Decreased bioefficacy of long-lasting insecticidal nets and the resurgence of malaria in Papua New Guinea

The study investigates whether reduced bioefficacy of long-lasting insecticidal nets (LLINs) contributed to the resurgence of malaria in Papua New Guinea (PNG) after 2015. PNG accounts for over 80% of malaria cases in the WHO Western Pacific Region, and national LLIN distribution (PermaNet 2.0, deltamethrin-treated) began in 2006 with 12.8 million nets distributed between 2010 and 2019. Initial distributions coincided with a marked decline in malaria prevalence (from 15.7% in 2008/09 to <1% in 2013/14), but prevalence rose to 7.1% by 2016/17. Potential drivers include antimalarial drug stock-outs, vector behavior (exophagy, opportunistic host selection, earlier evening biting), increased nocturnal human activity due to electrification, and concerns about insecticide resistance; however, regular monitoring has shown no evidence of emerging pyrethroid resistance in PNG Anopheles populations. Previous work (2000–2009) showed PermaNet 2.0 maintained high bioefficacy even after prolonged use. The present study tests the hypothesis that LLIN bioefficacy prior to use declined in nets manufactured from 2013 onward, potentially contributing to the resurgence.

Prior studies in PNG (Katusele et al.) indicated PermaNet 2.0 LLINs remained highly effective, killing nearly 100% of mosquitoes in WHO cone bioassays even after 5 years of use, with only slight declines after >7 years. Comparable African studies from the same era reported PermaNet 2.0 met or exceeded WHO requirements. LLINs should achieve ≥80% 24 h mortality or ≥95% 60 min knockdown within 3 years of use per WHO guidelines. Reports exist of substandard LLIN distributions in other countries (Rwanda 2015; Solomon Islands 2014). While pyrethroid resistance reduces LLIN efficacy in various settings, monitoring in PNG Anopheles has not detected resistance to deltamethrin or other tested insecticides. Some more recent reports are mixed: PermaNet 2.0 from 2017 in northern Tanzania reportedly met requirements, whereas a report from Iran suggested failure. The literature underscores the importance of bioefficacy testing and potential impacts of resistance and manufacturing differences.

Study design: Cross-sectional bioefficacy testing of PermaNet 2.0 LLINs using WHO cone bioassays on pyrethroid-susceptible Anopheles mosquitoes, including new/unused LLINs (n = 192) across manufacturing years 2007–2019 and used LLINs (n = 40) with owner-reported duration of use. Sampling: Unused LLINs manufactured in 2018–2019 (n = 49) were provided by Rotarians Against Malaria (RAM) PNG from consignments for different provinces. Unused LLINs from 2007–2017 (n = 143) were obtained from villages or provincial health authorities; all unused nets were in original unopened packaging. In total, 192 unused LLINs from 78 batches intended for or distributed to 15 PNG provinces were tested. Used LLINs (n = 40) were collected from communities in Madang and Gulf Provinces in 2018–2019; owners reported duration of use (1–3 years vs >3 years). Mosquito strains: Tests used fully pyrethroid-susceptible Anopheles farauti (PNGIMR laboratory colony originally from Rabaul, periodically back-crossed with local An. farauti; last in 2012) and locally reared wild An. farauti from Madang Province. WHO tube assays confirmed susceptibility to deltamethrin, lambda-cyhalothrin, DDT, bendiocarb, and malathion. For confirmatory tests, An. gambiae s.s. G3 (standard susceptible strain) was used at Liverpool School of Tropical Medicine (LSTM). WHO cone bioassay procedures: Conducted per WHO guidelines. Mosquitoes aged 3–5 days; 5 mosquitoes per cone; 3-minute exposure. For unused LLINs, one side panel section (30×30 cm) was tested per net with 5 cones (25 mosquitoes per LLIN). For used LLINs, 4 cones were used (20 mosquitoes per LLIN). Ambient tropical conditions in Madang met temperature and humidity requirements. Positive controls were 2012-manufactured LLINs known to yield 100% 24 h mortality; negative controls were untreated netting. Assays were invalidated if negative control 24 h mortality exceeded 10%. Abbott’s formula was applied when negative control mortality was >0% and ≤10%. Outcomes recorded were 60 min knockdown (KD60min) and 24 h mortality (M24h). LSTM confirmatory assays used 10 cones (50 mosquitoes) per LLIN with similar SOPs. Storage temperature assessment: Temperatures inside a shipping container filled with LLIN bales in Port Moresby were logged over 5 days at 4 locations (immediately beneath ceiling; center of topmost bale layer; beneath topmost bale layer; center of container) using USB data loggers. Heat-stress simulation: three LLINs (manufactured 2012, confirmed 100% bioefficacy) were stored at 60 °C in an oven for 6 weeks; bioassays were conducted weekly to assess any changes. Data handling and analysis: Data captured via Epicollect 5; analyzed using Microsoft Excel 2016 and GraphPad Prism 8. Main endpoints were proportions for M24h and KD60min, with exact (Clopper-Pearson) 95% CIs. Summary statistics include medians and IQRs for proportion plots. Correlations used Pearson and linear regression with 95% confidence bands. Group comparisons (e.g., 2007–2012 vs 2013–2019) used chi-squared tests.

