Comparisons between opioid agonist maintenance, detoxification and opioid antagonist approaches

Few studies comparing naltrexone treatments with opioid agonist maintenance treatments were found. The only randomized trial to compare methadone or buprenorphine with naltrexone was conducted in intravenous buprenorphine users[142]. There were no studies comparing long-acting naltrexone formulations to opioid agonist treatment approaches.

No studies assessed the potential benefit to populations of having more than one treatment available (e.g. buprenorphine and methadone, or opioid antagonists and opioid agonists).

No randomized trials were conducted in specific populations (e.g. young people, people with short histories of dependence, people who do not inject and pregnant women) to compare approaches based on opioid detoxification with approaches based on opioid agonist maintenance treatment. Although evidence strongly favours opioid agonist approaches, opioid detoxification remains the most requested treatment in many settings.

Impact of treatment on HIV transmission

Well-conducted observational studies to evaluate the impact of pharmacological treatment on HIV transmission would be useful, because most randomized trials did not collect data on HIV risk practices and HIV transmission. In particular, more data on the impact of buprenorphine treatment on HIV transmission is needed, because this treatment is unsupervised in many settings.

Optimal doses

Although there are sufficient studies on the optimal doses of methadone to make recommendations, the same is not true for buprenorphine. Research is also needed on the optimal doses of methadone and buprenorphine in non-injecting drug users (e.g. opium smokers).

Supervised dosing in agonist maintenance treatment

For both methadone and buprenorphine, surprisingly little research has been conducted on the impact of dosing supervision on treatment outcome. This is a particularly important question given the widespread use of unsupervised buprenorphine treatment in several countries.

Antagonist treatment in different settings

There is a discrepancy between the modest benefits of oral naltrexone treatment in developed countries and reports from clinicians of the usefulness of naltrexone in developing countries. Further research on opioid antagonists in developing countries is needed.

Prescription opioid dependence

In many countries, the number of people with prescription opioid dependence now exceeds the number of people with illicit opioid dependence. More research is required on the use of opioid detoxification and agonist maintenance in this population, including the required degree of supervision.

Adolescence

The relative lack of research in young people and people with brief histories of opioid dependence is concerning, because this population may have the greatest capacity for change. More research is needed on psychosocial assistance, including family-based approaches, and on the relative merits of opioid agonist treatment and withdrawal, and antagonist treatment.

Pregnancy and breastfeeding

More research is required to establish the safety of buprenorphine and naltrexone for pregnant and breastfeeding women.

Outcomes of planned cessation of agonist maintenance treatment

More research is needed on when opioid agonist treatment can be stopped without leading to high rates of relapse. Studies comparing methods of cessation of agonist maintenance treatment are also required.

Psychosocial support

More research is needed on various psychosocial approaches to treatment, particularly the more social approaches, such as employment and residential programmes.

Study size

Many studies are too small to adequately address the questions being asked. This may reflect the difficulties in funding clinical trials in this population, and the relatively small sizes of many treatment centres. Larger studies are required on most topics. In addition, many studies replicate earlier trials despite there remaining gaps in the evidence. Treatment networks at international levels may be needed to coordinate large trials to answer simple questions, a method that has been used in many fields of medicine in recent decades.

Study duration

The follow-up period of most studies is too short, given that opioid dependence is a chronic condition.

Study design

Most randomized trials considered in this review did not use intention-to-treat analyses, and many did not document allocation concealment. Although it is difficult to blind study participants in studies of psychoactive medication, greater consideration should be given to the use of blinding in outcome assessment and in the statistical analysis.

Outcome measures

Intention-to-treat analysis includes both “in treatment” and “out of treatment” patients. Most calculations assume that all people who drop out of treatment return to baseline levels of drug use. In reality, some people will start another form of treatment and other people will no longer require treatment. Although treatment retention is an important proxy measure in the absence of other relevant outcomes, continuous retention in treatment and treatment status at follow-up should be regarded as a measure of exposure to the intervention, rather than as a health outcome.

Studies should instead focus on drug use, related risk behaviours (e.g. injection, sexual activity that could lead to disease transmission) and health outcomes. Health outcomes should include measures of psychological health and well-being, function or dysfunction, quality of life, mortality and, where appropriate, specific health conditions (e.g. HIV, hepatitis C, sexually transmitted diseases).

Given that developments are taking place in pharmacogenetics research, consideration should be given to the storage of blood samples for such analysis from clinical trial participants, with their consent.

The impact of drug use beyond the individual (e.g. family, carers and society) should also be considered. Data for social cost estimates (e.g. criminal activity, health-care use, income and receipt of government benefits) are lacking for many settings. Cost-effectiveness and cost–benefit analyses are also possible if the cost of treatment is measured.