Effectiveness and cost-effectiveness of adjunctive personalised psychosocial intervention in treatment-resistant maintenance opioid agonist therapy: a pragmatic, open-label, randomised controlled trial

BACKGROUND: Opioid use disorder (OUD) is a chronic, debilitating and costly disorder, and is increasingly prevalent in many countries. Maintenance opioid agonist therapy (OAT) with oral methadone or sublingual buprenorphine is the first-line, empirically-supported treatment. However, many patients do not stop illicit drug use while enrolled in OAT. To address this, we developed a personalised psychosocial intervention (PSI) using a toolkit of behaviour change techniques as an adjunct to continued OAT. The aim was to estimate if the PSI was effective and cost-effective.

METHODS: This was a pragmatic, open-label, randomised controlled trial at a specialist NHS community addictions clinic in England. We recruited adults meeting Diagnostic and Statistical Manual of Mental Disorders IV criteria for opioid and/or cocaine dependence, enrolled in OAT for a median of 26 weeks (IQR 10-88), voluntarily seeking continued OAT, who were treatment-resistant. Treatment resistance was defined as using illicit or non-prescribed opioids and/or cocaine on one or more days in the past 28 days at study screening, verified by positive urine drug screen. Participants were allocated (1:1) by a web-accessed randomisation sequence (stratified by OAT medication, current cocaine use, and current injecting) to receive continued standard OAT (treatment-as-usual, TAU) or standard OAT and a personalised PSI. Treatment was provided open-label. Outcome data were collected by independent research assistants. The primary outcome was treatment response at 18 weeks, defined as abstinence from illicit and non-prescribed opioids and cocaine in the past 28 days recorded by the Treatment Outcomes Profile and negative urine drug screen. Taking a societal cost perspective, an evaluation of cost-effectiveness was done by taking a wide range of values of willingness-to-pay (WTP) for a unit improvement in the probability of treatment response, and EQ-5D-3L derived quality adjusted life years (QALYs). The planned analysis was intention-to-treat (ITT), including all those who were randomly allocated. This trial is registered with the ISRCTN registry, number ISRCTN69313751.

FINDINGS: Between June 7, 2013 and December 21, 2015, we randomly allocated 136 patients (50%) to the PSI group and 137 patients (50%) to the TAU group. The trial database was locked for analysis on April 19, 2017. In error, we re-randomised three participants. These cases were excluded from all analysis. Due to this error, the analysis is classified as a modified ITT (mITT). All other randomised participants were included. In the mITT analysis, treatment response was greater in the PSI group (22 [16·3%] of 135) compared to the TAU group (9 [6·7%] of 135; adjusted log odds 1·20, 95% CI 0·01 to 2·37, p-value=0·005). The PSI had a higher probability of being cost-effective than TAU. There was a probability range of 47% to 87% for WTP thresholds of £0 to £1,000. QALYs were higher in the PSI group than the TAU (mean difference 0·048, 95% CI 0·016 to 0·080, p-value=0·004), with a 60% and 67% probability of cost-effectiveness at the NICE willingness to pay thresholds of £20,000 and £30,000 per QALY, respectively.

INTERPRETATION: Compared to maintenance opioid agonist treatment as usual, a personalised PSI for treatment resistant patients achieved more abstinence from opioids and cocaine with a higher probability of being cost-effective. Patients who are treatment-resistant to OAT should be offered a personalised PSI.