In reference to the article published in Revista Española de Cardiología,1 it is accurate to note that the overall unadjusted treatment interruption rate was higher with clopidogrel than with ticagrelor due to greater physician-guided discontinuation (P=.003, Table 4 of the supplementary material)1, and was not associated with a higher risk of major adverse cardiovascular events (MACE) (P=.079). In contrast, the rate of disruptions was proportionally higher with ticagrelor (P=.003, Table 4 of the supplementary material)1, especially in the 90 days following the index acute coronary syndrome (P<.001, Table 4 of the supplementary material). Taking this into consideration, disruption was indeed associated with a higher risk of MACE (P=.001), particularly when it occurred within the first 90 days of treatment (adjusted hazard ratio, 3.83; P<.001, Figure 7 of the supplementary material), unlike what was seen with physician-guided discontinuation. Therefore, after adjustment for potential differential nonadherence based on the P2Y12 receptor inhibitor used, the time to interruption (earlier with ticagrelor than clopidogrel: 22 vs 53 days; P=.035, Table 5 of the supplementary material)1, and the interaction between these 2 variables on an additive scale, the association between mode/timing of treatment interruption and MACE risk according to the type of P2Y12 receptor inhibitor did not reach statistical significance (Table 5 of the supplementary material)1 in contrast to the results in all previous studies.
Concerning the bleeding rates, the studies cited by Martínez-Sellés used intention-to-treat (ITT) analyses, whereas this investigation was based on the on-treatment1,2 principle. Current evidence indicates that in observational studies, ITT analysis leads to biased effect estimates when there are differential time-dependent adherence rates.3,4 Notably, a recent study reported for the first time that ITT simulation consistently generated biased estimates of higher bleeding risk with ticagrelor than with clopidogrel in a real-world setting.2
FUNDINGData collection for this subanalysis were partially funded through an unrestricted grant from the pharmaceutical company AstraZeneca España S.A. (ESR-17-13127). The company was in no way involved in the design, collection, interpretation, or subsequent analysis of the data, in writing the manuscript, or in the decision to publish.
DECLARATION REGARDING THE USE OF ARTIFICIAL INTELLIGENCEArtificial intelligence was not used in the preparation of this article.
AUTHORS’ CONTRIBUTIONSAll authors have contributed to the conception/design, writing, critical revision, and final approval of the manuscript. The authors assume responsibility for all aspects of the article and will investigate and resolve any issues related to the accuracy and veracity of any part of the study. M. Almendro-Delia and Juan C. García-Rubir were responsible for acquisition, analysis, and interpretation of the data.
CONFLICTS OF INTERESTM. Almendro-Delia has received speaker fees from Eli Lilly and Company, Daiichi Sankyo, and AstraZeneca, as well as remuneration for consulting work for Daiichi Sankyo and AstraZeneca. The remaining authors declare no conflicts of interest.