Data analysis and interpretation of results in DATAS trial

Regarding the DATAS randomized study by Almendral et al. , 1 we would like to comment on methodological issues that may limit the validity of the results of this trial. The investigators have created clinically significant adverse events (CSAEs) scoring system 2 based on assigning arbitrary weights to actual events that may artificially change the value of the analysed events, thus creating the potential of converting statistically insignificant result into a significant one. It is understandably difficult to model the outcome death, using such a CSAE scoring system and therefore the results could potentially underestimate the lethality of the interventions. Another source of modifying the observed difference between the two groups is the method of ‘penalizing’ for crossovers. Adding scores to both groups may not fully compensate for right censored data, but could artificially modify the difference between groups. Furthermore, as paroxysmal atrial tachycardia was not an exclusion criterion, and dual-chamber (DC) implantable cardioverter defibrillators (ICDs) offered atrial anti-tachycardia functions, not surprisingly patients with single-chamber (SC) devices may experience a higher rate of long-duration atrial tachycardia. However, if some patients in the DC-ICD group received anti-tachycardia pacing or atrial shock, it is unclear whether this adverse event was not a reason for some crossovers. Given higher rates of both appropriate and inappropriate shocks in the DC-ICDs compared with SC-stimulated devices reported previously, 3 analysis of the causes of inappropriate shocks would be interesting. Considering primary indications for ICD implantation, current registries show different rates of primary indications in different countries. 4 The DATAS sample population results may therefore have limited generalizability to other populations with conventional indications for ICD implantation. Finally, as quality of life was to be evaluated as a secondary objective according to the study design, 2 it would be interesting to the readers to see these data published.