Background:  An aggressive “test and treat” approach to HIV prevention requires high coverage of yearly home-based door-to-door HIV testing campaigns to significantly reduce HIV incidence. The uptake of such campaigns in Sub-Saharan countries has only been documented in single campaigns; it is unknown whether individuals will repeatedly participate (or refuse to participate) in consecutive campaigns. Mathematical models used to assess the potential impact of a test and treat approach assume either perfect uptake of HIV testing and counseling (HTC) every year or independence between individuals’ decisions to participate in consecutive HIV testing campaigns.

Methods:  In Likoma (Malawi), 2 consecutive home-based HTC campaigns were conducted 18 months apart between February 2006 and September 2007. The uptake of HTC was assessed during surveys conducted before the first campaign and after the second campaign. The use of facility-based HTC services between home-based campaigns was assessed, as well as reasons for never testing.

Results:  The uptake of HTC during home-based HTC campaigns increased over time, from 70% (597/852) in the first campaign to slightly more than 80% (1717/2035) in the second campaign. Eight hundred and one island inhabitants targeted during the first campaigns were re-visited in 2007. Among those, we found that a large proportion repeatedly accepted home-based HTC (453/801, 56.5%). But we also found that a larger than expected proportion of the target population never accepted to participate in HTC provided during these campaigns (97/801, 12.1%). The proportion of “never testers” was higher among men and increased with age. Close to 70% (62/94) of those who never tested in any of the home-based campaigns had not previously tested at a health facility or during an outreach event. Never testers were more likely to have 1 current or past sexual partner who was infected with HIV at the time of the last survey than the rest of the population, and thus were at a high risk of undiagnosed HIV infection.

Conclusions:  These findings imply that pockets of undetected HIV infection (and thus possibly high viral loads) may persist even if the uptake of HTC during door-to-door campaigns is high, and the cost-effectiveness of a test and treat approach may have been over-estimated. Future studies should focus on strategies to quickly identify and target never testers during future periodical door-to-door HTC campaigns. These may, for example, include the provision of incentives to undergo HTC. Alternative approaches to identifying undiagnosed HIV cases (e.g., contact tracing) could also usefully be explored.