Clotting complications in patients with mechanical

Transcription

1 Anticoagulation in Children With Mechanical Valve Prostheses Scott M. Bradley, MD, Robert M. Sade, MD, Fred A. Crawford, Jr, MD, and Martha R. Stroud, MS Division of Cardiothoracic Surgery, Medical University of South Carolina, Charleston, South Carolina Background. Clotting complications in patients with mechanical valve prostheses can be prevented with either warfarin sodium (Coumadin; DuPont, Wilmington, DE) or antiplatelet agents. In children, it is not known whether one treatment regimen is more effective or safe than the other. Methods. We prospectively followed up 64 children and young adults (aged 18 years or younger at implantation) with a mechanical valve on the left side of the heart, from October 1986 through October Forty-eight patients were treated with Coumadin and 16 with aspirin and dipyridamole. The two groups were similar in age, sex, valve location and size, mean length of follow-up, and operative indication. There has been a total follow-up of 272 patient-years on Coumadin and 116 patient-years on aspirin and dipyridamole. Results. There was no difference between the two groups in survival or freedom from thromboembolism. Bleeding occurred more often in the patients taking Coumadin, but this difference was not statistically significant. Analysis of the literature showed thromboembolism and bleeding rates to be similar in the patients receiving Coumadin and those receiving antiplatelet agents. Conclusions. Coumadin and the combination of aspirin plus dipyridamole provided similar protection against complications in this group of children and young adults with left-sided St. Jude (St. Paul, MN) mechanical valves. The choice between the two regimens may depend on other factors, such as patient preference and convenience. (Ann Thorac Surg 1997;64:30 6) 1997 by The Society of Thoracic Surgeons Clotting complications in patients with mechanical heart valve prostheses can be prevented with either warfarin sodium (Coumadin; DuPont, Wilmington, DE) or antiplatelet drugs. In adults, there is evidence that Coumadin is more effective [1]. In children, there is no consensus that one treatment regimen is better than the other. Some groups favor antiplatelet drugs [2 6], and others favor Coumadin [7 13]. Further, the use of Coumadin in pediatric patients poses particular problems. Children may be more susceptible to trauma, and therefore to anticoagulant-related hemorrhage. The teratogenic effects of Coumadin are of concern to female patients who are, or will be, of child-bearing age. Finally, teenagers may be prone to poor compliance in taking medications, especially ones that significantly limit their activity. To determine whether Coumadin or antiplatelet drugs are preferable in children with mechanical valve prostheses, we prospectively followed up all children and young adults (aged 18 years or younger at the time of implantation) who received anticoagulation for a left-sided mechanical heart valve from October 1986 through October This article compares the rates of late death, thromboembolism, and bleeding events in the patients given Presented at the Forty-third Annual Meeting of the Southern Thoracic Surgical Association, Cancun, Mexico, Nov 7 9, Address reprint requests to Dr Bradley, Division of Cardiothoracic Surgery, Medical University of South Carolina, 171 Ashley Ave, Charleston, SC Coumadin with those in the patients given antiplatelet drugs. It also provides a pooled analysis of the currently available literature on this topic. Material and Methods This study includes the 64 patients, aged 18 years or younger, who survived the implantation of a mechanical prosthetic valve in the aortic or mitral position. Three patients who underwent replacement of a tricuspid valve functioning as the systemic atrioventricular valve (2 after atrial repair for transposition of the great arteries and 1 with congenitally corrected transposition) were included. No patient was excluded from the study for any reason other than age at the time of operation. All valves were St. Jude Medical (St. Paul, MN) and were implanted at the Medical University of South Carolina. Dates of valve replacement were from August 1979 to January Thirty-two of the 64 patients were operated on before October These patients usually were treated with no anticlotting prophylaxis until October 1986, when they were given either Coumadin or aspirin and dipyridamole [14]. The 32 patients operated on after October 1986 were started on anticlotting prophylaxis at the time of operation. Thus, all patients were treated with either Coumadin or aspirin and dipyridamole since October The choice between the two regimens was made by the patient s surgeon or referring physician. The dosage of Coumadin was adjusted to keep the prothrombin time about 1.5 times control. The dosage of aspirin 1997 by The Society of Thoracic Surgeons /97/$17.00 Published by Elsevier Science Inc PII S (97)

