| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Ann Thorac Surg 1996;62:1567-1570
© 1996 The Society of Thoracic Surgeons
Department of Surgery, University of Toronto The Toronto Hospital Eaton Bldg, NW, 13th Floor, Rm 222 200 Elizabeth St Toronto, ON M5G 2C4 Canada
Department of Surgery Johns Hopkins University, School of Medicine 600 N Wolfe St, Blalock 618 Baltimore, MD 21287-4618
Division of Hematology-Oncology Emory University School of Medicine 1003 Woodruff Memorial Bldg Atlanta, GA 30322
23 Spanish Bay Circle Pebble Beach, CA 93953
Department of Surgery, Rm 736 Ziegler Research Bldg University of Alabama at Birmingham Birmingham, AL 35294
Hamilton General Hospital 237 Barton St E Hamilton, ON L8l 2X2 Canada
To the Editor:
In August 1995, we were requested by CarboMedics, Inc, to act as an independent scientific committee to determine the relative frequency of thromboembolic, thrombotic, and anticoagulant-related bleeding complications occurring in patients with mechanical cardiac valves currently in use in North America. After extensive review and analysis of all relevant data, in regard to the three most commonly used models (CarboMedics, Medtronic-Hall, and St. Jude valves), we concluded that there was no evidence of significant differences among these mechanical valves in relation to these complications. Although the CarboMedics valve had not yet been introduced in the United States at the time of Grunkemeier, Starr, and Rahimtoola's pivotal publication [1], and thus was not included in their evaluation, our conclusion is in accord with their statement: "... it is difficult to determine the true complication rate for a given valve, or to establish a difference in rates between valve models. In general, there is more variation among series using the same model than between series with different models, probably because of non-valve factors such as patient selection, followup methods, and definitions."
Near the end of our deliberation, the article "Results with Mechanical Cardiac Valvular Prostheses" by Akins appeared [2]. Because his conclusions differed from ours, we have fully reviewed the issue and offer the following comments. Akins indicates that The Society of Thoracic Surgeons/The American Association for Thoracic Surgery Ad Hoc Committee Guidelines [3] suggest that early (30-day) events be included in the calculation of linearized rates. The guidelines actually note that "these rates should not be used unless the hazard function for the complication under study is proven to be constant during the entire time interval considered." It is well known that this is often not true for early postoperative events. The articles used in Akins' review were not consistent in the use of early follow-up events. He recalculated the rates for one article [4] using a second reference source [5] for early events. We found it was often difficult to determine if early events were included in the various articles' linearized rate calculations, and in some cases it could be determined only by direct conversation with the authors. We found at least four references [6–9] used in Akins' analysis did not include early events, but were not similarly recalculated.
In our analysis, we used the same articles as Akins [4, 6–30]; additionally, because CarboMedics valves are relatively new in this country, the six published articles presenting data suitable for analysis were added [31–36]. We used a consistent method of data extraction in analysis including multiple and late events only (all data presented in Figures 1 through 6
exclude early events). We have applied the general format used by Grunkemeier and associates [1]. This approach allows the reader to visually appreciate the wide variability among studies. Additionally, we included the Food and Drug Administration (FDA) objective performance criteria (OPC) from the 1993 valve guidelines. Valves applying for FDA certification must demonstrate statistical significance below twice the OPCs. It is interesting to note that the FDA specifically excluded the first 30 days from their rates. Also, we added an overall mean rate of morbid events calculated directly from the individual articles.
|
|
|
|
|
|
illustrate the results of our analysis. The results from each article are plotted as filled circles; the squares indicate the combined mean rate for each valve. The horizontal dotted lines are OPCs for the morbid event. These plots illustrate several important points. First, the majority of the rates for morbid events from the individual articles and all of the combined rates are below the FDA OPCs. Second, the variability among the articles for each valve type is large, and third, the mean value for the three valves are similar for each morbid event. This is further confirmed by the fact that formal statistical analysis indicates no difference among the different models' complication rates (all p values are greater than or equal to 0.1). In conclusion, in light of present available evidence, there is no detectable difference in the risk of complications of valve thrombosis, thromboembolism, or anticoagulant-related bleeding in either aortic or mitral positions among the CarboMedics, Medtronic-Hall, or St. Jude mechanical cardiac valves. The reason Akins arrived at a different conclusion is that he did not perform a structured analysis to formally compare the valve rates using and interpreting p values.
Accepted for publication May 17, 1996.
References
Cardiac Surgical Unit Massachusetts General Hospital White 503 32 Fruit St Boston, MA 02114
To the Editor:
I appreciate the opportunity to comment on the letter from the CarboMedics Scientific Committee concerning my recent review of mechanical prostheses. My principal concerns with their presentation relate in part to the methods used and in part to the appropriateness of applying formal statistical methods to these studies.
