HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Alert me to new issues of the journal Add to Personal Folders Download to citation manager Author home page(s): Dieter Horstkotte Cary W. Akins Permission Requests Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Horstkotte, D. Articles by Akins, C. W. Search for Related Content PubMed Articles by Horstkotte, D. Articles by Akins, C. W.

Ann Thorac Surg 1996;62:1565-1567
© 1996 The Society of Thoracic Surgeons

---

Correspondence

Results With Mechanical Cardiac Valvular Prostheses

Division of Cardiology Pneumology Department of Medicine University Hospital Benjamin Franklin The Free University of Berlin Hindenburgdamm 30 D-12200 Berlin Germany

To the Editor:

In a recent issue of The Annals Akins [1] tries to analyze the thrombogenic potential of mechanical prostheses currently approved by the Food and Drug Administration. If we take into account that the thrombogenicity of an implanted device is only one of a multitude of factors responsible for the occurrence of thrombotic and thromboembolic events, additional aspects may be important for optimization of anticoagulation management in an individual patient carrying a valve prosthesis, including site of implantation, thrombogenicity related to cardiac morphology and physiology, primary unbalanced or temporarily (eg, during infections) altered patient-related coagulability, and quality of the anticoagulation management [2]. Due to a lack of randomized comparisons between different prosthetic devices, large-scale multicenter trials have been conducted recently in Europe to prospectively address the impact of these factors on the optimal anticoagulation treatment after valve replacement [3, 4]. These studies may overcome the shortcomings of nonrandomized, single-center observations, which do not allow meaningful comparisons between thromboembolic and bleeding rates and are particularly inappropriate to show superiority of one device over another with respect to their thrombogenic potential [5].

The problem of different thromboembolic risks due to different intensities of anticoagulation may be overcome by comparing thromboembolic and bleeding rates, as Akins has done using a composite thromboembolic bleeding index [1], a technique that has been reported earlier [2, 5]. The major remaining problems are publication bias [6] and invalid reporting of complications due to inappropriate follow-up techniques [7, 8]. Currently accepted guidelines for reporting morbidity and mortality after cardiac valvular operations require reporting of all thrombotic/thromboembolic complications as well as all anticoagulant-related hemorrhages [9].

The reported background incidence for spontaneous-that is, without anticoagulation treatment-bleedings in patients 64 years or older is approximately 0.8% per year and for transient plus intransient cerebral ischemic attacks the rate is close to 1.3% per year if unselected men aged 65 to 74 years are taken into account [10, 11]. It is thus evident that all complications reporting less than approximately 2.0% thromboembolic plus bleeding complications per year are lacking proper follow-up techniques, as not even the background incidence of such complications in the general population is detected. Due to the underlying cardiac disease and the implantation of artificial materials, the thromboembolic hazard for patients with a cardiac valve prosthesis should be much higher. Moreover, the bleeding incidence in patients under anticoagulation should be higher than in a non-anticoagulated population. For these reasons it appears to be hardly defensible on scientific grounds to take publications into account reporting levels of thromboembolism and anticoagulant-related hemorrhage below the background incidence and then draw meaningful conclusions about the thrombogenicity of the respective implants. The attempt to carry out the most comprehensive possible postoperative observation, as has been done in the study published by us [12] and cited by Akins, on the other hand, inevitably leads to a considerably greater number of documented thromboembolic and bleeding complications, because minor, transient events such as nosebleeds are also recorded.

Finally, it should be borne in mind that the data from our group [12] used by Akins for his calculations were not exactly analyzed. We observed 127 thromboembolic or bleeding complications in patients with St. Jude aortic prostheses, 252 in patients with mitral prostheses, and 103 in patients with aortic plus mitral valve replacements. These would correspond to linearized annual percentages of 4.37% (aortic), 13.55% (mitral), and 13.99% (mitral plus aortic). In this as in the majority of other observation series, a constant complication hazard was, however, not found during the first 3 postoperative months. It was therefore necessary to calculate linearized complication rates only for the period later than the third postoperative month. Nevertheless, the total complications including those that occurred in the early postoperative period were also given in the mentioned article [12].

