The overall prevalence of HIV includes the number of diagnosed as well as undiagnosed people infected with HIV. In the UK, estimates of undiagnosed infections rely on data from unlinked anonymous (UA) surveys conducted in three selected adult populations: pregnant women, people who inject drugs and sexual health clinic attendees. These data, in addition to other surveillance and sexual behaviour data, are fed into a Multi-Parameter Evidence Synthesis (MPES) model which provides national prevalence estimates.
National estimates of the number of PLWHA in the UK are obtained from a complex statistical model, fitted to a collection of census, surveillance and survey-type prevalence data. The model is based on a subdivision of the target UK population aged 15+ years into key mutually exclusive exposure subgroups, and produces estimates by geographical region, infection diagnosis status and gender: these are organised in linked Tables 1-2.
Details of the model structure and its underlying assumptions can be found in Goubar et al. (2008) and Presanis et al. (2008); a more extensive analysis of trends in HIV infections and prevalence can be found in Presanis (2010). Data sources used to inform the model include the Unlinked Anonymous (UA) sero-prevalence surveys, the Survey of Prevalent HIV Infections Diagnosed (SOPHID), the Genito-Urinary Medicine (GUM) Clinic Activity Dataset (GUMCAD), HIV and AIDS New Diagnoses and Deaths surveillance and the Gay Men's Sexual Health Survey.
The estimated ratio of diagnosed to undiagnosed HIV infection among 15 to 44 years old is applied to the older1 population to obtain an estimate of the number of PLWHA aged 15+ years in the United Kingdom. This estimate is presented in our latest HIV annual report (HIV & STI Department, 2012).
The UA monitoring surveys provide data on HIV prevalence in selected adult populations (i.e. pregnant women and people who inject drugs). The primary objective of the UA programme is to monitor trends in the prevalence of both diagnosed and undiagnosed HIV infection in selected sub-populations. Since the surveys began in 1990, approaching ten million samples have been irreversibly unlinked and anonymised from Patient Identifying Information and tested for HIV infection. These surveys include not only individuals who have already been diagnosed with HIV infection, but also people who live unaware of their infection and would therefore not be captured by other surveillance systems.
The UA survey of HIV prevalence among GUM clinic patients, which informed HIV prevalence and PLWHA estimates in previous years, was not carried out in 2010; a redesigned version is planned for implementation in 2011.
The GUMCAD registry records patient-level information on all STI episodes occurring in any GUM clinic across England. Given the unavailability of the corresponding UA survey, information on HIV diagnoses among newly admitted patients in 2010 was extracted from GUMCAD to derive, together with appropriate modelling assumptions used in separate work (Conti et al., 2011), estimates of undiagnosed HIV prevalence in the corresponding sub-population.