Article Text

Reduction of malaria case incidence following the introduction of clothianidin-based indoor residual spraying in previously unsprayed districts: an observational analysis using health facility register data from Côte d’Ivoire, 2018–2022
  1. Emily R Hilton1,
  2. Ndombour Gning-Cisse2,
  3. Auguste Assi2,
  4. Mathieu Eyakou2,
  5. John Koffi2,
  6. Barthelemy Gnakou2,
  7. Bernard Kouassi2,
  8. Cecilia Flatley3,
  9. Joseph Chabi3,
  10. Constant Gbalegba4,
  11. Serge Alex Aimain4,
  12. Colette Yah Kokrasset4,
  13. Mea Antoine Tanoh4,
  14. Sylvain N'Gotta4,
  15. Octavie Yao4,
  16. Hughes Egou Assi5,
  17. Philomène Konan5,
  18. Kelly Davis6,
  19. Edi Constant7,
  20. Allison Belemvire8,
  21. Patricia Yepassis-Zembrou9,
  22. Pascal Zinzindohoue10,
  23. Blaise Kouadio10,
  24. Sarah Burnett6
  1. 1PMI VectorLink Project, PATH, Seattle, Washington, USA
  2. 2PMI VectorLink Project, Abt Associates, Abidjan, Côte d'Ivoire
  3. 3PMI VectorLink Project, Abt Associates, Rockville, Maryland, USA
  4. 4Programme National de Lutte Contre le Paludisme, Abidjan, Côte d'Ivoire
  5. 5Direction de l'Informatique et de l'Information Sanitaire, Abidjan, Côte d'Ivoire
  6. 6PMI VectorLink Project, PATH, Washington, District of Columbia, USA
  7. 7Centre Suisse de Recherches Scientifiques en Côte d'Ivoire, Abidjan, Côte d'Ivoire
  8. 8US President's Malaria Initiative, US Agency for International Development, Washington, District of Columbia, USA
  9. 9U.S. President’s Malaria Initiative, Centers for Disease Control and Prevention, Abidjan, Côte d'Ivoire
  10. 10U.S. President’s Malaria Initiative, U.S. Agency for International Development, Abidjan, Côte d'Ivoire
  1. Correspondence to Ms Emily R Hilton; ehilton{at}


Background Indoor residual spraying (IRS) using neonicotinoid-based insecticides (clothianidin and combined clothianidin with deltamethrin) was deployed in two previously unsprayed districts of Côte d’Ivoire in 2020 and 2021 to complement standard pyrethroid insecticide-treated nets. This retrospective observational study uses health facility register data to assess the impact of IRS on clinically reported malaria case incidence.

Methods Health facility data were abstracted from consultation registers for the period September 2018 to April 2022 in two IRS districts and two control districts that did not receive IRS. Malaria cases reported by community health workers (CHWs) were obtained from district reports and District Health Information Systems 2. Facilities missing complete data were excluded. Controlled interrupted time series models were used to estimate the effect of IRS on monthly all-ages population-adjusted confirmed malaria cases and cases averted by IRS. Models controlled for transmission season, precipitation, vegetation, temperature, proportion of cases reported by CHWs, proportion of tested out of suspected cases and non-malaria outpatient visits.

Results An estimated 10 988 (95% CI 5694 to 18 188) malaria cases were averted in IRS districts the year following the 2020 IRS campaign, representing a 15.9% reduction compared with if IRS had not been deployed. Case incidence in IRS districts dropped by 27.7% (incidence rate ratio (IRR) 0.723, 95% CI 0.592 to 0.885) the month after the campaign. In the 8 months after the 2021 campaign, 14 170 (95% CI 13 133 to 15 025) estimated cases were averted, a 24.7% reduction, and incidence in IRS districts dropped by 37.9% (IRR 0.621, 95% CI 0.462 to 0.835) immediately after IRS. Case incidence in control districts did not change following IRS either year (p>0.05) and the difference in incidence level change between IRS and control districts was significant both years (p<0.05).

Conclusion Deployment of clothianidin-based IRS was associated with a reduction in malaria case rates in two districts of Côte d’Ivoire following IRS deployment in 2020 and 2021.

  • epidemiology
  • malaria
  • public health

Data availability statement

Data are available on reasonable request.

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See:

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


  • Indoor residual spraying (IRS) using non-pyrethroid insecticides deployed in communities using pyrethroid-based insecticide treated nets is associated with reduced malaria prevalence, and with reduced malaria incidence in some settings. The efficacy of new IRS formulations with the neonicotinoid clothianidin as an active ingredient has been demonstrated against vectors in laboratory and experimental hut trials. There is mixed evidence on the impact of clothianidin-based IRS on routinely reported malaria case incidence.