With survival analysis the occurrence of the event among other factors leads to right censoring for analysis. Possible multiple events like nonfatal hemorrhages or thromboembolisms will not be taken into account [13]. To overcome these limitations it is advisable to use the cumulative incidence of such events. Because events may lead to different consequences, the analysis should be stratified by severity and sequelae of the event. Akins uses the linearized complication rates per patient-year and includes the postoperative follow-up period for this calculation. This can be misleading and is not lege artis from a statistical point of view, because an approximately constant hazard rate over the total time period would be necessary. The inclusion of the first 3 to 4 months, with its higher complication rate, leads to a systematic bias of the results [14]. The extent of this bias as well as its potential influence on later misinterpretation mainly depend on two factors: (1) the difference between the long-term and short-term hazard rates and (2) the total length of the follow-up period. In the referenced studies the short-term hazard rates are different for both the heart valve prostheses and the miscellaneous studies. These biases could be of such an extent that the concluded modest advantage or disadvantage for a particular heart valve substitute cannot be adequately drawn from this review. Such a conclusion should be based on an a priori defined, quantitative metaanalysis.

To summarize, the attempt to draw conclusions about the thrombogenicity of various heart valve prostheses from nonrandomized studies seems exceptionally problematic in the absence of nonstandardized postoperative observation techniques, and even more problematic if studies are included that have reported thromboembolic and bleeding complication rates that are lower than those found in nonselected, non-anticoagulated populations. The worst operative results (high perioperative mortality and a consequent low number of high-risk patients in the late follow-up period) in combination with poor postoperative follow-up techniques (identification of only a fraction of the actual complication rates) may then result in publication of the most superior complication rate.

Accepted for publication May 15, 1996.

References

1. Akins CW. Results with mechanical cardiac valvular prostheses. Ann Thorac Surg 1995;60:1836–44.
2. Horstkotte D. Prevention of intracardiac thrombosis and systemic embolism following heart valve replacement. Haemostaseologie 1993;13:172–80.
3. Horstkotte D, Bergemann R, Althaus U, et al. German experience with low intensity anticoagulation (GELIA): protocol of a multicentre randomized prospective study with the St Jude Medical valve. J Heart Valve Dis 1993;2:409–19.
4. Acar J. Thromboembolic events in prosthetic valve recipients: what is a safe level of anticoagulation? J Heart Valve Dis 1993;2:395–7.
5. Grunkemeier GL, London MR. Reliability of comparative data from different sources. In: Butchart EG, Bodnar E, eds. Thrombosis, embolism and bleeding. London: ICR Publishers, 1992:464–75.
6. Berlin JA, Begg CB, Louis TA. An assessment of publication bias using a sample of published clinical trials. J Am Stat Assoc 1989;84:381–92.
7. Bodnar E, Horstkotte D. Potential flaws in the assessment of minor cerebro-vascular events after heart valve replacement. J Heart Valve Dis 1993;2:287–90.
8. Bodnar E. A critical assessment of thrombosis and embolism reporting methods. In: Butchart EG, Bodnar E, eds. Thrombosis, embolism and bleeding. London: ICR Publishers, 1992:476–84.
9. Edmunds LH Jr, Clark RE, Cohn LH, Miller DC, Weisel RD. Guidelines for reporting morbidity and mortality after cardiac valvular operations. Ann Thorac Surg 1988;46:257–9.
10. Horstkotte D, Trampisch HJ. Long-term follow-up after heart valve replacement. Z Kardiol 1986;75:641–5.
11. Bamford J, Sandercock P, Dennis M, et al. A prospective study of acute cerebral vascular disease in the community: the Oxfordshire Community Project-1981–1986. 1: Methodology, demography and incident cases of first-ever stroke. J Neurol Neurosurg Psychiatry 1988;51:1373–80.
12. Horstkotte D, Schulte HD, Bircks W, Strauer BE. Lower intensity anticoagulation therapy results in lower complication rates with the St Jude Medical prosthesis. J Thorac Cardiovasc Surg 1994;107:1136–45.
13. Cox DR, Oakes D. Analysis of survival data. London: Chapman and Hall, 1984.
14. Blackstone EH, Naftel DC, Turner ME Jr. The decomposition of time-varying hazard into phases, each incorporating a separate stream of concomitant information. J Am Stat Assoc 1986;81:615–24.

