Cluster randomized trial of ITWL
The cluster randomized trial of ITWL was planned for Muheza district. Muheza is located 35 km west of Tanga City, the capital of Tanga Region and 364 km north of Dar es Salaam. It had a 2012 population of 204,461 residents (100,843 males and 103,618 females). With an area of 1,498 km2, its density was 140 inhabitants per square kilometer.[15]
Malaria is endemic in the district with the main vectors An. gambiae s. l. and An. funestus s.l. The trial initially planned to use the pyrethroid product that had proved efficacious in Kenya. During the planning phase of the Muheza trial, however, entomological data showed that pyrethroid resistance had been documented, indicating that a pyrethroid product might not be effective.[16] In response, implementation was delayed while the ITWL supplier, Vestergaard (formerly Vestergaard Frandsen) in Switzerland, developed a new ITWL designed to address insect resistance to pyrethroids.
The final ITWL product was composed of a high-density polypropylene non-woven fabric containing a proprietary combination of two nonpyrethroid insecticides (0.25% abamectin and 1% fenpyroximate).[17] The side of this fabric that is attached to house walls is inactive, moisture resistant, dust free, and thermo-stable. The side which faces the house interior is the active one, functioning as a long-lasting insecticidal reservoir containing the insecticidal mixture embedded in a polymer. A pilot found that a prototype of this product (termed PermaNet Lining) was well accepted by rural African households.[3, 9]
The protocol was updated to ensure statistical power as a two-arm 44-cluster randomized trial comparing the existing standard of care (long-lasting insecticide-treated nets, LLINs) against the experimental intervention (ITWL plus LLINs). To ensure the accuracy of this comparison, in August 2015 all enumerated households in both experimental and control arms were given one Interceptor® net (BASF Corporation, Germany) for every two persons and instructed on their use. These nets contain contained alphacypermethrin (200 mg/m2) coated onto polyester fibers. The primary planned endpoint was the cumulative one-year incidence of parasitemia in children aged 6-59 months assessed through a malaria rapid diagnostic test administered at monthly household visits. The trial protocol has been published.[7] The trial, NCT02533336, was first posted on 26/08/2015.
Implementation in experimental clusters began with recruitment of 140 installers. Next, NIMR professionals and a consultant who had managed the previous Asembo trial conducted a 5-day training session for the installers. The installers needed to be residents of the study villages and most were males. The installers were supposed to be capable of manual labor and have a basic knowledge of carpentry, but had no specific educational requirement. The training provided an overview of the project, and instruction and practice on the installation of ITWL. This entailed measuring the rooms, cutting ITWL (which came on rolls about 2 meters wide), identifying standard intervals for nails, attaching the material to walls while avoiding damage to the houses, household items, and the environment; and documenting the work. As manpower needs grew, additional installers were added and given on-the-job training. Some of the original installers were promoted to team leaders or supervisors. Supervisors were required to be literate to ensure they could complete the necessary forms.
Supervisors visited households in experimental clusters ahead of the planned installation exercise to describe the process and request consent from the household head. They assigned the installation teams to specific houses who consented to have ITWL installed, monitored the installation process, and approved installers’ payment on confirmation of the completion of work. The team leaders guided teams, measured the walls, windows and doors of each house, and ensured timely completion of each day’s work. Throughout the installation, five full-time NIMR staff oversaw the work, including epidemiology, entomology, sensitization, installation, logistics, and finances.