2 Ann Thorac Surg BRADLEY ET AL 1997;64:30 6 VALVE ANTICOAGULATION IN CHILDREN 31 Table 1. Demographics Characteristics Coumadin Aspirin and Dipyridamole p Value No. of patients Age at operation (y) ( ) ( ) Sex ratio (male/female) 27/21 10/6 0.6 Valve location (aortic/ 18/27/3 8/7/1 0.7 mitral/both) Valve size (mm) Aortic Mitral Mean follow-up (y) Total follow-up (patientyears) was5to6mg kg 1 d, and the dosage of dipyridamole was6mg kg 1 d in three divided doses [2, 7]. All patients were followed up prospectively and actively beginning in October Follow-up was carried out yearly by a survey form completed by the patient or the patient s parents, or by direct contact with the patient, the patient s parents, or the referring physician. Such follow-up is done routinely for all patients, both pediatric and adult, who undergo valve operations at this institution. One patient was lost to follow-up. Therefore, follow-up was 98% complete (closing interval, January 1996 to October 1996) at a mean of years (range, 1 month to 10 years). Patient age at operation ranged from 6 months to 18 years (mean, years). Patient age at the start of follow-up ranged from 10 months to 25 years (mean, years). Patient age at the end of follow-up ranged from 3 to 35 years (mean, years). Forty-eight patients initially were anticoagulated with Coumadin and 16 with aspirin and dipyridamole. During the study, 3 patients crossed over from Coumadin to aspirin and dipyridamole (1 because of a bleeding event, 1 because of difficulty in regulating the prothrombin time, and 1 by physician choice); 4 patients crossed over from aspirin and dipyridamole to Coumadin (2 because of embolic events and 2 by physician choice). For the purposes of analysis, patients were censored at the time of crossover; only events that occurred before crossover were included in all analyses. Standard definitions of valve-related morbidity and mortality were used [15]. Patient variables were analyzed as predictors of mortality, thromboembolism, and bleeding. For categorical predictors, event-free estimates were derived for each level of the variable by actuarial techniques, and subgroups were compared by Mantel-Cox [16]. Continuous predictors were broken down into quartiles and evaluated in a similar fashion. The following variables were analyzed as predictors: sex, race, valve location, valve size, operative indication, date of operation, age at operation, age at initiation of study, and anticlotting regimen. By univariate analysis, male sex (p 0.03), mitral location (p 0.09), and congenital disease as the operative indication (p 0.05) were predictors of mortality. These variables were entered into a multivariate Cox proportional hazards model. Only male sex (relative risk, 7.4; 95% confidence interval, 0.9 to 59.1; p 0.02) remained as a predictor of mortality. By univariate analysis, no variable was a significant predictor of either thromboembolism or bleeding. Year of operation was analyzed further as both a two-level (operation before versus after October 1986) and a continuous variable. By neither analysis was it a significant predictor of thromboembolism or bleeding. Therefore, patients operated on before and after October 1986 were considered together in all analyses. Event rates in patients with aortic valve replacement were nearly identical to those in patients with mitral valve replacement (thromboembolism, 2.5% versus 2.7% per patient-year, respectively, p 0.9; bleeding, 0.6% versus 1.2% per patient-year, respectively p 0.6). This is in agreement with previously published information on St. Jude valves in adults [1]. Therefore, patients with aortic and mitral valve replacements were considered together in all analyses. Data are presented as means plus or minus standard deviations. Comparisons between the two anticoagulation groups (Tables 1, 2) are by unpaired Student s t test for continuous variables and by 2 for categorical variables. Time-related events (survival, thromboembolism, and bleeding) are presented as actuarial curves with the standard error of the estimate [16]. Comparison between groups is by Mantel-Cox. Time zero is the start date of a given anticoagulation regimen. Linearized event rates are presented as percent per patient-year, are shown with 70% confidence limits, and are compared by the likelihood ratio test. Calculation of power is by noncentral 2 distribution. Statistical significance is defined as a p value less than Results There was no significant difference between the patients who received Coumadin and those who received aspirin and dipyridamole in age, sex, valve location, valve size, mean length of follow-up, or operative indication (see Tables 1, 2). There was a total follow-up of 272 patientyears on Coumadin and 116 patient-years on aspirin and dipyridamole. Table 2. Operative Indications a Indication Coumadin (n 48) Aspirin and Dipyridamole (n 16) Congenital disease 24 5 Rheumatic disease 14 3 Porcine valve failure 3 5 Endocarditis 3 3 Other 4 0 a p 0.06 (Coumadin versus aspirin and dipyridamole; 2 ).