Reply

Cardiac Surgical Unit Massachusetts General Hospital White 503 32 Fruit St Boston, MA 02114

To the Editor:

I read with interest the letter by Dr Horstkotte focusing on the complexity of efforts to analyze the thrombogenic potential of mechanical prostheses and was pleased that his letter corroborates the comments I made concerning this issue in my review.

He suggests that any study reporting less than approximately 2% thromboembolic plus bleeding complications per year should be ignored. Unfortunately, although his suggestion may be appropriate for people over the age of 65 years in western Europe, according to the studies cited by him, many of the articles in my review reported mean patient ages that were substantially lower, indeed, several decades lower, for example, from South Africa. His contention that it is hardly defensible scientifically to use such publications to draw meaningful conclusions is probably correct; that is why I applied no statistical comparisons to the cumulative numbers for the purpose of drawing meaningful conclusions.

Horstkotte correctly points out that linearized rates are only of value when the hazard function of a complication is relatively constant during the period of follow-up. He suggests that the higher early complication rates shortly after implantation can bias the long-term numbers. Presumably, with increasing length of follow-up the impact of higher early complication rates will be diluted by the longer period of follow-up. Unfortunately, the published literature for the last 15 years on mechanical valves has not taken into account this issue. In fact, virtually all publications include early and late events together. In the best of all worlds his comments are true, but to create a review one can only use what literature is available.

In summary, I agree with Horstkotte's contention that an attempt to draw statistically founded conclusions about such a comparison is inappropriate, and that is why I did not do it. After reviewing the quality of all these reports, I believe quantitative metaanalysis still cannot adequately account for publication bias, selection bias, and data extraction bias, let alone avoid the problem of combinability. Metaanalysis is best used for combining multiple small, prospective, randomized trials not retrospective, observational reports focusing on only one therapeutic choice, in this case a specific heart valve prosthesis [1]. The review was meant to be observational about the available data in the literature and not statistically conclusive.

Reference

1. Sacks HS, Berrier J, Reitman D, Pagano D, Chalmers TC. Meta-analyses of randomized control trials. An update of the quality and methodology. In: Bailar JC III, Mosteller F, eds. Medical uses of statistics. Boston: NEJM Books, 1992:427–42.

This article has been cited by other articles:

				D. Hering, C. Piper, R. Bergemann, C. Hillenbach, M. Dahm, C. Huth, and D. Horstkotte Thromboembolic and Bleeding Complications Following St. Jude Medical Valve Replacement: Results of the German Experience With Low-Intensity Anticoagulation Study Chest, January 1, 2005; 127(1): 53 - 59. [Abstract] [Full Text] [PDF]

				P Kvidal, R Bergstrom, T Malm, and E Stahle Long-term follow-up of morbidity and mortality after aortic valve replacement with a mechanical valve prosthesis Eur. Heart J., July 1, 2000; 21(13): 1099 - 1111. [Abstract] [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]

				T. O. Cheng Henry Swan II, MD Ann. Thorac. Surg., May 1, 1997; 63(5): 1516 - 1516. [Full Text]

This Article Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Alert me to new issues of the journal Add to Personal Folders Download to citation manager Author home page(s): Dieter Horstkotte Cary W. Akins Permission Requests Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Horstkotte, D. Articles by Akins, C. W. Search for Related Content PubMed Articles by Horstkotte, D. Articles by Akins, C. W.