The linked study tries to assess the number of chronic pain prescriptions and comes up with 1%, which is where they get the claim that 20% indicates addiction.

The linked study tries to assess the number of chronic pain prescriptions and comes up with 1%, which is where they get the claim that 20% indicates addiction. That said, the methodology for this (checking how many scripts are for long-lasting, high-dose opioids) looks dubious at best, and misses non-chronic, recurring illnesses that might lead to multiple prescriptions.