Hi, I understand that you discarded 30 items from the original version.

Your shortened version (20 items) can be justified if the shortened version satisfies the following conditions:

1. The shortened version represents the theoretical structure (dimensions) of the measured variable (construct).

1. At least three items have substantial loadings on each dimension.

2. The reliability for each dimension is at least satisfactory.

Hope my answer helps.

This is not strange at all. The profit of the EFA is precisely to reduce the dimensionality in a set of items (construct). When the items saturate below 0.30 in the output of the EFA, they can be discarded because they do not contribute in a relevant way to reducing the dimensionality.

The very purpose of EFA is to do that. Items that have factor loadings below 0.5 on any construct must be ideally removed. Some have even lower standards. You can refer to this useful document on EFA- https://www.open-access.bcu.ac.uk/6076/1/__staff_shares_storage%20500mb_Library_ID112668_Stats%20Advisory_New%20Statistics%20Workshops_18ExploratoryFactorAnalysis_ExploratoryFactorAnalysis4.pdf