high-intensity groups, respectively. Randomization was substratified for the presence or absence of atrial fibrillation. The patients remained in the trial until death or the termination of the study in September 1986.

Prothrombin Time

Blood (nine volumes) was collected in silicon-coated, evacuated glass tubes containing one volume of 3.8 percent sodium citrate (total volume, 4.5 ml) (Vacutainer, Becton Dickinson, Rutherford, N.J.) and platelet-poor plasma separated after centrifugation at 2000×g and 4°C for 10 minutes. The prothrombin time of the platelet-poor plasma was measured with use of Simplastin (General Diagnostics, Morris Plains, N.J.), a rabbit-brain thromboplastin with an international sensitivity index of 2.4, and an automated, optical-end-point, blood-coagulation instrument (Coag-a-mate X2, General Diagnostics).

The results were reported as a prothrombin-time ratio without conversion to an international normalized ratio, and they were available in the anticoagulation clinic when patients were seen for adjustment of their dose of warfarin.

Anticoagulation Therapy

The aim of treatment in the moderate-intensity group was a prothrombin-time ratio of 1.5 (international normalized ratio, 2.65), and the acceptable range was 1.3 to 1.7. In the high-intensity group the target was a ratio of 2.5 (international normalized ratio, 9.0), and the acceptable range was 2.3 to 2.7.

Oral anticoagulant therapy was begun within 10 days of surgery. All the patients were seen at the anticoagulation clinic within one week of discharge, and then every few days until their prothrombin-time ratio stabilized in the intended range. Thereafter, the patients were seen every four weeks, or more frequently if their prothrombin-time ratio was outside the intended range. None were treated initially with antiplatelet drugs. Dipyridamole (75 mg three times a day) was given later to patients who had a thromboembolic episode.

All the patients were advised of the risks of treatment and provided with a pamphlet stressing the adverse effects of warfarin and emphasizing the dangers of drug interaction, particularly with aspirin and other nonsteroidal antiinflammatory drugs. They were also told not to take any unprescribed medication. All the patients were also given a card with a list of their medications, as well as instructions to call the clinic or go to the emergency room if bleeding or symptoms suggestive of a thromboembolic episode developed. A list of the manifestations of thromboembolism or bleeding was included. The patients were told to carry the card at all times and to show it to any doctor who saw them between visits to the clinic.

At the anticoagulation clinic, a record of the prothrombin-time ratio was kept, and bleeding or thromboembolic episodes were also recorded.

A diagnosis of cerebral embolism was accepted for the purposes of the study if there was a residual neurologic deficit and if a computerized tomographic scan of the brain excluded bleeding as a possible cause. Other systemic embolization was confirmed by angiography or surgery. Suspected coronary-artery embolism was confirmed by coronary angiography or by the onset of myocardial infarction as evidenced by typical electrocardiographic changes and elevated levels of cardiac enzymes in a young person with coronary arteries that had been normal on angiography.

Bleeding was considered major if the patient died or required transfusion, hospitalization, or surgery. All other bleeding was defined as minor.

Statistical Analysis

The data were analyzed on a TANDON ATcompatible microcomputer with use of the SOLO statistical software package. The results of logistic-regression analysis were confirmed with use of the BMDP statistical package. The variables investigated were the intensity of anticoagulation, the duration of therapy, and the patient's age, sex, and history of smoking, diabetes, and atrial fibrillation. Because none of the patients had hypertension, it was not included in the final analysis. Relative risks were derived from the logistic-regression coefficient.

Results

Patients

Table 1. Table 1. Rate of Accession, Number of Patients Assigned to Treatment and Followed up, and Length of Follow-up. Table 2. Table 2. Age and Sex of Patients, Type and Site of Valves, and Prevalence of Atrial Fibrillation.

The characteristics of the patient groups are shown in Tables 1 and 2. We randomly assigned 258 patients to the treatment groups, but 11 patients (5 from the moderate-intensity and 6 from the high-intensity group) were lost to follow-up. Our analyses therefore involved 247 patients. All the patients were normotensive on entry and remained so during the follow-up period.

Of the 11 patients who were lost to follow-up, 9 were citizens of Yemen, to which they returned six months after their operations. None had had bleeding or thromboembolic episodes. The other two (one from each group) lived in Riyadh but were lost to follow-up seven and nine months after their operations, without any earlier complications.

