Supervisors info:
Βάνα Σύψα, Επίκουρη Καθηγήτρια, Ιατρική, ΕΚΠΑ
Άγγελος Χατζάκης, Καθηγητής, Ιατρική, ΕΚΠΑ
Δημήτριος Παρασκευής, Επίκουρος Καθηγητής, Ιατρική, ΕΚΠΑ
Summary:
Title: Methods to estimate the size of hard-to-reach populations with an application to the population of people who inject drugs in Athens.
Background: Estimates of the population size of people who inject drugs (PWID) are important for designing public health programs, assessing the coverage of harm reduction services and the burden of infectious diseases. Estimation is difficult, as PWID are a hidden/ hard-to-reach population. In Greece, extant estimates are based on capture-recapture applied yearly to three drug treatment sources (estimation from the Greek Documentation and Monitoring Centre for Drugs). The aims of this thesis are: 1) to review methods for estimating the size of hard-to-reach populations, and 2) to apply these methods to estimate the size of PWID in Athens using data from a community-based program implemented during an HIV outbreak in this population in 2012-2013.
Methods: The method used primarily for estimating the size of hard-to-reach populations is capture-recapture. Depending on the availability of data, other methods can be used such as the multiplier method. Data from the ARISTOTLE program were used in the thesis. The ARISTOTLE program was conducted in 5 consecutive Respondent-Driven Sampling (RDS) rounds. Participants had history of injecting drug use in the past 12 months and during the program they were tested for HIV and were interviewed with a structured questionnaire. Two methods for estimating the size of the target population were used: 1) Capture- recapture based on the five RDS rounds of ARISTOTLE program, 2) Multiplier method based on the number of newly diagnosed HIV cases and the estimated HIV prevalence. In the capture-recapture method, a log-linear model was used in order to estimate the target population and to control for dependency among sources. Both the Akaike and the Bayesian information criterion were used for model selection. The resulting estimates were compared with the existing estimates from treatment sources. For the multiplier method, the number of newly diagnosed cases of HIV among PWID reported to the national surveillance system during 2009 to 2013 and the prevalence of HIV infection as estimated by ARISTOTLE program were used (HIV prevalence was negligible before 2009).
Results: The estimated number of PWID who reported heroin as the main substance of use during the last 12 months was 4,731 for 2013 (95% CI: 4,175-5,287). The number of persons with injecting drug use in the past month was 4,367 (95% CI: 3,893-4,840) and ranged from 4,146 to 4,325, depending on the covariates included in a model. The total size of the population as estimated by the multiplier method for the period 2009-2013 was 6,964 (95% CI: 6,455 - 7,559). The number of persons with injecting drug use in the past 30 days was two-fold higher compared to the existing estimate from drug treatment data (4,367 versus 1,956, respectively). The coverage of needle and syringe programs in Athens in 2013 is estimated to be 97 syringes/PWID/year when the revised estimate is used as denominator versus 216 syringes/PWID/year based on the existing population size estimate.
Conclusions: The application of the capture-recapture method to ARISTOTLE data resulted to a higher estimate of the population size of PWID who injected drugs in the past months in Athens, compared to the existing estimates from drug treatment programs. Consequently, the coverage of harm-reduction programs in this population (syringe distribution and opioid substitution programs) is expected to be lower and needs to reduce the burden of infectious diseases are increased.
