Eave tubes for malaria control in Africa: a modelling assessment of potential impact on transmission

Abstract

Background: Novel interventions for malaria control are necessary in the face of problems such as increasing insecticide resistance and residual malaria transmission. One way to assess performance prior to deployment in the field is through mathematical modelling. Modelled here are a range of potential outcomes for eave tubes, a novel mosquito control tool combining house screening and targeted use of insecticides to provide both physical protection and turn the house into a lethal mosquito killing device.

Methods: The effect of eave tubes was modelled by estimating the reduction of infectious mosquito bites relative to no intervention (a transmission metric defined as relative transmission potential, RTP). The model was used to assess how RTP varied with coverage when eave tubes were used as a stand-alone intervention, or in combination with either bed nets (LLINs) or indoor residual spraying (IRS).

Results: The model indicated the impact of eave tubes on transmission increases non-linearly as coverage increases, suggesting a community level benefit. For example, based on realistic assumptions, just 30 % coverage resulted in around 70 % reduction in overall RTP (i.e. there was a benefit for those houses without eave tubes). Increasing coverage to around 70 % reduced overall RTP by >90 %. Eave tubes exhibited some redundancy with existing interventions, such that combining interventions within properties did not give reductions in RTP equal to the sum of those provided by deploying each intervention singly. However, combining eave tubes and either LLINs or IRS could be extremely effective if the technologies were deployed in a non-overlapping way.

Conclusion: Using predictive models to assess the benefit of new technologies has great value, and is especially pertinent prior to conducting expensive, large scale, randomized controlled trials. The current modelling study indicates eave tubes have considerable potential to impact malaria transmission if deployed at scale and can be used effectively with existing tools, especially if they are combined strategically with, for example, IRS and eave tubes targeting different houses.

Keywords: Anopheles; Eave tube; Eaves; Housing; Malaria; Novel intervention; Population model; Vector control.

Figures

Fig. 1

Model structure is summarized in this flow diagram, beginning with a mosquito entering the model and searching for a property, and ending with the completion of a cycle

Fig. 2

The effect of eave tubes on transmission potential of the vector population relative to no intervention (relative transmission potential, RTP). The different lines indicate different human host categories, with the lowest line (blue, least RTP) being the population protected by eave tubes, the uppermost line (red, greatest RTP) the population without any protection, and the middle line (purple) an overall RTP of the entire human host population. All plots use the same model assumptions: no other existing interventions, no mosquitoes are assumed killed on exit from an eave tube house, and 20 % deflection (without kill) by eaves tubes. Of those not deflected, mortality is estimated based on experimental data that used deltamethrin-treated eave tubes and showed either 70 % (a) of An. gambiae, or 52 % (b) of An. arabiensis were killed by the eave tubes (Snetselaar et al. in prep). Plots for eave tube-protected humans begin at 10 % coverage, as this is an empty category with 0 % eave tube coverage

Fig. 3

Effects of mosquito deflection and mortality on the impact of eave tubes on relative transmission potential (RTP). Plots show RTP across all combinations of deflection away from eave tube protected properties (x axis) and mortality for vectors attempting to enter (y axis). The different colored contours indicate different levels of RTP. a, b assume 70 and 30 % of properties are protected with eave tubes, respectively. No other interventions (LLINs or IRS) are assumed

Fig. 4

Effects of eave tube deflection and mortality assumptions on relative transmission potential (RTP) for people in houses with and without eave tubes. Plots show overall RTP across the population (purple, middle line), together with values for people with (blue, lowest line) and without (red, uppermost line) eave tube protection. Mortality for vectors attempting to enter eave tube protected property is assumed to be 70 % as a baseline, with 70 % (a) and 30 % (b) of properties protected by eave tubes. No LLINs or IRS are assumed. The green lines indicate the no-intervention value. Note the different vertical axis scales in a and b

Fig. 5

a–c Effect of altering coverage of households with eave tubes and LLINs on relative transmission potential (RTP). The x axis represents the percent of properties with eave tubes; the y axis represents the percent of properties with LLINs. It is assumed LLINs are used by 70 % of occupants in properties with LLINs. Eave tubes are assumed to cause 20 % deflection and 70 % mortality for remaining vectors attempting to enter an eave tube protected property (as described in Table 1). The distribution of each intervention differs among the plots. The different colored contours indicate different levels of RTP. a The distribution of each intervention is assumed random and not affected by the presence of the other intervention. b LLINS are present only in percentage of eave tube protected properties. c LLINS present only in percentage of non-eave tube protected properties

Fig. 6

a–c Effect of altering coverage of households with eave tubes and IRS on relative transmission potential (RTP). The x axis represents the percent of properties with eave tubes; the y axis represents the percent of properties with IRS. IRS is assumed to deflect 50 % of vectors that attempt to enter a property, and to kill 40 % of resting vectors as in Table 1. Eave tubes are assumed to deflect 20 % of vectors and cause 70 % mortality in the remainder attempting to enter an eave tube protected property. The distribution of each intervention differs among the plots. The different coloured contours indicate different levels of RTP. a The distribution of each intervention is assumed random and not affected by the presence of the other intervention. b LLINS are present only in percentage of eave tube protected properties. c LLINS present only in percentage of non-eave tube protected properties