The global database on insecticide resistance in malaria vectors was established in 2014 to help track insecticide resistance worldwide. Following a large data consolidation exercise conducted in 2019, which included the integration of separate WHO regional databases and data from other public sources, it has become the largest database on available at present. It contains data on the status of insecticide susceptibility of Anopheles mosquitoes in more than 88 countries. The contents of the database are provided by Member States, partners (including research and academic institutes) as well as nongovernmental organizations, and are extracted from published and non-published reports on insecticide bioassays.
The database for the Malaria Threat Maps is up-loaded on an ad hoc basis. The database provides the most up to date and comprehensive information regarding insecticide resistance on all continents. The data can be presented in a graphical format as maps detailing individual studies and site-level data for all the threat to vector control. Dashboards depicting epidemiological profile, vectors, insecticide resistance status and spread can be developed for all countries reporting malaria. The databases are interactive and data can be added to the database as it becomes available. Insecticide resistance maps and dashboards can be created for all countries, according to specific insecticides, resistance intensity and mechanisms of resistance. The changing insecticide resistance profile can be viewed spatially or temporally and provides a platform for countries to share insecticide resistance data as the data becomes available. Data reporting tools are also provided on the Global Malaria Programme.
These datasets provide information on decadal trends in insecticide resistance and enable comparison between countries. The data provided are on an ad hoc basis at a national and site-level scale.
The data can be accessed on the Global Malaria Programme page on the WHO website. The data has a web page which includes useful contextual information as well as additional resources. The data can be viewed in the original report in excel format as it is usually preferable for modelling to have an excel spreadsheet.
Database description
The data contained in the database can be explored and downloaded through the interactive data visualization platform Malaria Threats Map. The platform provides thematic maps showing the status and intensity of insecticide resistance, the status of detection of various resistance mechanisms and the level of involvement of metabolic mechanisms.
The database includes results from bioassays conducted following the standard WHO and CDC procedures for insecticide resistance monitoring in malaria vector mosquitoes with validated, or tentatively validated, discriminating concentrations, as well as results from biochemical and molecular assays. Results of bioassays conducted with non-standard discriminating concentrations are excluded.
Data sources
The data are provided by WHO Member States, as well as partner institutions (including research institutes and nongovernmental organizations), some of which run partner platforms such as Vector Base or IR Mapper and are regularly extracted from publications in international peer-reviewed scientific journals.
Data quality control
Data reported by WHO Member States are checked for validity and consistency at WHO country, regional and headquarters levels before being included in the database. Data from partner institutions are checked at WHO headquarters level.
What does the data look like?
The data is in excel spreadsheets used to generate maps of insecticide distribution as well as dashboard summaries of the country situation.
Description
The database contains results from:
Insecticide susceptibility bioassays (i.e. WHO tube test, WHO bottle bioassay and United States Centers for Disease Control and Prevention [CDC] bottle assays)
Intensity bioassays
Synergist-insecticide bioassays
Biochemical and molecular assays to detect the presence of various resistance mechanisms (i.e. mono-oxygenases, esterases, GSTs, kdr L1014S, kdr L1014F, Ace1R).
The information included typically looks like this. Data are available for each country and annually. Some datasheets contain a few years of data, while others report data for the most recent year. To make a longer time series of data, you can access older databases such as the African Network for Vector Resistance (ANVR) from the WHO. Be careful to ensure that the definitions indicate whether the normal concentrations were tested and whether intensity assays were conducted as well as information on methodology to be sure that data sets can be compared.
Key points to consider
The data include assays conducted with fewer than the minimum recommended number of mosquitoes (i.e. 100 female anopheline mosquitoes tested in ) as in the absence of other data these results are informative despite the small sample size.
Data from WHO susceptibility tests and CDC bottle bioassays are included. However, these measure different outcomes and results are therefore not directly comparable to those from WHO tests.
Mosquito collection durations may span two calendar years; hence, the year indicated for a test result in the database is the year when collection of mosquitoes began for that test.
Insecticide resistance status varies spatially and temporally. Annual national data, provided for summary purposes, are not necessarily reflective of local situations. For decision-making purposes, the most recent data and those at the lowest possible geographical level should be used.