  • This study presents evidence of the impact of clothianidin-based IRS on reducing the incidence of malaria cases reported in health facility registers in Côte d'Ivoire.


  • The results reported here expand the evidence base around the effectiveness of new non-pyrethroid formulations for IRS, which will support National Malaria Programme decision-making for rotation of IRS products in contexts of increasing vector resistance to pyrethroids. This work additionally highlights the value of high-quality health facility data for evaluating intervention impact.


Following two decades of steadily declining malaria mortality from 2000 to 2019, global progress in the fight against malaria has recently stalled and, in some regions, reversed.1 Vector control interventions including indoor residual spraying (IRS) and insecticide-treated nets (ITNs) have contributed significantly to progress against malaria2 and remain cornerstone malaria prevention methods.

IRS consists of applying a residual insecticide to the interior walls of homes to kill mosquitoes that encounter it. Despite the substantial evidence of the impact of IRS in reducing malaria incidence,3 a key challenge to its continued efficacy is the development of malaria vectors’ resistance to insecticides used in IRS. New IRS formulations with the neonicotinoid clothianidin as an active ingredient became WHO prequalified vector control products in 2017 and 2018.4 These include SumiShield 50WG (water dispersible granules), a clothianidin-based product and Fludora Fusion WP-SB (wettable powder in a water-soluble bag), which contains a mixture of clothianidin and deltamethrin (a pyrethroid). Although the biological efficacy of both products at killing mosquitoes has been demonstrated in lab and experimental hut trials,5–11 evaluation of their impact on clinically relevant outcomes in a ‘real-world’ setting is limited.12 13

Prior to the addition of neonicotinoids to the WHO prequalification listing, the list included only four chemical classes of IRS: carbamates, organophosphates, organochlorines and pyrethroids. Together, these four classes comprise only two modes of action,14 which increases the likelihood of cross-resistance between products with the same mode of action.15 With widespread resistance to pyrethroids well documented,16 neonicotinoids expanded the options available for insecticide rotation as recommended by the WHO Global Plan for Insecticide Resistance Management.17 The residual activity of SumiShield has been found to produce >80% vector mortality up to 9.5 months postspray,8 and between 8 and 10 months postspray for Fludora Fusion.11

Malaria remains a leading cause of morbidity and mortality in Côte d’Ivoire, accounting for around 33% of outpatient visits at health facilities. In its National Malaria Strategic Plans 2016–2020 and 2021–2025, Côte d’Ivoire included IRS as a key vector control intervention for reducing malaria burden in high transmission districts.18 The US President’s Malaria Initiative (PMI) began supporting preliminary entomological surveillance and insecticide susceptibility testing in Côte d’Ivoire in 2019, in preparation for the first IRS implementation in 2020.19 Two previously unsprayed districts, Nassian and Sakassou, were selected based on entomological monitoring and insecticide susceptibility data, as well as being among the highest malaria incidence districts in the country. IRS targeting decisions were also supported by a geographical reconnaissance study that assessed operational suitability based on factors such as population size and accessibility of structures to be sprayed.20 21 A high level of resistance to pyrethroids, organophosphates and carbamates was observed in both target districts through insecticide resistance monitoring,22 23 leading to the selection of clothianidin and mixed clothianidin with deltamethrin-based products for deployment.

In this study, we present a retrospective time series (2018–2022) analysis of the epidemiological impact of IRS campaigns that took place in 2020 and 2021 in two districts (Nassian and Sakassou) of Côte d’Ivoire. This analysis relies on routine health data abstracted directly from health facility consultation registers with the aim of expanding the evidence base around the effectiveness of clothianidin-based IRS products on malaria incidence at health facilities.


Study setting

Côte d’Ivoire is a coastal West African country with a generally warm and humid climate that ranges from equatorial on the southern coast to semiarid in the north. This study included two districts where IRS was implemented in 2020 and 2021: Nassian (population 60 622) and Sakassou (population 109 913); and two neighbouring control districts that did not receive IRS: Béoumi (population 178 145) and Dabakala (population 238 748). Population data were provided by district health offices from the 2021 national census.