3 32 BRADLEY ET AL Ann Thorac Surg VALVE ANTICOAGULATION IN CHILDREN 1997;64:30 6 Fig 1. Actuarial survival. (A D aspirin and dipyridamole; C Coumadin.) Fig 3. Freedom from bleeding. (A D aspirin and dipyridamole; C Coumadin.) Mortality There was no difference in survival between the patients who received Coumadin and those who received aspirin and dipyridamole (Fig 1). There were nine deaths during the study, seven in patients who took Coumadin and two in patients who took aspirin and dipyridamole. Three of the deaths in the Coumadin group were considered to be valve-related [15]: one was due to a bleeding event resulting in massive hemothorax and two were due to sudden death. The other four deaths in patients who took Coumadin resulted from ventricular dysfunction. Neither of the deaths in the patients who took aspirin and dipyridamole was valve-related. One occurred at a non valve-related reoperation; the other resulted from ventricular dysfunction. Thromboembolism There was no difference between the two groups in actuarial freedom from thromboembolism (Fig 2). There were seven thromboembolic events in the patients who received Coumadin and two in those who received aspirin and dipyridamole. Of the events that occurred in the patients who received Coumadin, three were neurologic (none resulted in a permanent deficit), three were valve thromboses (all were treated successfully with lytic therapy), and one was a myocardial infarction caused by a coronary arterial embolism (documented at cardiac catheterization). The two events that occurred in the patients who received aspirin and dipyridamole were neurologic (neither resulted in a permanent deficit). Bleeding Actuarial freedom from bleeding was somewhat better in the patients who received aspirin and dipyridamole, although this difference was not statistically significant (Fig 3). There were four bleeding events in the patients who received Coumadin: two episodes of vaginal bleeding, one intracerebral bleed (resulting in a mild, permanent deficit), and one hemothorax (resulting in death). There were no bleeding events in the patients who received aspirin and dipyridamole. Linearized Rates of Thromboembolism and Bleeding Events Analysis of thromboembolism and bleeding as linearized event rates, rather than by actuarial occurrence, gave similar results (Table 3). There was no significant difference in thromboembolism rates between the two groups. The rate of bleeding was lower in the patients who were given aspirin and dipyridamole, a difference that approached but did not reach statistical significance (see Table 3). Table 3. Linearized Event Rates a Event Coumadin Aspirin and Dipyridamole p Value Fig 2. Freedom from thromboembolism. (A D aspirin and dipyridamole; C Coumadin.) Thromboembolism ( ) (0.6 4) Bleeding ( ) (0 1.6) Both ( ) (0.6 4) a Rates are percent per patient-year (70% confidence limit).

4 Ann Thorac Surg BRADLEY ET AL 1997;64:30 6 VALVE ANTICOAGULATION IN CHILDREN 33 Table 4. Reports Included in the Literature Analysis Author Literature Analysis To summarize available information on the subject of anticoagulation in children with mechanical valve prostheses, an analysis of the literature was carried out. This analysis included reports published in English from 1985 to 1996 that contained information on thromboembolic and bleeding rates in children with mechanical valve prostheses who were treated with either Coumadin or antiplatelet agents. To derive pooled, linearized event rates, only studies that provided discrete numbers of events and patient-years of follow-up were included. Twelve reports (including the current one) satisfied these criteria (Table 4). These reports included 497 patients with 1,277 patient-years of follow-up with Coumadin and 433 patient-years of follow-up with antiplatelet agents. The pooled, linearized event rates for thromboembolism, bleeding, and the two combined were similar in the patients who received Coumadin and those who received antiplatelet agents, with none of the differences approaching statistical significance (Table 5). The sample size in this pooled analysis was large enough that a difference of about 2% in linearized event rates between the two treatments should have been detected with 80% power. Comment Year No. of Patients Follow-up (patient-years) Bradley et al [7] C 23 A D 17 Milano et al [8] C 247 Verrier et al [2] A( D) 155 McGrath et al [9] A D 22 Schaffer et al [10] C 65 Ilbawi et al [3] A D 21 Cornish et al [4] C 140 A 32 Stewart et al [11] C 211 Harada et al [5] C 208 A D 16 LeBlanc et al [6] C 5 A D 54 Ibrahim et al [13] C 106 Bradley et al C 272 A D 116 Total 497 C 1,277 A D 433 A aspirin; C coumadin; D dipyridamole. Coumadin and the combination of aspirin and dipyridamole were equally effective in this group of patients with mechanical valve