Eligibility of articles was determined independently by two investigators (KS and OV). Using a standardised data-extraction form (KS and BH) independently extracted data on study characteristics and outcomes, with input from OV. Any disagreements were resolved by consensus. Potential citations (published articles and conference papers after removal of duplicates) which were identified from the search strategy were reviewed for suitability. Citations which were on conference abstracts or were unrelated to community approaches of HIV testing in sub-Saharan Africa—for instance studies from other countries, laboratory studies or articles which were not reporting primary research—were excluded. Titles and abstracts were then examined and excluded if they did not report HIV testing, linkage to care or were on facility based HIV testing. Full text articles were then reviewed for full inclusion and exclusion criteria as detailed above.
Markers of study quality were examined [9 (link)] and strengths and limitations of the studies are presented along with propensity for bias. The latter was examined using a modified Cochrane Collaboration approach for assessing risk of bias.[10 ] We focused on three main domains in relation to our study objectives—selection bias (eg whether those who were already in care were excluded from linkage to care outcome calculations), outcome ascertainment (eg whether objective measures such as clinic records were sought to determine LTC) and attrition (with a cut-off of ≥20% loss-to-follow up as high attrition)—before summarising if the risk of bias in a study was low, medium or high overall. Studies were not excluded for quality reasons using formal criteria for reporting scientific data, not least because a large proportion of the available data came from operational delivery of HTC services and authors presented data as were available from the programmes.
Ethical approval was not required as only published literature was included for review.
Markers of study quality were examined [9 (link)] and strengths and limitations of the studies are presented along with propensity for bias. The latter was examined using a modified Cochrane Collaboration approach for assessing risk of bias.[10 ] We focused on three main domains in relation to our study objectives—selection bias (eg whether those who were already in care were excluded from linkage to care outcome calculations), outcome ascertainment (eg whether objective measures such as clinic records were sought to determine LTC) and attrition (with a cut-off of ≥20% loss-to-follow up as high attrition)—before summarising if the risk of bias in a study was low, medium or high overall. Studies were not excluded for quality reasons using formal criteria for reporting scientific data, not least because a large proportion of the available data came from operational delivery of HTC services and authors presented data as were available from the programmes.
Ethical approval was not required as only published literature was included for review.
Free full text: Click here
