RESUMEN
OBJECTIVE: To predict trends in diagnosed HIV prevalence by extrapolation to 2004 using data from the annual surveys of individuals receiving HIV-related care in England, Wales and Northern Ireland from 1996 to 2001. METHODS: Data from the annual surveys of prevalent HIV infections diagnosed (SOPHID) were adjusted for under-reporting and non-attendance and separately extrapolated for infections acquired homosexually, heterosexually and by other routes. The data were extrapolated using negative binomial and linear regression models based on the 1996 to 2001 annual surveys. RESULTS: The negative binomial model predicted an increase of 56% in diagnosed HIV prevalence in England, Wales and Northern Ireland between 2001 and 2004. The linear model predicted an increase of 25% for the same time period. The predicted increases are mostly driven by the large rise in the number of new diagnoses, in particular in individuals infected heterosexually. CONCLUSION: Increases in HIV prevalence in England, Wales and Northern Ireland have diverged from a linear trend. Negative binomial modelling of the data predicts that large rises in prevalence will continue during the early 2000s.