prostheses. Mortality and thromboembolism rates were similar in the two groups (see Figs 1, 2; Table 3). Bleeding complications occurred more frequently in the patients who received Coumadin (see Fig 3; Table 3). However, this difference did not achieve statistical significance. Thus, neither anticoagulation regimen provided significant benefit over the other in our population of children and young adults with left-sided St. Jude valve prostheses. This study has several strengths and weaknesses. One strength is the method used to acquire data. All patients were followed up prospectively and actively for periods extending to 10 years. The data, therefore, avoid the incompleteness that can result from retrospective review and passive data acquisition. The total number of patient-years of follow-up is at least as great as that in most other studies in the literature (see Table 4). Another strength is the homogeneity of the valve replacement devices. All the valves used in these patients were of a single type (St. Jude Medical). Therefore, the study avoids variations that can result from the inclusion of data on different valve types. On the other hand, our results apply only to this valve type, and may not necessarily be applicable to other valves. The major weakness of this study is the small number of patients, particularly in the aspirin and dipyridamole group. This small sample size limits the power of the study and means that there may be real differences between the two groups that we were unable to detect. To address this limitation in part, we performed the literature analysis, which provided information on a total of 497 patients (see Table 4). This analysis confirmed the apparent equivalence of treatment with either Coumadin or antiplatelet agents in children with mechanical valves (see Table 5). Another weakness of this study is its lack of randomization. Although the clinical characteristics of the two groups appear to be equivalent (see Tables 1, 2), it is impossible to exclude a hidden bias in the assignment of patients to the treatment groups. A further, potential weakness is the fact that 32 of the patients (50%) were operated on before the start of the study. Most of these 32 patients were maintained without anticoagulation until the study start date, October 1986, when they were begun on one of the two anticoagulation regimens. It is possible that these patients would have different event rates than the patients who received anticoagulation from the time of operation. However, separate analysis of the patients operated on before and after the study start date produced results that did not differ significantly. The literature analysis pooled information available in the English literature since 1985 on thromboembolism Table 5. Linearized Event Rates for the Literature Analysis a Event Coumadin Aspirin With or Without Dipyridamole p Value Thromboembolism ( ) (2 3.8) Bleeding ( ) (0 0.8) Both ( ) ( ) a Rates are percent per patient-year (70% confidence limit).

5 34 BRADLEY ET AL Ann Thorac Surg VALVE ANTICOAGULATION IN CHILDREN 1997;64:30 6 and bleeding rates in children with left-sided mechanical valves. This analysis showed no detectable differences in event rates with the use of Coumadin or antiplatelet agents (see Table 5). Several reports were excluded from this analysis because they did not contain enough information to derive discrete event numbers or patient-years of follow-up on a given treatment regimen. Two of the excluded reports contained follow-up information on a large number of patients (396) [17, 18]. One found no difference in event rates between Coumadin and antiplatelet agents [17]. A second favored antiplatelet agents in patients with aortic valve replacements, and Coumadin in patients with mitral valve replacements [18]. Although the information in these reports cannot be combined with that from other studies to give pooled, linearized event rates, it does not appear to be at odds with our analysis (see Table 5). Patient compliance in taking medication will affect the results of any study of anticoagulation for valve prostheses. We attempted to obtain information on compliance in our follow-up surveys. However, in practice, it proved difficult to obtain either reliable or complete information on the adequacy of anticoagulation. Thus, we were unable to factor compliance into our analyses. A question remains: If Coumadin is a more effective anticoagulant in adults with mechanical valves [1], why is it not also more effective in children? One possibility is that the coagulation system is intrinsically different in children and adults. Information on this possibility is conflicting, but most indicates that this probably is not the case [19]. Another possibility is that hemodynamic differences between children and adults account for the difference [2, 