4 Evidence and interpretation

The Appraisal Committee (appendix A) considered evidence from a number of sources (appendix B).

4.1 Clinical effectiveness

4.1.1 The review identified 17 studies of the clinical effectiveness of naltrexone treatment: 1 systematic review (Cochrane review), 13 randomised controlled trials (RCTs) and 3 non-randomised comparative studies. None of the studies were conducted in the UK. The length of follow-up varied in the RCTs from 20 days to 1 year. The RCTs and comparative studies evaluated the effectiveness of naltrexone treatment in a total of 1269 people and reporting was of poor quality (for example, randomisation was not adequately reported in most RCTs). Two of the RCTs were conducted in a prison setting (parolees or probationers), and seven included various psychosocial therapies in both arms of the trials (for example, twice-weekly drug counselling, psychotherapy and behavioural therapy).

4.1.2 In addition to the 17 studies of the clinical effectiveness of naltrexone treatment, 9 RCTs studied the effectiveness of different strategies to improve retention on naltrexone treatment. Three of these RCTs assessed reward with incentive vouchers for treatment compliance (one also included relationship counselling), four RCTs evaluated psychosocial therapies and two RCTs examined the effectiveness of pharmaceutical agents. These were of poor to moderate quality (the methods of randomisation were not adequately described and intention-to-treat analyses were not performed in most of the trials).

4.1.3 The main effectiveness outcomes reported in the RCTs were retention, relapse rates (opioid use) and re-incarceration of parolees or probationers.

Effectiveness of naltrexone compared with control treatment

4.1.4 Retention on treatment was reported in the Cochrane review and seven RCTs. The Cochrane review showed no significant difference in retention for people treated with naltrexone and adjunctive psychosocial therapy compared with psychosocial therapy alone (risk ratio [RR] 1.08; 95% confidence interval [CI], 0.74 to 1.57). Six of the seven RCTS (three of which included adjunctive psychosocial therapy in each arm) reported no significant difference in retention with naltrexone treatment. One trial reported a significant improvement in retention on naltrexone treatment compared with psychosocial therapy (RR 0.66; 95% CI, 0.43 to 0.93). A fixed-effect meta-analysis of all seven RCTs conducted by the Assessment Group showed no difference in retention on treatment with naltrexone, with a RR of stopping treatment of 0.94 (95% CI, 0.84 to 1.06) and a hazard ratio (HR) from five RCTs of 0.90 (95% CI, 0.69 to 1.17). However, a fixed-effects model was not the most appropriate method of meta-analysis because heterogeneity between trials was found. A random-effects meta-analysis gave a RR of stopping treatment of 0.90 (95% CI, 0.55 to 1.48).

4.1.5 Relapse rates (assessed by the presence of opioids in urine samples) were reported in the Cochrane review and six RCTs. The Cochrane review showed a significant reduction in illicit diamorphine use (RR 0.72; 95% CI, 0.58 to 0.90). Of the six RCTs that reported relapse rates, three included adjunctive psychosocial therapy in each arm. Five of the six RCTs reported no statistically significant differences in relapse rates with naltrexone treatment. One RCT reported a statistically significant improvement in relapse rates with naltrexone treatment (RR 0.41; 95% CI, 0.21 to 0.74). Pooled analysis of relapse rates in the six RCTs showed a statistically significant reduction in the risk of opioid use with naltrexone compared with placebo; the RR of relapse was 0.72 (95% CI, 0.58 to 0.90), with a HR of 0.53 (95% CI, 0.34 to 0.82) and a number needed to treat (NNT) of eight people to prevent one opioid relapse. One trial reported the number of people who were opioid free as a proportion of people retained on treatment. This trial showed no difference in the number of opioid-free people treated with naltrexone compared with psychosocial therapy over 26 weeks (statistical significance was not reported).

4.1.6 Two small RCTs reported re-incarceration rates of parolees or probationers having naltrexone treatment compared with placebo, with adjunctive psychosocial therapy in each arm. Pooled analysis showed a significant reduction in re-incarceration in favour of naltrexone (RR 0.50; 95% CI, 0.27 to 0.91).

4.1.7 Mortality was not reported in any of the trials. A retrospective study in the USA examined the number of deaths in people who had received naltrexone treatment over a period of 2 years (n = 1196) and in people who had not received naltrexone treatment (total number not reported). Diamorphine overdose resulted in 21 out of 33 (64%) deaths in the naltrexone group and 71 out of 96 (74%) deaths in people not being treated with naltrexone.

4.1.8 The National Coronial Information Service reported 32 naltrexone-related deaths in the period 2000–2003 in Australia. The mortality rate was estimated at 1 per 100 person years during naltrexone treatment and 22.1 per 100 person years during the first 2 weeks after treatment.

4.1.9 Nine RCTs evaluated the effectiveness of different strategies to increase retention on naltrexone treatment (incentive vouchers, psychosocial therapy and pharmaceutical agents). The length of follow-up in these trials ranged from 12 to 52 weeks. Pooled analysis of the effect of adjunctive psychosocial therapies on compliance with naltrexone treatment in six trials reported a significant improvement of 19% with adjunctive psychosocial therapy compared with naltrexone treatment alone.