Dose of Warfarin and Anticoagulant Effect

The mean (±SD) doses of warfarin in the moderate- and high-intensity groups were 5.9±2 and 8.5±2.3 mg, respectively. The anticoagulant effects were within the intended ranges (prothrombin-time ratios, 1.3 to 1.7 and 2.3 to 2.7, respectively) at 86 percent of the visits.

Thromboembolic Episodes and Prothrombin-Time Ratio

Table 3. Table 3. Thromboembolic Episodes in the Two Treatment Groups.

There were 33 thromboembolic episodes among the 247 study patients (3.85 per 100 patient-years), 17 in the moderate-intensity group and 16 in the high-intensity group, corresponding to embolism rates of 4.0 and 3.7 episodes per 100 patient-years, respectively. Table 3 shows the 95 percent confidence intervals for the observed incidence of systemic embolism and the sites of thromboembolism. There was no difference in the incidence or relative risk of thromboembolism between the two groups. None of the thromboembolic episodes occurred very early after the insertion of the prosthetic valve — i.e., before proper anticoagulant therapy. One third of the episodes occurred within the first 12 months of therapy.

Figure 1. Figure 1. Prothrombin-Time Ratios at Time of Patients' Presentation with Thromboembolic Events.

Prothrombin-time ratios measured at the time of the patients' presentation with thromboembolism are shown in Figure 1. The ratio in six patients with systemic embolism (three from each group) was 1.0. All six had discontinued anticoagulant therapy of their own volition before their thromboembolic episode. The prothrombin-time ratio in 12 patients (7 from the moderate-intensity group and 5 from the high-intensity group) was low, between 1.1 and 1.3. Ratios in the remaining 15 patients with thromboembolism ranged between 1.4 and 2.2.

Bleeding Episodes, Possible Predisposing Factors, and Prothrombin-Time Ratio

Table 4. Table 4. Bleeding Episodes in the Two Treatment Groups.

Table 4 shows the sites of major and minor bleeding episodes in both groups. There were 79 bleeding episodes (9.2 per 100 patient-years); 66 were minor (7.7 per 100 patient-years) and 13 were major (1.5 per 100 patient-years), including 2 fatal episodes of intracranial bleeding.

Forty-four minor and nine major episodes of bleeding (including both fatal episodes) occurred in the high-intensity group, which had a significantly greater risk of any bleeding (P<0.002) and of minor bleeding (P<0.01). The trend toward a greater risk of major bleeding was not statistically significant (P>0.05). The relative risk of any bleeding in the high-intensity group as compared with the moderate-intensity group was 2.5 (95 percent confidence interval, 1.4 to 4.4), and of minor bleeding 2.25 (95 percent confidence interval, 1.2 to 4.7).

Five of the 13 patients with major bleeding (including the 2 with fatal intracranial bleeding) had no apparent predisposition other than warfarin therapy, and all 5 were in the high-intensity group. Another four patients (two from each group) had possible predisposing factors for major bleeding: the recent intake of aspirin or a nonsteroidal antiinflammatory drug in three and a possible duodenal ulcer in one. The prothrombin-time ratio was above 3.0 in all nine of these patients. Probable contributing factors in the remaining four patients with major bleeding (two from each treatment group) included local trauma in two, a kidney stone in one, and a renal tumor in one; their prothrombin-time ratios at the time of bleeding were between 1.6 and 2.1.

Discussion

This open, randomized trial compared the incidence of systemic thromboembolism and bleeding complications in patients with mechanical heart-valve prostheses who received warfarin at one of two levels of anticoagulant intensity (moderate prothrombin-time ratio, 1.5, and high prothrombin-time ratio, 2.5; target international normalized ratios, 2.65 and 9) selected to correspond to the extremes of the accepted therapeutic range in many centers.15 , 16

Rates of thromboembolism were similar in the moderate- and high-intensity groups, but the overall risk of bleeding and the incidence of minor bleeding episodes were significantly greater during high-intensity therapy. The 95 percent confidence intervals for the observed rates of systemic embolism in the two groups remained wide (about 8 to 20 percent), so a moderate but real difference in the efficacy of the two warfarin regimens is still possible, but several considerations led us to terminate the trial at this stage.