20]. Compared with adults who receive anticoagulation for mechanical heart valves, children would be expected to have higher resting heart rates and lower incidences of atrial arrhythmias, atrial enlargement, and depressed ventricular function. These differences may make children less susceptible in general to thromboembolic events than adults. Another question is at what age should children with mechanical valves be considered adults, and be switched from antiplatelet agents to Coumadin? There is no exact answer to this question. Rather, the transition probably takes place over a period of years. In this study, 2 of the 4 patients who crossed over from aspirin and dipyridamole to Coumadin did so because their primary physician believed they were old enough (at the ages of 17 and 21 years) to be considered adults. In conclusion, Coumadin and aspirin plus dipyridamole provided similar protection against complications in this group of children and young adults with left-sided St. Jude mechanical valves. Bleeding occurred somewhat more often in the patients who received Coumadin, but this difference was small. These findings are in agreement with the information currently available in the literature. Further useful information would be provided by a prospective, randomized study of children with mechanical valves treated with either Coumadin or antiplatelet agents. In the absence of such a study, we suggest that it is reasonable to consider the preferences of the patient, family, and physician, the lifestyle of the patient, and the convenience of medication dosing and prothrombin time monitoring in choosing anticlotting prophylaxis for children with mechanical valves. References 1. Edmunds LH. Thrombotic and bleeding complications of prosthetic heart valves. Ann Thorac Surg 1987;44: Verrier ED, Tranbaugh RF, Soifer SJ, Yee ES, Turley K, Ebert PA. Aspirin anticoagulation in children with mechanical aortic valves. J Thorac Cardiovasc Surg 1986;92: Ilbawi MN, Lockhart CG, Idriss FS, et al. Experience with St. Jude Medical valve prosthesis in children. J Thorac Cardiovasc Surg 1987;93: Cornish EM, Human DG, de Moor MMA, et al. Valve replacement in children. Thorac Cardiovasc Surg 1987;35: Harada Y, Imai Y, Kurosawa H, Ishihara K, Kawada M, Fukuchi S. Ten-year follow-up after valve replacement with the St. Jude Medical prosthesis in children. J Thorac Cardiovasc Surg 1990;100: LeBlanc JG, Sett SS, Vince DJ. Antiplatelet therapy in children with left-sided mechanical prostheses. Eur J Cardiothorac Surg 1993;7: Bradley LM, Midgley FM, Watson DC, Getson PR, Scott LP III. Anticoagulation therapy in children with mechanical prosthetic cardiac valves. Am J Cardiol 1985;56: Milano A, Vouhé PR, Baillot-Vernant F, et al. Late results after left-sided cardiac valve replacement in children. J Thorac Cardiovasc Surg 1986;92: McGrath LB, Gonzalez-Lavin L, Eldredge WJ, Colombi M, Restrepo D. Thromboembolic and other events following valve replacement in a pediatric population treated with antiplatelet agents. Ann Thorac Surg 1987;43: Schaffer MS, Clarke DR, Campbell DN, Madigan CK, Wiggins JW, Wolfe RR. The St. Jude Medical cardiac valve in infants and children: role of anticoagulant therapy. J Am Coll Cardiol 1987;9: Stewart S, Ciancotta D, Alexson C, Manning J. The longterm risk of warfarin sodium therapy and the incidence of thromboembolism in children after prosthetic cardiac valve replacement. J Thorac Cardiovasc Surg 1987;93: Duran CMG, Gometza B, Martin-Duran R, Saad E, Al- Halees Z. Performance of 96 Carbomedics valve replacements in 75 patients less than twenty-one years of age. Ann Thorac Surg 1994;58: Ibrahim M, Cleland J, O Kane H, Gladstone D, Mullholland C, Craig B. St. Jude Medical prosthesis in children. J Thorac Cardiovasc Surg 1994;108: Pass HI, Sade RM, Crawford FA, Hohn AR. Cardiac valve prostheses in children without anticoagulation. J Thorac Cardiovasc Surg 1984;87: Edmunds LH, Clark RE, Cohn LH, Grunkemeier GL, Miller DC, Weisel RD. Guidelines for reporting morbidity and mortality after cardiac valvular operations. Ann Thorac Surg 1996;62: Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958;53: El Makhlouf A, Friedli B, Oberhansli I, Rouge JC, Faidutti B. Prosthetic heart valve replacement in children. J Thorac Cardiovasc Surg 1987;93: Rao PS, Solymar L, Mardini MK, Fawzy ME, Guinn G. Anticoagulant therapy in children with prosthetic valves. Ann Thorac Surg 1989;47: Lee GR, Bithell TC, Foerster J, Athens JW, Lukens JN, eds. Wintrobe s clinical hematology, 9th ed. Philadelphia: Lea & Febiger, 1993: Ebert PA. Discussion of Gardner TJ, Roland JMA, Neill CA, Donahoo JS. Valve replacement in children. J Thorac Cardiovasc Surg 1982;83:

