Correspondence
Coronary-Artery Bypass Grafting versus Stent Implantation
N Engl J Med 2005; 353:735-737August 18, 2005
- Article
To the Editor:
The analysis by Hannan and colleagues (May 26 issue)1 comparing coronary-artery bypass grafting (CABG) and percutaneous coronary intervention (PCI) is seriously flawed. The data were derived retrospectively from two separate New York State registries. The 12 risk factors for increased mortality that made up the propensity model are more common among patients undergoing CABG than among those undergoing PCI. Differences between the two groups in variables such as the incidence of carotid-artery disease (14.0 percent vs. 3.5 percent, respectively) and chronic obstructive pulmonary disease (COPD) (16.4 percent vs. 5.9 percent) have not been previously reported in registries that prospectively assessed risk before revascularization was chosen.2,3 Among patients in the Bypass Angioplasty Revascularization Investigation registry who underwent CABG or PCI, there was a similar likelihood of peripheral vascular disease (15 percent vs. 14 percent) and COPD (5 percent vs. 4 percent).
The clinical assessment conducted before CABG is more extensive than that before PCI. Ultrasonography and spirometry of the carotid artery are often performed before CABG but not before PCI. Thus, in the study by Hannan et al., risk-ascertainment bias may have caused the substantial differences between the unadjusted and adjusted mortality rates. Furthermore, the public reporting of risk-adjusted, physician-specific mortality rates for CABG and PCI in New York State strongly influences case selection and the reporting of risk factors.4,5
5 ReferencesJames D. Flaherty, M.D.
Charles J. Davidson, M.D.
Northwestern University Feinberg School of Medicine, Chicago, IL 60611
cdavidso@nmh.org 1
Hannan EL ,Racz MJ ,Walford G , et al. Long-term outcomes of coronary-artery bypass grafting versus stent implantation. N Engl J Med 2005;352:2174-2183
Full Text | Web of Science | Medline2
Brooks MM ,Jones RH ,Bach RG , et al. Predictors of mortality and mortality from cardiac causes in the Bypass Angioplasty Revascularization Investigation (BARI) randomized trial and registry. Circulation 2000;101:2682-2689
Web of Science | Medline3
Mark DB ,Nelson CL ,Califf RM , et al. Continuing evolution of therapy for coronary artery disease: initial results from the era of coronary angioplasty. Circulation 1994;89:2015-2025
Web of Science | Medline4
Narins CR ,Dozier AM ,Ling FS ,Zareba W . The influence of public reporting of outcome data on medical decision making by physicians. Arch Intern Med 2005;165:83-87
CrossRef | Web of Science | Medline5
Moscucci M ,Eagle KA ,Share D , et al. Public reporting and case selection for percutaneous coronary interventions: an analysis from two large multicenter percutaneous coronary intervention databases. J Am Coll Cardiol 2005;45:1759-1765
CrossRef | Web of Science | Medline
To the Editor:
The study by Hannan et al. comparing bypass surgery with percutaneous revascularization provides new insight into the potential efficacy of these two procedures for patients with multivessel coronary artery disease. In all observational studies, however, a multitude of potential biases can trip up the investigator. In 1991,1 a colleague and I observed marked differences in behavioral changes in patients who had undergone bypass surgery as compared with those who had undergone angioplasty: whereas 55 percent of patients who underwent bypass surgery quit smoking and had not resumed after one year, only 25 percent of patients who underwent angioplasty did so. Changes in smoking status may reflect multiple behavioral changes by patients that are difficult to quantify and that are more profound after surgery than after angioplasty; such changes may also reflect closer physician follow-up and pharmacologic intervention. All these factors might attenuate risk in the surgical group.
