Estimating intervention effectiveness in trials of malaria interventions with contamination

Abstract

Background: In cluster randomized trials (CRTs) or stepped wedge cluster randomized trials (SWCRTs) of malaria interventions, mosquito movement leads to contamination between trial arms unless buffer zones separate the clusters. Contamination can be accounted for in the analysis, yielding an estimate of the contamination range, the distance over which contamination measurably biases the effectiveness.

Methods: A previously described analysis for CRTs is extended to SWCRTs and estimates of effectiveness are provided as a function of intervention coverage. The methods are applied to two SWCRTs of malaria interventions, the SolarMal trial on the impact of mass trapping of mosquitoes with odor-baited traps and the AvecNet trial on the effect of adding pyriproxyfen to long-lasting insecticidal nets.

Results: For the SolarMal trial, the contamination range was estimated to be 146 m ([Formula: see text] credible interval [Formula: see text] km), together with a [Formula: see text] ([Formula: see text] credible interval [Formula: see text]) reduction of Plasmodium infection, compared to the [Formula: see text] reduction estimated without accounting for contamination. The estimated effectiveness had an approximately linear relationship with coverage. For the AvecNet trial, estimated contamination effects were minimal, with insufficient data from the cluster boundary regions to estimate the effectiveness as a function of coverage.

Conclusions: The contamination range in these trials of malaria interventions is much less than the distances Anopheles mosquitoes can fly. An appropriate analysis makes buffer zones unnecessary, enabling the design of more cost-efficient trials. Estimation of the contamination range requires information from the cluster boundary regions and trials should be designed to collect this.

Keywords: Contamination; Contamination range; Effective coverage; Malaria; Sigmoid random effects model; Stepped wedge cluster randomized trial.

Fig. 1

Schematic description of the effect of mosquito movement and the arising contamination in the boundary region of a CRT or SWCRT. On the front face of the rhomboid, the smooth decrease in prevalence between a control and intervention cluster (on the surface) based on the distance of a household to its nearest discordant household is visualized

Fig. 2

Schematic description of the effective coverage. For two households, one in the intervention arm (upper right corner) and one in the control arm (lower left corner), the area from which the effective coverage is calculated is shaded in grey. The closer a household is to one of these two households, the bigger its impact on the effective coverage (darker shade of grey). For the household in the control arm, the effective coverage is close to zero, since no intervention households are close. For the household in the intervention arm, the effective coverage is more than $50\%$

Fig. 3

The effectiveness of Solar-powered Mosquito Trapping Systems (SMoTS) (on the y-axis) in terms of the effective coverage is visualized in black, with credible intervals in grey. The effectiveness was estimated with the model including a random effect for the household effects (Sigmoid RE + hh)

Fig. 4

The effect of adding pyriproxyfen (PPF) to long-lasting insecticidal nets (LLINs) (on the y-axis) in terms of the effective coverage is visualized in black, with credible intervals in grey

Fig. 5

Illustration of the sigmoid relationship between the nearest discordant household and the effective coverage for the SolarMal trial in grey. In black, the sigmoid curve for ${\widehat{\mathcal{R}}}_{\mathit{ijk}}$ that approximates the effective coverage is illustrated. To increase the readability, only households within 2 km of the nearest discordant household are plotted on the x-axis