In this prospective cohort study of injection drug users, current enrolment in addiction treatment was negatively associated with attaining stable housing, even after adjustment for potential confounders including relationship status, employment status and drug use. The negative association of addiction treatment prior to attaining stable housing is somewhat unexpected, but may reflect the inadequacy of appropriate treatment exposure to meaningfully impact long-term housing outcomes. These findings also suggest that accessing addiction treatment may be a marker of instability (i.e., periods of extreme vulnerability during which other basic necessities including food and shelter take precedence over obtaining stable housing). Of note, the vast majority of the addiction treatment services offered in Vancouver do not have a formal linkage with permanent housing placement 
. It may also reflect the reality of the exceedingly low rental vacancy rate in Vancouver 
, as it is currently very challenging for persons with addictions to access independent apartments and houses, especially if they are actively using drugs as observed in our cohort and in other studies 
Over one-third of our study participants reported engagement in methadone maintenance therapy at baseline (36.1%, N
358), which targets opiate addiction, while exposure to other modalities that are not drug specific such as recovery houses, residential treatment, addiction counseling and self-help groups was low in this cohort. In fact, ongoing daily crack use negatively predicted attaining stable housing. This may reflect the many barriers that active drug users face in attempting to obtain or maintain stable housing and the adverse effects of ongoing drug use on efforts to do so. North et al.
followed 400 homeless persons over two years and also found that ongoing cocaine use was negatively associated with the attainment of stable housing 
. Unlike heroin addiction, there are no efficacious pharmacologic therapies for cocaine addiction and the accepted approach is cognitive behavioral therapy 
. Given the prevalence of daily crack and injection drug use in our cohort, the exposure to addiction counselling (4.4%) is very low and it appears that most users are not accessing meaningful cocaine addiction treatment 
. Marsden et al recently reported the effectiveness of community treatments (pharmacologic and psychosocial) for heroin and crack cocaine addiction in England and found that at six months, there was 37% complete abstinence from heroin and 52% complete abstinence from crack cocaine. They also noted that pharmacological treatment was less effective among users of both heroin and crack cocaine, who comprised 51% of our cohort 
The linear treatment intervention tested in the Birmingham model was comprised of abstinent-contingent housing for 6 months with behavioral treatment and employment training for 6–8 hours per day for homeless, cocaine-dependent treatment-seekers. 
In the third trial, clients who were assigned abstinent-contingent housing had a higher proportion in stable housing at 6 months (42% of 45) compared to clients who had housing not contingent on abstinence (33% of 54) and participants in treatment who had to find their own accommodation (26% of 39) 
. This comparison did not reach statistical significance, likely due to the small sample size. In contrast, we found a negative association of addiction treatment and stable housing; unlike these trials we did not have a linked housing intervention. In Birmingham and in other jurisdictions, there is little ability within the housing stock to accommodate persons who are unable to achieve abstinence and although the housing status does improve for many, a substantial proportion are unable to access stable housing, highlighting the need for the integration of addiction treatment services and supportive housing to target persons who are unable to achieve abstinence 
In Kertesz' review of the strengths and weaknesses of linear and Housing First approaches, he notes that they target different primary problems, namely housing retention vs. addiction and the achievement of abstinence 
. Studies of these two approaches have recruited different sub-populations of the chronically homeless. Most Housing First trials included persons with severe and persistent mental illness 
, with the exception of the Chicago trial of the homeless with chronic medical conditions accessing the emergency department 
and the Seattle study of severe alcoholics 
. The economic benefits found in these latter studies were related to the inclusion of chronically homeless persons who were high users of health and other public services and this may not be generalizable to all homeless persons 
. The linear approach trials included homeless cocaine-dependent persons seeking addiction treatment with the intervention goal being abstinence 
. The transition to market housing and long-term housing retention can be challenging as not all clients were able to remain abstinent 
Interestingly, we observed that a high proportion of our study participants who did not have prior addiction treatment exposure achieved stable housing over follow-up, suggesting that this group was more capable of accessing housing services that may have helped achieve this outcome and are quite separate from addiction treatment. This highlights the need for supportive housing with integrated addiction treatment services for the chronically homeless that are seeking treatment, given that housing stability is an important functional outcome.
Our study had several limitations. This was an observational study and the addiction treatment reflected what cohort participants accessed during the study period and likely represents usual care for persons who are active illicit drug users in our setting. As a result, our definition of addiction treatment was broad and may not have been stringent enough to provide sufficient exposure and duration to impact drug use and thereby improve the participant's housing status. However, our study did consider addiction treatment exposure during a six-month period, and other studies have reported positive outcomes following the provision of addiction treatment over six months 
. The negative association may also reflect the selection of the most heavy drug users who accessed addiction treatment and are less likely to be housed because of the intensity of their addiction. Like all observational studies, residual confounding may be present in this instance. However, it should be noted that our analyses included adjustment for a range of potential confounders, including intensity of drug and alcohol use. Finally, the study population was a non-random sample and our findings may have limited generalizability to other injection drug user populations.
In summary, we found that injection drug users who accessed addiction treatment services were less likely to attain stable housing compared to those who did not. In our study, addiction treatment exposure may have been a marker of life instability. Frequent (i.e., daily) drug use was prevalent among our study participants, particularly crack cocaine and injection heroin, and this was negatively associated with attaining stable housing. The exposure to addiction treatment services may not have been potent enough to reduce drug use sufficiently for participants to be able to access stable housing. Future studies should evaluate the formal linkage of addiction treatment and supportive housing services as a strategy to improve the health and housing status of this vulnerable population, as addiction treatment in our study did not positively impact the attainment of stable housing.