Simulation models suggest there are areas of Bioko Island in which high proportions of malaria prevalence can be attributed to infections acquired while traveling to higher burden areas.7,13,14 A recent model suggested that drastically reducing the number of imported infections between Bioko Island and the mainland could significantly reduce prevalence in areas with a high proportion of travelers.14 However, prior to 2020, there were, understandably, no intervention studies nor other data to definitively support the model simulations. Travel restrictions imposed as a measure to control the spread of SARS-CoV-2 in 2020 provided an opportunity to directly evaluate the impact of imported infections. We observed that in the absence of travel, prevalence in historically high travel areas decreased by 2%, while prevalence in low travel areas increased by 5% over the same period. This suggests that, assuming parallel trends, in the absence of the travel moratorium, one would have expected Pf prevalence to be 7% higher in high travel areas than what was observed. When adjusting for spray coverage and time one went indoors, the difference in trends decreased to 5% but remained significant.
In 2019, prior to the travel restrictions, odds of malaria infection were two to three times higher in areas of Bioko Island with a historically high proportion of travelers, relative to areas of historically low travel. This finding is similar to a 2013 analysis that showed infection risk was greater in children living in areas with the highest proportion of travelers.13 In 2020, when the movement of individuals was restricted, there was no difference in risk of malaria infection observed in high travel areas compared to low travel areas. This observation both further supports the hypothesis that a significant fraction of the Pf prevalence observed in high travel areas could be explained by imported infections,7 while also suggesting that malaria risk in these areas is not solely driven by importation.
Previous analyses have suggested there may be areas where malaria prevalence is solely attributable to imported infection,7 as several of these locations are in urban centers where there is generally improved infrastructure and fewer mosquitoes. If this were true, it would have been anticipated that the risk in high travel areas would be substantially less relative to low travel areas once importation was eliminated. However, this analysis showed no difference in odds of infection in the absence of travel. One explanation for this finding is that there was still residual travel occurring, even with the travel moratorium, which allowed infections to continue to be imported during 2020. While a small percentage of individuals did report travel in the past 8 weeks in the 2020 MIS (1%), it seems unlikely that this would sustain the observed prevalence in the population of high travel areas. Another explanation is that infections in high travel areas were acquired through travel to other areas of the island. There is often frequent travel within the island, especially between Malabo and areas in the periphery, where the force of infection is significantly higher.12,14 In 2019, residents in both high and low travel areas reported an average of 1 trip to another part of the island in the past 8 weeks (range: 1–5 trips), which did not substantially change in 2020. Therefore, it is possible that the remaining prevalence in high travel areas is from within-island parasite movement. However, there was no significant difference seen between prevalence in those who reported within island travel and those who did not in 2020 (Supplementary Table 1), suggesting this does not offer a full explanation. A final possibility is that high travel areas are receptive to local transmission, and levels of endemic transmission persist even when imported infections are removed. This is supported by a 2019 incidence study conducted in Malabo, which suggests local transmission is occurring in peri-urban areas in Malabo district.20 In that study, while travelers tended to be more likely to have an infection, the no incident infections related to travel were identified, supporting the hypothesis that there is local transmission occurring, even in areas where travel is common. In addition, recent entomological monitoring in urban Malabo using human landing catches21 and larval collections have confirmed the presence of anopheline vectors showing varying levels of human biting rates and larval densities across the city (Supplementary Fig. 3). Therefore, the results of this analysis suggest that control strategies that aim to reduce the malaria burden in travelers, either by reducing the burden in the areas where they travel, and/or by treating returning travelers, would impact the overall prevalence in several communities in Bioko. Additionally, control measures that aim to reduce local transmission, such as IRS, distribution of LLINs, and larval source management should be continued, even if additional interventions that target imported infections are introduced. Further analyses are needed to better understand the role of importation and local transmission at a more granular level.
The interpretation of our results depends on several assumptions. First, it is assumed that the change in prevalence in low-travel areas is a valid estimate of the change we would expect in the high-travel areas in the absence of imported infections. That is, the two areas had parallel trends prior to the elimination of travel.19 While it is difficult to definitively verify this assumption, comparing data from the 2018 MIS to that from 2019 and 2020 suggested that prior to the halt in travel, there were similar trends over time in the high and low travel groups. Additionally, the final model adjusted for measured variables that impact malaria transmission and changed over time within travel groups (time variant factors). However, it is possible that there are unmeasured factors that were not accounted for in the model. Most notable would be changes in the ecological landscape. For example, an outbreak occurred in 2019 in a low-travel area in the south of the Island, because of recent construction that created additional breeding sites.22 There was no precise measure of land use over the study period, so it was not possible to adjust for this variable in the analysis. However, there were seven EAs known to have major changes in land use, including the one where the 2019 outbreak occurred, and one where an outbreak occurred in 2020. Three of these EAs were in our analysis, and two areas showed large prevalence differences from 2019 to 2020, suggesting there may have been additional transmission due to the changes in the ecological landscape. When these EAs were removed from the analysis, the difference in differences and ratio of ratios were slightly attenuated, but still of similar magnitude and significance. Therefore, if it were possible to precisely measure changes in the ecological landscape and include them in the model, we may expect a slightly lower prevalence difference, but the conclusions would remain the same. Another assumption of a difference-in-differences model is that secular trends are consistent over time and have the same impact on both travel areas. While there have been changes in the monthly amount of rainfall overtime on Bioko, there have been increases both in high and low travel areas, and the assumption is that this would equally impact malaria transmission potential in these areas.
The emergence of SARS-CoV-2 in early 2020 disrupted health systems around the world. As countries closed borders, limited movement, and restricted activities to curtail the initial spread of COVID-19, other public health programs were impacted. This is especially true for many malaria endemic countries, in which COVID-19 restrictions and global supply chain issues resulted in disruptions in the distribution of long-lasting insecticide nets, application of insecticides, and availability of anti-malarial medicines.15,23,24 The World Health Organization modeled the potential impact of disruptions to malaria interventions and estimated these disruptions could increase cases by upwards of 20% and deaths by greater than 50%, especially in scenarios where access to treatment was disrupted.25 Similar impacts were seen during the Ebola outbreak in 2014–2015 when health systems were disrupted.26–29 However, in these models and analyses, the potential impact of reducing importation and movement of Plasmodium infections was not considered.30,31 This analysis shows that on Bioko Island, where malaria control interventions remained largely uninterrupted during the pandemic15 travel restrictions resulted in a decrease in malaria prevalence in areas with high travelers. It is possible that other areas with high proportions of imported infections may also have seen these decreases because of the travel restrictions, despite other interruptions to the health care system. This analysis suggests that the impact of COVID-19 on malaria burden may be underestimated in areas with a high prevalence of travelers. Additionally, as borders are now open and imported infections return, malaria control strategy discussions should include interventions that target these infections to reduce burden.