Doi:10.1016/j.socscimed.2007.11.049

Social Science & Medicine 66 (2008) 1675e1677 Arguments prove nothing unless verified. A commentary on Kaufman’s ‘‘Epidemiologic analysis of racial/ethnic disparities: Some fundamental issues and a cautionary example’’ a 3330 Hospital Drive NW, Room G208, Calgary, AB, Canada T2N 4N1 b Box 508 UMHC, 420 Delaware St SE, Minneapolis, MN 55455-0374, USA Keywords: Therapeutic efficacy; Heart failure; Study design; Race; ACE inhibitors ‘‘The strongest arguments prove nothing so long as ‘‘Studies of Left Ventricular Dysfunction’’ (SOLVD) the conclusions are not verified by experience. Ex- data was not an attempt to replicate a randomized con- perimental science is the queen of sciences and trolled trial but to improve, albeit with remaining in- equalities, our ability to identify potential reasons forpreviously reported differences in outcome Roger Bacon (1214e1294), philosopher and advo- ). Our analysis was never suggested to be more than hypothesis generating. The over-interpreta- Jay Kaufman has attempted to provide a scholarly tion by Kaufman and others served their purposes, assessment of pitfalls in attempting to assign the cause not ours. In addition, his criticisms of our analytic ap- of disease outcomes to apparent racial differences Our statements in the New England Journal of Med- in his arguments, he has continued to follow the same icine article were intentionally worded to discourage path as other authors, whom he appears to deride, in mis-interpretation or over-interpretation of our findings.
misinterpreting the conclusions of our 2001 paper, Since prior heart failure trials were conducted predom- ‘‘Lesser response to angiotensin-converting-enzyme in- inantly in white subjects and possible differences of re- hibitor therapy in black as compared with white patients sponse in blacks were uncovered it was correct to state with left ventricular dysfunction’’ published in the New that ‘‘the overall population of black patients with heart failure may be underserved by current therapeutic rec- ommendations’’ and ‘‘it seems appropriate to consider Despite Kaufman’s assertions, the matched-cohort current therapeutic recommendations as applying to design we used in our retrospective analysis of white patients but not necessarily to black patients’’.
These comments are clearly in the context of a hypothe-sis, not a definitive conclusion. Our recommendation for * Corresponding author. Tel.: þ1 403 220 3219; fax: þ1 403 210 ‘‘clinical trials in black patients that are designed pro- spectively to evaluate therapeutic responses.’’ ( 0277-9536/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.socscimed.2007.11.049 D.V. Exner, J.N. Cohn / Social Science & Medicine 66 (2008) 1675e1677 p. 1357) further emphasized our recognition responsible for the differences in outcome that were ob- for the need of additional, prospective randomized data.
served between the two groups. The abstract to our pa- We recognize that our matched-cohort design, as per specified ‘‘no significant change in the risk of death with any study of this kind, was imperfect. Yet, many was observed in association with enalapril therapy in ei- of Kaufman’s criticisms are unmerited. Our matching strategy was described as ‘‘questionable because race were further highlighted in Table 3 of our paper and specific comments were made that ‘‘mortality was sim- method used was both appropriate and scientifically ilar among the black patients and the matched white pa- valid. The comment that ‘‘once the data are already col- tients regardless of treatment assignment’’ and that ‘‘no lected, however, one can’t generally do better by throw- significant alteration in mortality was observed in asso- ing away a large proportion of these data’’ ( ciation with enalapril therapy’’ ).
is erroneous. In fact, a paper used to support Thus, it is neither surprising nor unexpected that subse- Kaufman’s criticisms states that ‘‘matching can be ex- pected to increase efficiency’’ when both the matching variables, and exposure, self-identified race, are nega- The analysis we reported in 2001 never attempted to That was the situation for our analysis.
identify genetic or environmental factors that might Despite Kaufman’s assertions related to our statistical contribute to our observations. Studies demonstrating models, we were cognizant of issues related to residual lesser antihypertensive potency of ACE inhibitors in confounding and misclassification and used great care black than in white hypertensive patients ( to deal with these issues as best we could. We acknowl- also have made no such attempt. On two points edged these limitations in our paper by stating that ‘‘no we agree with Kaufman. The findings from our paper degree of statistical adjustment can ensure complete have been both over-interpreted and mis-interpreted comparability’’ (, p. 1357). Further, by others. Our analysis was conducted to investigate the comment that ‘‘it is well appreciated in the theoret- whether observed differences in outcome could, in ical epidemiologic literature that groups with higher part, be explained by differences in therapeutic re- baseline risk will in general have more modest response sponse. If confirmed, we planned to conduct additional, definitive research in this area aimed at improving the lives of patients with heart failure. This research has theoretical concept is not universally accepted ( been completed. We also agree that a ‘‘randomized Moreover, it is well known that patients with left controlled trial (RCT) is widely considered to be ventricular systolic dysfunction and a higher baseline the gold standard for establishment of causality in risk derive greater benefit from therapeutic interven- tions, in terms of absolute benefit, than do patients been followed by at least one prospective randomized trial designed to address therapeutic response in a self-identified black population with heart failure It is clear that a therapeutic reduction of hospitaliza- tion rate is a particularly sensitive guide to efficacy in apy that reduces mortality, reduces hospitalization, and sicker patients, as evidenced by its usefulness in recent enhances quality of life in these patients. Such data are studies in advanced Classes III and IV heart failure critical both in providing valid evidence and in assisting physicians in treating individual patients.
tality as a guide to severity of heart failure, even thoughdifferences in mortality, comparing white and black pa- tients, may be a result of health management disparities.