4.1.10 In summary, the small evidence base reported conflicting results. The quality of the studies was poor and randomisation was not adequately reported in RCTs. One trial reported a significant improvement in retention and relapse rates with naltrexone treatment, but the majority of RCTs reported no significant difference between naltrexone and control treatment. When the results from these trials were pooled, naltrexone was associated with a significant reduction in relapse, but not a difference in retention on treatment compared with control treatment. Adverse events reported in two trials showed no difference in mortality between naltrexone and control treatment. Although useful data for the comparison of adverse events were not reported in the majority of the RCTs, mortality data from the USA showed no difference in mortality rates with naltrexone compared with non-naltrexone treatment.

4.2 Cost effectiveness

4.2.1 No published economic evaluations of the cost effectiveness of naltrexone treatment were identified. The manufacturer did not submit evidence for this appraisal.

4.2.2 The Assessment Group developed a decision analytical model to assess the cost effectiveness of naltrexone plus psychosocial support compared with placebo plus psychosocial support (psychosocial support alone). The model estimated costs and outcomes from a National Health Service (NHS) perspective. Costs were based on estimates of resource use including a daily dose of 50 mg naltrexone, counselling sessions, monitoring of treatment, GP visits, emergency department visits, inpatient hospital stays, outpatient mental health appointments and inpatient mental health admissions. The time horizon of the model was limited to 12 months. This was because of the length of follow-up in the trials, and clinical advice that people are not retained on naltrexone treatment in the long term.

4.2.3 Data on retention on treatment at 2, 6, 13 and 25 weeks and 12 months were included in the model. Data on this endpoint for naltrexone (with or without psychosocial therapy) compared with placebo treatment (with or without psychosocial therapy) was based on the meta-analysis of five RCTs (pooled HR 0.90; 95% CI, 0.69 to 1.17). This HR was applied to the survival curve for naltrexone treatment to generate an estimate of retention on psychosocial treatment alone. Data on the proportion of opioid-free people retained on treatment was based on a single RCT, which reported little difference between the naltrexone arm (84% opioid-free people) and the control arm (86% opioid-free people). Data from the 'National treatment outcome research study' (NTORS) were used to inform the proportion of drug-taking people who were injecting or not injecting.

4.2.4 Health outcomes were expressed as quality-adjusted life years (QALYs). In the absence of published data on quality of life associated with drug misuse, the Assessment Group obtained health-related utility data from a panel of members of the public. Average QALYs were calculated for people retained on naltrexone treatment and psychosocial therapy, people retained on psychosocial therapy alone, and people not retained on treatment. People retained on naltrexone treatment gained fewer QALYs than those retained on control treatment, based on evidence from one RCT that showed that a higher number of people relapsed on naltrexone. The total QALYs associated with each treatment arm were then determined by different retention rates.

4.2.5 The base-case analysis demonstrated that naltrexone plus psychosocial therapy is associated with an incremental cost-effectiveness ratio (ICER) of £42,500 per QALY gained. This is based on the assumption of a slightly higher (non-significant) proportion of people relapsing with naltrexone plus psychosocial therapy compared with psychosocial therapy alone.

4.2.6 Data on the proportions of opioid-free people in each arm of the model was based on one trial. A one-way sensitivity analysis that assumed the same proportion of people taking opioids in both arms reduced the ICER to £34,600 per QALY gained.

4.2.7 A probabilistic sensitivity analysis around the base-case analysis was also undertaken. The probability of naltrexone being cost effective does not rise much above 50%, regardless of the willingness to pay for an additional QALY. This is because of the uncertainty in the estimates of clinical effectiveness reported in the RCTs.

4.2.8 An analysis that included the costs of the criminal justice system and victims of crime was also performed. Costs to victims of crime included the costs of increased security measures and the direct costs of material or physical damage. The results of this analysis found that naltrexone plus psychosocial therapy dominates psychosocial therapy alone because it is less costly and more effective. The dominance of naltrexone therapy is driven by the reduction in costs to the victims of crime in the naltrexone arm, because of the increased retention on treatment and the associated reduction in crime. A one-way sensitivity analysis that excluded the costs to the victims resulted in an ICER of approximately £51,000 per QALY.

4.3 Consideration of the evidence

4.3.1 The Committee reviewed the data available on the clinical and cost effectiveness of naltrexone, having considered evidence on the nature of the condition and the value placed on the benefits of naltrexone by people who are dependent on opioids, those who represent them, and clinical experts. It was also mindful of the need to take account of the effective use of NHS resources.

4.3.2 The Committee was persuaded of the likely clinical effectiveness of naltrexone as a disincentive for opioid use in people following detoxification because of their understanding of the pharmacological basis of its action; that is, naltrexone blocks the receptors responsible for the euphoric effects of opioids.