Sakassou and Béoumi are located in a mosaic forest/savannah zone in central Côte d’Ivoire; Nassian and Dabakala are further north in arid zones of Sudanese and sub-Sudanese savannah. The climate throughout is tropical. The primary malaria parasite is Plasmodium falciparum and the primary vector is Anopheles gambiae sensu lato (s.l.).22–26

The two control districts were selected based on having similar malaria transmission patterns to the IRS districts, including vector biting and infection peaks around the rainy season, as described in detail by Kouassi et al.20 The rainy season typically extends from April to August in Sakassou and Béoumi; and from June to September in Nassian and Dabakala. The 2016 Malaria Indicator Survey reported 28.0% malaria prevalence in children 6–59 months in the Centre North region (location of Sakassou, Béoumi and Dabakala) and 46.9% in the North East region (location of Nassian). Prevalence measures were not reported disaggregated to the district level.27

Figure 1 illustrates the location of the study districts along with the geocoordinates of health facilities where data collection took place.

Figure 1

Map of study site showing (A) a map of Côte d’Ivoire with the study districts highlighted; (B) IRS (red) and control (grey) districts with the locations of health facilities where data collection for this study took place (black x-marks). IRS, indoor residual spraying.

Mass distribution of pyrethroid-only ITNs took place in all four study districts in 2017 and again in April 2021. During the 2021 ITN distribution, Sakassou, Béoumi and Nassian received deltamethrin-based nets. Alphacypermethrin-based ITNs were distributed in Dabakala.28 The 2021 Demographic and Health Survey (DHS) conducted from September to December 2021 reported that 70.8% of households in the Centre North and 77.8% of households in the North East possessed at least one ITN. A reported 48.4% of households in the Centre North and 53.6% of households in the North East had at least one ITN per two people.29

In Sakassou, Béoumi and Dabakala, community health workers (CHWs) conduct rapid diagnostic tests (RDTs) and provide treatment to patients with confirmed malaria cases. In these three districts, CHW coverage was reported to be between one per 500–1500 population in 2017. CHWs are also active in Nassian, however they do not conduct RDTs and instead refer suspected malaria cases to the nearest health facility. CHW coverage in Nassian was reported to be one per 250–500 population in 2017.30 31


In 2020 and 2021, PMI VectorLink carried out IRS spray campaigns in Nassian and Sakassou. The 2020 campaign took place from 10 August to 12 September, and the 2021 campaign took place from 2 August to 4 September. Both years, a clothianidin IRS product (SumiShield 50WG) was deployed in Nassian, and a combination clothianidin with deltamethrin product (Fludora Fusion WG-SB) was deployed in Sakassou. Both years, the IRS campaigns were initially planned to take place in April and May, immediately preceding the peak malaria season and to correspond with rising vector populations as recommended by the WHO.32 However, both campaigns were delayed by several months due to disruptions caused by the COVID-19 pandemic in 2020 and the diversion of operational resources to the mass ITN campaign in 2021.

In 2020, a total of 53 962 structures were sprayed across both districts, representing 91.9% of structures found, and the campaign protected a reported population of 193 935 people residing in treated households. The 2021 campaign sprayed a total of 60 496 structures (96.7% of found) and covered a population of 201 178. Reported populations protected from both campaigns would indicate higher than 100% population coverage when compared with census population estimates, which may be explained by differences in population enumeration between IRS spray teams and the census. District-level IRS coverage indicators are presented in table 1. Both years, the primary reasons for not spraying a found structure were refusals and locked structures.33 34

Table 1

Characteristics and summary statistics of the four study districts during the period from September 2018 to April 2022

Insecticide resistance and IRS residual efficacy

Insecticide susceptibility testing carried out by the PMI VectorLink project from 2018 to 2021 detected An. gambiae s.l. resistance to pyrethroids at sentinel sites nationwide, although vectors continued to be susceptible to non-pyrethroids including clothianidin. Insecticide residual efficacy testing conducted after each IRS campaign yielded high vector mortality rates (>80%) for at least 7 months post-campaign on mud and cement walls.33 34

Study design

This study employed a retrospective quasi-experimental design using routine data to assess the impact of the 2020 and 2021 IRS campaigns on reported all-ages population-adjusted malaria cases confirmed by RDT or microscopy in two districts that received IRS and two comparator districts that did not. The study period encompassed September 2018 to April 2022, providing a baseline of 24 months before the first IRS campaign in 2020; 12 months between the 2020 and 2021 IRS campaigns; and 8 months following the 2021 IRS campaign.

