The sudden increase in malaria morbidity in Riaba district in the first third of 2019 coincided with a significant increase in rainfall patterns and anopheline HBR that were observed across the island (Fig. 2). Even though other areas had seen an increase in confirmed malaria cases, namely in Luba district and in certain isolated communities in the West of Malabo district, in none was the spike in cases as dramatic as the one observed in Riaba district. The response from the malaria control teams to diagnose and treat positive individuals with anti-malarials was certainly effective and successfully curbed malaria cases by May 2019. Malaria prevalence and HBR, however, remained high, despite vector control interventions (Figs. 2 and 3). The significantly higher malaria prevalence in 2 to 10 year-old children measured almost four months later during the 2019 MIS (Fig. 4) could be explained by higher parasite densities in children that are detectable for longer periods [14, 15], by poorer adherence to treatment in this age-group following the test and treat intervention [16–18] or by a persistently high force of infection  driven by the high HBR observed, as is suggested by the data (Fig. 2C). The increased rainfall recorded in 2019, which favoured the availability of breeding habitats, could explain part of the increase in malaria transmission levels observed in Riaba district, but not all.
Notably, major road and real-estate development projects had been ongoing in Riaba for two to three months before the outbreak. Urban development and construction sites can potentially affect local vector ecology and thus require responsible management by sectors outside health care . In fact, the main finding of the entomology teams during the intervention was the substantial number of anthropogenic mosquito breeding habitats created by these projects. These larval habitats were not only numerous but large and presented a significant challenge for LSM activities (Fig. 8). Although efforts were made to treat as many of them as possible, there were simply too many habitats spread over large areas, exceeding the manpower available to tackle them. Following the interventions, malaria prevalence decreased in most communities, but higher PR was seemingly clustered in those nearer to the construction sites (Figs. 6 and 7). A notable exception was Patio López, where PR measured during the MIS was higher than during the outbreak survey (35.7% vs. 3.3%), though only 14 people were sampled during the MIS at this community. Therefore, the confidence limits of this estimate were wide (95CI 12.8 - 64.9%) and true prevalence could well have been on the lower bound. On the other hand, the HBR estimated in Patio López, despite spiking at the beginning of the year alongside HBR from most of the other entomological monitoring sentinel sites on the island, dropped by May 2019 and remained low thereafter. Conversely, the HBR in Patio Balboa remained high despite the interventions, which could be explained by the fact that this site is located in the surroundings of the anthropogenic breeding habitats created by construction projects. Though, historically, Patio Balboa has yielded particularly high vector densities, the current alterations to the local ecology may explain the persistently high HBR observed in this location.
Other, more pernicious drivers may have aggravated the outbreak in Riaba district. First, human mobility and parasite importation from mainland Equatorial Guinea, where malaria transmission is significantly more intense , have been identified as important contributors to the parasite prevalence observed on Bioko Island [22–24]. Apart from the generation of vector breeding habitats, the construction sites in Riaba demanded a high influx of migrant workers from mainland who could have been malaria infected and, therefore, could have increased the local parasite pool. Also, the presence of military camps in Riaba with highly rotating personnel determines a constant flux of people to and from mainland, potentially resulting in higher parasite importation. Second, the limited uptake of interventions by the population despite universal coverage may have also contributed to the problem. Data from annual MIS indicate that LLIN ownership and access constantly decrease from year to year, a problem that is ubiquitous across Bioko Island . The last mass distribution campaign on Bioko took place in 2018, six months before the outbreak. During that campaign, LLINs were distributed to virtually all households on the island . LLIN population access (i.e. availability of at least one LLIN for every two people) in Riaba, however, was estimated at 75.2% two months after distribution and only at 53.5% in 2019. Moreover, LLIN use, regardless of access, remains sub-optimal with 49.3% and 48.3% of the population surveyed in Riaba in both years reporting to have slept under a LLIN the night before, decreasing the effectiveness of this vector control intervention . Finally, changes in host-seeking behaviour of anopheline vectors as a response to indoor vector control interventions have been observed on the island . In light of increased vector densities, outdoor biting could have amplified transmission.
In 2019, the malaria control strategy of the NMCP/BIMEP was redefined towards the goal of malaria elimination. Given the local vector ecology, however, malaria receptivity on the island remains high . In 2019, EIR in Patio Balboa was more than ten-fold higher than in 2018 and, given that sporozoite rates did not increase in the same magnitude, this was mostly attributed to significant increases in vector densities. This suggests that, despite the great reductions in transmission intensity across Bioko in general, and in Riaba district in particular, malaria receptivity in these areas is indeed important. The outbreak in Riaba exposed several vulnerabilities that probably combined to produce the observed surge in malaria clinical cases. It showed that, if ecological conditions were to change or if interventions were relaxed, there will be a very high risk of malaria resurgence. Changing ecological conditions were revealed by the increase in HBR following rainfall anomalies and the increased availability of vector breeding habitats driven by the presence of large construction projects. Relaxing of interventions resulted from logistical constraints at the beginning of 2019 that delayed the start of malaria control activities, consequently weakening the monitoring and response capacities of the NMCP/BIMEP. The need of sustained malaria interventions with an emphasis on vector control cannot be underscored enough in this context, as cannot be the importance of multi-sector participation as an essential component of integrated strategies when it comes to the fight against malaria. This case study signals the heavy challenges ahead in the difficult path to malaria elimination on Bioko Island.