As clinicians, we are well aware that patients with prosthetic heart valves are complex ones and that their management should not be merely restricted to a problem of prothrombin time ratios. However our conclusion regarding the utility of an anticoagulation clinic in preventing thromboembolic events and haemorrhagic complications is correct. Of course any retrospective analysis has well known limits; nevertheless, the point raised by Girolami et al. was completely answered in the Data Analysis Section of our paper. The comparison of events occurring in patients in different periods of time (before and after enrollement in the anticoagulation clinic) was not carried out by means of the standard actuarial curve analysis, but through the method developed by Kurtze in 1989 as referred to in the paper, that overcomes this possible bias. According to the latter, the computation considers the curves with " a tale of two censors", taking into account the effect of time from the surgery on. So, the shape of the two curves (Fig. 2) is quite different.

No abstract