Translating Mosquito Insecticide Resistance into Operational Decisions: A Critical Review of Bioassays, Mechanisms, Thresholds, and Control Failure

INTRODUCTION

Mosquito insecticide resistance has become a geographically extensive but biologically heterogeneous challenge for vector control. Resistance phenotypes occur across major Aedes vectors and African malaria-vector populations, but their prevalence, intensity, mechanisms, and temporal trajectories vary among species and settings [1, 2]. This heterogeneity matters because a resistant classification does not represent a uniform biological state. The same mortality result may arise from different combinations of target-site alteration, detoxification, reduced penetration, physiological condition, or assay exposure, while populations carrying apparently similar mechanisms may experience different operational insecticide doses.

The central interpretive problem is therefore not whether resistance exists, but what a particular measurement establishes. Diagnostic-dose mortality can indicate that a population no longer responds like a susceptible reference population under a specified protocol. It does not directly quantify the strength of resistance, identify the responsible mechanism, predict the efficacy of a particular formulation, or determine whether an intervention will fail to protect people. These are distinct constructs situated at different biological and operational scales. Treating them as interchangeable compresses uncertainty and encourages decisions that are more precise than the evidence permits.

Field evidence reinforces this non-equivalence. Long-lasting insecticidal nets may retain protective value in areas where phenotypic resistance is detected, while their effectiveness may also be weakened by resistance in combination with poor physical condition, inconsistent use, inadequate coverage, altered vector behaviour, ecological change, or declining insecticide availability. Field effectiveness therefore cannot be inferred from bioassay mortality alone because net use, physical integrity, coverage, mosquito behaviour, and transmission ecology jointly shape outcomes [3, 4]. Conversely, preserved epidemiological protection in one context does not show that resistance is operationally unimportant elsewhere or that increasing resistance intensity will remain inconsequential.

This review develops a decision-centred interpretation of mosquito insecticide-resistance evidence. It critically compares bioassay constructs, target-site, metabolic, behavioural, and cuticular mechanisms, resistance intensity, operational exposure, product performance, and control outcomes. Its central argument is that translation from laboratory phenotype to operational action requires an explicit chain of evidence. Bioassay resistance is not equivalent to operational control failure; resistance frequency is not equivalent to resistance intensity; a detected mechanism is not equivalent to its field-level contribution; and laboratory exposure is not equivalent to the dose, behaviour, coverage, and environment of operational use.

What mosquito resistance bioassays measure

A resistance bioassay measures mosquito survival or mortality after a defined experimental exposure. Its result is conditional on the insecticide, concentration, exposure duration, delivery surface, route of uptake, post-exposure observation period, mosquito species, age, physiological state, and handling conditions. Diagnostic-dose assays are useful for identifying reduced susceptibility relative to a standardized expectation, whereas escalating-dose assays probe the strength of survival beyond that diagnostic exposure. Escalating-dose intensity assays consequently add information that frequency-based classification cannot provide, but their categories remain laboratory phenotypes rather than direct epidemiological thresholds [5]. A population with many survivors at the diagnostic dose may contain resistance of different strengths, and identical resistance frequencies can conceal markedly different dose–response relationships.

Assay design also determines the exposure construct being measured. Tests based on treated papers, coated bottles, direct topical application, or contact with insecticidal netting differ in how insecticide reaches the mosquito and in the uniformity of the delivered dose. Alternative contact formats can reveal effects of long-lasting insecticidal nets that may be underestimated or obscured by short standardized exposures [6]. Similarly, comparisons among the Centers for Disease Control and Prevention bottle assay, the World Health Organization tube assay, and topical application demonstrate differences in delivered exposure and mortality variability; their classifications should not be treated as interchangeable simply because all report mortality [7]. Agreement between methods may strengthen confidence in a phenotype, but disagreement may reflect assay geometry, dose distribution, absorption route, or experimental variability rather than a biological contradiction.

Interpretation is further weakened when mortality percentages are converted into coarse categories without retaining dose–response shape or uncertainty. Continuous probabilistic analysis can preserve information about the estimated dose–mortality relationship and the uncertainty surrounding resistance intensity, whereas broad low, moderate, or high labels may imply boundaries that have not been calibrated to product performance or disease outcomes [8]. A bioassay result should therefore be reported as a method-specific observation, not as an autonomous decision rule. At minimum, interpretation requires the assay protocol, delivered exposure, mosquito characteristics, replication, uncertainty, and the operational question to which the measurement is being applied. The evidence dimensions and interpretive boundaries for mosquito resistance bioassays measure are summarized in Table 1.

Table 1. What Mosquito Resistance Bioassays Measure: Vector Systems, Biological Mechanisms, Exposure Pathways, Evidence Requirements, Uncertainty, and Interpretive Boundaries

Target-site, metabolic, behavioural, and cuticular mechanisms

Target-site resistance modifies the mosquito protein upon which an insecticide acts, thereby reducing toxicodynamic sensitivity. Even within this apparently discrete category, interpretation is not straightforward. Population-genomic evidence in African Anopheles gambiae and Anopheles coluzzii shows that target-site resistance is structured by multiple variants and haplotypes whose phenotypic meanings depend on species, genetic background, and geography [9]. Allele frequency can therefore support mechanistic surveillance, but it is not a substitute for phenotypic testing, resistance-intensity measurement, or product evaluation. A frequent allele may be weakly predictive in one background and more consequential in another, particularly when metabolic or penetration mechanisms co-occur.