Health facility surveillance and data collection

A 2021 audit of key malaria indicators that are reported into the country’s health management information system (HMIS) found low levels of reporting completeness and agreement between HMIS data and health facility registers.35 Based on these findings, HMIS data were deemed of inadequate quality to support this study, and data were instead collected directly from health facility registers in the four study districts.

All primary care health facilities and hospitals in the four study districts were eligible for inclusion in the data collection. Eligibility for inclusion in the final analysis was based on availability of data for at least 9 out of 12 months for each ‘spray year’, defined as September to August.

Patient data were abstracted from consultation registers (registres de consultations curatives) for the period September 2018 to April 2022. For each patient, the diagnosis, RDT result and microscopy result were documented using a smartphone-based tool developed for this activity in KoboCollect. A suspected malaria case was defined as any written diagnosis indicating malaria (including phrases such as ‘suspected malaria case’, ‘simple malaria case’, ‘severe malaria case’), such that parasitological testing by RDT or microscopy would be expected, as per WHO guidance.36 A confirmed malaria case was defined as any positive RDT or microscopy result. Patients for whom a positive RDT or microscopy result was recorded, but where the clinician had not written ‘malaria’ as the diagnosis, were included in the total counts for suspected and confirmed cases. A photograph of a blank consultation register page is provided in the online supplemental material.

Supplemental material

The data collection took place from 8 August 2022 to 8 September 2022. A total of 184 data collection agents were recruited locally based on familiarity with facility registers and ability to use a smartphone. Agents attended a 1-day training where they received instruction on the use of the data collection tool, interpretation of facility registers and protocols to protect patient confidentiality. The trainings took place by district and were facilitated by three supervisors in each of the four study districts (12 total supervisors). During the data collection itself, agents at each facility received at least three supervision visits to ensure adherence to protocols and quality of data collection. Supervisors ensured that (1) available consultation registers for the entire study period had been assembled; (2) consultation registers from all facility departments (eg, maternity) were included; (3) agents adhered to protocols and correctly and consistently documented each indicator and (4) the number of patients in the registers matched the number in the database. Supervisors conducted at least one audit of collected data at each facility. The audit consisted of a supervisor randomly selecting 1 month of data that had already been collected by the agent and repeating the collection. The results for each indicator were compared between the audit and the real data collection. The audit allowed a 5% margin of error. Where differences greater than 5% were observed, the agent was required to repeat the data collection.

Monthly confirmed malaria cases were calculated as the sum of all patients that tested positive by RDT or microscopy. Patients for whom both RDT and microscopy positive results were recorded were counted as a single confirmed case.

RDT stockout data were obtained from district offices for facilities in Béoumi and Nassian. These data were not available in Dabakala and Sakassou. The percentage of facilities in Béoumi and Nassian that reported a stockout each month was visually compared with monthly trends in RDT-confirmed cases to detect potential impacts of stockouts.

Monthly RDT case data from CHWs was obtained from district offices of each district except for Nassian, where CHWs do not conduct RDTs. Due to the lack of an official archival system for CHW registers, these data were only available through district and national reporting systems. CHW-reported cases are collated by facility staff and sent to the district for data entry. In Béoumi and Dabakala, CHW data were pulled from the national HMIS DHIS2 (District Health Information Systems 2). In Sakassou, the data were pulled from the Système d’Information et de Gestion (SIG) report, a paper-based reporting tool that is transmitted from facilities to district offices for data entry into the DHIS2 platform.

Facility catchment population data based on the 2021 census were provided by district health offices. Populations were projected forward to 2022 and backward to 2018 assuming a 3% annual growth rate, based on the growth rate used in census projections. Data collected from facilities in urban areas that shared catchment populations were merged into single catchment areas.

Control variables

Environmental variables (precipitation, Enhanced Vegetation Index (EVI), temperature) were included to control for seasonal influences on malaria cases. Monthly precipitation data were pulled from the Climate Hazards Group InfraRed Precipitation with Station dataset.37 EVI and temperature data were pulled from the Moderate Resolution Imaging Spectroradiometer.38 The final climate dataset used in modelling included total monthly precipitation in millimetres, mean monthly EVI and mean monthly temperature in degrees Celsius. These were aggregated spatially at the commune level (administrative level below health district), scaled to have a mean of 0 and standard deviation (SD) of 1 and lagged by 2 months.

A dummy binary variable corresponding to the high malaria transmission season (June–October) was included to adjust for seasonality not accounted for by the environmental covariates alone.