Furthermore, Kaufman claims that in our paper ‘‘the Cohn, J. N., Julius, S., Neutel, J., Weber, M., Turlapaty, P., & Shen, Y., null finding for mortality is largely ignored.’’ ( et al. (2004). Clinical experience with perindopril in African- In doing so he fails to appreciate that in American hypertensive patients: a large United States community the SOLVD Prevention Trial, which was the source of trial. American Journal of Hypertension, 17, 134e138.
most of our black patients, mortality was not reduced Dawid, A. P. (2002). Counterfactuals: help or hindrance? Interna- tional Journal of Epidemiology, 31(2), 429 ). We also clearly reported that differences Dries, D. L., Exner, D. V., Gersh, B. J., Cooper, H. A., Carson, P. E., in rates of hospitalization for heart failure were & Domanski, M. J. (1999). Racial differences in the outcome of D.V. Exner, J.N. Cohn / Social Science & Medicine 66 (2008) 1675e1677 left ventricular dysfunction. The New England Journal of Medi- Packer, M., Fowler, M. B., Roecker, E. B., Coats, A. J., Katus, H. A., & Krum, H., et al. Carvedilol Prospective Randomized Cumula- Dries, D. L., Strong, M. H., Cooper, R. S., & Drazner, M. H. (2002).
tive Survival (COPERNICUS) Study Group (2002). Effect of car- Efficacy of angiotensin-converting enzyme inhibition in reducing vedilol on the morbidity of patients with severe chronic heart progression from asymptomatic left ventricular dysfunction to failure: results of the carvedilol prospective randomized cumula- symptomatic heart failure in black and white patients. Journal tive survival (COPERNICUS) study. Circulation, 106, 2194.
of the American College of Cardiology, 40(2), 311e317.
Shekelle, P. G., Rich, M. W., Morton, S. C., Atkinson, C. S., Tu, W., Exner, D. V., Dries, D. L., Domanski, M. J., & Cohn, J. N. (2001).
& Maglione, M., et al. (2003). Efficacy of angiotensin-converting Lesser response to angiotensin-converting-enzyme inhibitor ther- enzyme inhibitors and beta-blockers in the management of left apy in black as compared with white patients with left ventricular ventricular systolic dysfunction according to race, gender, and dysfunction. The New England Journal of Medicine, 344(18), diabetic status: a meta-analysis of major clinical trials. Journal of the American College of Cardiology, 41(9), 1529e1538.
Greenland, S., & Morgenstern, H. (1990). Matching and efficiency Sheldon, R., Connolly, S., Krahn, A., Roberts, R., Gent, M., & in cohort studies. American Journal of Epidemiology, 131(1), Gardner, M. (2000). Identification of patients most likely to benefit from implantable cardioverter-defibrillator therapy: the Canadian Kaufman, J. (2008). Epidemiologic analysis of racial/ethnic dispar- Implantable Defibrillator Study. Circulation, 101(14), 1660e1664.
ities: Some fundamental issues and a cautionary example. Social SOLVD Investigators (1992). Effect of enalapril on mortality and the Science & Medicine, 66(8), 1659e1669.
development of heart failure in asymptomatic patients with re- Maldonado, G., & Greenland, S. (2002). Estimating causal effects.
duced left ventricular ejection fractions. [See comments]. The International Journal of Epidemiology, 31(2), 422e429.
New England Journal of Medicine, 327(10), 685e691, [published Moss, A. J. (2000). Implantable cardioverter defibrillator therapy: the erratum appears in The New England Journal of Medicine 1992, sickest patients benefit the most. Circulation, 101(14), 1638e1640.
Packer, M., Coats, A. J., Fowler, M. B., Katus, H. A., Krum, H., & Taylor, A. L., Ziesche, S., Yancy, C., Carson, P., D’Agostino Jr., R., Mohacsi, P., et al.Carvedilol Prospective Randomized Cumula- & Ferdinand, K., et al.for the African-American Heart Failure tive Survival Study Group (2001). Effect of carvedilol on survival Trial Investigators (2004). Combination of isosorbide dinitrate in severe chronic heart failure. The New England Journal of and hydralazine in blacks with heart failure. The New England Journal of Medicine, 351(20), 2049e2057.

Source: http://health-equity.pitt.edu/4093/1/Arguments_prove_nothing_unless_verified._A_commentary_on_Kaufman's_%E2%80%9CEpidemiologic_analysis_of_racial_ethnic_disparities.pdf