CORRESPONDENCE: LETTER TO THE EDITOR
Reply
Satish Arora, MD*,
Pål A. Jenum, MD, PhD,
Pål Aukrust, MD, PhD,
Halvor Rollag, MD, PhD,
Arne K. Andreassen, MD, PhD,
Svein Simonsen, MD, PhD,
Einar Gude, MD,
Arnt E. Fiane, MD, PhD,
Odd Geiran, MD, PhD and
Lars Gullestad, MD, PhD
* Rikshospitalet Medical Center, Cardiology, Sognsvannsveien 20, Oslo, Oslo 0027, Norway (Email: satish.arora{at}medisin.uio.no).
We agree with the suggestion by Dr. Flores-Ríos and colleagues that our findings may have significant implications for the management of heart transplant recipients, particularly in geographical regions where Toxoplasma gondii seropositivity is virtually endemic. Although our article (1) has shown an independent association between seropositivity and long-term outcome, our de novo findings do need to be corroborated by other centers.
Flores-Ríos et al. write that our article does not list candidate variables that were not included in the Cox analysis. We agree that some of the methods could have been more thoroughly described, but this omission was at least partly attributable to space limitation. Nevertheless, we refer to the Statistical Analysis section of our article and would like to emphasize that all variables listed in Table 1 were candidate variables for the Cox analysis. All of these variables were initially tested using the Kaplan-Meier method, and if this found a significant association (p < 0.05) with the end point (mortality or allograft vasculopathy), the variable was included in the multivariate Cox analysis.
We collected data on a large number of covariates (n = 26), and to avoid overzealous modeling, it is important to identify unimportant covariates. Simulation work has shown that at least 10 events need to be observed for each covariate included in multivariate modeling to allow an accurate analysis (2). Given such statistical constraints, it is common to use a p value <0.05 on univariate analysis to guide selection of covariates for multivariate analysis (3,4). We do not dispute the suggestion that the inclusion of known prognostic factors is relevant, but this should not be done at the expense of overmodeling or exclusion of statistically significant variables. Dr. Flores-Ríos and colleagues claim that factors such as ischemia time, rejection episodes, and cytomegalovirus infection should be taken into account regardless of statistical significance. However, we believe that it is not appropriate to include nonsignificant parameters in multivariate analyses, particularly in relatively small study populations. Furthermore, not all of the suggested variables by Dr. Flores-Ríos and colleagues are established to be equally relevant. For example, in the recent study be Hussain et al. (5), cytomegalovirus infection was not found to be a risk factor for mortality after heart transplantation. Hence, we believe that the statistical approach applied in our article is appropriate and valid and allows more accurate replication by others.
 |
References
|
|---|
- Arora S, Jenum PA, Aukrust P, et al. Pre-transplant Toxoplasma gondii seropositivity among heart transplant recipients is associated with an increased risk of all-cause and cardiac mortality J Am Coll Cardiol 2007;50:1967-1972.[Abstract/Free Full Text]
- Peduzzi P, Concato J, Feinstein AR, Holford TR. Importance of events per independent variable in proportional hazards regression analysis. II. Accuracy and precision of regression estimates. J Clin Epidemiol 1995;48:1503-1510.[CrossRef][ISI][Medline]
- Lietz K, Miller LW. Improved survival of patients with end-stage heart failure listed for heart transplantation: analysis of organ procurement and transplantation network/U.S. United Network of Organ Sharing data, 1990 to 2005 J Am Coll Cardiol 2007;50:1282-1290.[Abstract/Free Full Text]
- Lin HM, Kauffman HM, McBride MA, et al. Center-specific graft and patient survival rates: 1997 United Network for Organ Sharing (UNOS) report JAMA 1998;280:1153-1160.[Abstract/Free Full Text]
- Hussain T, Burch M, Fenton MJ, et al. Positive pretransplantation cytomegalovirus serology is a risk factor for cardiac allograft vasculopathy in children Circulation 2007;115:1798-1805.[Abstract/Free Full Text]
Related Article
-
Pre-Transplant Toxoplasma gondii Seropositivity Among Heart Transplant Recipients and Mortality
- Xacobe Flores-Ríos, María J. Paniagua-Martín, Javier Muñiz-García, and María G. Crespo-Leiro
J. Am. Coll. Cardiol. 2008 51: 1827.
[Full Text]
[PDF]
Top
References