Estimating the prevalence, incidence and cessation of injecting drug use in Glasgow 1960-2000: combining expert opinion with capture-recapture prevalence data

Estimation of and trends in the size of injecting drug user (IDU) populations is fundamental to the planning and development of health-care services for this group. In Glasgow, log-linear modelling of capture-recapture data estimated 8490 (95% CI 7490-9720) and 7190 (6090-8620) current IDUs in 1990 and 2000, respectively. Hitherto, however, no consensus had been sought on the size and shape of its IDU epidemic curve. A modified Delphi approach was used to elicit one for Glasgow. Twelve experts were asked to provide their opinion on the prevalence, incidence and cessation of injecting for quinquennia during 1960-2000. Instead of the usual iterative process to refine experts' consensus, the elicitation of IDU incidence and cessation provided an opportunity, not previously explored, to combine these data and examine coherence with capture-recapture IDU prevalence estimates. Scrutiny of consensus data indicated that experts had underestimated incidence during 1985-1990, and overestimated prevalence during 1995-2000 and cessation during 1985-2000. Coherent median estimates indicated that prevalence (149-557), incidence (28-49) and cessation (1-2%) remained low and stable during 1960-1975, rose steeply between 1975 and 1990 (prevalence from 557 to 8238; incidence from 49 to 1375; cessation from 2 to 10%), and by 2000 there had been a decline in prevalence (6809) and incidence (1195) but a further rise in cessation (15%). Prevalence estimates from capture-recapture studies were essential to anchor experts' consensus and derive coherent data. Further efforts are required to collect data which allow accurate estimation of the incidence and cessation of IDU.