Typically, what happened was that a user had say 1 BTC on paperwallet AddressA. They would import that key into a wallet, then they would create a transaction sending say 0.5 BTC to AddressB... The scenarios for what happens to the change were:
1. The wallet software sends the change back to AddressA. (What most users thought would happen)
or
2. The wallet software creates a new, random private/public keypair, derives AddressX from that keypair and sends the change there. (What a lot of wallet software did). The user would then uninstall the wallet and/or delete the newly created wallet file (without realising they needed a backup of this new wallet and/or keypair)
So in scenario 1, the user is fine, they still have the paperwallet with the private key and access to the "change". However, in scenario 2, if the user destroyed the wallet file (without any backup), they lost access to the new, randomly generated keypair that they need to access the change that is now in AddressX.
It doesn't really matter which way wallet software deals with "change"... as long as the user knows and takes the necessary steps to maintain access/control of the necessary private keys, then they won't lose anything.
There is a reason paperwallets are not recommended for new or inexperienced users