Supplementary appendix This appendix formed part of the original submission and has been peer reviewed. We post it as supplied by the authors. Supplement to: Rich JD, McKenzie M, Larney S, et al. Methadone continuation versus forced withdrawal on incarceration in a combined US prison and jail: a randomised, open-label trial. Lancet 2015; published online May 29. http://dx.doi.org/10.1016/ S0140-6736(14)62338-2.
Appendix - Health Economic Methods Health care utilization and costs The base case economic analysis takes the perspective of the healthcare payor examining direct medical care costs for 30 days post criminal justice system discharge. Methadone treatment costs were obtained from administration cost from each methadone treatment center. Hospital information system records (Eclipsis now Allscripts) provided utilization and financial data about ambulatory care, emergency visits, hospitalizations and total cost (not charge) at The Miriam Hospital and Rhode Island Hospital in the Lifespan health system. Based on the hospital information system records, physician costs were estimated from Medicaid reimbursement assuming level III reimbursement. Lastly, our base case analysis of direct medical costs did not consider short-term or long-term societal costs (drug dependence, criminal justice, productivity). Consequently, the above cost estimates likely underestimate cost benefits of continued methadone. Assuming that forced opiate withdrawal leads to lower methadone maintenance use and hence higher medical utilization, these limitations bias the analysis in favor of forced methadone withdrawal and against continued methadone. Health economic literature data were obtained by using search strategies recommended by the National Institute for Health and Care Excellence (NICE) health technology assessment guidance. We limited literature to those in the United States given international differences in health care systems and opioid delivery systems in various countries. Cost-effectiveness analysis Using a health care payor perspective, costs were based on drug administration fees for methadone from the dispensing centres and direct medical care costs for physician, ambulatory, emergency and hospitalization care. The time horizon was 30 days to match the primary clinical outcome. Given the short-time time horizon, costs were not discounted. We calculated the total care costs and the incremental cost-effectiveness ratio with effectiveness as the proportion of individuals enrolled in methadone maintenance treatment post-release at 30 days: Cost Forced Withdrawal Cost Methadone Continuation Methodone Entry Forced Withdrawal Methodone Entry Methadone Continuation Because of the non-parametric cost data, we generated 10,000 bootstrap samples to estimate the uncertainty in the incremental cost-effectiveness ratio. 21-23 In the base-case analysis, 4 individuals had missing economic data, 2 in each arm of the data. Consistent with the bias against Methadone Entry Forced Withdrawal (see Health care utilization and costs section), costs were assumed to be nil for individuals missing data in the base-case. In
sensitivity analysis, we substituted the mean cost or a regression-based cost for each intervention arm for missing data. Lastly, in an additional sensitivity analysis, we included societal costs (drug dependence, criminal justice, productivity but not HIV transmissions avoided) as estimated from a US study providing annual non-medical costs inflated to June 2014 using the United States Consumer Price Index (CPI) and adjusted for the one month time horizon. 52,53 Based on a previously published meta-analysis involving 15 studies and 1191 individuals, 52 methadone continuation to 30 days resulted in clean urines in 47.4% and reduced opioid use in 52.6%. Based on the same study, 52 monthly non-medical societal costs inflated to June 2014 costs equalled $3448 with continued methadone and $10,004 without methadone (assuming that the clean urines occurred in one-half that of individuals receiving methadone or 23 7%). Quality of life estimates were obtained from the literature and included 0 863 for heroin free, 0 6583 for reduced heroin use and 0 633 for untreated or relapse. 52 Results Continued methadone treatment resulted in a significantly higher methadone treatment costs that were more than offset by physician and medical care savings post-release resulting in a significantly lower 30-day total cost (Appendix Table 1). Because continued methadone treatment also resulted in higher likelihood of the primary outcome measure (methadone maintenance at 30 days), it dominated forced methadone withdrawal in deterministic analysis (Appendix Table 2). Sensitivity analysis Probabilistic sensitivity analysis (Appendix Figure 3) found that continued methadone treatment instead of forced withdrawal reduced costs by $19 (sd $1706) per individual with a 21% likelihood of being cost-saving and was cost-effective for willingness to pay thresholds exceeding $70,000 on the cost-effectiveness analysis frontier. 54 When incorporating societal costs, continued methadone treatment reduced costs by $1632 (sd $4057) per individual with a 47% likelihood of being costsaving and was optimal for all willingness to pay thresholds on the cost-effectiveness analysis frontier in probabilistic sensitivity analysis. Substituting mean or regression-based costs for the 4 individuals with missing cost data did not affect the results (available upon request). In the treatment as received on release sensitivity analysis, continued methadone treatment resulted in a significantly higher methadone treatment costs that were not completely offset by physician and medical care savings post-release resulting in a significantly higher 30-day total cost (Appendix Table 1). Because continued methadone treatment versus forced methadone withdrawal also resulted in higher likelihood of the primary outcome measure (methadone maintenance at 30 days), it had an incremental cost-effectiveness ratio of $17,482 per quality-adjusted life year gained, falling within the typical cost-effective range in deterministic analysis (Appendix Table 2).
In the treatment as received on release sensitivity analysis, probabilistic sensitivity analysis found that continued methadone treatment instead of forced withdrawal increased costs by $165 (sd $502) per individual with an 18% likelihood of being cost-saving and optimal for willingness to pay above $26,000 on the cost-effectiveness frontier analysis in probabilistic sensitivity analysis. When incorporating societal costs, continued methadone treatment reduced costs by $2978 (sd $4142) per individual with a 60% likelihood of being cost-saving and was optimal for all willingness to pay thresholds on the cost-effectiveness frontier analysis in probabilistic sensitivity analysis. Substituting mean costs or regression-based estimated costs for the 4 individuals with missing cost data did not affect the results (available upon request).
Appendix Table 1 - One month post-release mean outcomes per patient in a randomized controlled trial of methadone continuation versus forced withdrawal upon incarceration Continued Methadone Forced Methadone Withdrawal Incremental Difference Intention to treat Total Cost $609 $637 $28 Methadone maintenance 96% 80% +16% Treatment as received on release Total Cost $667 $521 +$146 Methadone maintenance 90% 41% +49% Limitations Limitations include the unavailability of actual methadone drug doses and other medication use. The absence of methadone dose information however should be minor given the minimal cost of methadone ($0.03 per mg of methadone based on average wholesale cost (REDBOOK 2014)). Medical care utilization was limited to care within the Lifespan health care system. Level III Medicaid reimbursement likely underestimates physician cost by using lower rates and by not including additional physicians (e.g., consultants) and higher levels of reimbursement for patient complexity (e.g., hospitalized patients).