|
| HOME | SUBSCRIPTIONS | CURRENT ISSUE | PAST ISSUES | CARDIOSOURCE | SEARCH | HELP | FEEDBACK |
|
|
|
|||||||||
|
|||||||||||
|
J Am Coll Cardiol, 2005; 45:1908, doi:10.1016/j.jacc.2005.03.006 © 2005 by the American College of Cardiology Foundation |
* Athens Euroclinic, 9 Athanassiadou St., Athens 11521, Greece (Email: dkatrits{at}otenet.gr).
The researchers do not provide the absolute number of deaths; however, it can be indirectly inferred from the presented total number of patients and event rates that their entire study includes only 10 deaths among patients with an unsuccessful procedure and about 100 more deaths in patients with successful procedures. The main inferences are practically based on 10 deaths. Such small numbers do not allow any meaningful modeling in CK-MB strata or multivariate analyses.
Furthermore, the use of percentages is misleading here; for example, the 5% death rate on patients with type 2 myocardial infarction (MI) and unsuccessful procedure is based on a single death, and the 9.1% death rate on patients with type 1 MI is based on only two deaths! Even the unadjusted analyses are based on extremely thin information. The successful procedure data are also limited: the observed differences of 0.4% to 0.7% excess death rate with small CK-MB elevations are certainly not statistically significant given the small number of events, but they may be clinically important when extrapolated to the millions of patients who undergo "successful" procedures.
Moreover, the overall one-year death rate in the study by Jeremias et al. is only 2%, whereas it has been about 4% in other studies evaluating peri-procedural MI (2). The same relative risk increase would have a larger absolute magnitude in a population at higher baseline risk of death.
We are also concerned that the details of the study protocols and justification of data pooling as described in the study by Jeremias et al. (1) are not fully described in the cited reference (see reference 8 in their report). The definition of "unsuccessful" procedure is not standardized in the published data. The definition of Jeremias et al. comprises five different elements (stenosis, flow grade, dissection, repeat revascularization, stent thrombosis). Each of these may be selected with different cutoffs (e.g., stenosis >50% or >30%; dissection 
D or C; Thrombolysis In Myocardial Infarction [TIMI] flow grade <3 or other criteria). Some investigators may use only some (3,4), but not all, of these criteria, or may add other parameters (5). In a database of 6,186 patients, it is almost certain that definition changes can always allow identification of a minuscule group of about 100 patients where 10 deaths have occurred and thus support a claim that this is the high-risk group par excellence.
Overall, such claims about high-risk groups are primarily hypothesis-generating speculations. It is important to try to replicate these findings in other patient cohorts using the exact same definitions. Chances are that the greater the degree of selection and strict data-fitting in the original investigation, the less likely the findings are replicated elsewhere (6). Nevertheless, given the great clinical importance of the observations made by Jeremias et al. (1), such validation is essential.
 |
References |
|---|
 Top References |
|---|
Related Article
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | SUBSCRIPTIONS | CURRENT ISSUE | PAST ISSUES | CARDIOSOURCE | SEARCH | HELP | FEEDBACK |
