Only keep required fields.

Fields created in data steps can be dropped or kept by using the DROP or KEEP statements. Often fields are created which aren’t necessary for the output data set such as do loop counts or file pointer controls.
Using these statements for output datasets are used fairly widely, but they are not always used for input datasets. When merging data sources together, consider whether every variable in all datasets are required for this merge - if not, drop the data as soon as possible.