Background: Between 1990 and 2006 in Birmingham, Alabama USA, 4 separate randomized controlled studies, called "Homeless 1" through "Homeless 4", treated cocaine substance abuse among chronically homeless adults, largely black men, many with non psychotic mental health problems. The 4 studies had 9 treatment arms that used various counseling methods plus, in some arms, the provision of housing and work therapy usually with a contingent requirement of urine-test verified abstinence from substances. Participants in the abstinent -contingent arms who lapsed on abstinence were removed from housing and sent to an evening public shelter from which they were daily transported to day treatment until they returned to abstinence.
Aims of the Study: This paper compares the cost effectiveness of the treatment arms.
Methods: Societal cost per participant (in 2014 dollars) for each arm is defined as direct treatment cost plus cost of jail or hospital plus societal expense of public shelter use by lapsed participants. An untreated Base Case is defined as 5 percent abstinence with 95 percent usage of a public shelter. Incremental Cost Effectiveness Ratios (ICERs) for paired arms are defined as the change in cost per participant divided by the change in abstinence. Bootstrapping estimates confidence intervals.
Results: Average cost per participant at the end of 6 months of active treatment in 7 arms with comparable data ranged from $10,447 to $36,194 with corresponding average weeks abstinent ranging from 6.1 to 15.3 out of a possible 26 weeks. In contrast, the Base Case would cost $6,123 for 1.3 weeks of abstinence. Compared to the Base Case, the least expensive "DT2" treatment has an ICER of $901 (95% CI = $571 to $1,681) per additional week of abstinence and the most expensive "CMP4" has an ICER of $2,147 (95% CI = $1,701 to $2,848). Additionally, the Homeless 3 study found that the abstinent contingent housing (ACH3) treatment compared to the Non Abstinent Contingent Housing (NAC3), analogous to "Housing First", achieved better abstinence (12.1 v. 10 weeks) at higher average cost ($22,512 v. $17,541) yielding an ICER for this comparison of ($2,367, 95% CI = $-10,587 to $12,467). Similar results are found at 12 months (6 months after active treatment).
Discussion: More intensive methods of counseling improved abstinence but 4 of the 7 treatments were inefficient ("dominated"). Bootstrapping shows that results are sensitive to which individuals were randomly assigned to each arm. A limitation of the analysis is that it does not consider the full societal cost of lost wages, crime costs beyond jail expenses and deterioration of neighborhood quality of life. Additionally, populations treated by Housing First programs may differ from the Birmingham Homeless studies in the severity of addiction or co-occuring psychological problems.
Implications for Treatment: The Homeless studies show that abstinent contingent safe housing with counseling can substantially improve abstinence for homeless cocaine abusers. Incremental costs rise sharply with more intensive counseling; modest programs of counseling may be more cost effective in a stepped treatment strategy.