Trends in parasite prevalence following 13 years of malaria interventions on Bioko island, Equatorial Guinea: 2004–2016

Despite a large scale-up in access to long-lasting insecticide-treated nets (LLIN) and indoor residual spraying (IRS), malaria remains a major public health burden across much of sub-Saharan Africa. As intervention coverage increases across the continent, monitoring and evaluation of control programmes become paramount, although reliable indicators are often lacking and can be contradictory [1]. Several recent studies have used repeated community surveys to analyse long-term trends of malaria transmission in areas of declining exposure to malaria parasites [2‐4]. Community surveys have advantages over passively collected data in that they are able to estimate the true prevalence within a population, regardless of symptoms and health-seeking behaviour. Accurately tracking changes in malaria transmission following implementation of malaria control can help to ensure that interventions are efficiently targeted.

Islands are considered ideal settings for effective malaria control due to the limited population migration and well-defined geographical borders. Several island nations have achieved substantial reductions in malaria transmission attributed to successful interventions [5‐8], with Sri Lanka recently being certified malaria-free [9]. Bioko Island, Equatorial Guinea, with historically high malaria transmission [10], has been subject to extensive interventions including intensive vector control, improved case management, intermittent preventative treatments (IPT) and behavioural change interventions since 2004 through the Bioko Island Malaria Control Project (BIMCP) [11]. The BIMCP is a public–private malaria control project funded by Marathon Oil, the government of Equatorial Guinea and other partners. It has been implemented by Medical Care Development International (MCDI), along with researchers and key stakeholders. Early evaluations indicated that malaria transmission had dropped considerably, resulting in a 60% reduction in child mortality following the first 4 years of interventions [12], and a corresponding 50% drop in parasite prevalence in children aged between 1 and 14 years [13]. However, despite its island status, a study in 2013 demonstrated that travel to the mainland of Equatorial Guinea increased the risk of infection in travellers themselves, and in non-travellers living in neighbourhoods with high proportions of travellers [14], highlighting the impact of population movement on malaria transmission levels [15‐18].

Malaria indicator surveys (MIS) take place annually within sentinel sites across the island as part of an enhanced monitoring and evaluation programme to assess the progress of the BIMCP towards its ultimate goal of malaria elimination. Transmission trends over more than a decade of vector control were examined using the time series of 13 consecutive cross-sectional assessments of parasitaemia and anaemia prevalence.

In the thirteen surveys between 2004 and 2016, 35,146 household visits were conducted, ranging from 1043 to 4721 households per year (Table 1). The median household size was 6, with an average of 2 children (aged between 1 and 14 years) tested per household.

Reported spray coverage ranged from 56% (in 2015 when targeted focal IRS was introduced) to 80% (in 2005). Spray coverage varied substantially between sites. Sites in Malabo and the North West consistently did not achieve spray coverage over 80%-in part due to a large proportion of refusals from homeowners. A low proportion of houses reported universal net coverage, ranging from 13% (in 2011) up to 57% and 42%, following the LLIN campaigns in 2007/08 and 2014/15, respectively. In the subsequent year, following the first distribution, universal coverage had already substantially reduced (21%) whilst coverage remained similar in the year following the second distribution (40%). The percentage of children aged 1–14 years sleeping under a net the night before the survey was below 55% in all survey years except in 2008 when it increased to 81% (Table 1). By 2009, usage had already decreased to 40%. Utilization increased to 54% following the 2nd mass distribution in 2014/15, though it remained substantially lower than immediately after the 1st mass distribution in 2007/08. Utilization of nets was higher in houses that had universal coverage, with usage varying between 76% and 92% over the study period (Table 1). Within these households, priority for sleeping under nets appeared to be given to the youngest children, with 85% of under 5 s sleeping under a net, compared to 82% and 78% of 5–10 and 10–14 year olds, respectively (p < 0.001). In houses without universal coverage, priority was also given to children aged between 1 and 5 years, with 30% reporting sleeping under a net, compared to only 23% and 20% of children aged 5–10 and 10–14, respectively (p < 0.001). In participants over the age of 14 years, 27% reported using a net in households without universal coverage, compared to 78% in households with universal coverage.

Monitoring and evaluation play a key role in malaria control campaigns and are essential for assessing the impact of interventions and guiding programmes in adapting operational approaches. The World Health Organization (WHO) Global Technical Strategy [20] emphasizes surveillance as a core intervention for reducing malaria burden and achieving elimination. On Bioko, comprehensive surveillance through annual MIS have allowed for detailed examination of the impact of interventions and control strategies.

The BIMCP has been an overall success, achieving a reduction in parasitaemia of over 75% over 13 years. In the early years of implementation, there was heterogeneity of impact between sites. Differential impact of universal interventions have been documented in other studies [4, 21] and has been attributed in part to uneven distribution or uptake of interventions [11, 22], which was seen on Bioko with varying coverage of IRS and use of LLINs. There was evidence that living in a site that had initial high prevalence resulted in a higher risk of infection overall, although, this was not universally true, with Bilelipa (with a pre-intervention prevalence of 49%) recording a 97% reduction in prevalence by 2016, whilst Central (pre-intervention prevalence of 28%) achieved a more modest 57% reduction. More notable was the difference in reduction by region with the North Eastern sites (Bakake Grande, Baney, Rebola) showing an average decrease of 94% compared to an average decrease of 65% in the North West (Basacato, Punta Europa, Sacriba, Sampaka). In Punta Europa, a site which consistently showed higher prevalence, this was attributed initially to low IRS coverage with infection prevalence decreasing once IRS coverage increased.