- New, unused LLINs manufactured 2007–2012 (n = 25) largely achieved WHO thresholds: 84.2% (21/25) had 100% 24 h mortality; all 25 met either ≥80% 24 h mortality or ≥95% 60 min knockdown. - New, unused LLINs manufactured 2013–2019 (n = 167) showed markedly reduced bioefficacy: only 17% met WHO criteria (≥80% 24 h mortality or ≥95% KD60min). Mean KD60min and M24h across 2013–2019 were 41.23% (95% CI 39.74–42.73) and 40.12% (95% CI 38.65–41.62), respectively, versus 96.48% KD60min and 98.72% M24h in 2007–2012. - Year-specific averages for new LLINs (examples): 2013 M24h 48.20%; 2014 M24h 28.89%; 2016 M24h 20.00%; 2018 M24h 58.16%; 2019 M24h 39.45% (see Table 1 summary within article). - Statistical significance: The difference between 2007–2012 and 2013–2019 groups was large and statistically significant (Chi-squared = 75.4; p < 1e-5; z = 8.442). - Batch analysis: 78 distinct batches were represented. In 2013, one batch (#1 258 13 (3), n = 5) met 100% KD and 100% mortality; other 2013 batches showed diminished bioefficacy. After 2013, no multi-sampled batch met criteria across all sampled nets, indicating a change circa 2013. - Used LLINs (n = 40): KD60min proportions were ~51% (1–3 years use) and ~52% (>3 years); M24h was 48% (1–3 years) and 63% (>3 years); differences were not statistically significant. Overall, only 37% (95% CI 23–54%) of used LLINs met the WHO performance criterion, and compliance did not correlate with reported duration of use. - Cross-species confirmatory assays: Strong concordance between An. farauti and An. gambiae G3 assays (R^2 = 0.80 for 24 h mortality), indicating results generalize to susceptible Anopheles in other settings. - Storage temperature findings: Container temperatures exceeded 50 °C only near the inner ceiling and for just 6% of the time; otherwise remained ≤40 °C. Heat-stress simulation (60 °C for 6 weeks) on 2012 nets with 100% baseline efficacy showed no reduction in KD60min or M24h, suggesting storage heat unlikely explains observed reductions. - Quality control context: Independent QA (chemical content, wash index by Crown Agents and TÜV SÜD) reported insecticide content within specifications across 2007–2019; however, bioefficacy was poor post-2013, implying potential issues with surface bioavailability rather than total content.

The data demonstrate a sharp decline in the bioefficacy of PermaNet 2.0 LLINs manufactured from 2013 onward, with only 17% of unused nets meeting WHO thresholds, compared with 100% compliance among nets manufactured 2007–2012. This temporal shift aligns with the resurgence of malaria in PNG after 2015. While multiple factors likely contributed to increased transmission (drug stock-outs, vector and human behavior changes), pyrethroid resistance in PNG Anopheles does not appear to explain the findings, given sustained susceptibility in monitoring assays. Experimental and field temperature data argue against storage heat as a primary cause. QA certifications verifying insecticide content suggest that inadequate bioavailability of active ingredient on net surfaces may have reduced functional performance. The findings indicate that decreased LLIN bioefficacy likely contributed to diminished protection and increased transmission risk, potentially also fostering selection pressure for resistance through exposure to sublethal doses. The concordance between An. farauti and An. gambiae results supports broader relevance beyond PNG. Routine inclusion of bioefficacy testing in pre-delivery inspections and operational monitoring is warranted to ensure LLIN batches meet performance standards before distribution.

This study provides evidence that PermaNet 2.0 LLINs manufactured from 2013 to 2019 and distributed in PNG frequently failed to meet WHO bioefficacy standards prior to use, in stark contrast to nets manufactured 2007–2012. The timing coincides with a significant resurgence of malaria in PNG, suggesting reduced LLIN bioefficacy likely contributed to increased transmission. The work highlights a critical need to strengthen quality assurance by adding standardized bioefficacy testing to pre-delivery inspections and implementing routine operational monitoring across malaria-endemic countries. Future research should identify the underlying causes of reduced bioefficacy (e.g., changes in manufacturing, active ingredient formulation or distribution within fibers, surface bioavailability), assess whether similar issues affect other LLIN brands or countries, and evaluate mitigation strategies, including improved product validation and alternative control tools where appropriate.

- Limited availability of unused LLINs from 2007–2012 reduced sample size for early years, though differences versus 2013–2019 were large and statistically robust. - Only one section (side panel) per LLIN was tested; within-net variability was not accounted for, potentially affecting precision. For used nets, testing side panels (subject to greater wear) may overestimate failure rates compared to roofs. - Fewer mosquitoes per LLIN were used than in phase 3 trials (25 vs 100 for unused; 20 for used), though the study tested many more nets across many batches and years. - The study did not determine the precise mechanistic cause of reduced bioefficacy (e.g., formulation changes, fiber properties, insecticide migration to surface). - Owner-reported usage duration for used LLINs may be subject to recall bias.