Getting to zero: micro-foci of malaria in the Solomon Islands requires stratified control

Background The Solomon Islands has made significant progress in the control of malaria through vector control, access and use of improved diagnostics and therapeutic drugs. As transmission is reduced there is a need to understand variations in transmission risk at the provincial and village levels to stratify control methods. Methods A cross-sectional survey of malaria in humans was conducted in the Solomon Islands during April 2018. Nineteen villages across 4 provinces were included. The presence of Plasmodium species parasites in blood samples was detected using PCR. Results Blood samples were analysed from 1,914 participants. The prevalence of DNA of Plasmodium falciparum was 1.2 % (n = 23) and for Plasmodium vivax was 1.5 % (n = 28). 22 % (n = 5/23) of P. falciparum DNA positive participants were febrile and 17 % of P. vivax DNA positive participants (n = 5/28). The prevalence of both P. falciparum and P. vivax was extremely spatially heterogeneous. For P. falciparum, in particular, only 2 small foci of transmission were identified among 19 villages. Plasmodium falciparum infections were uniformly distributed across age groups. Insecticide-treated bed net use the night prior to the survey was reported by 63 % of participants and significantly differed by province. Conclusions Malaria transmission across the Solomon Islands has become increasingly fragmented, affecting fewer villages and provinces. The majority of infections were afebrile suggesting the need for strong active case detection with radical cure with primaquine for P. vivax. Village-level stratification of targeted interventions based on passive and active case detection data could support the progress towards a more cost-effective and successful elimination programme. Supplementary Information The online version contains supplementary material available at 10.1186/s12936-021-03779-y.


Background
Since the beginning of the millennium, substantial progress was made to reduce the global incidence of human malaria, with 12 new countries being certified as malaria-free since 2007 [1,2]; however, these gains have plateaued over recent years. Countries making significant progress to reduce transmission are committing to malaria elimination by 2030 [3]. As malaria transmission reduces, malaria cases become more spatially heterogeneous [4,5]. In response, national malaria programmes are encouraged to use local evidence to design and implement a mix of interventions by transmission stratum, rather than utilizing a one-size-fits-all approach [6]. In countries controlling malaria, interventions can be targeted to entire villages with higher malaria incidence until only individual episodes of malaria remain and programmes have the capacity to investigate and respond focally [3].
The Solomon Islands has made significant progress to control malaria through vector control, access and use of improved diagnostics and therapeutic drugs. Vector control has been the principal preventative intervention to reduce malaria transmission over the past two decades with long-lasting insecticidal nets (LLINs) and IRS, both globally and in the Solomon Islands. These strategies have seen Plasmodium falciparum replaced by Plasmodium vivax as the dominant malaria parasite in the Solomon Islands [7].
Here the annual parasite incidence (API) was reduced from over 200 per thousand in 2003 to just below 30 per thousand in 2014. However, since 2015, the API has increased reaching 107 per thousand in 2019. This resurgence has delayed the goal of malaria elimination with the priority of the 2021-2025 Strategic Plan now seeking to reach zero cases by 2034 [8]. In 2019, four provinces contributed to approximately 86 % of the malaria burden in the country: Central Islands (10.5 %), Guadalcanal (27.1 %), Honiara (15.2 %), and Malaita (33.1 %) [8]. The malaria caseloads, in the other six provinces (Choiseul 0.7 %, Isabel 0.5 %, Makira 4.2 %, Rennell-Bellona 0 %, Temotu 2.5 %, and Western 6.0 %) were responsible for only 14 % of the national malaria burden [8]. This heterogeneity argues for a stratified approach to malaria control. Here, a cross-sectional survey of malaria in humans was undertaken to understand the heterogeneity in malaria transmission risk at the provincial and village levels.

Study site
The cross-sectional study was conducted in the Solomon Islands (− 8.0° S, 157.0° E) in April 2018. The Solomon Islands is hot and wet with an annual rainfall of 2,005 mm (mean for 1999-2017 at Henderson Airport, Guadalcanal Island). Rain falls year round with a peak during January to March. The mean daily coastal temperature ranges between 24 and 30 °C with a mean of 26 °C.
The four study sites-Honiara City Council as well as Guadalcanal, Isabel and Malaita Provinces-were selected in consultation with the Ministry of Health and Medical Services, Solomon Islands. Survey sites were selected to cover a range of transmission scenarios from low to high transmission, and to include areas with P. falciparum cases based on passive case detection at health facilities. Inclusion criteria required villages to have a minimum resident population of 200, and to be accessible by sea or road. The sites encompassed 19 suburbs and villages ( Fig. 1), hereforth termed villages.

