The ALPHA Network is a collaboration between 10 longitudinal population-based studies in Kenya, Malawi, South Africa, Tanzania, Uganda and Zimbabwe. These studies collect data on HIV infection alongside demographic, behavioural, socio-economic and clinical data from residents of the study areas. The Network harmonises these data and conducts comparable and pooled analyses on the epidemiology and demographic impact of HIV. The results are used to improve understanding of the HIV epidemic in these communities and the wider region, and to inform policy. The Network also aims to strengthen the analytical capacity of its member organisations and to foster collaboration between them. ALPHAs Scientific Advisory Committee sets the research agenda and decides on funding applications.
ALPHA HIV Incidence and Mortality (ALPHA HIM) data collection
The datasets available here are derived from ALPHAs standardised data exchange protocols called the ALPHA data specifications. Only specifications on Residency and HIV tests were used to create the HIV incidence and mortality data. The said specifications have been transformed into two analysis-ready datasets, one for HIV incidence and one for all-cause mortality by HIV status. There are two major steps involved in generating the shared ALPHA HIM data:
1. Preparation of the ALPHA Specifications from the source data for each study. This process is described in the accompanying documentation for the shared ALPHA study data document for each of the 10 member studies.
2. Preparation of the shared incidence and mortality data from the ALPHA specifications. This is described in annotated Stata do-files which are available on request.
The data used for ALPHAs own analyses are analysis-ready datasets with a similar structure to the ALPHA shared data but with more detail, such as dates of birth and other events, and a larger set of variables than the shared data. As such, they contain potentially sensitive identifying information and are therefore unsuitable for sharing on this repository. The analysis-ready datasets provided here do not contain sensitive information and have been simplified compared to the originals, principally by removing dates of events. There will be minor differences between estimates made using these shareable data and those published estimates in ALPHA papers on the same topic. Our assessment is that these differences are trivial, but users are advised to contact one of the ALPHA team if they find larger discrepancies or struggle to replicate the published findings. Data users should also contact this e-mail address if they have questions relating to accessing the datasets.