First, previously reported rates of thromboembolism are similar to those we observed.1 These rates,4 expressed as episodes per 100 patient-years of follow-up, vary between 1.2 and 9.3 for ball valves, 0.7 and 5.6 for Björk–Shiley valves, and 0.6 and 3.9 for St. Jude valves. Overall, the previously reported incidence is about 4 thromboembolic episodes per 100 patient-years, as compared with 3.85 in our study. Second, it would take a trial of impractical size or duration to seek even the largest difference in rates of systemic embolism between treatment groups that remains consistent with our results (e.g., between 20 and 10 percent, corresponding to event rates of 5.8 and 2.9 per 100 patient-years). Finally, it would be difficult to justify continuing the study in the face of the statistically significant excess of bleeding episodes in the high-intensity group, including two fatal episodes.

Major bleeding has been reported in about 2.4 percent of patients during long-term therapy with oral anticoagulants after valve replacement, and fatal bleeding in 1.7 percent.17 These figures can be compared with those for our patients: major bleeding (none fatal) in 3.3 percent of the patients in our moderate-intensity group and 7.2 percent of those in our high-intensity group (1.6 percent fatal). The consideration of rates of major bleeding per 100 patient-years allows more direct comparisons. The rate was about 1.7 in previous surveys,17 as compared with 0.95 and 2.1 in our moderate- and high-intensity groups, respectively.

This study confirms the presence of a direct relation between the intensity of anticoagulation and the risk of bleeding. Of the 13 major bleeding episodes in our patients, 9 occurred when the prothrombin-time ratio was above 3.0 (international normalized ratio, >10.0); the remaining four, in which the ratio was between 1.6 and 2.1, had recognizable precipitating factors.

There was also a relation between thromboembolism and the prothrombin-time ratio, since 18 of the 33 embolic events (54 percent) were associated with a ratio of 1.3 or less at onset. Factors other than the adequacy of warfarin therapy may have a role at higher levels of anticoagulation, such as the type of valve prosthesis — thromboembolism was most frequent in patients with Beall valves and least frequent in those with St. Jude valves.

Eleven patients were lost to follow-up. All had been seen for at least six months after surgery, and none had had thromboembolic or bleeding episodes.

The study was not double blind, since the investigators who saw the patients during follow-up needed to adjust the warfarin dosage to attain the targeted prothrombin-time ratio. To avoid bias, all the patients were asked a uniform set of questions about thromboembolism and bleeding at every visit to the anticoagulation clinic.

Controlled trials have shown that secondary prevention after venous thromboembolism requires a less intense anticoagulant effect (international normalized ratio, 2.0 to 2.5) than was previously thought.14 More recently, it has become apparent that increasing the target range of the international normalized ratio from between 2.0 and 2.25 to between 2.5 and 4.0 in patients treated with warfarin for three months after the implantation of a tissue heart valve confers no therapeutic benefit, but carries an increased risk of bleeding.18 We now demonstrate that a moderate warfarin effect is also sufficient for the long-term treatment of patients with mechanical heart valves.

Funding and Disclosures

We are indebted to W. Greer, Ph.D., chairman of the Department of Biomedical Statistics and Scientific Computing, for his valuable assistance with the statistical analysis.

Author Affiliations

From the Department of Medicine, King Faisal Specialist Hospital and Research Centre (J.N.S., J.O.S.), and King Saud University (L.A.R.M.), Riyadh, Saudi Arabia, and Flinders Medical Centre, Bedford Park, Australia (A.S.G.). Address reprint requests to Dr. Saour at the Department of Medicine, King Faisal Specialist Hospital and Research Centre, P.O. Box 3354, Riyadh 11211, Saudi Arabia.

References (18)

  1. 1. Stein PD, Collins JJ Jr, Kantrowitz A. . Antithrombotic therapy in mechanical and biological prosthetic heart valves and saphenous vein bypass grafts . Chest 1986; 89:Suppl:46S–53S.

  2. 2. Dale J. . Arterial thromboembolic complications in patients with Starr-Edwards aortic ball valve prostheses . Am Heart J 1976; 91:653–9.