The proportion of suspected malaria cases that received parasitological testing at facilities was calculated to control for the potential effect of RDT stockouts on confirmed malaria case counts. To control for varying levels of CHW activity, the proportion of cases reported by CHWs out of all confirmed cases was calculated. To control for patterns in health-seeking behaviour, the number of non-malaria outpatient visits was calculated. These three indicators were all scaled to have a mean of 0 and SD of 1 and included as modelling covariates.

Analytical approach

An interrupted time series with control (ITSc) modelling strategy was used to detect changes in the level (intercept) and trend (slope) of malaria case incidence following each IRS campaign.39 The dependent (outcome) variable was all-ages monthly confirmed malaria cases, with health facility catchment population as the offset. A generalised estimating equation model was fit to control for correlation within each facility catchment area. A negative binomial distribution was selected to account for hypothesised overdispersion in count data, and a first-order autoregressive correlation structure was specified to account for linearly associated correlation between repeated measures from the same facility over time.

Separate models were fit to compare the postspray period of each IRS campaign against the 24-month baseline period (September 2018–August 2020). Model 1 included September 2020–August 2021 (12 months) as the post-IRS period. Model 2 included September 2021–April 2022 (8 months) as the post-IRS period.

Malaria cases averted by IRS were estimated by generating counterfactual scenarios where IRS was not implemented and comparing them to the true scenarios. To accomplish this, the binary IRS intervention term was set to equal 0 throughout the study, and model predictions were generated using coefficients generated from the true situation model. The number of cases averted was calculated as the difference between the model predictions from the true situation and the counterfactual model predictions where IRS was absent.

Data were cleaned, transformed and joined in R V. Model fitting and simulations were run in Stata/SE V.17.0 (StataCorp). Data visualisations were created in R and Tableau Desktop V.2022.2 (Tableau Software, Washington, USA).


A total of 94 primary care facilities, including 4 district hospitals (1 in each study district), were identified and included in the data collection. No primary care facilities in the study districts were excluded from the data collection.

Four facilities in urban centres were found to share catchment populations with the district hospital (two in Sakassou, one in Dabakala and one in Nassian) and merged, yielding a total of 90 facility catchment areas. A total of 89 facility catchments and 90.8% (3596) of facility months were included in the analysis, based on eligibility criteria of facilities reporting at least 9 out of 12 months each spray year. Of these, 37 (41.1%) facility catchment areas were located in IRS districts, and 53 (58.9%) were in control districts (table 1).

Data collection teams observed 51 (1.4%) facility months where it appeared that only a partial month was available in the consultation registers. These months were treated as complete months, due to the infeasibility of systematically determining that the registers for all other months were not also missing some records that could not be detected. Reasons for data unavailability and incompleteness included registers that were missing or destroyed (eg, damage from water or termites), and periods when facilities were non-operational.

Across the 4 districts, a total of 1 010 035 patient visits were recorded in the consultation records during the study period. A total of 775 066 suspected malaria cases were reported, of which 709 185 (91.5%) were tested by RDT or microscopy. A total of 697 211 cases (98.3% of tested cases) were confirmed. Of the confirmed malaria cases, 221 785 (31.8%) were reported from facilities in IRS districts, and 81 064 (11.6%) were reported by CHWs.

Statistical results

Figure 2 displays modelled predictions of confirmed all-ages malaria case incidence per 1000 population plotted with observed values for the two fitted models, along with predicted case incidence under a counterfactual ‘no IRS’ scenario. Full model output including incidence rate ratios (IRRs) for all covariates is presented in table 2.

Figure 2

Monthly all-ages confirmed malaria case incidence per 1000 population. Vertical dashed lines indicate IRS campaign dates. Solid lines represent modelled predictions from controlled interrupted time series models and points represent observed values. IRS districts are in red and control districts are in grey. Red dashed lines represent the predicted incidence of cases under a counterfactual ‘no IRS’ scenario, and green shaded areas represent the estimated incidence of cases averted by IRS. The x-axis counts the number of months starting September 2018. (A) The time period includes the pre-IRS period from September 2018 to August 2020, and the post-IRS period from September 2020 to August 2021. (B) The time period includes the pre-IRS period from September 2018 to August 2020, and the post-IRS period from September 2021 to April 2022. IRS, indoor residual spraying.

Table 2

Multivariate incidence rate ratios (IRR) for population-adjusted all-ages laboratory-confirmed malaria cases

Both models found that the month-to-month trend in confirmed malaria cases was decreasing during the preintervention baseline period in control districts (model 1: IRR 0.991, 95% CI 0.987 to 0.996; model 2: IRR 0.991, 95% CI 0.988 to 0.994) and in IRS districts (model 1: IRR 0.992, 95% CI 0.983 to 1.002; model 2: IRR 0.992, 95% CI 0.982 to 1.002). The trend in IRS districts was not found to be significantly different from control districts in either model (p>0.05).

