Low-Dose International Normalized Ratio Self-Management: A Promising Tool to Achieve Low Complication Rates After Mechanical Heart Valve Replacement

Transcription

1 Low-Dose International Normalized Ratio Self-Management: A Promising Tool to Achieve Low Complication Rates After Mechanical Heart Valve Replacement Heinrich Koertke, MD, Armin Zittermann, PhD, Kazutomo Minami, MD, Gero Tenderich, MD, Otto Wagner, Mahmoud El-Arousy, MD, Arno Krian, MD, Juergen Ennker, MD, Uwe Taborski, MD, Wolf Peter Klövekorn, MD, Rainer Moosdorf, MD, Werner Saggau, MD, Michiel Morshuis, MD, Jan Koerfer, MD, Dirk Seifert, MD, and Reiner Koerfer, MD Department of Thoracic and Cardiovascular Surgery, Heart and Diabetes Center North-Rhine-Westphalia Bad Oeynhausen, Clinic of the Ruhr University Bochum, Bochum; Evangelisches Johanniter Klinikum, Duisburg; Heart Center Lahr, Lahr; Kerckhoff- Klinik, Bad Nauheim; Heart Center Marburg, Clinic of the Philipps-University Marburg, Marburg; and Klinikum Ludwigshafen, Clinic for Heart Surgery, Ludwigshafen, Germany Background. International normalized ratio (INR) selfmanagement can significantly reduce INR fluctuations, bleeding, and thromboembolic events compared with INR control managed by general practitioners. However, even patients with INR self-management may have an increased risk of bleeding if their INR value is above 3.5. This study evaluated the compliance, clinical complications, and survival of patients after mechanical heart valve replacement with low-dose INR self-management compared with conventional-dose anticoagulation. Methods. Group 1 (n 908) received low-dose anticoagulation with a target INR range of 1.8 to 2.8 for aortic valve replacement and 2.5 to 3.5 for mitral or double valve replacement. Group 2 (n 910) received conventional-dose anticoagulation with a target INR range of 2.5 to 4.5 for all heart valve prostheses. Results. In groups 1 and 2, 76% and 75% of INR values, respectively, were in the target range. Results did not differ according to schooling and age. The rate of thromboembolic events per patient year was 0.18% in group 1 and 0.40% in group 2 (p 0.210). The rate of bleeding complications was 0.74% for group 1 and 1.20% for group 2(p 0.502). In most patients with clinically relevant bleeding, these complications occurred although their measured INR values were below 3.5. The survival rate did not differ between the study groups (p 0.495). Conclusions. Low-dose INR self-management is a promising tool to achieve low hemorrhagic complications without increasing the risk of thromboembolic complications. INR self-management is applicable for all patients in whom permanent anticoagulation therapy is indicated. Even INR values below 3.5 can bear the risk of bleeding complications. (Ann Thorac Surg 2005;79: ) 2005 by The Society of Thoracic Surgeons During recent decades, heart valve replacement has become a frequent intervention in cardiovascular medicine. Mechanical heart valves account for approximately 70% and biological heart valves for approximately 30% of totally implanted heart valves. Because biological heart valves are durable for only 8 to 12 years in the aortic position and only 6 to 8 years in the mitral position, these implants are restricted to elderly patients [1]. Compared with biological heart valves, mechanical heart valves bear a much higher risk of thromboembolism; thus, anticoagulation therapy after mechanical heart valve replacement is mandatory. Hemorrhage is a possible complication of anticoagulation therapy. Bleeding and thromboembolism account for 75% Accepted for publication Sept 7, Presented at the Forty-first Annual Meeting of The Society of Thoracic Surgeons, Tampa, FL, Jan 24 26, Address reprint requests to Dr Koertke, Herz- und Diabeteszentrum NRW, Georgstrasse11, Bad Oeynhausen, Germany; hkoertke@ hdz-nrw.de. of all complications after mechanical heart valve replacement. The risk of complications in conjunction with ongoing anticoagulation therapy is considerably higher when international normalization ratio (INR) values fluctuate strongly. When anticoagulation-induced complications occur, as many as 60% of the coagulation values controlled are outside the therapeutic range [2, 3]. INR self-management allows a close-meshed control of the intensity of anticoagulation. The Early Self-Controlled Anticoagulation Trial I (ESCAT I) showed that selfmanagement could reduce the INR fluctuations. Approximately 78% of the INR values of patients under INR self-management were within the therapeutic range compared with only 60% of the INR values in those patients managed by their family doctors [4]. Moreover, ESCAT I demonstrated a significant reduction in complications and in the mortality rate of patients with INR self-management compared with conventionally treated patients [5]. Until recently, a therapeutic INR range from 2.5 to 4.5 has been regarded as necessary to prevent thromboembolism; however, this range includes a zone of higher risk for bleeding 2005 by The Society of Thoracic Surgeons /05/$30.00 Published by Elsevier Inc doi: /j.athoracsur