Data quality is dependent on the correct implementation of the standard insecticide resistance test procedures. Ambient conditions such as temperature and humidity, as well as the procedures followed to rear and handle the mosquitoes during the test, may cause significant variations in test results that hamper the comparability of results in time and space. Countries and institutions are encouraged to strictly follow the WHO standard operation procedures for resistance monitoring.
Examples of data use in literature
WHO: Global vector control response 2017-2030. Geneva: World Health Organization; 2017.
Wilson AL, Courtenay O, Kelly-Hope LA, Scott TW, Takken W, Torr SJ, Lindsay SW: The importance of vector control for the control and elimination of vector-borne diseases. PLOS Neglected Tropical Diseases 2020, 14(1):e0007831.
Golding N, Wilson AL, Moyes CL, Cano J, Pigott DM, Velayudhan R, Brooker SJ, Smith DL, Hay SI, Lindsay SW: Integrating vector control across diseases. BMC medicine 2015, 13(1):249.
Dengela D, Seyoum A, Lucas B, Johns B, George K, Belemvire A, Caranci A, Norris LC, Fornadel CM. Multi-country assessment of residual bio-efficacy of insecticides used for indoor residual spraying in malaria control on different surface types: results from program monitoring in 17 PMI/USAID-supported IRS countries. Parasit Vectors. 2018 Jan 30;11(1):71. doi: 10.1186/s13071-017-2608-4. PMID: 29382388; PMCID: PMC5791726.
Bisanzio D, Ally M, Ali AS, Kitojo C, Serbantez N, Kisinza WN, Magesa S, Reithinger R. Modelling Insecticide Resistance of Malaria Vector Populations in Tanzania. Am J Trop Med Hyg. 2022 Jul 5;107(2):308-314. doi: 10.4269/ajtmh.21-0262. PMID: 35895397; PMCID: PMC9393459.
How to use this data
Here is an example illustrating how to download the data, unzip it and read in a sheet. Sometimes the data needs to be prepared before it can be plotted.
Table 1. The pyrethroid resistance intensity data for KwaZulu-Natal in South Africa (2016)
Data has been extracted for KwaZulu-Natal in South Africa looking at pyrethroid resistance in Anopheles arabiensis, the local vector mosquito. Insecticide intensity studies looking at deltamethrin and permethrin showed that there was no significant resistance to Deltamethrin at 5X the discrimination dose and for Permethrin at 10X the discriminating dose. At normal concentration, there is some resistance as indicated by the low levels of mortality. Intensity assays show that this has not translated to operational resistance.
How to plot this data
Figure 1. The distribution of pyrethroid resistance in South Africa.
This is a graphical representation of Table 1. This shows the distribution of pyrethroid resistance in KwaZulu-Natal for 2010-2024. The resistance is to a specific type of pyrethroid, namely, deltamethrin and the vector that was tested was Anopheles arabiensis.
Figure 2. The distribution of pyrethroid resistance on the African continent.
Resistance to has been identified from most countries in Africa but there is a scarcity of data on insecticide resistance intensity. The above figure shows the distribution of the intensity of pyrethroid resistance on the African continent which ranges from low to moderate in the east and south to moderate to high in the west. This map is made up of 1445 bioassays with the majority being in north-western Africa.
How can this data be used in disease modelling?
We use the results of discriminating concentration bioassays in Mozambique to inform the levels of insecticide resistance. These bioassays measure the concentration of insecticide required to differentiate between susceptible and resistant vectors. This is valuable in translating how vector resistance affects transmission potential.
Preparing the data
We calculate the proportion of confirmed resistance to calculate the selection pressure, and the mean mortality across all studies done in Mozambique. The data glossary states that the adjusted mosquito mortality is either the result of applying Abbott’s formula to the average observed mosquito mortality accross all replicates (when control mortality is between 5-20%) or the average observed mosquito mortality across all replicates (when control mortality is below 5%).
