Fig 5. Trial sequential analysis of one-year mortality.

The risk of type Ⅰ error was set at 5% with a power of 80%. The variance was calculated from the data obtained from the included trials. The relative risk reduction (RRR) was set at 20%.