Updating to current epidemiologic and economic conditions
To facilitate the interpretation of this study, we have adjusted all epidemiologic and economic information to values for 2019, the most recent year with comprehensive data. All economic data are expressed in current 2019 US dollars. We converted monetary amounts to 2019 USD through three steps: (1) We converted costs reported in US dollars to the equivalent in Tanzania shillings in the same year using the applicable exchange rate. The applicable rate for the organization installing the wall liner, the National Institute for Medical Research, was the net rate received by its bank (i.e., 2074.1 in 2015 and 2065.0 in 2016) while the applicable rate for other sources was the official rate of 1572.1 Tanzania shillings (TZS) per US dollar in 2011 [18]; (2) We converted TZS in any year prior to 2019 to 2019 TZS based on the Tanzanian official GDP deflator in that year and in 2019 [19] (i.e. 72.333 in 2011, 100.000 in 2015, 107.472 in 2016, 121.006 in 2019) [19]; (3) We converted 2019 TZS to 2019 USD based on the 2019 exchange rate at US$1 equals 2288.1 TZS.[20]
For comparative indicators, we used Tanzania’s 2019 per capita GNI ($1080).[21] We adjusted epidemiologic information based on malaria’s 2019 burden of 2,038 disability-adjusted life years (DALYS) lost per 100,000 population per year.[22]
ITWL intervention phases
Installation of ITWL in experimental clusters was conducted in three phases totaling 204 days, characterized by distinct modes of sensitization and payment. The first installation phase, August 10, 2015 through December 4, 2015 (117 days), used large community meetings to try to sensitize residents about the desirability of ITWL. This phase involved 352 workers (220 installers, 110 team leaders, and 22 supervisors). Village leaders, community health workers (CHWs), and researchers used these meetings to try to inform residents about ITWL, the installation schedule. These meetings also sought to raise awareness, enhance participation in the trial, and increase the use of bed nets. In this phase, installers and team leaders were each paid 10,000 TZS per day. ($4.82 in 2015, equivalent to $5.29 in 2019 USD).
The second installation phase, December 5, 2015 through January 15, 2016 (42 days), aimed at raising the number of community members participating in the project. This phase introduced better personal protective equipment (flexible gloves) for installers, added a megaphone so that project staff could better attract residents’ attention, and initiated door-to-door visits and sensitization to explain the product and answer questions in detail. It also sought to improve the efficiency of the installation. This phase had only 242 workers (22 supervisors and 220 installers). The payment system was changed from a daily wage to an output-based payment, with each supervisor receiving 1500 TZS ($0.72 in 2015USD, equivalent to $0.79 in 2019USD) and each team of installers receiving 7,000 TZS ($3.37 in 2015, equivalent to $3.70 in 2019 USD) for each installed house. This piecework mode of payment was implemented after noting that the daily wage system appeared to create a perverse incentive for installers to work slowly so as to maximize their number of work days.
The third installation phase, January 16, 2016 through February 29, 2016 (45 days), involved additional sensitization approaches. It added the distribution of brochures with photographs and simply-worded Swahili explanations of ITWL benefits. Members of the project’s socio-economic team continued to make announcements throughout the village with a portable megaphone to increase residents’ willingness to have the product installed in their homes. While maintaining the previous phase’s piecework payment modality, the third phase sought to further reduce costs per household by lowering the number of personnel to 154 workers (14 supervisors and 140 installers).
At two months after installation, an entomological trial in experimental huts in Zeneti, near Muheza, of the incremental benefit of alternative ITWL products over LLINs alone found no benefit of the pyrethroid product due to insecticide resistance but a small, though not statistically significant incremental benefit of the non-pyrethroid product on mosquito mortality.[23] Noting that results at two months were not necessarily predictive of longer term results, the investigators initiated the cluster randomized epidemiologic trial in Muheza with this non-pyrethroid product in 2015. However, the entomological results from this trial that emerged in May 2016 showed that this wall liner was no longer effective. Cone bioassay tests at 90 days after installation found that ITWL no longer killed mosquitoes in residents’ houses in Muheza district, perhaps due to issues with degradation of chemical content and/or bioavailability of the insecticides in the ITWL, such that mosquitoes did not obtain a lethal dose upon contact”. Entomological studies on mosquito age confirmed the lack of efficacy.[16]
As a result, the study’s data safety monitoring board, investigators and sponsors determined that the study needed to be terminated prematurely. Collection of epidemiologic data was stopped. Since ITWL was no longer beneficial and potentially harmful, they concluded later in 2016 that the ITWL should be removed from residents’ houses where possible.[8]
De-installation of ITWL material lasted from September 21, 2016 through October 6, 2016 (16 days). It involved 13 regular NIMR staff. The de-installation phase began by a three-day training by NIMR staff of 220 de-installers and 22 cluster supervisors. Next, community residents were invited to sensitization sessions, aided by project staff, community leaders and CHWs, to explain why ITWL was being removed prematurely.
