Rationale and cost-efficiency compared for urine or saliva testing and behavioural inquiry among UK offender populations: injectors, arrestees and prisoners.

Abstract

Background: Costs, methodology and efficiency at obtaining behavioural information and biological sample to be tested for blood-borne viruses or illegal drugs are compared for UK surveys of offender populations: injectors in the community or attending drug agencies, arrestees and prisoners.

Methods: Arrestee surveys use extensive behavioural interview + urine sample and measure a key performance indicator in UK's Drugs Strategy. They have low efficiency (urine sample for drugs testing available from under 60% of eligible arrestees) at high cost (pound sterling 110-190 or 350 per urine sample and at least pound sterling 500-800 per injector). Random mandatory drugs testing of prisoners has the highest efficiency (because refusals are punishable), but the cost is high (pound sterling 110-120 per urine sample and pound sterling 300-500 per injector) and behavioural data are lacking. Prisoner surveys use self-completion questionnaire + saliva sample. They guarantee demonstrable anonymity in estimating the prevalence of blood-borne viruses and prisoners' associated risk behaviours, have high efficiency (saliva sample from over 80% of inmates) at low cost (pound sterling 30 per saliva and pound sterling 70-110 per injector), but behavioural data are limited to risk factors for blood-borne viruses. Low cost also characterises comprehensive interview + saliva sample from injectors in the community (pound sterling 90 per saliva sample, all from injectors) but efficiency cannot be assessed because the sampling frame of eligible injectors is not known. Voluntary unlinked anonymous surveys of injectors at drug agencies use self-completion questionnaire + saliva sample to be tested for blood-borne viruses. They are the least costly at pound sterling 43 per injector and moderately efficient with two-thirds volunteer rate by eligible injectors.

Discussion: For scientific added-value, we recommend co-location of survey types geographically and temporally; a common core set of behavioural questions; saliva sample as well as (and, eventually, instead of) the less acceptable urine sample. Survey methodologies for measuring key performance indicators should stand up to scrutiny in terms of openness, design credentials, statistical power, and costs. We examine how participation bias or inadequate survey size can compromise the effective monitoring of a key performance indicator.