Balancing rigour and acceptability: the use of HIV incidence to evaluate a community-based randomised trial in rural Uganda

Recent debate about the evaluation of community based, HIV/AIDS behavioural interventions has focused on the appropriateness of the randomised controlled trial (RCT) design, and the difficulty of obtaining reliable outcome measures. A community based HIV/AIDS behavioural change RCT, recently conducted in rural Uganda, used HIV incidence as the principal outcome measure. This paper examines the acceptability of the trial from the community perspective. It asks whether, in a rural African setting, it is possible to implement a scientifically rigorous evaluation without compromising acceptability of the trial to the community. Opinions of the trial held by community members working as trial field workers were collected by semi-structured interview (n = 37), and focus group discussions (4). Community opinions of the trial were ascertained through 10 focus groups. For both field workers and the community. the sero-survey was more salient than the intervention, and the source of many rumours and disputes. Despite intensive mobilisation and close monitoring of field workers. it was impossible to ensure the veracity of explanations about the survey at ground level, and to protect each individual from coercion. The community expected a reward in return on their blood. Although the introduction of incentives at the final survey round increased the acceptability of the trial, they not only created jealousies and tensions, but also led to expectations of greater rewards in future. We conclude that RCTs in poor, rural communities are feasible, but the challenges involved should not be underestimated. Obtaining community support for the trial, respecting established hierarchies, and close supervision of field workers are all essential, but even then, controversies should be anticipated. There is an urgent need for relevant guidelines to help researchers navigate the complex ethical issues involved. (C) 2002 Elsevier Science Ltd. All rights reserved.