Metabolic resistance acts before sufficient insecticide reaches its target, commonly through increased detoxification or sequestration. Functional evidence demonstrates that particular cytochrome P450 alleles can increase pyrethroid survival and reduce mortality caused by pyrethroid-treated nets [10]. Such evidence provides a stronger causal chain than expression data alone because it connects a defined genetic variant to phenotype and product response. Nevertheless, the field-level contribution of a validated allele remains population- and product-dependent. Multiple detoxification enzymes may be co-expressed, substrate specificity may differ among active ingredients, and apparent metabolic resistance may interact with target-site or cuticular processes. Mechanism detection consequently indicates biological plausibility, not a universal or quantitatively fixed contribution to control loss.

Cuticular and behavioural processes further complicate mechanism attribution. Reduced penetration can delay insecticide uptake while enzymatic detoxification removes part of the absorbed dose, allowing the two pathways to act jointly rather than independently [11]. Measuring only enzyme activity may consequently overattribute resistance to metabolism, whereas measuring cuticular change without uptake kinetics may not establish its functional importance. Behaviour operates at a different point in the exposure pathway by altering whether, when, where, and for how long mosquitoes contact treated surfaces. Physiological resistance and shifts in biting or resting behaviour have been observed together after long-lasting insecticidal-net deployment, but temporal co-occurrence does not identify a single causal mechanism [12]. Environmental change, host availability, vector-species turnover, or human behaviour may produce similar patterns. Forced-contact bioassays cannot measure this behavioural component of operational exposure.

Resistance intensity and operational exposure

Resistance frequency describes the proportion of mosquitoes surviving a specified diagnostic exposure; resistance intensity describes how survival changes as exposure increases. The distinction is operationally important because two populations with similar diagnostic-dose mortality may differ in their susceptibility at higher concentrations and in their response to particular formulations. In Accra, high pyrethroid-resistance intensity occurred alongside low mortality in tests of pyrethroid-only long-lasting insecticidal nets, providing a coherent entomological warning signal [13]. However, this concurrence does not independently establish loss of population protection. The evidence remains specific to the tested vector population, products, exposure methods, and ecological setting, and it does not quantify intervention use, coverage, physical barrier effects, or disease outcomes.

Intensity is also dynamic rather than a fixed property of a species or region. Surveillance across Kinshasa and other provinces of the Democratic Republic of Congo showed substantial variation among insecticides, locations, and survey periods, demonstrating that diagnostic-dose frequency cannot stand in for a stable or spatially uniform estimate of resistance strength [14]. Variation may represent different selection histories, vector composition, mechanism combinations, or sampling conditions. Mosquito physiological state further modifies observed intensity: age and blood feeding can change apparent resistance and subsequent longevity [15]. Comparisons made with non-equivalent age structures, feeding states, rearing histories, or environmental conditions may therefore misclassify biological differences as geographic or temporal changes in resistance.

Operational exposure is the insecticide dose actually encountered by a mosquito during intervention use. It depends on contact probability, duration, route, repetition, product condition, active-ingredient availability, household use, coverage, and mosquito behaviour. Laboratory intensity assays deliberately standardize these factors to improve comparison, but standardization does not create equivalence with field contact. Longitudinal evidence from Malawi linked increasing resistance intensity partly to greater expression of P450 alleles and declining entomological performance of tested nets, while leaving the magnitude of population-level epidemiological effect unresolved [16]. The defensible interpretation is therefore exposure–response based: resistance intensity identifies the strength of survival under specified laboratory challenge, whereas operational significance emerges only when that phenotype is connected to product-specific dose, natural contact, intervention condition, coverage, vector ecology, and an explicitly defined control objective.

From laboratory phenotype to control failure

Translation from a resistant laboratory phenotype to operational failure requires several evidential transitions. The phenotype must first affect the performance of a relevant product under a realistic contact pathway; the resulting reduction in killing, deterrence, or feeding inhibition must then persist at the coverage achieved by the programme; and that change must be sufficient to alter an explicitly defined entomological or epidemiological objective. In Kinshasa, reduced net-induced mortality co-occurred with high Plasmodium infection in resistant vectors, a concerning field pattern that does not alone establish a causal resistance-to-transmission pathway [17]. Figure 1 integrates the sequential evidence transitions linking laboratory resistance measurements, interacting biological mechanisms, realized operational exposure, product performance, and the probability that a vector-control objective will not be achieved.

