Author + information
- John E. Sanderson, MD⁎ ()
- ↵⁎Birmingham University, Cardiovascular Medicine, The Medical School, University of Birmingham, Birmingham, West Midlands B15 2TT, United Kingdom
I thank Dr. Schaber for his clarifications and comments on my commentary (1). I have no doubt that the PROSPECT (Predictors of Response to CRT) trial was designed with the best of intentions, but sometimes there are unforeseen and unintended consequences of clinical trials. The importance of centers having implantation experience is clearly a sine qua non, but my point was that the echocardiography experience at these centers might not have been so critically evaluated. This is illustrated by Dr. Schaber's statement that “enrolling centers were not required to analyze tissue Doppler imaging (TDI); this was the responsibility of the core echocardiography laboratories…hence, training the centers to analyze tissue Doppler images was not necessary.” This might be strictly true, but technical skill is required for the acquisition of the TDI images as well as analysis. It is clear from the study design report (2) that TDI measurements were a critical part of the study from the start (see Table 1 of the article, which lists all the standard and TDI indexes to be tested). In addition, it was intended that these would be a major part of the study: page 601: “We will test each echocardiographic predictor against predetermined primary and secondary response outcome measures, each with predefined cutoff values.” In light of this, it is surprising that Dr. Schaber states “The PROSPECT trial was not originally powered to assess the validity of TDI measure, and as such, sites were not required to have echocardiography machines with this capability.” Indeed, one of the problems in retrospect was the lack of power. A large number of echocardiographic predictors were evaluated, but individual sample sizes were not calculated; rather, power calculations were done on an initial sample size of 250 patients, but there is no justification of how this number was decided upon. Furthermore, if there is wide variation in the ability to acquire high-quality images, then analysis will be difficult no matter how good the training is, and this will affect the power of the study—as subsequently became obvious in the published results (3). For example, even the results of the relatively simple measurement of left ventricular volume were often widely different in the implanting centers compared with the core laboratories, as I pointed out in my commentary. Thus, a potentially useful tool has been discredited and now practically all but discarded, which might not be in the best interest of potential patients or the community paying for the excess number of implanted devices.
- American College of Cardiology Foundation