Strengths of our analysis include its sample size, random allocation of statin therapy, and masked ascertainment of incident events. Our analysis plan was also highly conservative in several respects, an approach we took on an a-priori basis so as not to underestimate potential hazards of treatment. For example, we used the observed HR for diabetes within JUPITER of 1·25 rather than the smaller HRs reported in the most compre- hensive recent meta-analysis of 1·18 for rosuvastatin and 1·09 for all statins.3 We also elected to conservatively include all incident cases of physician-reported diabetes that occurred during the trial (including those reported between the time of study completion and the last patient closeout visit)
การวิเคราะห์ของเรารวมถึงขนาดของตัวอย่างการจัดสรรแบบสุ่มของการรักษาด้วยยา statin และ