Indicator 15.1: "HIV prevalence among young people aged 15-24"
Percentage of young people aged 15-24 who are HIV infected.
Population-based surveys (DHS/AIS), behavioral surveillance survey (BSS). Originally data were supposed to be derived from WHO guidelines for HIV sentinel surveillance (Sentinel Surveillance, Sero-survey with biomarkers).
What It Measures
Community-based surveys are potentially the best source of data on HIV prevalence among young people in the general population. However, they may not provide good estimates for subgroups of the young population, e.g. IDUs, whose behaviour would place them in categories at high risk of HIV infection, because such surveys are unlikely to find sufficient people in these categories to provide representative samples. The efficiency of such surveys depends on the prevalence of HIV in the general population. In low or concentrated epidemics the numbers infected are not sufficient to give valid results. Even in generalized epidemics, with prevalence rates below 5%, implementers should carefully consider the value of conducting population-level surveys.It is less sustainable to collect data for this indicator than it is to obtain ANC surveillance data. Such surveys are costly and complex and should only be considered in situations where the quality of the surveys can be assured. In order to provide robust estimates of prevalence trends they must be repeated at regular intervals in a comparable manner. If such surveys can only be conducted at infrequent intervals the findings can be compared with the results of ANC surveillance.
How to Measure It
This indicator should be reported as percentages for males and females and the age groups 15-19, 20-24 and 15-24 years. The unweighted sample sizes and non-response rates (separately for absenteeism and refusal) should be given for each category. The HIV testing protocol should also be given.
Strengths and Limitations
The findings of a general population survey can be taken at face value if the survey is truly representative of the population in which it was carried out. General population surveys approach participants, while most other methods of data collection rely on participants presenting themselves at the place where HIV testing is being conducted. This means that selection and participation bias should be less significant in these surveys. If the sampling frame is inaccurate, however, or if the survey is badly implemented, there may still be some selection bias. Participation bias is potentially a greater problem. The extent of participation bias is influenced by the topic of the survey and the protocol under which testing is performed. The factors of particular concern are those that may relate to the HIV status of the potential respondent, e.g. high-risk sexual behaviour. If the individuals who are absent from the survey or those who choose not to participate differ markedly in such characteristics from the persons who participate the accuracy of the prevalence estimates may be affected. In countries with relatively low levels of adult HIV prevalence (between 1 and 3%), general population-based surveys are likely to underestimate prevalence levels. Indeed, people at higher risk of HIV are more likely to missed by general population-based surveys, either because they are typically excluded from the sample (e.g. military or police living in barracks, sex workers working in brothels), or because they live outside of households for reasons related to their risk behaviour (sex workers, IDU), or because of their mobility (e.g. truckers, fishermen, other mobile groups). In addition, where prevalence is low (e.g. in the 0-3% range) it will be difficult to observe significant changes in prevalence over time, unless a unusually large sample size is used,If basic information is collected from the people who do not take part in the survey or part of it, participation bias can be detected and adjusted for at the analysis stage. Response rates for the survey should always be reviewed and presented.A lack of continuity is a potentially serious limitation of community-based survey data. Because surveys are expensive and time-consuming, the scope and format of successive surveys may vary. This introduces an unquantifiable error into the estimates. The collection of reliable data on HIV prevalence over time requires a series of comparable surveys to be carried out periodically in the same population.