Field procedures
All residents in selected villages, over the age of 5 years, were invited to participate in the study. The only exclusion criteria was the resident's unwillingness to participate in the study. Residents were equally encouraged to participate across age categories and genders. After enrolment, demographic information and data on possible risk and protective factors associated with mosquito-borne diseases of participants were collected. Data collected included: (1) name, age, gender, household number, (2) fever history, (3) malaria diagnosis and treatment history and (4) access/use of mosquito protection measures. The typanic temperature of participants was measured (Welch Allyn Braun ThermoScan Pro 6000) and any febrile individuals (temperatures > 38 °C) were immediately referred to the nearest health facility.
Each participant provided a ≤ 10 ml blood sample by venepuncture using vacutainers, drawn by a nurse trained and employed by the Ministry of Health and Medical Services. Five µl of selected blood samples was immediately tested for malaria using the AccessBio Car-eStart rapid diagnostic test (RDT) (G0161) according to the manufacturer's instructions. Concurrently, 3 × 50 µl blood spots were placed onto cellulose chromatography papers (2 × 7 cm; Whatman ® Grade 3MM) and dried under ambient conditions. Dried blood spots were stored in individual snap-lock bags and sealed in an airtight container with silica gel.
Serum was separated by centrifugation at 1,500g for 10 min. Serum and clots were initially stored at 4 °C, then frozen at − 20 °C within 4 days, until shipped internationally on dry ice and subsequently stored at − 80 °C until analysed. A unique code was assigned to each participant and their associated samples.

PCR detection of Plasmodium spp. parasites
Dried blood spot (DBS) samples were pooled (4 samples/ pool) for extraction using the Chelex/saponin method [adapted from 9]. For each sample, 2 × 3 mm punches were used, being the equivalent of 2 × 2 µl of blood. Each sample was eluted to 150 µl, from which 5 µl of template was used in both nPCR and qPCR. Extracted nucleic acid were amplified in a nested PCR targeting the pan Plasmodium 18 S gene [10,11]. Positive pools were de-pooled and each sample in the pool was individually extracted as above and further analyed by pan genus and species-specific nested PCR [10,11]. Where all de-pooled samples were negative in the nested PCR, blood clots were analysed by real-time PCR [12] targeting genus-sepcific 18 S Plasmodium followed by a multiplex reaction targeting P. falciparum, P. vivax, Plasmodium malariae and Plasmodium ovale.

Statistical analysis
The relationship of province on use of insecticide-treated nets (ITNs), other vector control measures, domestic and international travel history and malaria positivity were The relationship between malaria PCR positive explanatory variables was analysed using a generalized linear model (GLM; package MASS) with a binomial distribution, a binary dependent variable (i.e., negative or positive).
Step-forward multi-model inference (MMI), based on ranking the value of the Akaike's Information Criterion (AIC), was used to select the explanatory variables that best described malaria PCR positivity. The relative strength of evidence for each model within the set of alternatives was assessed using Akaike weights (wAIC) where the wAIC for each model is interpreted as the probability for the most likely model, with support varying from 0 (no support) to 1 (total support) [13][14][15]. The most parsimonious model from the final set of nested models was compared with the likelihood ratio test and compared with the Χ 2 distribution [16,17]. The explanatory variables were village, gender, temperature, age, ITN use and malaria history. Analyses were performed using the R package (v3.5.1).

Study population
A total of 1,977 individuals (215 from Honiara, 221 from Guadalcanal, 392 from Western Malaita, and 996 from Isabel) participated in the study. Participants had a median age of 30 years, with 62 % being female ( Table 1). The average tympanic temperature of participants during the survey was 37.1 °C, and 34 people had a temperature exceeding 38 °C at the time of the survey, with the maximum temperature recorded being 40.6 °C.

Vector control
Insecticide-treated bed net use (predominately longlasting insecticide treated nets (LLINs)) the night prior to the survey was reported by 63 % of participants and significantly differed by province (χ 2 = 141.49 df = 4, p < 0.0001; Fig. 2), ranging from 48 % in Malaita Province to 75 % in Isabel Province. The use of ITNs also varied by villages from 29 % in Lilisiana, Malaita to 95 % in Hovukoilo Village in Isabel (Additional file 1: Table S1). There was limited use of other mosquito protective measures; 10 % of study participants used mosquito coils (usually a volatile pyrethroid) and 5 % of survey participants' houses had window screens. Topical repellents were used by only 0.8 % of survey participants. The use of other mosquito control measures was significantly related to the province (χ 2 = 301.27 df = 4, p < 0.0001), with higher mosquito coil use in Honiara and Guadalcanal (39 and 29 %, respectively), and 16 % of participant houses in Western Malaita having window-screens (Fig. 2).

