Response to Letter by Roquer
We thank Dr Roquer for the comments on our article. The study by Roquer et al was designed to identify independent predictors of 30-day mortality in intracerebral hemorrhage (ICH) out of a large number of variables.1 In contrast to the results of our study,2 they found previous use of antiplatelets to be an independent predictor of 30-day mortality in 194 consecutive ICH patients. In our study, univariate analysis initially suggested the same result, but multivariate logistic regression showed this to be subject to bias in terms of age and prehospital disability.
What might explain the discrepant results? In the study by Roquer et al, we suspect relevant differences in important baseline variables between patients with and without previous antiplatelet use. In detail, 26 of 47 antiplatelet users (55%!) had received antiplatelets as “secondary stroke prevention”. The proportion of patients having had a previous stroke in the nonantiplatelet group is not given but was likely substantially lower. In addition, whereas patients with a premorbid modified Rankin Scale score >2 were excluded, no adjustment was made for premorbid modified Rankin Scale scores between 0 to 2. Finally, the analysis focused solely on 30-day mortality and would have been strengthened by additional data on outcome of ICH survivors.
Regarding the specific points of criticism on our study, we do not believe that it is necessary to include initial clinical severity as a parameter into the multivariate model. When we designed our analysis, we anticipated that antiplatelet therapy could be associated with severe neurological deficits at hospital admission, as it is the case for oral anticoagluation. Therefore, initial clinical severity constitutes a dependent variable for which adjustment is not appropriate. Furthermore, we strongly feel that it is a positive and not a negative aspect of our study to have used a large stroke registry with data from more than 100 hospitals. This reflects current stroke care in a more general way than possible by studies performed by one or just a few centers. The question, whether medical procedures, length of hospital stay and mortality rates are different between the hospitals, is of comparatively limited relevance, because all hospitals were located in one federal state and were participating in a quality assurance program.
In summary, we are well aware of the fact that conflicting data have been published on this issue. However, based on our study, which analyzed by far the largest patient population so far, the effect of previous antiplatelet therapy on outcome after ICH appears to be either small or nonexistent. However, ultimately only large prospective studies sufficiently powered to control for all relevant parameters will be able to unequivocally resolve the issue.