Batter V
Scientific Institute of Public Health-Louis Pasteur, Brussels, Belgium.
AIDS. 2000 May 5;14(7):871-80. doi: 10.1097/00002030-200005050-00014.
To monitor trends in HIV seroprevalence among sentinel populations of patients with new sexually transmitted disease episodes in 17 networks of 15 European countries.
Time trends were estimated by logistic regression for homo-/bisexual men, injecting drug users (IDU) and non-IDU heterosexuals. The networks were included as independent variables to account for different prevalence levels across Europe. Socio-demographic and behavioural data were also included in the model. The interactions of time and each factor in the model were evaluated to determine whether trends differed across networks or population subgroups.
Overall, more than 150,000 sexually transmitted disease episodes were registered in this study. In the European network as a whole, the prevalence decreased significantly among IDU [n = 2619; odds ratio (OR) for annual change, 0.86; 95% confidence interval (CI), 0.80-0.93]. It also decreased among homo-/bisexual men (n = 11,809; OR, 0.92; 95% CI, 0.89-0.94). In both groups, trends did not differ statistically between networks. Among non-IDU heterosexuals (n = 114,024) the prevalence increased significantly (OR, 1.08; 95% CI, 1.04-1.13), especially among women (OR, 1.13) and the time trends differed statistically across networks. A significant increase was observed in four networks, whereas no specific change was detected in the others.
By applying a standardized protocol, trends in HIV prevalence could be compared across networks and estimated at a more global level. For the validity of HIV trends in such surveys, it is essential to minimize the number of patients not being tested in networks where voluntary testing is required.
