Author + information
- Xacobe Flores-Ríos, MD⁎ (, )
- María J. Paniagua-Martín, MD,
- Javier Muñiz-García, MD, PhD and
- María G. Crespo-Leiro, MD, PhD
- ↵⁎Hospital Juan Canalejo, Cardiology, As Xubias s/n, A Coruña, Galicia 15009, Spain
Arora et al. (1) recently reported that pre-transplant Toxoplasma gondii (T. gondii) seropositivity was associated with increased risks of advanced cardiac allograft vasculopathy, mortality attributable to cardiac allograft vasculopathy, and all-cause mortality. These investigators are to be congratulated on recognizing the possible relevance of chronic T. gondii infection, a hitherto underinvestigated aspect of the response to heart transplant (HT). If their conclusions are corroborated, they may have significant implications for HT patient management, especially in centers such as ours, where the prevalence of T. gondii seropositivity among HT patients (75%) is much higher than the 27% reported by Arora et al. (1).
It is nevertheless disappointing that the investigators did not provide more information on the analyses that led them to their conclusions. They state that they used stepwise Cox regression analyses including all variables with p values <0.05 in the univariate analyses, but candidate variables not included in the Cox analysis are not named and the strengths of the univariate associations are not given.
Furthermore, in adopting this combination of a purely statistical criterion for variable inclusion in the regression model, in opposition to a clinically oriented analysis, variables of established clinical relevance have been left apart. Although factors such as pre-transplantation coronary artery disease, ischemia time, donor age, recipient age, diabetes, cytomegalovirus infection, or previous rejection episodes may not have differed significantly between the T. gondii seropositive and seronegative groups in this study, they affect HT outcome (2) and should have been taken into account regardless of their statistical significance. Minor differences working in the same direction could explain a substantial part of the reported relationship, and their exclusion calls into question the accuracy of the association between T. gondii seropositivity and end points.
The failure to tell the reader which variables were tested by univariate analysis severely limits the ability of other researchers to compare the findings of Arora et al. (1) with their own results, and infringes the principle that the description of research methods should suffice to allow replication by others. We would really appreciate if the investigators could provide detailed information about statistical methods in associated online repositories.
- American College of Cardiology Foundation