2 1910 KOERTKE ET AL Ann Thorac Surg LOW-DOSE INR SELF-MANAGEMENT 2005;79: complications, beginning with an INR that exceeds 3.5. In addition, there is evidence from recent data analyses on oral anticoagulation and risk of death that an INR target value of 3.5 may already be too high [6]. In that earlier study, a minimum risk of death was attained at 2.3 INR for patients with mechanical heart valve prostheses. With an increase of 1 U of INR above 2.5, the risks of death from cerebral bleeding and from any cause were nearly doubled. Recently, target ranges of 2.0 to 3.0 INR for aortic valve prostheses and 2.5 to 3.5 INR for mitral valve prostheses have been recommended [7]. Nevertheless, these guidelines are based on only a few studies, and controlled prospective trials on complication rates in patients who are on low-dose INR self-management are almost completely lacking. Encouraged by the high percentage of INR values lying within the therapeutic range in the ESCAT I study, we began ESCAT II to eliminate the high-risk zone for bleeding. ESCAT II was thus designed to assess the effects of a low-dose INR therapeutic range accompanied by self-management. ESCAT II is an ongoing prospective, controlled, randomized multicenter study with 3300 patients. The study began in October We recently presented the first results of a planned interim analysis indicating that INR selfmanagement permits lower anticoagulation levels after mechanical heart valve replacement [8]. In a very recent meta-analysis, however, it was assumed that patients with mechanical heart valves would benefit from a treatment strategy with a target INR higher than 3.0 [9]. We therefore performed an extended analysis of our interim data to further evaluate our strategy of low-dose INR self-management. Material and Methods Patients The interim analysis comprised 1818 patients, 18 to 82 years of age. All participants underwent a mechanical aortic, mitral, or double valve replacement. Exclusion criteria were a contraindication to phenprocoumon (eg, allergy), known ulcerous disease with bleeding tendency, hypocoagulability or hypercoagulability (medical history), and an age less than 18 years. All patients gave written informed consent to the study procedures. The study protocol was approved by the ethics committee of each of the six participating centers. Study Protocol Implanted were 840 Medtronic Hall heart valves (Medtronic GmbH, Düsseldorf, Germany), 895 St. Jude Medical heart valves (St. Jude Medical GmbH, Nürnberg, Germany), and 83 other heart valves from different manufacturers (38 Carbo- Medics valves [Austin, TX]; 29 ATS Medical valves [Minneapolis, MN], 2 Mosaic valves [Medtronic, Minneapolis, MN], 2 Omnicarbon valves [MedicalCV, Inner Grove Heights, MN], 8 On-x valves [MCRI, Austin, TX], and 4 Pyrolite valves [Sulzer- Medica, Austin, TX]). After heart valve replacement, the patients were randomly assigned to either the low INR range (group 1) or the conventional INR range (group 2) group. The conventional group (n 908) had a target INR range of 2.5 to 4.5. The low-dose group (n 910) had a target range of 1.8 to 2.8 for aortic valve recipients and 2.5 to 3.5 for mitral or double valve recipients. The study began in October For this interim analysis, only data that were available until December 2001 were included. All study participants joined a training course for INR self-management at our clinics which is based on the assumption that patients who depend on long-term anticoagulation therapy are able to selfresponsibly determine the INR values and correct the dose of anticoagulants. Initial training began between postoperative day 6 and day 11. Training lasted 5 hours and was terminated by a short test. INR measurements were performed with a coagulation monitor (CoaguCheck S, Roche Diagnostics, Mannheim, Germany). During the training course, the patients were instructed that any INR value between the two extremes was acceptable. However, the patients of the low-dose group were also instructed that the desired INR value should be close to 2.3 (aortic valve recipients) or 3.0 (mitral or double valve recipients), whereas the patients of the conventional dose group were instructed that the desired INR value should be close to 3.0. INR values were usually determined weekly. The patients had to check their INR values once a week during the first study year and once a fortnight during the second study year. Results had to be transmitted to the study center every month. All patients had to visit their study center for