The main preventative intervention employed on Bioko is IRS. Whilst IRS can protect the sprayed household through a repellent effect (depending on the insecticide used), its main effect is through killing action resulting in community-wide protection when a sufficiently high spray coverage is reached [23]. The WHO recommends that > 80% of structures need to be sprayed for IRS to have a community effect [24]. In this study, living in a site with spray coverage over 80% resulted in lower risk of infection but the association was not significant. A previous study collating data from Equatorial Guinea and Malawi demonstrated that higher spray coverage resulted in decreased infection prevalence [19].

In addition to IRS, two mass LLIN distributions took place; however, MIS which took place within 6 months of the distributions suggest that if high coverage was originally achieved, it had quickly reduced. This may be because some of the nets that were distributed were subsequently taken to the mainland where little malaria control is in place and malaria transmission is substantially higher. Studies are ongoing to establish why attrition of nets was so high. In households with universal coverage, the use of nets was high, indicating that the barrier to higher LLIN utilization is the lack of access to nets, not unwillingness to use them.

Net use was shown to be protective against infection, with children reporting sleeping under a net the previous night (and not living in a sprayed house) having 30% lower odds of being infected. A treated net provides a physical barrier against biting mosquitoes and protection through reduction of the mosquito population by insecticidal action. Several sites saw steep falls in parasite prevalence between 2007 and 2008, which may have been due to increased net use. The proportion of people with access to nets post-2008 reduced to approximately 20%, but despite this, there was no sign of a rebound in malaria prevalence (or severe anaemia), suggesting other factors were also contributing to maintenance of the new lower equilibrium. The highest infection prevalence was consistently detected in children aged between 5 and 14 years old, who were also least likely to be using a net.

IRS and LLIN are often used in combination rather than alone. As with previous studies in Equatorial Guinea [25, 26], and elsewhere [27, 28], this study demonstrated an additive effect of the combination of both interventions. The effect of the combination of the two interventions is likely to depend on intervention coverage, initial transmission levels, the insecticides used, and on the local vector population [29‐31].

In recent years, population movement to the mainland of the country increased, perhaps due to more transport options, or an overall increase in wealth. Travel to the mainland, where malaria is higher [26], has been shown to be a risk factor for infection on Bioko previously [14]. More people reported travel in the three sites demonstrating the least impact, compared to the three with the highest impact, further highlighting the importance of this risk factor. Infection importation remains a persistent and significant challenge for malaria control on Bioko and a potential barrier to further reduction of transmission on the island. Despite being an island, population travel to and from higher risk areas makes Bioko vulnerable to importation of infection. Border screening and treatment, targeted ‘Malaria Awareness’ campaigns or chemoprophylaxis may help to alleviate this issue. Expansion of control activities to include the mainland area of Equatorial Guinea are likely to have an additional impact.

Whilst interventions have reduced the risk of infection, substantial variation in transmission over the study period remains unexplained. A factor which may have resulted in reduced transmission over time is an increase in the quality of housing, which has been shown in Bioko [32], and elsewhere, to be associated with lower malaria incidence [33, 34]. Bioko has seen improvements in socio-economic status during the project timeline which has probably also contributed to a reduction in overall transmission. In addition, access to free treatment with artemisinin-based combination therapy was introduced in 2005 and provision of drugs through private vendors has also increased over the study period. A reason for the differential impact could be due to population behaviour, such as population movement between higher and lower transmission areas or increasing outdoor behaviour in the evenings due to the provision of electricity and lighting resulting in heightened exposure to outdoor biting. Data collected several years ago on Bioko suggested that outdoor biting was not associated with malaria infection at the time, mainly because peak biting occurs at a time when most people are indoors [35]. However, as human behaviour changes, outdoor biting may become a significant risk for infection.

Although there has been a decrease in prevalence in most years of the project, 2013 stood out as an anomaly, when prevalence increased in most sites, following a particularly low prevalence recorded in 2012. This may have been influenced by the choice of insecticide used that year. In 2004, the insecticide used for spraying was switched from deltamethrin to bendiocarb following the detection of the kdr mutation in the Anopheles gambiae population [36]. Subsequent analysis showed that despite the presence of kdr, the 2004 pyrethroid spray round nevertheless resulted in a sharp reduction in sporozoite infectivity in the main vector after spraying, and there was no apparent association between kdr and phenotypic resistance. In 2013 deltamethrin spraying was re-introduced as part of a programme of insecticide resistance management through rotation [37]. Following the change back, a sharp increase in parasite prevalence was recorded in the subsequent household survey of 2013, despite no obvious decrease in intervention coverage, prompting fears of resistance of local vectors to deltamethrin [38]. The presence of resistance markers is currently being monitored across the island.

Despite the heterogeneous impact of interventions in Bioko, transmission has declined substantially overall. Understanding the factors that have contributed to differential impact will help the BIMCP to prioritize strategies and to continue the downward trajectory in parasite rate. Indeed, several areas, which have now reached pre-elimination levels, are now implementing further strategies, such as reactive case detection, to attempt elimination of local transmission. This study has shown that vector control with IRS and LLINs remained the most important protective intervention on Bioko, whilst travel to the mainland constituted a substantial risk for malaria infection. The low proportion of houses with universal coverage of LLINs, and data suggesting that the age group most likely to be infected are least likely to be using a net (10–14 yr olds) is concerning. Strategies to ensure a higher proportion of the population have access to nets have been implemented, but the lack of LLINs on the mainland and the high rate of associated leakage of nets, may have undermined the ability to maintain high coverage levels. Controlling parasite importation from the mainland, and maintaining universal access to the core prevention, namely LLIN use, represent two clear challenges to further reducing malaria transmission on Bioko.

Whilst the gains over the past 13 years have been impressive, the slow, but steady progress highlights the importance of sustaining malaria control interventions on a long-haul basis. Long-term commitment and surveillance are necessary to chart the downward trend which may not necessarily advance in an uninterrupted linear manner.