4.3.3 The Committee considered the quality and outcomes of the RCTs. The Committee heard from clinical experts that the most important effectiveness outcome for naltrexone treatment is relapse prevention (continued abstinence from taking illicit opioids). Experts advised that retention on treatment is problematic as an outcome of effectiveness because it is primarily a measure of treatment compliance (taking naltrexone medication) and noted that people who do not wish to remain in contact with drug treatment services may still be compliant in taking naltrexone. The Committee noted that most of the studies showed that naltrexone therapy resulted in no significant difference in retention on treatment, although pooled analysis of the RCTs showed a lower rate of relapse to illicit opioid use compared with control treatment. The Committee was persuaded by the experts' testimony that, in clinical practice, naltrexone therapy was often associated with dramatic improvements in abstinence from opioid use, which is the principal aim of treatment. The Committee considered that mortality in people retained on naltrexone treatment was likely to be reduced specifically because of the lower relapse rate to illicit opioid use. The Committee additionally concluded that the reduction in opioid use could lead to substantial improvements in overall quality of life for those people retained on naltrexone treatment.

4.3.4 The Assessment Group and experts raised concerns regarding the external generalisability of the RCTs, none of which were conducted in the UK and which involved people with differing patterns of opioid misuse. The Committee heard from experts that naltrexone is usually only prescribed following an initial clinical assessment for people who are highly motivated to remain in an abstinence programme, who have been fully informed of the potential adverse effects and the benefits of treatment, and for whom an abstinence programme is judged to be appropriate based on initial clinical assessment. Experts also noted that none of the RCTs were conducted in the UK, and the degree of supervision of naltrexone administration was likely to be variable, whereas adequate supervision is currently recommended as best clinical practice in the UK. The Committee was persuaded that close monitoring is particularly important when naltrexone treatment is initiated because of the higher risk of fatal overdose at this time. In addition, the Committee understood that discontinuation of naltrexone may be associated with an increase in inadvertent overdose from illicit opioids. The Committee therefore considered that supervision of naltrexone administration is important for continued compliance with medication, in order to maximise retention on treatment and abstinence from illicit opioid use, and to minimise the adverse effects of treatment. The Committee was also aware of two uncontrolled studies that reported higher effectiveness of naltrexone in highly motivated people. The Committee was convinced that effectiveness outcomes for retention on treatment and abstinence from opioid use are likely to be higher in clinical practice than reported in the RCTs, because treatment is targeted to motivated people who have expressed a preference for an abstinence programme. The Committee concluded that adequate supervision of treatment is likely to lead to further improvements in the effectiveness of naltrexone treatment over and above that seen in the RCTs.

4.3.5 The Committee considered the personal statements and testimonies of patient experts on the adverse effects of naltrexone treatment. The Committee heard that people taking naltrexone often experience adverse effects of unease (dysphoria), depression and insomnia, which can lead to relapse to illicit opioid use while on naltrexone treatment, or failure to continue on treatment. Experts advised that adverse effects may be caused by either withdrawal from illicit drugs or by the naltrexone treatment itself, and stressed the importance of prescribing naltrexone as part of a care programme that includes psychosocial therapy and general support. The Committee noted the importance of patients and carers being fully informed of the potential adverse effects as well as the benefits of naltrexone treatment, and considered psychosocial therapy to be instrumental in some people remaining on naltrexone treatment.

4.3.6 The Committee considered the uncertainties relating to the external generalisability of the effectiveness evidence used in the economic model and the consequent uncertainties in interpreting the ICERs. The Assessment Group advised that for the purpose of economic modelling, the rate of relapse to illicit opioids used in their assessment was based on a small study that reported no significant difference in relapse rates between naltrexone and control treatment. The Committee noted that this was contrary to the pooled analysis of relapse rates reported in three RCTs (HR 0.53; 95% CI, 0.34 to 0.82), which showed a significant reduction in relapse with naltrexone compared with control treatment. The Committee concluded that the model may have underestimated the reduction in relapse to opioid use, and therefore also underestimated the cost effectiveness of naltrexone treatment on the basis of potential improvements in both quality of life and mortality.

4.3.7 The Committee further considered the base (reference) case ICER of £42,500 per QALY for adjunctive naltrexone treatment plus psychosocial therapy compared with psychosocial therapy alone. The Committee considered that this ICER was based on data on the effectiveness of adjunctive naltrexone treatment compared with psychosocial therapy alone in a general population of drug misusers. The Committee was persuaded that this ICER was a conservative figure that underestimated the cost effectiveness of naltrexone in people who were highly motivated to remain abstinent and who were enrolled in a supervised treatment programme. The Committee was persuaded that for these people, the ICER for naltrexone treatment would be substantially lower than the base case, principally because of the factors outlined in sections 4.3.4 and 4.3.6.

4.3.8 In summary, the Committee was convinced of the clinical effectiveness of naltrexone treatment in a selected, highly motivated group of people. The Committee concluded that for people who preferred an abstinence programme, who were fully informed of the potential adverse effects and benefits of treatment, and who were highly motivated to remain on treatment, naltrexone treatment would fall within acceptable cost-effectiveness limits.