2020 IRS campaign

In the month immediately following the 2020 IRS campaign, confirmed malaria case incidence in IRS districts decreased significantly by 27.7% (IRR 0.723, 95% CI 0.592 to 0.885, p=0.002) from the month prior, a significant reduction compared with control districts (IRR 0.625, 95% CI 0.494 to 0.790, p=<0.001). The level of malaria incidence in control districts did not change significantly over the same period (IRR 1.158, 95% CI 0.975 to 1.376, p=0.095). The month-to-month trend in incidence did not change in IRS (IRR 1.025, 95% CI 0.974 to 1.078, p=0.346) or control districts (IRR 1.015, 95% CI 0.983 to 1.048, p=0.367) post-IRS.

An estimated 7910 (95% CI 4410 to 12 721) malaria cases were averted in Sakassou, and 3078 (95% CI 1285 to 5462) in Nassian in the 12 months after the 2020 IRS campaign (figure 2). These cases averted represent an estimated 15.8% reduction in Sakassou, and a 16.0% reduction in Nassian compared with if IRS had not been implemented.

2021 IRS campaign

The level of malaria case incidence in IRS districts decreased significantly by 37.9% the month following the 2021 IRS deployment (IRR 0.621, 95% CI 0.462 to 0.835, p=0.002), a decline that was significantly greater than in control districts (IRR 0.562, 95% CI 0.351 to 0.901, p=0.017). The level of case incidence in control districts did not change significantly post-IRS (IRR 1.105, 95% CI 0.870 to 1.405, p=0.413). The trend in monthly cases increased significantly in all districts post-IRS, with an average 6.2% increase each month in IRS districts (IRR 1.062, 95% CI 1.048 to 1.077, p<0.001) and an average 2.4% increase each month in control districts (IRR 1.024, 95% CI 1.006 to 1.042, p=0.008). The increase in monthly trend was significantly higher in IRS districts than control districts (IRR 1.037, 95% CI 1.033 to 1.042, p<0.001).

An estimated 11 213 (95% CI 10 301 to 11 987) cases were averted in Sakassou in the 8 months following the 2021 IRS campaign, representing a 28.9% reduction. In Nassian, an estimated 2957 (95% CI 2832 to 3038) cases were averted, representing a 20.4% reduction (figure 2) compared with if IRS had not been implemented.

RDT stockouts and proportion of tested malaria cases

The average proportion of tested out of suspected malaria cases reported in facility registers was high (>89%) across all study districts during the study period (table 1). A prominent drop was observed in all districts in 2020, indicating an overall decrease in testing. In Sakassou and Béoumi, the decline was observed primarily in July and August 2020. In Dabakala, the decline extended from July to September 2020, and in Nassian from October to November 2020. In Nassian, a prominent drop was also observed in September 2018, the first month of the study period. In Béoumi and Nassian, districts where facility RDT stockout data were available, there was an observed correlation between the months of low proportion of tested cases with high percentage of facilities reporting stockouts (figure 3).

Figure 3

Monthly proportion of all-ages suspected malaria cases reported in health facility consultation registers that went on to receive testing by rapid diagnostic test (RDT) or microscopy in the four study districts. In Béoumi and Nassian, for which monthly health facility stockout data were available, the coloured squares represent the monthly proportion of health facilities reporting a stockout of RDT kits, where darker orange represents a higher proportion. The vertical dashed lines indicate the dates of IRS campaigns. IRS, indoor residual spraying.

Both ITS models detected a strongly significant association between the proportion of tested malaria cases and increased rates of confirmed cases (model 1: IRR 2.022, 95% CI 1.855 to 2.204, p<0.001; model 2: IRR 1.858, 95% CI 1.714 to 2.015, p<0.001) (table 2).


Results from the Côte d'Ivoire 2021 DHS indicate that malaria prevalence across the country has declined since 2016. In the Centre North region (including Sakassou, Béoumi and Dabakala districts), prevalence fell from 28.0% in 2016 to 18.8% in 2021. The North East region (Nassian) experienced a more modest drop in prevalence, from 46.0% to 44.3%.27 29 The findings presented in this study suggest that IRS may have conferred additional case reduction benefits within the broader regional context of declining malaria transmission.