Plasmodium prevalence by PCR
Blood samples from 1,914 participants were analysed by PCR for Plasmodium spp. DNA (henceforth, DNA malaria positive participants are referred to as malariapositive and used to determine malaria prevalences): 46 participants were malaria positive with 17 individuals positive for P. falciparum, 22 with P. vivax, 6 with both P. falciparum and P. vivax, and a single individual with P. ovale (Table 2). Of Plasmodium PCR positive individuals, seven were febrile (temperature ≥ 38 °C): 2 were positive for P. falciparum, 2 were positive for P. vivax and 3 were PCR positive with both P. falciparum and P. vivax;   Table 3). Thus, the percentage of febrile or symptomatic participants for P. falciparum was 22 % (n = 5/23) and for P. vivax was 17 % (n = 5/28). For P. falciparum, the base GLM model was most substantially improved by adding village (100 % wAIC support). Sequentially the model was improved by adding temperature (93 % wAIC support, Fig. 3). These explanatory variables of village and temperature were significant (log-likelihood ratio test) and were included in the most parsimonious model ( Table 4). None of the other remaining candidate factors were able to further improve model fit. Of note is that although the prevalence of P. falciparum was reduced almost by half from 1.9 to 0.9 % by using an ITN (Table 3), this factor did not explain sufficient variation to be included in the final model.
Much of the variation in P. falciparum prevalence was explained by village, and this species was extremely heterogeneous across the provinces, with only 2 small foci of transmission identified among 19 villages surveyed (Fig. 1). Plasmodium falciparum DNA positive samples were extremely localized, there was 1 mixed positive sample from Guadalcanal (n = 119) and the remaining 22 positive participants were from only 2 villages in Malaita.   Overall, 69 % of the P. falciparum positives were from a single village, where the prevalence was 15.5 % (Fig. 1). When compared at the provincial level, P. falciparum prevalence was highest in Malaita (Table 2). For P. vivax, the base GLM model was most substantially improved by adding village (99 % wAIC support). Sequentially the model was improved by adding temperature (87 % wAIC support, Fig. 3), bed net use (94 % wAIC support) and gender (51 % wAIC support). These explanatory variables of village, temperature, bed net use and gender were significant (log-likelihood ratio test), and were included in the most parsimonious model (Table 4). Adding age or malaria history was unable to further improve the model fit. While age was not included in the most parsimonious P. vivax model, a univariate GLM did pick up an influence of age on infection (χ 2 = 9.63, df = 1, p = 0.0019; Fig. 3). For gender, males were more likely to be positive for P. vivax ( Table 3).
The P. vivax positive samples were heterogeneous across the villages, but they were not as extremely localized as the P. falciparum infections. Plasmodium vivax DNA was detected in participants from Guadalcanal (n = 4), Honiara (n = 4) and Malaita (n = 20), across 7 villages. Overall 63 % of the P. vivax positive individuals were found in 2 villages, where the prevalence was 8.0 and 9.7 % (Fig. 1). Plasmodium spp. DNA was not detected from Isabel province participants.

