In the Platelet Inhibition and Patient Outcomes (PLATO) trial, a prespecified subgroup analysis showed a significant interaction between treatment and region (P=0.045), with less effect of ticagrelor in North America than in the rest of the world.Reasons for the interaction were explored independently by 2 statistical groups. Systematic errors in trial conduct were investigated. Statistical approaches evaluated the likelihood of play of chance. Cox regression analyses were performed to quantify how much of the regional interaction could be explained by patient characteristics and concomitant treatments, including aspirin maintenance therapy. Landmark Cox regressions at 8 time points evaluated the association of selected factors, including aspirin dose, with outcomes by treatment. Systematic errors in trial conduct were ruled out. Given the large number of subgroup analyses performed and that a result numerically favoring clopidogrel in at least 1 of the 4 prespecified regions could occur with 32% probability, chance alone cannot be ruled out. More patients in the United States (53.6%) than in the rest of the world (1.7%) took a median aspirin dose ?300 mg/d. Of 37 baseline and postrandomization factors explored, only aspirin dose explained a substantial fraction of the regional interaction. In adjusted analyses, both Cox regression with median maintenance dose and landmark techniques showed that, in patients taking low-dose maintenance aspirin, ticagrelor was associated with better outcomes compared with clopidogrel, with statistical superiority in the rest of the world and similar outcomes in the US cohort.The regional interaction could arise from chance alone. Results of 2 independently performed analyses identified an underlying statistical interaction with aspirin maintenance dose as a possible explanation for the regional difference. The lowest risk of cardiovascular death, myocardial infarction, or stroke with ticagrelor compared with clopidogrel is associated with a low maintenance dose of concomitant aspirin.URL: http://www.clinicaltrials.gov. Unique identifier: NCT00391872.
View details for DOI 10.1161/CIRCULATIONAHA.111.047498
View details for Web of Science ID 000293338400014
View details for PubMedID 21709065