a cardiologic check-up every 6 months. During that visit, the patients had to demonstrate that they were able to measure the INR values correctly and interpret the results accordingly. The third phase of the training consisted of telephone care and control and the possibility of a consultation around the clock. The patients had the opportunity to adjust their anticoagulation therapy with the help of specialized cardiologists. Patients were asked to report any complication to the study center immediately. Thromboembolism complications and bleeding events were graded as I, II, or III. Briefly, a grade III thromboembolism was defined as heart valve prosthesis thrombosis or severe thromboembolism that required inpatient treatment or caused long-term impairment, including transient ischemic attacks. A grade II thromboembolism required outpatient treatment but caused no permanent impairment, and a grade I thromboembolism (questionable events) required no medical treatment. Grade III bleeding was defined as severe bleeding that required transfusion, surgical or endoscopic intervention, inpatient care, or that caused long-term impairment. Grade II bleeding led to outpatient medical care but did not require surgical or endoscopic intervention, and grade I bleeding (mild bleeding) required no medical treatment. Only grade III complications were used for data analysis, as previously described [8]. Data collection was both passive and active: Patients were asked to inform their local study centers when a thromboembolism or bleeding complication had occurred. Moreover, they were asked during their follow-up visits whether a complication had occurred within the last 6 months. All reports of grade III complications were sent from the emergency hospital to the study center and were double-checked by a center cardiologist. Finally, the study center in Bad Oeynhausen contacted each local study

3 Ann Thorac Surg KOERTKE ET AL 2005;79: LOW-DOSE INR SELF-MANAGEMENT 1911 Table 1. Characteristics of Patients With Low-Dose International Normalized Ratio (INR) Self-Management and With Conventional Dose INR Self-Management Low Dose n 910 Conventional Dose n 908 Significance (p) Age (yrs) NS Male (%) Body weight (kg) Body height (meter) Smoker (n) NS Hypertension (n) NS Hypercholesterolemia (n) NS Atrial fibrillation (n) NS Red blood cells (10E12/I) NS Hemoglobin (g/dl) NS Hematocrit (%) NS Fibrinogen (mg/dl) NS Total cholesterol (mg/dl) NS HDL-cholesterol (mg/dl) NS LDL-cholesterol (mg/dl) NS Blood pressure, systolic (mm Hg) NS Blood pressure, diastolic (mm Hg) NS Heart rate (beats/min) NS LDL low-density lipoprotein; HDL high-density lipoprotein; NS not significant. center regularly to get information about the number of bleeding and thromboembolism events. Two years after valve implantation, the patients stopped sending their INR values to the study center. Only grade III complication and deaths that occurred within 24 months of heart valve replacement were used for data analysis. Statistical Analysis All statistical evaluations were performed with the Statistical Package for Social Sciences (SPSS), version 11 (SPSS, Chicago, IL). For comparative evaluations, the 2 test and the Student t test (normal distributed data) were used. Normal distribution of the data was tested by the Kolmogorov-Smirnov test. Freedom from grade III complications and survival rates were calculated with the Kaplan-Meier product-limit estimator. Differences in grade III complications and survival rates between the low-dose INR group and the conventional-dose INR group were tested with the log-rank test. Data analysis was based on the assumption that for a statistical power of 0.8 ( 0.05; 0.20), a reduction of grade III complications by 2% (eg, from 3% to 1%) or an increase by 3% (eg, from 3% to 6%) would require a study population of 1690 patients or 1310 patients. A study population of 3300 or 2360 patients would be necessary to achieve an value of 0.01 under otherwise similar assumptions. A p value less than 0.05 (two-tailed test) was considered statistically significant; p values of Kaplan-Meier survival analysis greater than 0.05 and less than 0.15 were considered borderline significant [10]. Data were expressed as mean values standard deviation unless otherwise stated. Results In the low-dose INR group, 81% of patients received aortic valve prostheses and 12.5% and 6.5% received double valve prostheses in mitral and double positions, respectively. The corresponding values for the conventional INR group were 77%, 16.5%, and 6.5%, respectively. Baseline characteristics of the study groups are given in Table 1. The conventional-dose INR group included a slightly higher percentage of women, had patients with a lower body weight, and the body height tended to be lower compared with the low-dose