Comparison of cone bioassay estimates at two laboratories with different Anopheles mosquitoes for quality assurance of pyrethroid insecticide-treated nets

Abstract

Background: Quality assurance (QA) of insecticide-treated nets (ITNs) delivered to malaria-endemic countries is conducted by measuring physiochemical parameters, but not bioefficacy against malaria mosquitoes. This study explored utility of cone bioassays for pre-delivery QA of pyrethroid ITNs to test the assumption that cone bioassays are consistent across locations, mosquito strains, and laboratories.

Methods: Double-blinded bioassays were conducted on twenty unused pyrethroid ITNs of 4 brands (100 nets, 5 subsamples per net) that had been delivered for mass distribution in Papua New Guinea (PNG) having passed predelivery inspections. Cone bioassays were performed on the same net pieces following World Health Organization (WHO) guidelines at the PNG Institute of Medical Research (PNGIMR) using pyrethroid susceptible Anopheles farauti sensu stricto (s.s.) and at Ifakara Health Institute (IHI), Tanzania using pyrethroid susceptible Anopheles gambiae s.s. Additionally, WHO tunnel tests were conducted at IHI on ITNs that did not meet cone bioefficacy thresholds. Results from IHI and PNGIMR were compared using Spearman's Rank correlation, Bland-Altman (BA) analysis and analysis of agreement. Literature review on the use of cone bioassays for unused pyrethroid ITNs testing was conducted.

Results: In cone bioassays, 13/20 nets (65%) at IHI and 8/20 (40%) at PNGIMR met WHO bioefficacy criteria. All nets met WHO bioefficacy criteria on combined cone/tunnel tests at IHI. Results from IHI and PNGIMR correlated on 60-min knockdown (KD60) (r_s = 0.6,p = 0.002,n = 20) and 24-h mortality (M24) (r_s = 0.9,p < 0.0001,n = 20) but BA showed systematic bias between the results. Of the 5 nets with discrepant result between IHI and PNGIMR, three had confidence intervals overlapping the 80% mortality threshold, with averages within 1-3% of the threshold. Including these as a pass, the agreement between the results to predict ITN failure was good with kappa = 0.79 (0.53-1.00) and 90% accuracy.

Conclusions: Based on these study findings, the WHO cone bioassay is a reproducible bioassay for ITNs with > 80% M24, and for all ITNs provided inherent stochastic variation and systematic bias are accounted for. The literature review confirms that WHO cone bioassay bioefficacy criteria have been previously achieved by all pyrethroid ITNs (unwashed), without the need for additional tunnel tests. The 80% M24 threshold remains the most reliable indicator of pyrethroid ITN quality using pyrethroid susceptible mosquitoes. In the absence of alternative tests, cone bioassays could be used as part of pre-delivery QA.

Keywords: Anopheles; Bioassay; Bioefficacy; Cone bioassay; ITN; Insecticide treated nets; LLIN; Long lasting insecticidal nets; Malaria; Mosquito; Pyrethroid; Quality assurance; Tunnel test.

Fig. 1

Rectangular whole net with five sides; net piece samples were cut from bottom side (A), middle side A (B), roof (C), middle side B (D) and top side (E)

Fig. 2

WHO cones fixed on plastic cone board held at 60° in IHI (A) and at 45° in PNGIMR (B)

Fig. 3

Correlation of cone bioassay tests results between IHI and PNGIMR testing facilities. Thick dashed lines are the WHO threshold 95% KD60 (A) and 80% M24 (B). Thin dashed lines indicate these assay-inherent 95% (lower) CIs of these thresholds. Large dots represent averages per sampled nets (4 per net type) and small dots represent all subsamples (5 per net)

Fig. 4

Bland-Altman Plot showing the mean difference (y axis) plotted against the average value from both sites (x) A KD60 and B M24. For KD60 Mean difference (limits of agreement) was 15.5 (−25.4–56.5) and for M24 Mean difference (limits of agreement) was −17.0 (−61.4–27.3). At lower mean values of knockdown, the agreement between the two testing facilities was lower than at higher mean values of knockdown but there was a consistent difference in mean difference in M24 measured at each testing facility

Fig. 5

Bioefficacy of the five ITNs that demonstrated discordant results of pass or fail between facilities. Each ITN passed efficacy criteria in IHI using the bioefficacy criterion of 95% KD60 A but did not reach the optimal bioefficacy criterion of 80% 24 h mortality B. Three of the nets showed mean 24 h mortality close to 80% at PNGIMR with confidence intervals that overlapped the optimal bioefficacy threshold of 80% mortality B. Dashed lines are the WHO thresholds for 95% KD60 and 80% M24

Fig. 6

Correlation between M24 and KD60 at IHI A and PNGIMR B. ITNs passing (green) or failing (red) based on stringent cut-off WHO cone bioassay criteria of 80% M24 and 95% KD60. ‘Borderline’ nets for which the mean KD or M24 values are within the margin of stochastic error (95% CI) inherent to WHO cone bioassays based on the total number of mosquitoes used (n = 100) are shown in amber. Thick dashed lines are the WHO thresholds 95% KD60 and 80% M24. Thin dashed lines indicate these assay-inherent 95% (lower) CIs of these thresholds

Fig. 7

Results from the literature review A Relationship between KD60 and M24 in WHO cone bioassays with pyrethroid ITNs (deltamethrin, alpha-cypermethrin and permethrin) using Anopheles mosquitoes. B Relationship between KD60 and M24 in ITNs grouped by production technology. Dashed lines are the WHO threshold 95% KD60 and 80% M24