The effect sizes reported in this study align with a recent Cochrane review of clustered randomised control trials that found that IRS may reduce malaria incidence by 14% on average, with the effect ranging from 35% significant case reductions to 32% non-significant increases.3 That a significant impact of IRS was observed despite the non-optimal timing of the intervention is encouraging. It is apparent in the epidemiological curves (figure 3) that the timing of both IRS campaigns missed the first weeks of the peak malaria transmission season. Had the campaigns been conducted in the recommended time frame,32 these results suggest that an even greater effect may have been observed. The increase in month-to-month trend observed in IRS districts after the second round of IRS suggests that a continued strategy of annual IRS application may be necessary to sustain the reductions in malaria case levels observed directly after the campaigns.

The results presented here are aligned with the observations of entomological monitoring conducted by PMI VectorLink in collaboration with the Institut Pierre Richet, the Centre d’Entomologie Médicale et Véterinaire, and the Institut National d’Hygiène Publique. In the 7 months following IRS implementation (September–March), the indoor entomological inoculation rate (EIR) in IRS districts dropped from a pre-IRS baseline of 6.05 (measured from September 2019 to March 2020) to 1.34 in 2020 (77.9% reduction) and 1.76 in 2021 (70.9% reduction). In control districts, indoor EIR increased from 0.23 to 0.50 (117.4% increase) and 0.43 (86.9% increase) over the same periods in 2020 and 2021, respectively (online supplemental figures 2 and 3).23–25

The modelling strategy employed in this study assessed the second round of IRS without accounting for the first round. Lingering product efficacy combined with the suppression of transmission caused by the first round of IRS may have contributed to the impact of the second round.33 Although studies have detected evidence of a potential accumulative impact of IRS after 4 years of annual deployment in Uganda41 and after 3 years in Madagascar,42 the relationship between successive rounds of IRS is not well understood.

Evidence on the impact of clothianidin-based IRS on malaria incidence is mixed. An observational study in Uganda observed a rebound in routinely reported cases that coincided with a switch to clothianidin-based IRS products after 4–5 years of sustained IRS using pirimiphos-methyl. This resurgence also coincided with distribution of pyrethroid ITNs in an area of high pyrethroid resistance.12 In northern Zambia, living in a house treated with clothianidin-deltamethrin-based IRS was not associated with decreased parasite prevalence compared with living in a house treated with pirimiphos-methyl, although community coverage with either product was associated with reductions in prevalence.13 Additional research will be crucial to understanding the impact of clothianidin in various transmission contexts, particularly as National Malaria Programmes (NMPs) seek to rotate IRS insecticide products.

In April 2021, pyrethroid ITNs were distributed in all study districts. The evidence on adding non-pyrethroid IRS in communities using pyrethroid ITNs is mixed,3 and the WHO recommends adding IRS to ITNs only as part of insecticide resistance management strategies.43 Prior to 2021, mass ITN distribution had last been carried out in 2017 and the baseline period for this study began at least 1 year later. The newly distributed nets may have had an additive effect with the 2021 IRS campaign, although their impact was likely limited in the context of pyrethroid resistance. In the absence of granular ITN coverage data, or of any control districts that did not receive ITNs, an independent effect of IRS and ITNs was not discernible.

The collection of patient data directly from facility consultation registers allowed this study to avoid data quality challenges that often accompany use of routine data sources, such as variable reporting rates, outliers resulting from data entry errors and changes in reporting indicator definitions over time. Given the resource-intense needs of carrying out the data collection, this approach may not be widely feasible for other vector control impact evaluations. However, this experience points to the potential benefits to be reaped from strengthening health information reporting and management. Routine HMIS data are frequently underutilised due to quality concerns, despite being an important longitudinal data source that can enable study designs not possible using intermittent cross-sectional community surveys.44–46 Indeed the Global Technical Strategy for Malaria 2016–2030 has emphasised the importance of high-quality routine data by redefining surveillance as a core intervention.47

Despite this, the reliance of this study on routine data constitutes a notable limitation. Bias in malaria incidence estimates can arise from changes in care-seeking, differences in access to parasitological diagnosis and incomplete registration of patients.44 These potential influences would tend to bias the results towards underestimation of malaria case rates but would not be expected to vary in association with the intervention of interest (IRS). To adjust for differences in care-seeking and access to healthcare, the modelling strategy controlled for non-malaria outpatient attendance. However, bias may still be introduced if there were unobserved overall differences between IRS and control districts.

