Cost-effectiveness implications of increasing the efficiency of the extended-release naltrexone induction process for the treatment of opioid use disorder: a secondary analysis