Buprenorphine was more effective and more costly for all primary and most secondary outcomes compared to naltrexone. Incremental cost-effectiveness ratios were small – below $50 for primary outcomes, mostly below $350 for secondary outcomes. Naltrexone was dominated by placebo for all secondary outcomes. Performance of the buprenorphine intervention fell short for AIDS Risk Inventory aggregated scores and AIDS Risk Inventory injection drug use items.
Because this is not a cost-utility analysis, we did not use quality-adjusted life years (QALYs) saved as an outcome measure and cannot directly compare our cost-effectiveness results with standard cost-utility thresholds such as the £20,000–30,000 per QALY saved threshold used by the National Institute for Health and Clinical Excellence to assess inclusion of treatments for the United Kingdom’s National Health Service, or the World Health Organizations’ per disability-adjusted life year averted, GDP per capita-based thresholds. However, we provide some context for assessing the cost-effectiveness of our interventions by looking at incremental cost-effectiveness ratios reported by other studies involving HIV prevention and substance abuse interventions, and find that our incremental cost-effectiveness ratios compare favorably. For example, compared to naltrexone, buprenorphine cost $233 per additional patient still abstinent from illicit opiates at 6 M; a randomized controlled trial in the US found a cost of $18,846 per additional alcohol abstainer at 4 M for an enhanced intervention compared to a standard intervention with a Well Woman Exam (Ruger JP, Abdallah AB, Luekens C, Cottler L, “Cost-effectiveness of peer delivered interventions for cocaine and alcohol abuse among women: a randomized trial,” under review). The $142 incremental cost-effectiveness ratio for each injection of heroin or other opiates prevented can be compared to the cost per HIV infection averted by a Thai needle and syringe program (between $82 and $165 2004 US dollars) 
, and is significantly less than the cost per HIV infection averted by a methadone treatment program in China (between $2,509–$4,609) 
. Another study reports incremental cost-effectiveness ratios of $212 and $166 for voucher-based and prize-based contingent management interventions, compared to standard outpatient treatment, for extending longest duration of cocaine and opioid abstinence by an additional week 
. In our study, the buprenorphine arm yielded incremental cost-effectiveness ratios of $197 (compared to placebo) and $291 (compared to naltrexone) for each additional week of maximum consecutive period of abstinence [data not shown], and $198 (compared to placebo) for an additional week from the last time heroin or opiates was used (longest duration abstinent). Although our incremental cost-effectiveness ratios from Malaysia are similar to those for comparable outcome measures from studies conducted in the United States, the financial burden may nonetheless be heavier for Malaysia as a less wealthy country.
Incremental costs were driven mainly by medication costs. If the cost of medication can be decreased – for example, if Malaysia obtained cheaper buprenorphine through discount – then buprenorphine can become even more cost-effective. The cost of buprenorphine can potentially also be lowered through local production in Malaysia, but the extent of savings would depend on the costs of licensing the medication from the patent holder and of building or modifying manufacturing facilities.