1 ReferencesJohn R. Crouse, III, M.D.
Wake Forest University School of Medicine, Winston-Salem, NC 27157
jrcrouse@wfubmc.edu 1
Crouse JR III ,Hagaman AP . Smoking cessation in relation to cardiac procedures. Am J Epidemiol 1991;134:699-703
Web of Science | Medline
To the Editor:
By searching a public database, www.health.state.ny.us/nysdoh/, for CABG and PCI, we found that the actual number of revascularization procedures performed in New York State from 1996 through 2000 was 75,271 for CABG and 137,798 for PCI. However, the article by Hannan et al. reports the survival for only 37,212 patients who underwent CABG (49 percent of the total) and 22,102 who underwent PCI (16 percent); patients who had prior revascularization, disease of the left main coronary artery, or early acute myocardial infarction were excluded, as were those from out of state.
Although the first three exclusions are appropriate for the comparison of mortality rates associated with these two revascularization procedures, the last seems unwarranted, since data on the deaths of out-of-state patients should be accessible. Regardless, the provision of numbers and clinical characteristics of patients who were excluded would assist the clinician in determining the extent and character of selection bias in this cohort analysis. The absence of this information limits the physician's ability to generalize and apply the study results to choices in coronary revascularization. In addition, the low percentage of cases for both procedures included in the study — particularly for the PCI group — needs to be explained.
Stanley A. Rubin, M.D.
Freny V. Mody, M.D.
Department of Veterans Affairs, Greater Los Angeles, Los Angeles, CA 90073
sarubin@ucla.edu Author/Editor Response
In response to Drs. Flaherty and Davidson: the reason we used a propensity analysis was to adjust for differences in the prevalence of risk factors, which are usually present in observational studies. If there were no differences, there would be no need for propensity analyses. It is not difficult to accept the finding of increased carotid-artery disease in patients who undergo CABG, given the concordant findings of increased rates of stroke, aortoiliac disease, and femoral or popliteal disease, all of which can be easily ascertained by the cardiologist before stenting and are primarily clinical diagnoses. In addition, the easily identified clinical findings of increased age, diabetes, three-vessel coronary disease, and renal failure support the likelihood of more severe carotid-artery disease in the CABG group. We disagree with the contention that public reporting results in the avoidance of PCI in patients at high risk. However, even if this were true, it would seemingly shift the sicker patients to CABG surgery, which undermines Flaherty and Davidson's earlier claim that ascertainment bias is an explanation for differences in risk factors.
Regarding Dr. Crouse's comments: it is possible that differences in the rate of smoking cessation between the CABG and PCI groups could contribute to the superiority of long-term outcomes with CABG. However, if this is the case, we do not regard it as a bias in the study but, rather, as a problem that must be dealt with by improving the outcomes of PCI through closer follow-up.
In response to Drs. Rubin and Mody: there were 137,798 patients who underwent PCI and 75,271 who underwent isolated CABG in New York from 1997 through 2000, the years of the study. After records were combined for patients who had multiple procedures and a relatively small number of patients without valid Social Security numbers were excluded, there were 106,551 patients who underwent PCI and 66,250 who underwent isolated CABG. The relatively low percentage of patients with PCI was a result of the elimination from both procedures of those with single-vessel disease, which resulted in 47,470 patients who underwent PCI and 59,441 who underwent CABG. Another 7255 patients who did not receive stents were removed from the PCI group, and another 17,279 with stents in the PCI group and 20,747 in the CABG group were removed because of previous revascularizations, left main coronary artery disease, an acute myocardial infarction within 24 hours before the procedure, or all of these. A relatively small number of patients (834 in the PCI group and 1482 in the CABG group) were excluded because they were from outside New York State. We did not have access to the National Death Index because of resource constraints, but earlier studies of ours have demonstrated that the absence of data from this source does not introduce a bias.1
1 ReferencesEdward L. Hannan, Ph.D.
University at Albany, State University of New York, Rensselaer, NY 12337
elh03@health.state. ny. us Gary Walford, M.D.
St. Joseph's Hospital, Syracuse, NY 13203Robert H. Jones, M.D.
Duke University Medical Center, Durham, NC 277101
Hannan EL ,Racz MJ ,McCallister BD , et al. A comparison of three-year survival following coronary artery bypass graft surgery and percutaneous transluminal coronary angioplasty. J Am Coll Cardiol 1999;33:63-72
CrossRef | Web of Science | Medline
