It's a comparison between two control groups and the amount that were infected.

X - vaccine - 25 people get it Y - no vaccine - 250 people get it

That would mean it's 1 - 25/250 * 100 = 90% effective.