We conducted systematic searches for cohort studies reporting
remission from dependence upon amphetamines, cannabis, cocaine,
or opioids. Stages followed, separately for each drug, were consistent
with the methodology recommended by the Meta-analysis of Observational
Studies in Epidemiology (MOOSE) group (Stroup et al.,
2000). The first stage involved a search of the peer-reviewed
literature. In consultation with a qualified archivist, three electronic
databases were chosen: Medline, EMBASE and PsycInfo. Broad search
strings, tailored to each database to have the best coverage of the
literature, were used: remission, cohort and specific drug type
(amphetamines, cannabis, cocaine and opioid). Opioid searches
were limited further by dependence so as to identify a manageable
number of articles (without the dependence limitation over five times
the number of articles would have been included). This limitation
excluded studies that focused on “opioid use” which would have
included more than opioid dependence and so may have overestimated
remission rates. Searches were limited to the publication
timeframe of January 1990 to March 2009 and to human subjects.
Additional articles were identified from: reference lists of review
articles; relevant articles that were identified in other searches
(prevalence, incidence and mortality searches) for the broader Global
Burden of Disease (GBD) study; and expert consultation. Email requests
for remission data were also sent to investigators conducting prospective
studies of persons who had met criteria for drug dependence.
Studies were excluded if they did not focus on the drugs of interest,
did not report remission data, did not include primary data (review
articles), comprised case studies, reported duplicate data, or comprised
treatment trials. All identified treatment samples were located
in high income countries and so were included because it is likely that
people who are drug dependent in high income countries will receive
treatment, especially so if they are dependent on opioids. Studies
were excluded if they had less than three years follow-up, since
shorter follow-up studies may overestimate remission by including cases with a temporary lull in the course of their disorder. Table 1 shows
the number of articles culled in this process.