Data collection agents worked with health facility staff to ensure that all available registers for the study period were assembled, and to record the reasons for missing registers. For a small number of facility months (51), the consultation registers were discernibly incomplete for reasons determined through discussion with facility staff (eg, registers were destroyed, or staff recalled being absent during part of the month). However, for the purposes of this analysis, such partial months were treated as complete months, due to the fact that it could not be systematically assured that the records for all other months were also not missing some partial data.

Another noteworthy limitation is the inclusion of CHW-reported case data that was extracted from DHIS2 and SIG reports, due to historical CHW registers being broadly missing or destroyed. Although a previous audit of DHIS2 and SIG reports uncovered serious data quality concerns,35 excluding these data may have led to underestimation of malaria cases. Having access to facility aggregated CHW data does not allow determination of CHW reporting completeness, nor the proportion of CHWs who reported each month. In Nassian (an IRS district), the practice by which CHWs refer suspected malaria cases to facilities for testing may result in underdiagnosis of cases compared with districts where CHWs perform RDTs themselves. Variations in CHW reporting and levels of activity were controlled for by adjusting the models for the proportion of confirmed cases that were reported by CHWs.

The environmental data included in the analysis models were spatially aggregated at the commune level instead of at the same granularity as our unit of analysis (health facility catchment area). This may introduce bias if environmental and climatic variations below the level of aggregation influenced case incidence differently in IRS versus control areas.

The most pressing limitation of this study is the occurrence of RDT stockouts in 2020 that severely impacted the primary outcome of confirmed malaria cases, and the unavailability of facility stockout data in two districts (Béoumi and Nassian). In the other two districts (Sakassou and Dabakala), stockout data were only available from district pharmacies and/or hospitals, which may not accurately reflect the RDT stock situation at facilities. To control for the effect of stockouts on the number of confirmed cases, the proportion of suspected malaria cases that were tested was included in the models. Based on the correlation with RDT stockouts at facilities in Béoumi and Nassian, this was a reasonable proxy measure. However, it may not fully capture the trends of RDT stockouts and may also reflect variations in malaria case management at health facilities.


The results presented here suggest a positive public health impact of IRS that contributes to the evidence base around the effectiveness of clothianidin-based IRS products. In the context of increasing resistance to the current arsenal of antivector chemical classes, particularly pyrethroids, expanding the evidence base around different chemical classes will inform NMPs who seek to rotate the insecticides being deployed. This work, interpreted with the limitations described above, highlights the value of health facility data for evaluating intervention impact, empowering evidence-based decision-making by NMPs.

Supplemental material

Data availability statement

Data are available on reasonable request.

Ethics statements

Patient consent for publication

Ethics approval

Study approval was granted by the ethics committee Comité National d’Ethique des Sciences de la Vie et de la Santé (CNESVS) in Abidjan, Côte d’Ivoire.


The authors would like to thank the Côte d’Ivoire Programme National de Lutte Contre le Paludisme for their support on this project. This work would not have been possible without the efforts of the data collection agents as well as the collaboration of health facility staff and the district health offices in Sakassou, Béoumi, Dabakala, and Nassian who provided access to facility registers and answered questions. Dale Rhoda and Caitlin Clary of Biostat Global Consulting provided crucial advice around the design of the statistical analysis of this study. Finally, we would like to expressly thank all the partners of the PMI VectorLink Project in Côte d’Ivoire for their support and contributions.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Handling editor Alberto L Garcia-Basteiro

  • Contributors ERH, NG-C, AA, ME, JK, BG, SN'G, OY, HEA and PK oversaw health facility data abstraction, including planning, development of data collection tools and quality control. ERH performed descriptive and statistical analysis of epidemiological data with the support of SB. BK oversaw collection of entomological data and KD performed descriptive analyses of entomological data. ERH drafted the final manuscript. CF, JC, CG, SAA, CYK, MAT, EC, AB, PY-Z, PZ and BK critically reviewed manuscript drafts and provided feedback. All authors read and approved the final manuscript. ERH is responsible for the overall content as the guarantor of this work.

  • Funding This study was supported by the PMI VectorLink Project (USAID/PMI contract AID-OAA-1-17-00008, task order AID-OAA-TO-17-00027).

  • Disclaimer The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the US Centers for Disease Control and Prevention and US Agency for International Development.

  • Map disclaimer The depiction of boundaries on this map does not imply the expression of any opinion whatsoever on the part of BMJ (or any member of its group) concerning the legal status of any country, territory, jurisdiction or area or of its authorities. This map is provided without any warranty of any kind, either express or implied.

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.