Indirect prevalence estimates of a national drug using population: The use of contact-recontact methods in Wales

Abstract

The changing discourses of risk entail more than shifts in cosmologies of misfortune: they also embrace technical changes in risk surveillance. Public health medicine has sought estimates of drug-using populations to plan drug services and to interpret data on the prevalence of infectious diseases. Concealment of risk behaviours had led to technical advances in indirect methods of population estimation, one such method being that of contact-recontact, where the size of the hidden population (unknown to agencies) is estimated by modelling the dependency relationships (the overlaps) between different agenices lists of clients. This study demonstrates the use of contact-recontact methods for estimating the national prevalence of serious drug use, a procedure only previously attempted for city-wide populations of drug users. Three- and four-sample contact-recontact models have been used and log linear regression techniques have been applied to assess dependencies between datasets. Results from the modelling exercise provide an estimate for serious drug use in Wales in 1994 of 8384 people (95% C.I. 5307-11 407). This gives a rate per thousand population of 2.87 (95% C.I. 1.8-3.9) and 5.34 (95% C.I. 3.4-7.3) per thousand aged 15-55. Prevalence estimates are also provided for former Welsh counties by age and sex. Prevalence estimates for injecting drug use are provided for a smaller number of former counties in Wales. Methodological problems have been encountered (and variously addressed) in respect of varying agency definitions of clients, matching definitions and availability of datasets.