Premium
Estimating the size and dynamics of an injecting drug user population and implications for health service coverage: comparison of indirect prevalence estimation methods
Author(s) -
Kimber Jo,
Hickman Matthew,
Degenhardt Louisa,
Coulson Tim,
Van Beek Ingrid
Publication year - 2008
Publication title -
addiction
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.424
H-Index - 193
eISSN - 1360-0443
pISSN - 0965-2140
DOI - 10.1111/j.1360-0443.2008.02276.x
Subject(s) - medicine , poisson regression , population , estimation , demography , poisson distribution , statistics , environmental health , mathematics , management , sociology , economics
Aims (i) To compare indirect estimation methods to obtain mean injecting drug use (IDU) prevalence for a confined geographic location; and (ii) to use these estimates to calculate IDU and injection coverage of a medically supervised injecting facility. Design Multiple indirect prevalence estimation methods. Setting Kings Cross, Sydney, Australia. Participants IDUs residing in Kings Cross area postcodes recorded in surveillance data of the Sydney Medically Supervised Injecting Centre (MSIC) between November 2001 and October 2002. Measurements Two closed and one open capture–recapture (CRC) models (Poisson regression, truncated Poisson and Jolly–Seber, respectively) were fitted to the observed data. Multiplier estimates were derived from opioid overdose mortality data and a cross‐sectional survey of needle and syringe programme attendees. MSIC client injection frequency and the number of needles and syringes distributed in the study area were used to estimate injection prevalence and injection coverage. Findings From three convergent estimates, the mean estimated size of the IDU population aged 15–54 years was 1103 (range 877–1288), yielding a population prevalence of 3.6% (2.9–4.3%). Mean IDU coverage was 70.7% (range 59.1–86.7%) and the mean adjusted injection coverage was 8.8% (range 7.3–10.8%). Approximately 11.3% of the total IDU population were estimated to be new entrants to the population per month. Conclusions Credible local area IDU prevalence estimates using MSIC surveillance data were obtained. MSIC appears to achieve high coverage of the local IDU population, although only an estimated one in 10 injections occurs at MSIC. Future prevalence estimation efforts should incorporate open models to capture the dynamic nature of IDU populations.