We investigate stochastic extinction in an epidemic model and the impact of random vaccinations in large populations formulated in terms of an optimal escape path. We find that different random vaccination strategies can have widely different results in decreasing expected time till extinction, for the same total amount of vaccines used. Vaccination strategies are considered in terms of two parameters: average frequency of vaccinations, given by $\gamma$, and the amplitude of the vaccinations, $\epsilon$, where $\epsilon \ll 1$ refers to the proportion of the population being vaccinated at some particular instant. It is found that while the average number of individuals vaccinated per unit time, $\gamma \epsilon$, is kept constant, the particular values of $\gamma$ and $\epsilon$ can play a highly significant role in increasing the chance of epidemic extinction. The findings suggest that expected time till extinction can be significantly shortened if less frequent vaccinations occur in larger groups, corresponding to low $\gamma$, high $\epsilon$ strategy.