Discussion
Historically in the Solomon Islands, P. falciparum was the dominant species with a higher prevalence in younger age groups [18,19], but as transmission declined, the proportion of P. vivax cases increased and the prevalence of submicroscopic malaria infections was high [20,21], with malaria transmission becoming increasingly fragmented, affecting fewer villages and provinces [7,20]. Such heterogeneity in low malaria transmission settings was previously documented [22][23][24][25]. In the current study, most infected individuals were afebrile with the prevalence of P. falciparum uniformly distributed across age groups. The epidemiology of low malaria transmission observed here is not dissimilar to that encountered at the early 1970 s during the DDT-based Malaria Eradication Programme when prevalence dropped to 1.4 % and P. vivax dominated [18]. Similar shifts in malaria epidemiology were also observed in Temotu and Isabel Provinces in the early 2010 s during elimination efforts in these two provinces [26,27]. In the current study, statistical hotspots were not defined by geospatial statistics due to the difficulty to show statistical significance at such very low transmission intensities where only isolated foci remain [5].
Malaria transmission intensity is a function of both receptivity of the environment to support anopheline vector populations and the presence of malaria parasites in an area or the risk of introduction of malaria parasites [28] with most variability in local exposure to malaria due to differences in mosquito populations densities [29]. Previous work in the Solomon Islands argued that the biting rate of the dominant malaria vector, Anopheles farauti, is a better predictor of malaria receptivity than the sporozoite rate or entomological inoculation rate [30]. Considering the fundamental role of mosquito densities and survivorship on malaria transmission, vector control has been directly responsible for large reductions in malaria transmission, not only in the Solomon Islands [31][32][33][34], but globally [35]. In the Solomon Islands, household-based LLIN distributions have been continuously implemented since the early 2000 s, with annual indoor residual spraying (IRS) in selected high burden areas Table 4 Final set of nested models evaluated to determine which best predicted the prevalence of Plasmodium DNA-positives Model comparison was made on the basis ∆AIC, wAIC and goodness-of-fit using maximum likelihood estimation. The full list of explanatory variables included village, gender, temperature, age, bed net use and malaria history Here, the impact of LLIN use on the prevalence of P. falciparum was not detected, and this is likely a consequence of the low number of infections that were detected and the difficulty associated with detecting statistical significance at low transmission. LLIN use did significantly reduce P. vivax transmission with 94 % wAIC support.
While malaria transmission has been reduced, it is important to highlight the fragility of these reductions and the ability of malaria transmission to quickly rebound. In the Solomon Islands, malaria incidence rebounded from 20/1,000 people in 1976 to 450/1,000 in 1992 after the Malaria Eradication Programme ended [33]. A recent systematic review across the years 1930-2000 identified 75 malaria resurgences across the globe, with 91 % of resurgences resulting from delays in implementing malaria vector control strategies [36]. Hence, the World Health Organization (WHO) recommends that malaria control efforts, including vector control, must be maintained in receptive areas even after transmission has been eliminated. In the Solomon Islands, malaria transmission across the country has increased during the short time since this survey was completed. The reasons are multi-faceted and may include delays in vector control during decentralisation of the malaria control programme, decreased bioefficacy of LLINs [37], a shortage of LLINs per household, withdrawal of IRS and minimal use of primaquine to treat P. vivax hypnozoites [8].
The current WHO recommendations for elimination and post-elimination settings calls for stratified targeting of control efforts by transmission intensity zones [3,38]. In response, the Solomon Islands Strategic Plan for Malaria Control and Elimination, 2021-2025 identified 24 high malaria burden health zones by passive case detection in the provinces of Central Islands, Guadalcanal, Honiara, Makira, Malaita, Temotu and Western. This study confirmed the locations of high malaria burden health zones in three of these provinces, defining the residual malaria foci as small in size (e.g., individual villages), and provides evidence supporting the implementation of the strategic plan to target these foci effectively.
The strategic plan outlines the strategies for vector control, case management and surveillance and response. For vector control, the aim is to achieve universal coverage with LLINs, and to use IRS to rapidly reduce incidence in high transmission and outbreak areas. Previous work demonstrated that the risk of being bitten by An. farauti, the dominant malaria vector, occurs early in the evening and pre-dominantly in the peri-domestic areas of house verandas and kitchens [39]. Hence, targeted IRS in these high risk structures could enhance malaria control in residual malaria foci [40]. For case management, and in particular primaquine use, the lack of G6PD testing has been a barrier and the aim is to roll out point-of-care G6PD testing to improve usage of primaquine. For surveillance and response, the strategic plan outlines the steps for creating an elimination-ready case-based surveillance system for use in Isabel Province using reactive case detection (RACD) based on the 2-4-7 model. Cases will be reported within two days, investigated within four days and appropriate remedial measure taken within a week. Such stratified malaria control will require improved capacity [41] and data management to rapidly detect and report infection foci as well as to respond with targeted interventions against the vector and parasite [42].

Conclusions
Pockets of malaria were detected in highly localized foci in the Solomon Islands. This highlights the need for stratified malaria control with increased vector control in high transmission areas and case-based surveillance using RACD in low transmission areas with anti-malarial treatment including radical cure for P. vivax. Village-level stratification of targeted interventions could support the progress towards a cost-effective and successful elimination programme.

Funding
This study was funded by both the Australian Government through the Department of Foreign Affairs and Trade and the Global Good Fund I, LLC. The views expressed in this publication are those of the authors and are not necessarily the views of the Australian Government. The Australian Government neither endorses the views in this publication, nor vouches for the accuracy or completeness of the information contained within the publication. The Australian Government, its officers, employees and agents, accept no liability for any loss, damage or expense arising out of or in connection with, any reliance on any omissions or inaccuracies in the material contained in this publication.

Availability of data and materials
The datasets supporting the conclusions of this article are available in the JCU Tropical Data Hub repository, (https:// doi. org/ 10. 25903/ 4054-3m75).

Declarations
Ethics approval and consent to participate Community meetings were held with all village residents prior to the survey, where the aims, the possible risks and potential benefits of the study were explained in Solomon Islands Pidgin to potential participants (residents 5 years of age or older). Participation was voluntary with written informed consent obtained prior to enrolment from each adult participant > 18 years. For children between 13 and 18 years, signed consent was obtained from both the minor and a parent or guardian; for children between 5 and 13 years, signed consent was obtained from a parent or guardian. Ethical approvals for the study were obtained from the National Health Research & Ethics Committee, Solomon Islands (HRE066/17), the James Cook University Human Research Ethics Committee, Australia (H7107) and the London School of Hygiene and Tropical Medicine Observational Research Ethics Committee (16,167). The field sampling and subsequent analyses was performed in accordance with relevant guidelines and regulations of these research boards, and as stipulated in the approvals.

Consent for publication
Not applicable.