The women made it work: fuzzy transitive closure of the results chain in a dengue prevention trial in Mexico

Background: A modified theory of planned behaviour (acronym CASCADA) proposes that Conscious knowledge precedes a change in Attitude, which in turn precedes positive deviations from negative Subjective norms, intention to Change, perception of Agency to change, Discussion of possible action, and Action itself. We used this as a results chain to investigate gender-specific behaviour dynamics in chemical-free dengue prevention. Methods: Secondary analysis of the Mexican arm of a cluster randomised controlled trial used household survey data on intermediate outcomes of dengue prevention behaviour. We used a matrix of odds ratios between outcomes, transformed to a symmetrical range (-1, 1), to compute fuzzy transitive closure of the results chain for control and intervention clusters, then for male and female respondents separately in each group. Transitive closure of a map computes the influence of each factor on each other factor, taking account of all influences in the system. Cumulative net influence was the sum of influences across the results chain. Results: Responses of 5042 women and 1143 men in 45 intervention clusters contrasted with those of 5025 women and 1179 men in 45 control clusters. Control clusters showed a distal block (negative influence) in the results chain with a cumulative net influence of 0.88; intervention clusters showed no such block and a cumulative net influence of 1.92. Female control respondents, like the overall control picture, showed a distal block, whereas female intervention responses showed no such blocks (cumulative net influence 0.78 and 1.73 respectively). Male control respondents showed weak distal blocks. Male intervention responses showed several new negative influences and a reduction of cumulative net influence (1.38 in control and 1.11 in intervention clusters). Conclusions: The overall influence of the intervention across the results chain fits with the trial findings, but is different for women and men. Among women, the intervention overcame blocks and increased the cumulative net influence of knowledge on action. Among men, the intervention did not reinforce prevention behaviour. This might be related to emphasis, during the intervention, on women's participation and empowerment. The fuzzy transitive closure of the CASCADA map usefully highlights the differences between gender-specific results chains.