INR group. Cardiovascular risk factors such as smoking, hypertension, and atrial fibrillation were equally distributed in both study groups, and the study groups had similar laboratory data. The average weekly phenprocoumon dose was 17.4 mg in the conventional-dose group and 16.5 mg in the low-dose group. The average observation time was 13 months/ patient (range, 2 to 24 months), leading to a total observation time of 2848 patient years. Of the 1818 patients, 239 (13.15%) terminated the study prematurely for a variety of reasons: they could not manage the device (n 54); they had more confidence in their physicians (n 65); measurements were stopped by illness (n 32); patients were noncompliant (n 44) or not interested in INR selfmanagement (n 9); they missed or had incomplete instructions for INR self-management (n 21); health insurance did not pay for the device (n 2), and travel costs to the place of reexamination were too high (n 3). Nine patients terminated with no comment. The percentage of study dropouts in the low-dose and the conventional-dose groups did not differ significantly (data not shown). The patients submitted 73,148 INR values to the study centers during the investigation period. Mean INR values in patients with aortic valve replacement were 2.41 in the low-dose group and 2.80 in the conventional-dose group, whereas in patients with mitral and double valve replace-

4 1912 KOERTKE ET AL Ann Thorac Surg LOW-DOSE INR SELF-MANAGEMENT 2005;79: Fig 2. International normalized ratio (INR) values within the therapeutic range in patients with low-dose (solid squares) or conventional-dose (solid circles) INR self-management over time (reha rehabilitation.) Fig 1. International normalized ratio (INR) values of patients with INR self-management: (A) Low dose of anticoagulation and aortic valve replacement. (B) Conventional dose of anticoagulation and aortic valve replacement. (C) Low dose of anticoagulation and mitral or double valve replacement. (D) Conventional dose of anticoagulation and mitral or double valve replacement. ment, they were 2.90 in the low-dose group and 3.10 in the conventional-dose group (Fig 1). Approximately 76% of the INR values measured in the low-dose group and approximately 75% of the INR values in the conventional-dose group were within group s target range. These values remained very stable in both outpatient groups throughout the study period (Fig 2). Of the INR values submitted, 3.86% were below the lowest target range (1.8 INR) and 1.08% were above the highest target range (4.5 INR). As expected, more INR values in the low-dose group were below 1.8 INR compared with the conventional-dose group (5.12% vs 2.57%). Unsurprisingly too, only 4.76% of the values in the low-dose group were above INR 3.5, whereas 13.76% of the values in the conventional-dose group were above this level (Fig 1). In Figure 3, the percentage of INR values in the target range is given in relation to schooling. In the German school system, comprehensive school is the lowest educational level and grammar school is the highest preuniversity educational level. Data show that educational level did not have a major influence. In Figure 4, the percentage of INR values in the target range is given according to age. Results showed no major difference between the age groups. During the entire observation period, 41 grade III complications occurred (32 bleeding events and 9 thromboembolisms). All but 3 complications occurred in the patients with an aortic valve replacement. Fifteen complications (3 thromboembolisms, 12 bleeding events) occurred in the low-dose INR group, whereas 26 complications (6 thromboembolisms, 20 bleeding events) occurred in the conventional-dose INR group. Fewer bleeding events and also slightly fewer cases of thromboembolism occurred in the low-dose group compared to the conventional-dose group (0.74% vs 1.20% and 0.18% vs 0.40%, respectively). In both cases the differences between the two study groups were not statistically significant (p and p 0.210, respectively). However, the total complication rate of bleeding and thromboembolic events tended to be lower in the low-dose group than in the conventional-dose group (Fig 5). In the patients with grade III thromboembolism, the mean INR values were 2.3 (range, 1.0 to 3.4) at week 4 preceding the insult, 2.5 (range, 1.0 to 3.7) at week 3, 2.4 (range, 1.4 to 4.4) at week 2, 2.3 (range, 1.4 to 3.1) at week 1, and 2.1 (range, 1.1 to 3.0) at week 0. Five of the 9 patients had at least one INR value below 1.8 during the 4 weeks before the complication occurred; however, in 2 patients INR values did not fall below 2.5. In those 15 patients with grade III bleeding events, where Fig 3. International normalized ratio (INR) values in patients with INR self-management according to schooling.