Show the code
# Load the data from Mozambiqueinsecticide_data <-read_excel("MTM_DISCRIMINATING_CONCENTRATION_BIOASSAY_20250330.xlsx", sheet =2)# Calculate selection pressure: Proportion of instances of confirmed resistanceres =sum(insecticide_data$RESISTANCE_STATUS =="Confirmed resistance")/nrow(insecticide_data)# Calculate mean mortality insecticide_data_summ <- insecticide_data |>mutate(MORTALITY_ADJUSTED =as.numeric(insecticide_data$MORTALITY_ADJUSTED)) |>group_by(YEAR_START, RESISTANCE_STATUS) |>summarise(mean_mortality =mean(MORTALITY_ADJUSTED), .groups ="drop") |>complete(YEAR_START =full_seq(YEAR_START, 1), RESISTANCE_STATUS) |>mutate(time = (YEAR_START -min(YEAR_START)) *365)# Assign colours from theme health radarstatus_colours <-c("Susceptible"= theme_health_radar_colours[5], "Possible resistance"= theme_health_radar_colours[6], "Confirmed resistance"= theme_health_radar_colours[13],"Undetermined"= theme_health_radar_colours[7])insecticide_data_summ |>ggplot() +geom_point(aes(x = YEAR_START, y = mean_mortality, colour =as_factor( RESISTANCE_STATUS))) +theme_health_radar() +scale_colour_manual(values = status_colours) +scale_y_continuous(limits =c(0, NA)) +labs(title ="Trend of mosquito mortality in Mozambique",subtitle ="Results from MTM discriminating concentration bioassay",x ="Year",y ="Adjusted mosquito mortality (%)",colour ="Resistance status",caption =str_wrap("We see that more mosquitoes survive in bioassays of confirmed resistance in Mozambique over time, indicating a worsening resistance profile. In 2021, mortality dropped to just 40%, highlighting a significant decline in susceptibility. Possible resistance remains somewhat stable (about 92% mosquito mortality). Source: Malaria Threats Map"))
We incorporate this data into the model in two ways:
Selection Pressure, as the proportion of vectors that start as resistant due to prior selection by insecticide use. From the data we find that in 115 of the 529 bioassays conducted resistance was confirmed, and set the selection pressure \(res\) at 21%.
Adjusting the effectiveness of interventions, \(effr_{level}\) to represent the reduced effectiveness of an intervention such as long-lasting insecticidal nets (LLINs) against resistant mosquitoes.
In a more complex model you may also have different rates of mosquito mortality \(\mu_m\), susceptibility to infection \(b\), or biting rates \(a\) (due to fitness costs) for insecticide-sensitive and resistant vectors.
In this simple model example, we assume the coverage of LLINs to be constant at 75%, further reduced by the usage rate. A more detailed example of the implementation of LLINs (including effectiveness, attrition and insecticidal decay) into a transmission model can be found here. We demonstrate the impact of vector insecticide resistance on the effective coverage of LLINs.
Show the code
# Time points for the simulationY =14# Years of simulationtimes <-seq(0, 365*Y, by =1)# Estimates of resistance levels over timeresdata <- insecticide_data_summ |>filter(RESISTANCE_STATUS =="Confirmed resistance") |>mutate(mean_mortality = mean_mortality/100) # convert into decimal# Make a time dependent variable of resistance levelseffr_level <-approxfun(resdata$time, resdata$mean_mortality, n =365*Y, ties ="ordered", method ="constant", rule =2)# SEACR-SEI modelseacr <-function(times, start, parameters, effr_level) { with(as.list(c(start, parameters)), { P = S + E + A + C + R + G M_s = Sm_s + Em_s + Im_s M_r = Sm_r + Em_r + Im_r M = M_s + M_r m = M / P# Seasonality seas.t <- amp*(1+cos(2*pi*(times/365-phi)))# Nets itn <- itn_cov*itn_use itn_r <- itn*(effr_level(times)) # Adjust effective coverage of ITNs based on resistance level# Force of infection Infectious = C + zeta_a*A # infectious reservoir lambda.v <- seas.t*a*M/P*b*(Im_r*(1-itn_r) + Im_s*(1-itn))/M lambda.h <- seas.t*a*c*Infectious/P*(1-itn)# Differential equations/rate of change# Insecticide-sensitive mosquito compartments dSm_s = (1-res)*mu_m*M - (lambda.h)*Sm_s - mu_m*Sm_s dEm_s = (lambda.h)*Sm_s - (gamma_m + mu_m)*Em_s dIm_s = gamma_m*Em_s - mu_m*Im_s# Insecticide-resistant mosquito compartments dSm_r = res*mu_m*M - (lambda.h)*Sm_r - mu_m*Sm_r dEm_r = (lambda.h)*Sm_r - (gamma_m + mu_m)*Em_r dIm_r = gamma_m*Em_r - mu_m*Im_r# Human compartments dS = mu_h*P - lambda.v*S + rho*R - mu_h*S dE = lambda.v*S - (gamma_h + mu_h)*E dA = pa*gamma_h*E + pa*gamma_h*G - (delta + mu_h)*A dC = (1-pa)*gamma_h*E + (1-pa)*gamma_h*G - (r + mu_h)*C dR = delta*A + r*C - (lambda.v + rho + mu_h)*R dG = lambda.v*R - (gamma_h + mu_h)*G dCInc = lambda.v*(S + R)# Outputlist(c(dSm_s, dEm_s, dIm_s, dSm_r, dEm_r, dIm_r, dS, dE, dA, dC, dR, dG, dCInc), itn=itn, itn_r=itn_r) })}# Initial values for compartmentsinitial_state <-c(Sm_s =5000000, # susceptible insecticide-sensitive mosquitoesEm_s =3000000, # exposed and infected insecticide-sensitive mosquitoesIm_s =1000000, # infectious insecticide-sensitive mosquitoesSm_r =4000000, # susceptible insecticide-resistant mosquitoesEm_r =2000000, # exposed and infected insecticide-resistant mosquitoesIm_r =1000000, # infectious insecticide-resistant mosquitoesS =3500000, # susceptible humansE =350000, # exposed and infected humansA =1300000, # asymptomatic and infectious humansC =650000, # clinical and symptomatic humansR =100000, # recovered and semi-immune humansG =100000, # secondary-exposed and infected humansCInc =0# cumulative incidence in humans infected by insecticide resistant mosquitoes)# Country-specific parameters should be obtained from literature review and expert knowledgeparameters <-c(a =0.3, # human biting rateb =0.4, # probability of transmission from mosquito to humanc =0.4, # probability of transmission from human to mosquitor =1/7, # rate of loss of infectiousness after treatmentrho =1/160, # rate of loss of immunity after recoverydelta =1/200, # natural recovery ratezeta_a =0.4, # relative infectiousness of of asymptomatic infectionspa =0.35, # probability of asymptomatic infectionmu_m =1/10, # birth and death rate of mosquitoesmu_h =1/(62*365), # birth and death rate of humansgamma_m =1/10, # extrinsic incubation rate of parasite in mosquitoesgamma_h =1/10, # extrinsic incubation rate of parasite in humansres =0.2174, # selection pressure, obtained from dataamp =0.5, # amplitude of seasonalityphi =200, # phase angle, start of seasonitn_use =0.58, # probability of sleeping under a netitn_cov =0.75# coverage of LLINs in the population at risk )# Run the modelout <-ode(y = initial_state, times = times, func = seacr, parms = parameters,effr_level = effr_level)# Post-processing model output into a dataframedf <-as_tibble(as.data.frame(out)) |>pivot_longer(cols =-time, names_to ="variable", values_to ="value") |>mutate(date =ymd("2010-01-01") + time)df |>filter(variable %in%c("itn", "itn_r"), date >"2010-01-01") |>mutate(`Insecticide Status`=case_when( variable =="itn"~"Sensitive", variable =="itn_r"~"Resistant")) |>ggplot(aes(x = date, y = value, colour =`Insecticide Status`)) +geom_line() +scale_y_continuous(labels = scales::percent, limits =c(0, 0.6)) +labs(x ="Year", y ="Effective coverage", title ="Effective coverage of LLINs in Mozambique 2010 to 2023",subtitle ="Changes due to confirmed resistance",caption =str_wrap("In this example plot, transmission from insecticide-sensitive mosquitoes is consistently protected due to the effective coverage of LLINs. However, when confirmed insecticide resistance increases, and more mosquitoes survive the MTM discriminating bioassays, LLIN effectiveness decreases, as is seen in 2012, 2020 and 2021. Increases in mosquito mortality in 2014 led to an increase in the effective coverage of LLINs. Source: Model output.")) +theme_health_radar()
Policy implications
The direct relationship between insecticidal efficacy and LLIN impact is shown in the example above. Declining effectiveness due to resistance highlights the urgent need for insecticide resistance management to maintain the returns on investment in vector control. This may lead to changes in the active ingredient in the LLINs used, rotations of the insecticide classes in IRS campaigns or subnational adaptations to vector control strategies. Ultimately insecticide resistance compromises the effectiveness of core interventions and threatens progress toward malaria elimination goals.
