**4. Discussion**

IRS is a widely applied vector control intervention. However, relatively little attention is given to the assessment of application rates whether at the level of control programmes and communities, or at the level of households. Tools aimed at improving consistency of application on a community level, such as the IK Smart Light [17], are being developed but are not ye<sup>t</sup> widely deployed. Apart from the implications for IRS campaign success, consistent delivery of insecticidal products is important when conducting experimental hut trials, which are reliant on the well-defined application rate of IRS products. In this study, we compared an automated track spray system for IRS to a well-trained human spray operator, to discern whether the track sprayer delivered a more consistent spray such that use in experimental hut trials could be implemented.

Large variation in spray deposits, measured by the amount of insecticide applied to a filter paper, have previously been reported for the three insecticides tested in this study, ranging from 0.31 to 3.78 times the recommended dose for Actellic 300CS [18], between 0.63 and 1.37 times the dose for Vectron T500 [12,13], and between 0.80 and 1.32 for Fludora Fusion [19,20]. A large proportion of this variation could be removed by using an automated spraying process, such as shown here with the Micron track sprayer. Analysis of the overall wall spray pattern using both fluorescein and IRS products demonstrated markedly less variation in the automated than in the manual spray. This confirms that a significant proportion of variation in spray application could be eliminated. Improved accuracy in insecticide application in trials would lead to more robust data and thus better-informed product development decisions, which could avoid unnecessary delays in bringing new products to the market.

With the exception of the study with BRF, the track sprayer resulted in a higher median concentration sprayed than manual spraying. For the insecticidal products, results were also compared to the target dose and showed underdosing with both spray methods. Although this underdosing was only significant for the track sprayer, it is likely that the

underdosing was a factor for both methods, whilst the larger variation in manual spraying masked the difference between actual and target dose. Underdosing can occur if the speed of spraying is too fast, i.e., not enough liquid is deposited on the filter papers, or if the distance from the wall is too large. As both the distance and the speed are regulated for the track sprayer, it is unlikely that these factors caused the lower-than-expected application rate on the filter papers. It is recommended that in future studies, the track sprayer is used in conjunction with enhanced filter paper analysis so that any differences in the dose delivered compared to the target dose can be identified and evaluated.

To correct for potential deviations from target dose in the spray solution prior to spraying, we compared the sprayed filter papers to the concentration in liquid samples taken from the spray solution. Theoretically, the amount of insecticide sprayed onto filter papers would reflect the concentration of insecticide in the spray tank, assuming the spray nozzle moves at a constant speed up or down a swath and that the distance of the nozzle from the wall is also constant. Analysis of the spray solution can indicate dilution or mixing errors, such as adding too much or too little water or product to the spray tank or not shaking the spray tank to thoroughly mix the product with the water before and during spraying. We found that the concentration of PMM in the spray tank was considerably lower than expected, which may explain the lower-than-expected concentration found on filter papers with both spray methods.

Analysis of fluorescein values stratified by horizontal swath position showed that the 8002 nozzle used did not provide a consistent application rate across the horizontal swath, with less fluorescein being deposited in the centre of a swath. This difference was significant in the track sprayer deposits, but less apparent in the manual spray deposits. Similar to the results with fluorescein, the insecticide dose sprayed with the track sprayer was lower but more consistent in the centre position compared to positions at the edges of the swaths. Whilst the same nozzle was used for the track and manual spraying, different nozzles were used for fluorescein spray and each of the insecticidal products, making the possibility that this result could be an artefact of individual nozzles less likely.

We discerned a difference in spray rhythms in upward and downward swaths with manual spraying, even when the overall swath spray times were consistent between the two directions. The study was conducted using only one spray operator for each experiment and did not have the aim of characterising the entire range of variation that might be present during manual spraying. However, it is interesting to note that, even with expert training, and when keeping to the overall requirement of spraying 2 metres per 5 s, differences in rhythm can exist that lead directly to inconsistent spray application.

Height position analysis demonstrated significant differences in the amount of fluorescein applied between different wall height positions in the manual spray, particularly in the lower half of the wall. This aligns with the observation that spray operators tend to move the spray lance slowly at the top of the swath and then speed up towards the bottom. Although one significant difference was detected in fluorescein applied by the track spray between the uppermost and lowermost wall heights, no significant differences were seen when the overlap positions were also included, demonstrating that the track sprayer delivers a much more consistent spray application than the manual spray. Variation in the amount of insecticide sprayed manually was generally larger per vertical position compared to the track sprayer, but the trend of decreasing concentrations with lower wall heights was not shown when spraying with insecticides.

Overall, the track sprayer delivered a more consistent deposit of spray solution, making it a potential methodological improvement to experimental hut evaluations of novel IRS formulations.