5 Ann Thorac Surg KOERTKE ET AL 2005;79: LOW-DOSE INR SELF-MANAGEMENT 1913 Fig 4. International normalized ratio (INR) values in patients with INR self-management in relation to age. the complication occurred was not induced by an external event, INR values were clearly below 3.5 INR, a value that is generally regarded as safe. In detail, mean INR values were 2.8 (range, 1.8 to 3.4) at week 4 preceding the insult, 2.8 (range, 1.8 to 3.4) at week 3, 2.9 (range, 1.8 to 4.0) at week 2, 3.0 (range, 1.7 to 4.8) at week 1, and 2.9 (range, 1.5 to 3.9) at week 0. Only 7 out of the 15 patients with a grade III bleeding event had at least one INR value above 3.5 during the 4 weeks before the complication occurred. In 3 patients a bleeding episode occurred although the measured INR values did not exceed 2.8. Twenty-eight patients died within 30 days of surgery (12 in the low-dose and 16 in the conventional-dose group). Twenty-one late deaths occurred in the low-dose group and 14 in the conventional-dose group. Only 1 patient (from the conventional-dose group) died because of an embolic event. The other causes of death were myocardial infarction (n 26), intraoperative death (n 6), unknown (n 4), multiorgan failure (n 11), atrial fibrillation (n 7), lung failure (n 1), renal failure (n 3), neoplasm (n 2), and other causes (n 2). Survival rate did not differ between study groups (Fig 6). Fig 5. Freedom from grade III complications (heart valve prosthesis thrombosis or severe thromboembolism that required inpatient treatment or caused long-term impairment, including transient ischemic attacks) in patients with low-dose or conventional-dose international normalized ratio self-management. Fig 6. Survival of patients with international normalized ratio self-management and low-dose or conventional dose of anticoagulation therapy. Comment This interim analysis of the ESCAT II study demonstrates that it is possible to maintain approximately 75% of the INR values in the target zone, even when the target range is narrow. Moreover, our data show that the incidence of thromboembolism and bleeding complications can be reduced to less than 2% per observation year in the conventional-dose INR group and to less than 1% per observation year in the low-dose INR group. It is encouraging that both types of complications tended to be lower in the low-dose group than in the conventional-dose group. Thus, our data do not support the results of a recent meta-analysis that indicated patients benefit from a target INR higher than 3.0 [8]. It should be mentioned, however, that our study design is based on a concept that enables the patients to perform INR measurements self-responsibly and also allows them to selfadjust the anticoagulation medication. The 24-hour care and consultation confirms the patients self-confidence, especially in those situations when questions and difficulties arise. Therefore, the results of our study and the data of that earlier meta-analysis must be compared with caution. We have already demonstrated in the ESCAT I study that patients whose INR values were measured by their family doctors had approximately 18% fewer INR values lying within the therapeutic range than did the self-management INR group (60% vs 78%) [4]. The incidence of hemorrhagic complications in the ESCAT I study was 2.6% in the patients whose INR measurements were managed by the family doctor and 1.7% in the group with conventional INR self-management. In the ESCAT II study, the incidence of hemorrhagic complications tended to be lower in both study groups (1.2% and 0.7%) compared with the ESCAT I study INR self-management group. These differences in hemorrhagic complications between ESCAT I and ESCAT II are most probably related to the differences in INR values. In those ESCAT II patients where almost all hemorrhagic complications occurred (patients with aortic valve prostheses), the mean INR values were only 2.8 in the conventional-dose INR group and 2.4 INR in the low-dose INR group; whereas in the ESCAT I study, the patients with INR self-management had a mean INR value of 3.0 [4].

