Our study shows a very large reduction over time on the adequate LLIN coverage in Lihir Islands, decreasing from 97% of individuals having access to LLIN after 2016 mass distribution (19) to less than 7% in 2-2.5 years (one year prior to the next mass distribution campaign). This reduction in LLIN coverage is rather striking, and significantly different to the relatively high bed net coverage maintained in previous distributions campaigns in the country. As an example, a 2011 study in selected sites in PNG showed that 88.8% of people still had access to LLIN in villages where the distribution had been conducted during the 2 years preceding the survey, whereas coverage decreased to 67.6% in those villages where the distribution had been done more than 2 years before the survey (5). Lack of LLIN maintenance is a common issue in other tropical areas: in Uganda, LLIN population coverage decreased from 65–18% three years after distribution (30); and in Tanzania, households with at least one LLIN for every two people were below 30% two years after LLIN distribution (31). Accordingly, we found that only less than 14% of the total population surveyed slept under LLIN the previous night; and were able to show that adequate LLIN coverage was strongly associated with LLIN use as previously reported (5, 32), confirming the importance to achieve high coverage and maintenance to sustain LLIN use.
Interestingly, while adequate LLIN coverage in PNG has improved over time, especially in the islands’ region where Lihir Islands are located (increasing from 46% in 2009 to 82% in 2011), use of LLIN still remained poor: 25% of individuals in this region reported sleeping under LLIN in 2009, which increased to only 40% in 2011 despite the substantial improve in LLIN coverage (33). In fact, the villages in the PNG islands’ region are those with better LLIN coverage but worse LLIN use (23). Hence, achieving adequate LLIN coverage during mass distribution is clearly not sufficient to ensure their use. Promoting LLIN maintenance over time is also key to enhance their use and impact on malaria transmission. Mass LLIN distribution campaigns in PNG, similar as to many other countries, achieve a high coverage in the minimum time possible using strategies that are proven to work such as use of coupons or training of distributors (34). Including additional strategies to target issues affecting long-term coverage could enhance maintenance and use of LLIN until the following mass distribution.
We showed that households with at least one child under 5 years-old and households with at least one adult woman, had a higher odds of owning at least one LLIN similar to other studies in PNG (33). The scale up of antenatal services in the country, where pregnant women are targeted for prevention strategies and receive LLIN (35), has likely contributed to increase LLIN ownership in their households and probably enhanced maintenance. Antenatal services provide awareness on the importance of pregnant women and children below 5 years old to sleep under LLIN to prevent malaria. Interestingly, sex and age were associated with LLIN use, with women more likely to sleep under LLIN as well as the younger members of the household.
However, the key factor for LLIN maintenance and use was the head of the household’s knowledge about LLIN preventing from malaria infection. Only 37% of the heads of the households reported that sleeping under a LLIN was effective to prevent malaria. These results suggest that education campaigns on malaria prevention tools targeting the heads of the households could further promote LLIN maintenance and use, as shown in previous studies (36). Such education campaigns could be included as part of the mass LLIN distribution strategy, which could also look for the support of community leaders, pastors and other influential community members to deliver key awareness messages.
In addition, although we did not observe an association between a household having at least one resident attending school and increased odds of owning one LLIN or sleeping under a LLIN, half of the households had at least one child between 5- and 14- years old attending school. Consequently, there is also a big opportunity to promote LLIN maintenance and use through frequent education campaigns in schools. In Tanzania, for example, nets were provided annually to children attending primary and secondary school, which resulted in high level of maintenance (50–75% of nets given) over the first four years of distribution, even in the absence of a mass distribution campaign (37).
On the other hand, addressing alternative uses and repurposing of LLIN could also enhance maintenance and use. In our study, half of the surveyed households admitted using LLIN for the alternative purposes. We were unable to determine if LLIN used for alternative purposes were those provided the distribution campaign in 2016 or those remaining from the 2013 campaign. However, since most households did not retain a single LLIN, it is likely that most nets used for other purposes were those given in 2016. The most common alternative use given to LLIN was fishing, which could be related to the low LLIN maintenance and use observed in the PNG islands’ region, among other factors. Another common alternative use was to protect seedlings and food. These common misuses have been also described in Sub-Saharan Africa (4, 12–14) and all relate to basic needs like ensuring food supply. LLIN distribution using mass campaigns are proven to increase LLIN ownership (38–40); however, especially when health education might not be sufficient to reduce alternative uses of LLIN when other basic needs are involved, some creative solutions could be used during mass LLIN distribution campaigns, such as facilitating access of target communities to suitable and without insecticide materials for fishing and gardening. In addition, ensuring high LLIN bioefficacy and teaching communities about the impact of LLIN in reducing mosquito population could further motivate communities to better maintain LLIN.
In our study, we used a community approach that allowed a massive outreach to the Lihirian population but also had some limitations. Although the goal of VMAs was to survey their entire village or assigned part of a village, full coverage of all villages was not possible due to logistical constraints. Hence, our study could be subjected to selection bias. However, all statistical models were adjusted for village in order to minimize bias arising from the non-representativeness of a few villages. In addition, our sample size was very large and spatio-geographically and demographically representative. Because data collection was conducted by the VMAs who are living in the same village, we avoided sensitive questions such as socioeconomic factors or enquiring about pregnancy status. These two pieces of information could have given important insight, such as LLIN use during pregnancy or the role of socioeconomic status contributing to repurposing of LLIN. We also observed that, while less than 7% of individuals had adequate access to LLIN, close to 14% were reported to sleep under them, which could be due to more than two individuals sharing a double net (as commonly seen for young children) and due to the social desirability bias inherent in self-reported measures. Finally, in order to maximize quality of the results, VMAs received an intensive training and close supervision, and data were carefully reviewed to recode impossible values and minimize missing values.