indication. When a physician chooses adrug, it is likely that the decision hasbeen influenced by the patient’s charac-teristics. Confounding by indicationwill occur if, as is often the case, thesecharacteristics are also risk factors forthe outcome. Confounding by indica-tion may create, or mask, an apparentassociation. For instance, even if drug Ais superior to drug B in randomizedclinical trials, this benefit might bemasked, in an observational study, ifdrug A is preferentially prescribed tothe sickest patients (i.e., those who havethe worst prognoses). This would leadto the erroneous conclusion that the 2drugs are “equivalent.” The problem iscompounded in database studies be-cause many of the potential confound-ing factors are not present in the data-bases and thus cannot be adjusted for inthe analysis.

A lack of difference does not estab-lish equivalence. The criteria for estab-lishing therapeutic equivalence between2 therapeutic products have been clearlyestablished by regulatory agencies inthe domain of generic bioequivalence:equivalence criteria require the 90%confidence interval of the ratio of thegeometric mean of a test referencedrug product to be within the bio-equivalence interval that must be es-tablished a priori.

Given the pitfalls of demonstratingthe beneficial effects of drugs with da-tabases, and given that Tu et al’s1 studywas not planned as an equivalencestudy and did not meet the criteria forthis type of study, their conclusionsshould not be used to guide medicalpractice or to establish policy.

Jacques J. Lelorier

Montreal, Quebec, Canada17 October 2006

1. Tu K, Gunraj N, Mamdami M. Is ramiprilreally better than other angiotensin-convertingenzyme inhibitors after acute myocardial in-farction? Am J Cardiol 2006;98:6–9.

doi:10.1016/j.amjcard.2006.10.011

The Impact of Previously OccludedNoninfarct Vessels in Patients WithAcute Myocardial Infarction

In patients with ST-elevation myo-cardial infarction (STEMI) who aretreated with percutaneous coronary in-tervention, those with multivessel dis-ease have poorer clinical outcomes.1,2

This seems to be related mainly to

worse Killip classes at presentation inthese patients but also to a higher clin-ical risk profile, a higher rate of recur-rent ischemia, and a higher incidence ofinfarct-related complications, such assevere mitral regurgitation and atrio-ventricular blockade.2

We read with great interest the reportby van der Schaaf et al,3 who describe aseries of patients with STEMI treatedwith percutaneous coronary interven-tion in whom multivessel disease wasassociated with higher mortality. In ourpoint of view, the main findings of thiswork are that (1) multivessel diseasewas not an independent risk factor ofhigher mortality in the multivariateanalysis; (2) the poor outcomes of pa-tients with multivessel disease were re-stricted only to those with chronic cor-onary occlusions in noninfarct-relatedcoronary arteries, whereas patients withmultivessel disease without noninfarct-vessel chronic coronary occlusions hadsimilar outcomes to patients with sin-gle-vessel disease; and (3) the presenceof a chronic coronary occlusion in anoninfarct vessel was an independentpredictor of mortality. This may have 2important clinical implications in pa-tients referred to primary percutaneouscoronary intervention for STEMI: (1)those in whom a noninfarct vessel chroniccoronary occlusion is documented shouldbe managed more aggressively (e.g.,with an intra-aortic counterpulsationpump), irrespective of other character-istics (e.g., even in absence of cardio-genic shock), and (2) those with mul-tivessel disease but without chronicocclusions in other vessels should notbe managed more aggressively thanthose with single-vessel disease.

In a previous report that also evalu-ated the influence of the presence of achronic occlusion in a noninfarct vesselin patients with STEMI treated withprimary percutaneous coronary inter-vention, noninfarct-vessel chronic coro-nary occlusions were associated with ahigher mortality rate in those with mul-tivessel disease, but this angiographicfinding was not an independent variableassociated with higher mortality onmultivariate analysis.4 One explanationfor this discrepancy might be the largernumber of patients included in the studyby van der Schaaf et al.3 Other differ-ences, such as a shorter duration of fol-low-up and a lower prevalence of non-infarct vessel chronic occlusions in the

work by van der Schaaf et al,3 may haveaccounted for the different results of the2 studies. However, the explanationcould be also related to the exclusion ofKillip class from the characteristicsstudied and thus from the variables in-cluded in the multivariate analysis (pa-tients were classified simply as havingor not having cardiogenic shock).3 Inour study,4 in patients with multivesseldisease, those with noninfarct vesselchronic occlusions not only had ahigher frequency of cardiogenic shock(30% vs 11%) but also had a higherfrequency of Killip class II or III (23%vs 14%).