Determining economic costs of the ITWL intervention
For analytic purposes, the unit of analysis was a household, defined as a group of people who live together and share food and expenses. In small villages each household generally owned its own house (a building). However, in large villages and small towns, multiple households could share a house. In most cases, the household members were related to one another, so they also constituted a family. The sample for the intervention arm consisted of the 5,666 households in experimental clusters who had received ITWL. The NIMR Office at Muheza provided aggregate financial expenditures for all project activities serving these households. Non-financial data on the other hand, represented the opportunity cost of contributed labor used in the intervention. Specifically, non-financial costs involved the average time and number of household members spent in the following activities: (i) attending sensitization meetings, (ii) consenting, (iii) removing household items before the installation took place, (iv) waiting for the installers to arrive and complete their work, (v) putting household items back after ITWL installation, (vi) removing household items before de-installation, (vii) waiting for the de-installers to arrive and/or assisting them or waiting for them to complete their work, and (viii) putting household items back after ITWL de-installation. Research staff from the project’s socio-economic team interviewed 136 households using questionnaires and recorded their observations. The authors obtained qualitative data through regular interactions with residents and community leaders.
We then imputed and valued the aggregate and average time for the 5,666 households. The numbers of households installed were derived from records of payments to installers. While Vestergaard donated ITWL products for this trial, we imputed its cost based on a related product. Vestergaard had previously marketed the pyrethroid ITWL used in the Asembo study under the name ZeroVector.[14] As both products had comparable purposes, settings (rural areas with mostly mud houses), and methods of supply (rolls about 2 meters wide and 100 meters long), we used the average cost per household in Kenya based on that product’s latest sales price per roll ($68.50) as the estimate for this study.[8]
To compute the cost of time households spent during sensitization, installation, and de-installation exercises, we used Tanzania’s 2015 daily wage rate of TZS 6,581 for a typical laborer from nearby sisal estates in Muheza district, obtained as part of the interviews in this study. Since none of the resources used in installing and de-installing ITWL involved capital inputs, we considered all costs occurring at the time of installation. We added 15% of the direct expenditure(s) for overhead based on the rate allowed by the main sponsor of this economic assessment for expenses in Tanzania.[24] These overhead costs covered utilities, facilities upkeep, and central administration.
Efficacy and cost of IRS in Tanzania
Information in the efficacy and financial costs of the comparative intervention (IRS) came from existing publications. Tanzania had previously hosted a cluster randomized trial of the incremental efficacy of IRS as an addition to bed nets. That study was conducted near Muleba, Tanzania (in the country’s Lake Region). The intervention arm, entailing rounds of spraying conducted prior to the long and short rains, followed the same two-round schedule as another intervention study in the same district.[25] In the randomized trial, IRS added to bed nets reduced malaria incidence by 57% compared to the control arm with only bed nets, a significant change, with a 95% confidence interval of 3%-81%.[26] We multiplied the best estimate of efficacy times the DALY burden in Tanzania to get the best estimate of the DALYs averted and used the confidence interval on efficacy to generate the confidence interval on DALYs averted.
To estimate the financial cost of IRS, we extracted information from a modeling study of the combination IRS plus LLIN and of LLIN alone in mainland Tanzania (including the vicinity of Muleba), reported in 2011 US dollars.[27] We used the version of their analysis that included the adjustment for increased insecticide resistance (2008-2012). Their analysis generally followed the framework of integrated vector management and incorporated the value of in-kind contributions from government personnel and community leaders.[28] However, their analysis did not include a non-financial component, the economic value of household time for IRS.
As an additional component of the ITWL economic evaluation, we designed and implemented a household survey. After obtaining the appropriate permission from the government and local leaders, we used a cluster sample to select 3 districts in the vicinity of Lake Victoria, Tanzania. From these districts, we randomly selected 29 wards, then randomly chose 5 households from each ward, and invited the chosen households to participate in this costing survey. From the 145 households invited to participate, 135 households (93%) completed the survey. A research team member interviewed a member of each participating household in person between December 2015 and January 2016. The survey collected information about the time household members spent attending informational meetings with government officials regarding IRS, providing 20 liters of water per household, removing and replacing furniture, awaiting the spraying operator, being present during spraying, waiting for two hours or more after spraying (with windows and doors open), and cleaning up dead insects. We valued their time at the 2015 hourly minimum wage of TZS513 ($0.2713 in 2019USD).
Cost-effectiveness framework for future ITWL products
The study was approved by the ethics committees of the National Institute for Medical Research, Tanzania, Kilimanjaro Medical College, the London School of Tropical Medicine and Hygiene, and the Committee for the Protection of Human Studies in Research at Brandeis University.