The first issue is that of linearized rates and the segregation of early and late events. The Society of Thoracic Surgeons/The American Association for Thoracic Surgery Ad Hoc Committee guidelines in fact do not really recommend the use of linearized rates. As a member of the committee that developed the early drafts of these guidelines, I was fully aware, as were all of the other participants, that linearized rates are only of value when the hazard function for a particular complication is a constant. However, after reviewing almost 20 years of English-language literature, I found that the use of actuarial event-free rates is not sufficiently consistent to provide appropriate comparisons.
In their comments about the inconsistencies of early and late events used by me in my review, the CarboMedics Scientific Committee state that of the 52 references used by me to calculate my composite linearized rates, in four cases they thought that the references did not include early events and were not recalculated, as I had done for the CarboMedics combined review (their reference 4). This is not strictly true. The data from Horstkotte and associates (their reference 6) include the early event rates in a separate table from the late event rates. For that study I combined the events from the early table and the late table to generate the complete linearized rate. Similarly, in the article by Fernandez and associates (their reference 7), the early rates are available in a separate table, and these were added to the late event rates to calculate a complete linearized rate. From the study published by Lund and colleagues (their reference 8), the only censoring of information would seem to be the exclusion of patients who died within the first 30 days after valve replacement. There is no evidence from the article that valve-related complications were censored at any time interval. Finally, in the study by Antunes and associates (their reference 9), the only possible early event exclusion listed in the article was that thromboembolism was counted if it occurred after postoperative day 7.
The CarboMedics Scientific Committee then reanalyzed some, but not all, of the articles used in my review and added to these six recently published articles concerning the CarboMedics valve. They implied that they used only multiple and late events and that they excluded early events in their analysis. (From my review of all of these studies, that would be a very difficult number to determine for many of them.) Of the six additional studies concerning the CarboMedics valve, those by Aagaard and associates (their reference 31) and Fiane and colleagues (their reference 33) report patients who came from institutions already included in Copeland's review of the international experience (their reference 4). Data from the study by de Luca et al (their reference 32) and by Fiane and colleagues (their reference 33) are not separated between aortic and mitral valves. Finally, the report from Yamauchi and colleagues (their reference 35) had not yet been published at the time of my review.
Finally, the CarboMedics Scientific Committee contend that I should have performed a structured analysis to compare the valve rates. My review clearly gives reasons why I believe that that was and is inappropriate. The reports come from widely disparate population groups in which extremely different anticoagulation protocols were followed and were summarized in reports that had marked degrees of variability in the intensity of follow-up. I continue to believe that the evaluation of late thromboembolism with mechanical prostheses is subject to so may patient and physician variables that to ascribe statistical significance to numbers related only to the prosthesis is to inappropriately use a univariate approach to a multivariate problem.
This article has been cited by other articles:
|
|
 |
|
  D. N. Salem, P. T. O'Gara, C. Madias, and S. G. Pauker Valvular and Structural Heart Disease: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines (8th Edition) Chest, June 1, 2008; 133(6_suppl): 593S - 629S. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. W. Emery, A. M. Emery, A. Knutsen, and G. V. Raikar Aortic Valve Replacement with a Mechanical Cardiac Valve Prosthesis Card. Surg. Adult, January 1, 2008; 3(2008): 841 - 856. [Full Text]
|
|
|
|
 |
|
 V. Pengo, G. Palareti, U. Cucchini, M. Molinatti, R. Del Bono, F. Baudo, A. Ghirarduzzi, C. Pegoraro, and S. Iliceto Low-Intensity Oral Anticoagulant Plus Low-Dose Aspirin During the First Six Months Versus Standard-Intensity Oral Anticoagulant Therapy After Mechanical Heart Valve Replacement: A Pilot Study of Low-Intensity Warfarin and Aspirin in Cardiac Prostheses (LIWACAP) Clinical and Applied Thrombosis/Hemostasis, July 1, 2007; 13(3): 241 - 248. [Abstract] [PDF]
|
|
|
|
|
|
D. N. Salem, P. D. Stein, A. Al-Ahmad, H. I. Bussey, D. Horstkotte, N. Miller, and S. G. Pauker Antithrombotic Therapy in Valvular Heart Disease--Native and Prosthetic: The Seventh ACCP Conference on Antithrombotic and Thrombolytic Therapy Chest, September 1, 2004; 126(3_suppl): 457S - 482S. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. W. Emery, G. J. Van Nooten, and P. J. Tesar The initial experience with the ATS Medical mechanical cardiac valve prosthesis Ann. Thorac. Surg., February 1, 2003; 75(2): 444 - 452. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. D. Stein, J. S. Alpert, H. I. Bussey, J. E. Dalen, and A. G.G. Turpie Antithrombotic Therapy in Patients With Mechanical and Biological Prosthetic Heart Valves Chest, January 1, 2001; 119(2009): 220S - 227S. [Full Text] [PDF]
|
|
|
|
|
|
A. E. Fiane, O. R. Geiran, and J. L. Svennevig Up to eight years' follow-up of 997 patients receiving the CarboMedics prosthetic heart valve Ann. Thorac. Surg., August 1, 1998; 66(2): 443 - 448. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
