Table 3

Linear probability model estimates* of the effect of RDT and ACT subsidies on malaria testing behaviour, defined as self-report of taking any malaria test†, for N=444 study participants

UnadjustedAdjusted (parsimonious)Adjusted (full‡)
Prespecified comparisons of interest (% differences)
 RDT-only subsidy (Group B vs Group D)21.4% (8.5% to 34.2%)18.8% (5.9% to 31.6%)18.6% (5.9% to 31.3%)
 ACT-only subsidy (Group C vs Group D)−2.9% (−16.2% to 10.5%)−3.5% (−16.9% to 9.9%)−2.8% (−16.5% to 11.0%)
 ACT subsidy when RDT is subsidised (Group A vs Group B)§0.35% (−11.3% to 12.0%)1.5% (−10.1% to 13.2%)2.7% (−8.6% to 14.1%)
Covariate effects (% differences)
 Wealth: poorest 40th centile−13.0% (−22.3% to −3.7%)−14.5% (−24.1% to −4.9%)
 Reference level (%) (Group D—no subsidies)52.4% (42.8% to 62.1%)58.2% (47.8% to 68.7%)62.0% (32.8% to 91.3%)
 Sample size435427427
  • *LPM models specified as a generalised linear model for a binomial outcome with an identity link function. All models include two factors (RDT subsidy and ACT subsidy) and their interaction to match the randomised 2×2 factorial study design in order to estimate the prespecified effects of interest.

  • †Self-report of any malaria test (RDT or slide) at CHW or health facility.

  • ‡Fully adjusted model includes age (of patient), gender (of patient), education level (of patient or guardian if patient <18 years), occupation (of patient or guardian), household size, wealth, sublocation of residence, and an interaction term between ACT and RDT subsidies. Only wealth was significant and is the only covariate retained in the parsimonious model. Full model results can be found in the online supplementary table 1.

  • §Defined as the coefficient of the main effect of the ACT subsidy plus the coefficient of the ACT×RDT interaction term.

  • ACT, artemisinin combination therapy; RDT, rapid diagnostic test.