6 Ann Thorac Surg BRADLEY ET AL 1997;64:30 6 VALVE ANTICOAGULATION IN CHILDREN 35 DISCUSSION DR CONSTANTINE MAVROUDIS (Chicago, IL): This was an excellent presentation of a very controversial topic in a relatively small group of patients. The issue is clearly important because of the significant problems that occur with anticoagulation in children. Also important is the risk of cerebral thromboembolism over time in patients who are not treated with Coumadin, which of course fuels the controversy. The difficulty is in planning a prospective clinical study to determine the comparative efficacy of antiplatelet therapy versus Coumadin therapy. Because such a study has not been and is unlikely to be performed, studies such as this one take on added importance. In those patients who underwent reoperation and replacement of an existing prosthetic valve, did you find surrounding tissue ingrowth, either above or below the valve, that might have interfered with valve function and contributed to clot formation? Under these circumstances, both antiplatelet therapy and Coumadin therapy would be ineffective, which could confound any results. Also, how do you manage patients with prosthetic heart valves who are taking aspirin and dipyridamole and have a stroke? Do you convert them to Coumadin therapy? Thank you, I enjoyed this paper. DR BRADLEY: Thank you very much for your comments. The problem of pannus formation is obviously a significant one. However, it is difficult to obtain accurate information on its incidence in the absence of reoperation. We do not have any specific information on pannus formation in the patients who are in this series. The one comment that I can make is in reference to the three episodes of valve thrombosis that we observed. All three of these episodes were treated successfully with lytic therapy, which indicates that at least a significant portion of the thrombosis was due to thrombosis per se and not to pannus formation in these particular patients. The 2 patients who had thromboembolic events while taking aspirin and dipyridamole crossed over to Coumadin. As Dr Mavroudis has mentioned, the significant limitation of this study is the small number of patients, particularly in the aspirin and dipyridamole group. This was compensated for somewhat by the fact that we have a reasonably large total number of patient-years of follow-up, even in the aspirin and dipyridamole group. One attempt to address this limitation was the literature analysis, which pooled information from a larger number of patients with more patient-years of follow-up. In the absence of a prospective, randomized study, which would be very difficult to put together, I think that ongoing follow-up of the patients in all these series is the best that we are going to be able to do. DR JOHN H. CALHOON (San Antonio, TX): Doctor Bradley, that was a lovely paper. I would like to echo some of Dr Mavroudis comments, and add one that you may have answered. Did you have any frozen valves that required reoperation because of the kind of pannus that he described? It sounds like you were able to get them to all free up with thrombolytics. I also have a couple of other points. This looked like an older group of patients for kids. Do you use, with such good results, the Ross procedure in the smaller kids, or will you start using it? Anecdotally, having operated on a couple of patients who were taking ticlopidine, we find that they do not clot very well after operation. It seems to be a great antiplatelet agent. Do you have any experience or knowledge of that being used in lieu of or in addition to aspirin and dipyridamole? This was really a beautifully presented paper. DR BRADLEY: Thank you very much. We did not have any incidence of valves that were frozen by pannus alone, so again I cannot comment on that particular problem. We also have no experience with the use of ticlopidine. It is the preference of our group, and particularly so in recent years, to use valve repair rather than replacement in the mitral position. In the aortic position, we prefer valve repair. When that is not possible, we use either the pulmonary autograft or, in select circumstances, aortic homografts for valve replacement. Nonetheless, there are situations in which none of these possibilities can be carried out successfully. I think there still will remain, therefore, a small number of patients who require valve replacement with mechanical valves, and this study was aimed specifically at deciding how to manage these patients. DR ROSS M. UNGERLEIDER (Durham, NC): I just have a quick question, and I think Dr Calhoon alluded to this. With the greater emphasis now on using pulmonary autografts or homografts in the aortic position, I suspect that in the future we will see mechanical valves being used more frequently in the mitral position in children. I wonder whether you looked at your data, and the outcomes in particular, and separated the outcomes into adverse outcomes