Predictors of human immunodeficiency virus (HIV) infection in primary care: a systematic review protocol

Benhildah N. Rumbwere Dube, Tom P. Marshall, Ronan P. Ryan

Antiretroviral therapies for human immunodeficiency virus are more effective if infected individuals are diagnosed early, before they have irreversible immunologic damage. A large proportion of patients that are diagnosed with HIV, in United Kingdom, would have seen a general practitioner (GP) within the previous year. Determining the demographic and clinical characteristics of HIV-infected patients prior to diagnosis of HIV may be useful in identifying patients likely to be HIV positive in primary care. This could help inform a strategy of early HIV testing in primary care. This systematic review aims to identify characteristics of HIV-infected adults prior to diagnosis that could be used in a prediction model for early detection of HIV in primary care. The systematic review will search for literature, mainly observational (cohort and case-control) studies, with human participants aged 18 years and over. The exposures are demographic, socio-economic or clinical risk factors or characteristics associated with HIV infection. The comparison group will be patients with no risk factors or no comparison group. The outcome is laboratory-confirmed HIV/AIDS infection. Evidence will be identified from electronic searches of online databases of EMBASE, MEDLINE, The Cochrane Library and grey literature search engines of Open Grey, Web of Science Conference Proceedings Citation Index and examination of reference lists from selected studies (reference searching). Two reviewers will be involved in quality assessment and data extraction of the review. A data extraction form will be developed to collate data from selected studies. A checklist for quality assessment will be adapted from the Scottish Intercollegiate Guidelines Network (SIGN). This systematic review will identify and consolidate existing scientific evidence on characteristics of HIV infected individuals that could be used to inform decision-making in prognostic model development. PROSPERO CRD42016042427.