新型冠状病毒疫苗作为新冠肺炎防控的有效手段，一直受到各界的广泛关注。但不同疫苗的临床试验报告中，疫苗效力及95%置信区间的计算方式不尽相同。本文选取辉瑞BNT162b2疫苗和国药两种疫苗（WIV04疫苗和HB02疫苗）的Ⅲ期临床试验报告，尝试还原其中采用的疫苗效力及95%置信区间的计算方法，并从过程和结果两个角度对每种方法进行分析。
两份数据报告都考虑到了随访时间的差异，通过人年发病率密度计算疫苗效力，更加科学且严谨。在疫苗效力95%置信区间的计算中，辉瑞文献采用了Beta二项分布模型和克洛珀-皮尔逊法两种方法，但计算过程较复杂，易产生误差；相较而言，国药文献采用Poisson回归模型，具有更高的合理性和可信度。
 

As an effective means of COVID-19 prevention and control, COVID-19 vaccine has been widely concerned by all walks of life. However, the vaccine efficacy and its 95% confidence intervals of different vaccines are not calculated in the same way. Based on the phase III clinical trial reports of the BNT162b2 vaccine developed by Pfizer and the two vaccines developed by Sinopharm (WIV04 vaccine and HB02 vaccine), this paper attempts to restore the calculation method of vaccine efficacy and 95% confidence interval, and analyzes the process and results of each method.
Both data reports took into account the difference in surveillance time, and calculated vaccine efficacy by incidence density per 1000 person-years, which was more scientific and rigorous. In the calculation of 95% confidence interval of vaccine efficacy, Pfizer adopted two methods: the beta-binomial model and the Clopper–Pearson method, but the calculation process was complicated and prone to errors. in comparison, Sinopharm used Poisson regression model, which has higher rationality and credibility.