Taking Diversity Into Account While Constraining Randomisation of Zambian and South African Communities for a Clinical Controlled Trial
Community-based controlled trials increasingly shape public health policy. How the diversity of a community influences the outcomes of community-based trials is poorly understood and methodologies that allow the social context of disease to be fed into a constrained randomisation process of communities are poorly developed.
This study qualitatively grades 24 African communities experiencing high levels of tuberculosis along a diversity continuum in order to contribute the social context of tuberculosis to the randomisation process of a community-based control trial.
Four to six days of rapid fieldwork - labeled “broad brush surveys” (BBS) - combining participatory tools and observational methods (transect walks, informal conversations, daily time charts, structured observations, snapshot surveys, free-listing with community leaders or local health committees) aimed to capture the topography, population profile, local economy, mobility and general perceptions of tuberculosis in 16 communities in Zambia and eight communities in the Western Cape Province of South Africa. Two trained research assistants and two locally recruited fieldworkers conducted fieldwork.
Drawing on an open-closed typology model of urban systems developed by Wallman, six indicators of diversity comprising livelihood, topography, interest groups, interaction, mobility and housing were compared and enumerated across sites. Sites were then graded along an “open-closed” continuum of diversity; open being more heterogeneous and closed more homogenous.
All 8 South African sites and 10 Zambian sites were classified as relatively open; six Zambian sites as relatively closed. Bio-statisticians used this data alongside HIV estimates, Tuberculin Skin Test surveys and rural/urban location to constrain randomization. The allocation of interventions consequently meant only Zambia having open/closed comparisons in the same intervention.
Combining a model of diversity and rapid appraisal techniques presented an unusual opportunity to feed the findings of qualitative data from 24 African sites into a community randomization process.
Keywords: Diversity, Community, Constrained Randomisation, Public Health
Emma Jane Murray
Research Assistant, Desmond Tutu TB Centre