Missing values

So i ran an MCAR test on my data and had the result that my data is indeed missing completely at random . The following shows the percentage of missing values.

am i still supposed to replace missing values by using expected maximization or should i just replace all the missing values with 0?