  3. 3. Horstkotte D, Körfer R. The influence of prosthetic valve replacement on the natural history of severe acquired heart lesions: a comparison of complications and clinical and hemodynamic findings after implantation of Björk–Shiley, St. Jude Medical, and other heart valve prostheses. In: De Bakey ME, ed. Advances in cardiac valves: clinical perspectives. New York: Yorke Medical Books, 1983:47–86.

  4. 4. Chesebro JH, Ezekowitz M, Badimon L, Fuster V. . Intercardiac thrombi and systemic thromboembolism: detection, incidence, and treatment . Annu Rev Med 1985; 36:579–605.

  5. 5. Christo MC, Souza JM, Stortini MJ, et al. . Late complications of cardiac valve replacement with Lillehei-Kaster pivoting disc prosthesis . Artif Organs 1980; 4:199–201.

  6. 6. Yeh TJ, Anabtawi IN, Cornett VE, Ellison RG. . Influence of rhythm and anticoagulation upon the incidence of embolization associated with Starr-Edwards prostheses . Circulation 1967; 35:Suppl I:I-77–I-81.

  7. 7. Duvoisin GE, Brandenburg RO, McGoon DC. . Factors affecting thromboembolism associated with prosthetic heart valves . Circulation 1967; 35:Suppl I:I-70–I-76.

  8. 8. Akbarian M, Austen G, Yurchak PM, Scannell JG. . Thromboembolic complications of prosthetic cardiac valves . Circulation 1968; 37:826–31.

  9. 9. Friedli B, Aerichide N, Grondin P, Campeau L. . Thromboembolic complications of heart valve prostheses . Am Heart J 1971; 81:702–8.

  10. 10. Barnhorst DA, Oxman HA, Connolly DC, et al. . Long-term follow-up of isolated replacement of the aortic or mitral valve with the Starr-Edwards prosthesis . Am J Cardiol 1975; 35:228–33.

  11. 11. Fuster V, Pumphrey CW, McGoon MD, Chesebro JH, Pluth JR, McGoon DC. . Systemic thromboembolism in mitral and aortic Starr-Edwards prostheses: a 10–19-year follow-up . Circulation 1982; 66:Suppl I:I-157–I-161.

  12. 12. Poller L, Taberner DA. . Dosage and control of oral anticoagulants: an international collaborative survey . Br J Haematol 1982; 51:479–85.

  13. 13. Hirsh J, Deykin D, Poller L. . "Therapeutic range" for oral anticoagulant therapy . Chest 1986; 89:Suppl:11S–15S.

  14. 14. 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; 307:1676–81.

  15. 15. 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; 51:1537–41.

  16. 16. Czer LS, Matloff JM, Chaux A, De Robertis M, Stewart ME, Gray RJ. . The St. Jude valve: analysis of thromboembolism, warfarin-related hemorrhage, and survival . Am Heart J 1987; 114:389–97.

  17. 17. Levine MN, Raskob G, Hirsh J. . Hemorrhagic complications of long-term anticoagulant therapy . Chest 1986; 89:Suppl:16S–25S.

  18. 18. Turpie AGG, Gunstensen J, Hirsh J, Nelson H, Gent M. . Randomised comparison of two intensities of oral anticoagulant therapy after tissue heart valve replacement . Lancet 1988; 1:1242–5.

Citing Articles (253)

    Letters

    Figures/Media

    1. Table 1. Rate of Accession, Number of Patients Assigned to Treatment and Followed up, and Length of Follow-up.
      Table 1. Rate of Accession, Number of Patients Assigned to Treatment and Followed up, and Length of Follow-up.
    2. Table 2. Age and Sex of Patients, Type and Site of Valves, and Prevalence of Atrial Fibrillation.
      Table 2. Age and Sex of Patients, Type and Site of Valves, and Prevalence of Atrial Fibrillation.
    3. Table 3. Thromboembolic Episodes in the Two Treatment Groups.
      Table 3. Thromboembolic Episodes in the Two Treatment Groups.
    4. Figure 1. Prothrombin-Time Ratios at Time of Patients' Presentation with Thromboembolic Events.
      Figure 1. Prothrombin-Time Ratios at Time of Patients' Presentation with Thromboembolic Events.
    5. Table 4. Bleeding Episodes in the Two Treatment Groups.
      Table 4. Bleeding Episodes in the Two Treatment Groups.

      More Research