related to the position of the valve. Was there a difference in the outcomes related to the aortic versus mitral position? And, in looking at the data that way, can you make recommendations regarding the anticoagulation regimens depending on the position of the valve? Thank you again for your excellent paper. DR BRADLEY: Thank you very much for your comments. We did specifically break the data down into two groups according to whether the aortic or the mitral valve had been replaced. Not only were there no detectable differences in either thromboembolism or bleeding complications, but the rates were essentially identical in the two groups, and we therefore pooled the two groups in all these analyses. DR CHRISTOPHER J. KNOTT-CRAIG (Oklahoma City, OK): It seems to me that your conclusions are controversial. If the complications are that much lower in the aspirin and dypiridamole group, how strongly do you feel about converting all those patients taking Coumadin to therapy with aspirin and dipyridamole? That is my first question. My second question concerns patients with valve thrombosis. How do you decide how long to allow the cardiologists to attempt to resolve the clots with lytic agents, and what dosages are required for that? And finally, when we use Coumadin in patients, the Coumadin dosage varies with the size of the patient, and we use the international normalized ratio to standardize this. Are you dosing all the patients receiving aspirin and dipyridamole with a uniform dosage, or do you titrate the dosage against their size? DR BRADLEY: Thank you for your questions. The conclusions from our data are that the thromboembolism rates are very similar between the two groups, and that the bleeding rates appear to be somewhat less with aspirin and dipyridamole than with Coumadin. The data available in the literature indicate that rates of both these complication are indistinguishable between the two treatment regimens. I would not conclude from this information that aspirin and dipyridamole are better than Cou-

7 36 BRADLEY ET AL Ann Thorac Surg VALVE ANTICOAGULATION IN CHILDREN 1997;64:30 6 madin, but merely that the two regimens cannot be distinguished on the basis of current information. Lytic therapy was used in 3 patients with valve thromboses. All these patients had reasonable hemodynamics on presentation and, therefore, were candidates for lytic therapy rather than reoperation. The courses of lytic therapy lasted 24 to 48 hours. Indications for declaring lytic therapy a failure were hemodynamic deterioration or failure of normal valve function (assessed clinically and by echocardiography) to return after 48 hours of therapy. In regard to the dosage of the medications, in the early years of the study, Coumadin usually was dosed to maintain prothrombin times about 1.5 times normal. More recently, it has been dosed to maintain international normalized ratios around 2.5 to 3. Aspirin and dipyridamole dosing has been fairly standardized, and we use about 5 mg kg 1 d of aspirin and about 6 mg kg 1 d of dipyridamole in three divided doses. DR KIT V. AROM (Minneapolis, MN): I have one question. It seems to me that in the Coumadin group, the incidences of hemorrhage and thrombosis are higher in the adult population. Can you explain why? DR BRADLEY: The lower thromboembolism rates observed in children compared with adults may be due to hemodynamic differences. Compared with adults, children have higher heart rates and lower rates of atrial arrhythmias and depressed ventricular function. These differences may make children less susceptible to thromboembolism than adults. Lower bleeding rates may relate to the adequacy of anticoagulation, an issue on which we do not have completely accurate information. DR AROM: What year did you first use the international normalized ratio instead of the prothrombin time level? Was it recently or in the middle of your study? DR BRADLEY: The international normalized ratio became adopted fairly routinely about 4 or 5 years ago. Some of our patients are followed up by cardiologists in outlying centers, and I cannot exactly answer when they adopted the international normalized ratio. DR AROM: The prothrombin time ratio is not as reliable as the international normalized ratio. Bound volumes available to subscribers Bound volumes of the 1996 issues of The Annals of Thoracic Surgery are available only to subscribers from the Publisher. The cost is $ (outside US add $27.00 for postage) for volumes 61 and 62. Each bound volume contains a subject and author index, and all advertising is removed. The binding is durable buckram with the name of the journal, volume number, and year stamped on the spine. Payment must accompany all orders. Contact Elsevier Science Inc, 655 Avenue of the Americas, New York, NY 10010; or telephone (212) (facsimile: (212) ).