My (totally unsourced and straight from my arse) guess is that both the decrease in efficacy and the increase in dependency are going to occur at a rate perportional to the current dosage. As far as I can tell the user's desired dose will increase at an exponential rate - making the drugs thoroughly nsuitable for long term usage.
Further for conditions that must be managed because they are impossible to fix opiods are hard to replace. Opoids may provide a managable solution to managing pain for decades for some patients while singing koom ba ah, meditating, discussing your feelings or alternative therapies are largely snake oil.
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.
It's basically saying 20% of all opioid scripts are for chronic pain, which sounds like a pretty reasonable number to me.