Multiple Imputation (MICE) for small size?

Hi all,

I am wondering whether I should use Multiple Imputation by Chained Equations for my sample size. I have a small sample (n=50) and there is one important predictor which I hope to include in my logistic model. However, this predictor is only available among 40 participants (i.e. 20% missing).

So my question is: Should I include this predictor and run the MICE package or should I simply drop this predictor?

All the help & feedback are appreciated.