The combined use of indoor residual spraying (IRS) and long-lasting insecticidal nets (LLINs) for malaria reduction in endemic rural Tanzania

Executive summary
The project addresses a critical research question concerning the operational use of LLINs and IRS, namely: Can LLINs sustain transmission reduction gains made by IRS following withdrawal of IRS?

Muleba is an area that was holoendemic for P. falciparum malaria before PMI IRS intervention in 2007. PMI has been conducting IRS in Muleba for the past three years with a single pyrethroid spray round annually. This IRS intervention has already reduced malaria prevalence significantly.

The government of Tanzania’s commitment to universal coverage of LLINs to malaria endemic area, defined as two LLINs per household, has resulted in Kagera region being prioritized for universal coverage. We aim at scaling-up LLIN coverage and withdrawing IRS in one study arm whilst maintaining a single IRS round with high LLIN coverage in the other arm.

	First year	Intervention year
Arm 1	IRS+LLINs	IRS+LLINs
Arm 2	IRS+LLINs	LLINs

The trial design will involve randomized selection of stratified village clusters for 2 study arms. The primary outcome measure will be prevalence of parasitaemia and anaemia in cohorts of children aged 0.5-10 years. Secondary outcomes will include malaria transmission as measured by entomological inoculation rates (EIR) of the mosquito vector species and serological prevalence of malaria, as well as user acceptability LLINs compared with IRS.

The study location and the proposed trial design will provide PMI and national malaria control programmes with valuable answers to the key objectives. If IRS is withdrawn programmatically from areas currently given IRS and LLIN universal coverage, this study would provide assurance that it would be safe to do so without endangering a return to higher transmission levels. Findings from this study are expected to inform decision making so that resource utilization can be optimised.

Outcome Measures
Primary outcome:

1. Prevalence of malaria infection and anaemia in children 0.5-14 years, or
2. Incidence of confirmed malaria episodes in children 0.5-14 years through active case detection

Secondary outcomes:

1. Mean haemoglobin (g/dL) in children 0.5- 14 years.
2. Sero-conversion rates
3. Entomological Inoculation Rate (EIR) for each mosquito vector species.
4. Geometric mean density for each mosquito vector species
5. Usage and perception of LLIN in an environment of reduced IRS.
6. Detection and monitoring of resistance markers including kdr

Community Sensitization and local coordination
Detailed mapping of the trial area and house location in villages was the main technical activity during the Preparatory phase of Q1.2. Sensitization of village leaders was done prior to GPS mapping of village boundaries and houses. Sensitization meetings were chaired by members of the PKMPT team together with a representative of the Muleba District Office. Sensitization for phase 1 mapping proceeded without any major objections and every village sensitized has now been mapped.

Selection of clusters for pilot malaria prevalence survey
The trial area was defined in September during initial visits to Kagera by LSHTM, KCMC and NIMR scientists and after discussions and data collection with RTI and DMO. Final selection of the study clusters will be made according to the following criteria:

1. Accessibility – travelling time from PAMVERC office in Muleba town (<45 min) and the condition of access roads
2. Distances between clusters (≥1km)
3. Malaria prevalence (from January pilot survey)
4. Population size (>500 residents)

During initial visits we identified and geo-located 75 villages within the district that met criterion 1. This has since been extended to 113 villages across 29 wards. These villages lie in an area of approximately 2000km2 extending 70km North to South and 68km East to West, at an elevation ranging between 1110 to 1640m above sea level.

Of these villages 87 have been mapped to house level (including hamlet boundaries, health facilities and other landmarks) using GPS handheld units and ExpertGPS v3.8 (TopoGrafix) software . These 87 villages are comprised of 378 hamlets and 49,195 houses; an average of 4.3 hamlets and 565 houses in each village. This corresponds with the 2009 projected census estimates of 590 hamlets/village provided by the Muleba District Office. Figure 2 illustrates the distribution of health facilities (HF) within the northern wards of our study area and the density of houses served by those HF.

With these maps it becomes possible to identify clusters to use in the pilot prevalence survey. Each cluster will be comprised of at least one village, because the village is the smallest unit we can refer to when directing RTI where to conduct IRS during our intervention year. By dragging concentric circles over the maps we can identify (at random) suitable clusters of one or more villages that have a core sampling area of 1km radius surrounded by a buffer zone of an additional 1km.

Within the core, 100 households will be selected randomly and the resident children will be screened for parasites using RDTs, blood smear and PCR. As hamlet boundaries and tracks within the villages have also been mapped, we will attempt to limit the number of hamlets within the core area and to identify the most efficient routes between core sampling houses.

All villages entirely or partly within the buffer zone and core sampling area will receive the same treatment (continued IRS or replacement LLINs). Figures 3-5 exemplify the cluster selection and household sampling process.