Product-specific evidence can strengthen the translation chain without completing it. In southeast Côte d’Ivoire, poor conventional-net bioefficacy was associated with intense pyrethroid resistance and overexpression of several P450 genes, whereas next-generation nets retained greater activity [18]. In Mozambique, escalating P450-associated resistance reduced mortality even to a piperonyl-butoxide–pyrethroid net, demonstrating that a product intended to mitigate one resistance pathway can be overtaken by sufficiently strong or complex resistance [19]. These findings connect phenotype, mechanism, and product response, but they remain entomological evidence. They do not independently establish inadequate household use, reduced community coverage, increased human exposure, or higher disease incidence.

Epidemiological evidence can also diverge from entomological warning signals. Health-facility data from Benin did not reveal a simple association between standard phenotypic-resistance frequency and clinical malaria incidence, underscoring the limits of a single-marker failure rule [20]. Such a result should not be interpreted as evidence that resistance is operationally harmless; routine-data quality, ecological linkage, intervention coverage, immunity, healthcare access, climate, and vector composition can obscure an effect. Control failure should instead be defined as failure to achieve a specified objective under documented implementation conditions, with evidence identifying where the chain from product contact to population protection weakened. The evidence dimensions and interpretive boundaries for from laboratory phenotype to control failure are summarized in Table 2.

Table 2. From Laboratory Phenotype to Control Failure: Vector Systems, Biological Mechanisms, Exposure Pathways, Evidence Requirements, Uncertainty, and Interpretive Boundaries

Decision thresholds and evidence interpretation

A decision threshold is a rule linking evidence to an action, not a universal biological constant. The consequences of acting too early or too late, the availability of alternative products, the reversibility of the decision, and the uncertainty of the evidence should determine how much support is required. Decision rules should therefore distinguish technical resistance detected under standardized exposure from practical resistance that materially changes control performance in a defined programme context [21]. A diagnostic-dose result may appropriately trigger confirmation or intensified surveillance without automatically triggering product withdrawal.

Threshold interpretation also depends on the intervention being evaluated. A mortality value obtained from an insecticidal net can reflect net sampling, washing, storage, manufacturing variation, active-ingredient migration to the surface, textile condition, and mosquito susceptibility. These factors may change during transport and field use, so product and vector measurements must be interpreted jointly [22]. Separate thresholds are consequently needed for assay validity, resistance alert, product investigation, operational action, and post-decision review. Collapsing these functions into one mortality cutoff hides the evidential transition between detection and action.

Molecular and model-based indicators can extend surveillance but do not remove the need for calibration. Target-site allele-frequency maps can support anticipatory surveillance, although their operational meaning remains conditional on vector species, genetic background, phenotype, product, and local exposure [23]. Data-informed deployment models can combine resistance, ecology, intervention efficacy, cost, and resource constraints, but their rankings remain dependent on structural assumptions and setting-specific inputs [24]. Decision thresholds should therefore be tiered: an alert threshold initiates verification, an investigation threshold triggers product and exposure assessment, and an action threshold requires evidence that the expected benefit of changing strategy exceeds its costs and risks.

Critical methodological synthesis

The most consistent finding across the evidence base is non-equivalence between resistance detection and operational failure. Bioassays, molecular markers, product tests, field entomology, and epidemiological outcomes measure different parts of the causal chain. Across synthesis and cluster-randomized evidence, piperonyl-butoxide–pyrethroid nets can improve entomological or epidemiological outcomes in high-resistance settings, but the magnitude and persistence of benefit vary by product, setting, coverage, and time since deployment [25–27]. These findings demonstrate that mechanism-aware products can restore part of the effect lost to resistance; they do not establish that one product class is universally superior or evolutionarily durable.

Comparative intervention evidence further shows that next-generation nets cannot be treated as a homogeneous category. A four-arm trial in Tanzania found that some dual-active-ingredient nets improved malaria outcomes and cost-effectiveness relative to pyrethroid-only nets [28]. The inference is product- and setting-specific because active ingredients differ in mode of action, resistance vulnerability, wash durability, manufacturing characteristics, price, and dependence on sustained household use. A positive result for one combination cannot be transferred automatically to another combination or to a population with a different mechanism profile.

Time since deployment is a central but often undermeasured modifier. Third-year follow-up of a cluster-randomized trial showed that the advantage of piperonyl-butoxide nets weakened as net use and piperonyl-butoxide content declined [29]. This finding illustrates why laboratory susceptibility cannot be isolated from operational exposure. Textile deterioration, declining surface availability of active ingredients, loss of synergist, net attrition, inconsistent use, and replacement practices may alter effectiveness even when the underlying mosquito phenotype is unchanged.

The strongest supported inference is therefore conditional rather than threshold-based: resistance becomes operationally important when a defined phenotype and mechanism reduce the performance of a relevant product under realized exposure sufficiently to compromise a specified programme objective. The unresolved question is which combinations of assay phenotype, intensity, mechanism, product condition, contact behaviour, coverage, and ecology best predict that transition. Comparative models can help organize these variables, but their recommendations remain dependent on local inputs and assumptions [24]. The convergent findings, context-dependent results, methodological limitations, and remaining uncertainties are synthesized in Table 3.

Table 3. Critical Methodological Synthesis: Convergent Findings, Context Dependence, Methodological Limitations, Evidence Confidence, and Residual Uncertainty