Killip class has been validated as astrong and independent predictor ofshort- and long-term mortality in pa-tients treated with primary angioplasty.5In patients with STEMI treated with pri-mary percutaneous coronary interven-tion not having cardiogenic shock atadmission, those in Killip classes I, II,and III had 4%, 10%, and 28% 6-monthmortality, respectively, in a largestudy.5 Thus, all the studies evaluatingoutcomes in patients with STEMItreated with primary angioplasty shouldinclude such a simple and independentpredictor of mortality in patients withSTEMI.

Raul Moreno, MD

Maria J. Perez-Vizcayno, MD

Angel Sanchez-Recalde, MD

Luis Calvo, MD

Guillermo Galeote, MD

Fernando Alfonso, MD

Rosana Hernandez, MD

Madrid, Spain

Dominick J. Angiolillo, MD

Jacksonville, Florida

Esteban Lopez de Sa, MD

Carlos Macaya, MD

José L. López-Sendón, MD

Madrid, Spain30 October 2006

1. Kahn JK, Rutherford BD, McConahay DR,Johnson WL, Giorgi LV, Shimshak TM, LigonR, Hartzler GO. Results of primary angio-plasty for acute myocardial infarction in pa-tients with multivessel coronary artery disease.J Am Coll Cardiol 1990;16:1089–1096.

2. Moreno R, Garcia E, Elizaga J, Abeytua M,Soriano J, Botas J, Lopez-Sendon JL, DelcanJL. Results of primary angioplasty in patientswith multi-vessel disease. Rev Esp Cardiol1998;51:547–555.

3. van der Schaaf RJ, Vis MM, Sjauw KD, KochKT, Baan J, Tijsen JGP, de Winter RJ, Piek JJ,Henriques JPS. Impact of multivessel coronarydisease on long-term mortality in patients with

429Readers’ Comments

ST-elevation myocardial infarction is due tothe presence of a chronic total occlusion. Am JCardiol 2006;98:1165–1169.

4. Moreno R, Conde C, Perez MJ, Villarreal S,Hernandez R, Alfonso F, Bañuelos C, Angio-lillo DJ, Escaned J, Fernandez A, Macaya C.Prognostic impact of a chronic occlusion in anon-infarct vessel in patients with acute myo-cardial infarction and multi-vessel disease un-dergoing primary percutaneous coronary inter-vention. J Invasive Cardiol 2006;18:16–19.

5. DeGeare VS, Boura JA, Grines LL, O’NeillWW, Grines CL. Predictive value of the Killipclassification in patients undergoing primarypercutaneous coronary intervention for acutemyocardial infarction. Am J Cardiol 2001;87:1035–1038.

doi:10.1016/j.amjcard.2006.10.015

CorrectionIn the October 2, 2006, supplement

to the Journal, the article by Weber et al(Controlled-release carvedilol in thetreatment of essential hypertension.2006;98(suppl):32L–38L), Figure 2 hadan error in the placebo bar. The changein diastolic blood pressure in the pla-cebo arm was �0.38 (not �0.38). Thecorrected Figure 2 is presented.

doi:10.1016/j.amjcard.2006.10.012

Figure 2. Model-adjusted peak (3–7 hours) diastolic blood pressure (DBP)and systolic blood pressure (SBP) reductions measured by ambulatoryblood pressure monitoring, from baseline to study end point. Valuespresented are �SE. *p Values based on pairwise comparison of changefrom baseline with placebo. CR � controlled release.

430 The American Journal of Cardiology (www.AJConline.org)