This study supports the use of medications other than naltrexone in heroin treatment programs. Meta-analyses and reviews have found inadequate evidence for naltrexone’s effectiveness (though two Russian studies presented positive results in retention and relapse reduction with naltrexone compared to no treatment) 
. Both buprenorphine and methadone have been approved for maintenance treatment in Malaysia 
. Studies comparing buprenorphine and methadone find no significant difference in effectiveness between the two, or identify greater effects for methadone 
but comparative effectiveness is still in dispute 
. In terms of cost-effectiveness, one study finds that, when the costs of drug-related crime is included, buprenorphine has lower cost but fewer heroin-free days than methadone, but when the cost of crime is excluded, buprenorphine is dominated by methadone 
. Other CEAs have been favorable to buprenorphine 
Treating heroin use reduces HIV risk behaviors, especially through reduction of injection drug use 
. Though neither the naltrexone nor buprenorphine interventions in this study proved cost-effective in reducing injection drug use, they are still likely to prevent HIV cases. A 2010 literature review on the effect of drug treatment programs to reduce HIV transmission among drug users found that drug treatment improves access to and compliance with antiretroviral therapy, and patients in drug treatment are more likely to attain sustained viral suppression, reducing HIV transmission 
. Because no future costs were included in this study, cost-effectiveness was likely underestimated since treatments for substance abuse positively affect the housing and employment sectors 
, crime 
, and the children of users 
. We did, however, study changes in work-related functions. Both the buprenorphine and naltrexone interventions increased the mean number of days paid for working (past 30 days) by almost 30% between baseline and 6 M, compared to 6% increase for placebo. Only buprenorphine treatment increased the mean income from employment (past 30 days) between baseline and 6 M, by 22%; it also reduced illegal income [data not shown]. These findings are consistent with the results from a study of buprenorphine and methadone maintenance programs in the Ukraine, which indicated that, over 6 months, number of days employed increased while number of participants receiving illegal income decreased under both programs 
We examined outcomes pertaining to medical treatment and medical problems. The buprenorphine arm reported a 56% decrease in the mean number of days participants experienced medical problems (past 30 days) between BL and 6 M while naltrexone and placebo arms showed increases. In addition, the buprenorphine intervention produced an 8% increase in days of outpatient treatment for drugs or alcohol (past 30 days), versus substantial decreases in the naltrexone and placebo arms (data not shown). In a study of problems with alcohol use among mentally ill adults, increased participation in substance abuse treatments and decreases in psychiatric and medical symptoms precede remissions from alcohol dependence or abuse 
. This suggests that the buprenorphine treatment might be more able to bring about longer term improvements than the naltrexone or placebo interventions. Taken together, our results indicate buprenorphine is a cost-effective alternative to naltrexone that might enhance economic productivity and reduce drug use over a longer term, both of which have broader positive societal implications for Malaysia.
This study has some limitations. First, buprenorphine was compared only to oral naltrexone. Cost-effectiveness may vary depending on how effective and costly implantable naltrexone is relative to its oral formulation. A recent Australian clinical audit suggests that naltrexone-implanted patients had longer total treatment duration, more days in treatment per episode, longer mean treatment times, but fewer treatment episodes than buprenorphine 
. Second, our findings’ generalizability may be limited since the study was conducted through community recruitment and a single outpatient clinic in Muar, Malaysia. Treatment costs, population characteristics, and willingness to enter and adhere to treatment may differ in other places. Third, secondary outcome measures were self-reported and not independent verified; however, there is no reason to believe that inaccuracies from self-reporting systematically biased comparisons among the three intervention arms in this double-blind, double-dummy, randomized controlled trial. Fourth, the study attrition rates were quite high for the three invention arms, especially for the placebo (41%) and naltrexone (33%) arms (18% for the buprenorphine arm). The difference in attrition rates between the placebo and naltrexone arms is not statistically significant. Unfortunately, we had not collected data on the characteristics of participants who dropped out of the study. Given the sizable attrition rates, it is difficult to say if attrition was systematically driven by certain participant characteristics, or if it was a result of a lack of intervention effectiveness (especially of the placebo and naltrexone arms). If attrition was driven by participant characteristics, the secondary outcome measures – especially the Addiction Severity Index items for medical treatment and work-related functions () – may be biased by excluding study drop-outs who might have been sicker or who were less employable.
Buprenorphine maintenance treatment offers a potent way to tackle drug-related HIV infection and other public health concerns. One study used gross-costing to estimate the cost of implementing the Malaysian randomized controlled trial interventions in other countries 
. Incremental costs comparing buprenorphine to naltrexone were under $1,000 for most countries. Buprenorphine treatment can reach 10% of opiate users with $36 million in Afghanistan and an estimated 100% of users with $30 million in Lao PDR, two of the world’s largest opium producers. Per patient buprenorphine treatment costs vary – $834 in Iran, $2,863 in Botswana, $7,202 in the UK – yet such costs may still be less than what would be needed to address the consequences of drug abuse, HIV and other drug-related infections as well as other societal and future costs. These numbers suggest that buprenorphine, found to be more effective and potentially cost-effective compared to naltrexone in Malaysia, can be used to treat heroin dependence in even poor countries at a cost that can be within reach, especially if drug discounts and foreign aid are made available for this purpose.