6 1914 KOERTKE ET AL Ann Thorac Surg LOW-DOSE INR SELF-MANAGEMENT 2005;79: It should be mentioned that the INR zone that is necessary to minimize bleeding complications in patients who are on anticoagulation therapy is not well established. Hemorrhagic complications are reported to increase sharply when INR values are above 6.0 [11]. The number of INR values above this level was negligible (Fig 1) in our study. However, in a recent study a minimum risk of death occurred in patients with mechanical heart valve prostheses at an INR value of 2.3, while the risk of death from cerebral bleeding doubled with an increase of 1 U of INR above 2.5 [6]. Our data confirm the assumption that bleeding complications may occur in some patients with aortic valve prostheses who have relatively low INR values. Mean values in those patients with grade III bleeding complications were indeed only 2.9 INR within the last weeks before the complication occurred. Surprisingly, the values of some patients did not exceed 2.8 INR. Data indicate that the relatively low mean INR values in both ESCAT II groups may have contributed to the overall relatively low incidence of grade III bleeding complications. Consequently, a clear rationale exists to propagate a low range and narrow target zone of INR self-management. Moreover, the etiology of life-threatening hemorrhage under obviously adequate oral anticoagulation has to be further characterized in future studies. Without anticoagulation therapy, patients with mechanical prosthetic heart valves have an incidence of thromboembolism of 8.6% per observation year [12]. Some of our patients with thromboembolic events had INR values below 1.2, indicating effectively no anticoagulation. Nevertheless, it should be mentioned that in the ESCAT II study, even low-dose INR self-management was able to reduce the incidence of thromboembolism to below 0.5% per observation year. Data indicate that the target value of 1.8 INR is generally safe for the group with aortic valve prostheses and low-dose INR self-management. Moreover, this target range is close to the target range of 2.0 to 3.0 INR for aortic valve prostheses, which has recently been recommended [7]. An anticoagulation level of INR 3.0 to 4.0, as recommended in earlier years [13] for patients with heart valve prostheses, no longer seems to be the optimal anticoagulation therapy. Our data clearly demonstrate that low-dose INR self-management does not increase the risk of thromboembolism compared with conventional-dose INR self- management. Because the dropout rate of the study participants was less than 15%, it can be concluded that INR selfmanagement is generally well accepted by the patients. Our evaluation showed that the percentage of INR values within the target range was not influenced by age nor did educational level have a major impact. In conclusion, low-dose INR self-management is a promising tool to achieve low hemorrhagic complications without increasing the risk of thromboembolic complications. The combination of the low-dose of anticoagulants and the INR self-management may be responsible for these favorable results. The technique allows patients to avoid serious sequelae because of a phenprocoumon overdose. The concept of INR self-management is applicable, independent of age and educational level, for all patients in whom permanent anticoagulation therapy is indicated. References 1. Koerfer R, Koertke H. ESCAT early self-controlled anticoagulation trial. In: Krian A, ed. Advancing technology of bileaflet mechanical heart valves. Darmstadt, Germany: Steinkopf Verlag, 1998: Stein PD, Albert JS, Copeland JMG, Dalen JE, Goldman S, Turpie AGG. Antithrombotic therapy in patients with mechanical and biological prosthetic heart valves. Chest 1995; 108 (suppl):371s 9S. 3. Jafri SM, Gheorghiade M, Goldstein S. Oral anticoagulation for secondary prevention after myocardial infarction with special reference to the warfarin re-infarction study. Prog Cardiovasc Dis 1992;34: Koertke H, Minami K, Bairaktaris A, Wagner O, Koerfer R. INR self-management following mechanical heart valve replacement. J Thromb Thrombolysis 2000;9:S Körtke H, Minami K, Breymann T, et al. INR selfmanagement after mechanical heart valve replacement: ESCAT (early self-controlled anticoagulation trial). Z Kardiol 2001;90(suppl 6):VI/ Oden A, Fahlen M. Oral anticoagulation and risk of death: a medical record linkage study. Br Med J 2002;325: Stein PD, Alpert JS, Bussey HI, Dalen JE, Turpie AG. Antithrombotic therapy in patients with mechanical and biological prosthetic heart valves. Chest 2001;119(1 suppl):220s 7S. 8. Koertke H, Minami K, Boethig D, et al. INR self-management permits lower anticoagulation levels after mechanical heart valve replacement. Circulation 2003;108(suppl 1):II Vink R, Kraaijenhagen RA, Hutten BA, et al. The optimal intensity of vitamin K antagonists in patients with mechanical heart valves. J Am Coll Cardiol 2003;42: Kaplan E, Meier P. Nonparametric estimation from incomplete observation. J Am Stat Assoc 1958;53: Penning-van Beest FJ, Geleijnse JM, van Meegen E, Vermeer C, Rosendaal FR, Stricker BH. Lifestyle and diet as risk factors for overanticoagulation. J Clin Epidemiol 2002;55: Schulman S. Care of patients receiving long-term anticoagulant therapy. New Engl J Med 2003;349: Cannegieter SC, Rosendaal FR, Wintzen AR, van der Meer FJ, Vandenbroucke JP, Briet E. Optimal oral anticoagulant therapy in patients with mechanical heart valves. N Engl J Med 333;1995:11 7. DISCUSSION DR RALPH J. DAMIANO (St. Louis, MO): I have one question for you. It is a little hard for me to understand how maintaining a lower INR decreases the rate of thromboembolism, especially since both groups utilized self-management and maintained INR in a relatively tight range. It is pretty obvious that bleeding complications would be less, but how do you explain having a significantly lower risk of thromboembolism at a lower INR? DR SEIFERT: Since the variance is not so great, we feel that the variance from the top to the bottom actually causes the thromboembolism and not the sole value itself.