Effectiveness of three-year old piperonyl butoxide and pyrethroid-treated long-lasting insecticidal nets (LLINs) versus pyrethroid-only LLINs against malaria infection: results of a cluster randomised trial in Tanzania

Abstract

After decades of success in reducing malaria through the scale-up of pyrethroid long-lasting insecticidal nets (LLINs), malaria decline has stalled, coinciding with the rapid spread of pyrethroid resistance. A new class of net, treated with a mixture of a pyrethroid and a synergist, piperonyl butoxide (PBO), demonstrated superior efficacy compared to standard (std) pyrethroid LLINs against malaria in an area of intense pyrethroid resistance, reducing malaria prevalence by 44% over 2 years in the present trial. However, an important question is left unanswered regarding the performance of this PBO-LLIN over the World Health Organization recommended lifespan of 3 years for LLINs. We conducted a four-arm randomized controlled trial using a two-by-two factorial design that evaluated the effectiveness of PBO-LLIN arms (12 clusters PBO-LLIN alone and 12 clusters PBO-LLIN + Indoor Residual Spraying; IRS) compared to std-LLIN (12 clusters std-LLIN alone and 12 clusters std-LLIN + IRS) and IRS arms versus no IRS arms from January 2014 to December 2017 in Muleba, Tanzania. Malaria infection prevalence in 80 children, 6 months to 14 years, per cluster was measured twice a year and analysed in an intention to treat (ITT) and per protocol (PP) approach. Density of malaria mosquito vectors and entomological inoculation rate (EIR) were assessed monthly in 7 houses per cluster. Logistic regression allowing for within cluster correlation of responses was used to compare malaria prevalence between PBO-LLIN groups vs std-LLIN groups and IRS groups vs no IRS groups during the third-year follow-up at 28- and 33-months post-intervention. No further IRS was conducted after the first spray round in 2015; as yearly IRS is recommended by WHO, results need to be interpreted in light of this limitation. Vector density and EIR were analysed using negative binomial regression. Malaria results were available for 7471 children. At 28 months, malaria infection prevalence was lower in the PBO-LLIN groups (69.3%) compared to the std-LLIN groups (80.9%, Odds Ratio: 0.45, 95% Confidence Interval: 0.21-0.95, p value: 0.0364). The effect was weaker at 33 months post-intervention (OR: 0.60, 95%CI:0.32-1.13, p value: 0.1131), in the ITT analysis but still evident in the PP analysis (OR: 0.34, 95%CI: 0.16-0.71, p value: 0.0051). At this time point, net usage in household participants was 31% and PBO concentration in PBO-LLINs was reduced by 96% compared to those of new nets. A total of 17,451 Anopheles mosquitoes were collected during the 3150 collection nights done in the third year. There was no reduction in EIR (DR: 0.63, 95%CI: 0.25-1.61, p value: 0.3296) between the PBO groups and std-LLIN groups or between IRS and no IRS groups (DR: 0.7, 95%CI: 0.41-2.28, p value: 0.9426). PBO-LLINs no longer provided community protection from malaria infection, compared to std-LLINs by the third year of use due to low net usage. Children still sleeping under PBO-LLINs had lower odds of infection than those sleeping under a std-LLIN, however prevalence remained unacceptably high. It is urgent that net distribution frequencies and effective lifespan of this class of LLIN are aligned for maximum impact.