METHODS: A multi-host, multi-site transmission model was developed, taking into account the three areas (forest, farm, and village) where transmission is thought to occur. Latin hypercube sampling of model parameters was used to identify parameter sets consistent with possible prevalence in macaques and humans inferred from observed data. We then explore the consequences of increasing human-macaque contact in the farm, the likely impact of rapid treatment, and the use of long-lasting insecticide-treated nets (LLINs) in preventing wider spread of this emerging infection.

RESULTS: Identified model parameters were consistent with transmission being sustained by the macaques with spill over infections into the human population and with high overall basic reproduction numbers (up to 2267). The extent to which macaques forage in the farms had a non-linear relationship with human infection prevalence, the highest prevalence occurring when macaques forage in the farms but return frequently to the forest where they experience higher contact with vectors and hence sustain transmission. Only one of 1,046 parameter sets was consistent with sustained human-to-human transmission in the absence of macaques, although with a low human reproduction number (R(0H) = 1.04). Simulations showed LLINs and rapid treatment provide personal protection to humans with maximal estimated reductions in human prevalence of 42% and 95%, respectively.

RESULTS: A total of 1599 Anopheles specimens were collected in the village, of which about 90% were An. balabacensis. Anopheles balabacensis was present throughout the year and was the dominant Anopheles species in all habitat types. The shrub bushes habitat had the highest Anopheles species diversity while forest edge had the greatest number of Anopheles individuals caught. GLMM analysis indicated that An. balabacensis abundance was not affected by the type of habitats, and it was more active during the early and late night compared to predawn and dawn. PCR assay showed that 1.61% of the tested An. balabacensis were positive for malaria parasites, most of which were caught in oil palm estates and infected with one to two Plasmodium species.

METHODS AND ANALYSIS: A population-based case-control study will be conducted over a 2-year period at two adjacent districts in north-west Sabah, Malaysia. Confirmed malaria cases presenting to the district hospital sites meeting relevant inclusion criteria will be requested to enrol. Three community controls matched to the same village as the case will be selected randomly. Study procedures will include blood sampling and administration of household and individual questionnaires to evaluate potential exposure risks associated with acquisition of P. knowlesi malaria. Secondary outcomes will include differences in exposure variables between P. knowlesi and other Plasmodium spp, risk of severe P. knowlesi malaria, and evaluation of P. knowlesi case clustering. Primary analysis will be per protocol, with adjusted ORs for exposure risks between cases and controls calculated using conditional multiple logistic regression models.

METHODS: We did this population-based case-control study over a 2 year period in the state of Sabah in Malaysia. We enrolled cases with microscopy-positive, PCR-confirmed malaria who presented to two primary referral hospitals serving the adjacent districts of Kudat and Kota Marudu. We randomly selected three malaria-negative community controls per case, who were matched by village within 2 weeks of case detection. We obtained questionnaire data on demographics, behaviour, and residential malaria risk factors, and we also assessed glucose-6-phosphate dehydrogenase (G6PD) enzyme activity. We used conditional logistic regression models to evaluate exposure risk between P knowlesi cases and controls, and between P knowlesi and human-only Plasmodium spp malaria cases.

FINDINGS: From Dec 5, 2012, to Jan 30, 2015, we screened 414 patients and subsequently enrolled 229 cases with P knowlesi malaria mono-infection and 91 cases with other Plasmodium spp infection. We enrolled 953 matched controls, including 683 matched to P knowlesi cases and 270 matched to non-P knowlesi cases. Age 15 years or older (adjusted odds ratio [aOR] 4·16, 95% CI 2·09-8·29, p<0·0001), male gender (4·20, 2·54-6·97, p<0·0001), plantation work (3·50, CI, 1·34-9·15, p=0·011), sleeping outside (3·61, 1·48-8·85, p=0·0049), travel (2·48, 1·45-4·23, p=0·0010), being aware of the presence of monkeys in the past 4 weeks (3·35, 1·91-5·88, p<0·0001), and having open eaves or gaps in walls (2·18, 1·33-3·59, p=0·0021) were independently associated with increased risk of symptomatic P knowlesi infection. Farming occupation (aOR 1·89, 95% CI 1·07-3·35, p=0·028), clearing vegetation (1·89, 1·11-3·22, p=0·020), and having long grass around the house (2·08, 1·25-3·46, p=0·0048) increased risk for P knowlesi infection but not other Plasmodium spp infection. G6PD deficiency seemed to be protective against P knowlesi (aOR 0·20, 95% CI 0·04-0·96, p=0·045), as did residual insecticide spraying of household walls (0·52, 0·31-0·87, p=0·014), with the presence of young sparse forest (0·35, 0·20-0·63, p=00040) and rice paddy around the house (0·16, 0·03-0·78, 0·023) also associated with decreased risk.

INTERPRETATION: Adult men working in agricultural areas were at highest risk of knowlesi malaria, although peri-domestic transmission also occurrs. Human behavioural factors associated with P knowlesi transmission could be targeted in future public health interventions.

RESULTS: Using Open Data Kit GeoODK, we designed and piloted an electronic questionnaire for rolling cross sectional surveys of health facility attendees as part of a malaria elimination campaign in two predominantly rural sites in the Rizal, Palawan, the Philippines and Kulon Progo Regency, Yogyakarta, Indonesia. The majority of health workers were able to use the tablets effectively, including locating participant households on electronic maps. For all households sampled (n = 603), health facility workers were able to retrospectively find the participant household using the Global Positioning System (GPS) coordinates and data collected by tablet computers. Median distance between actual house locations and points collected on the tablet was 116 m (IQR 42-368) in Rizal and 493 m (IQR 258-886) in Kulon Progo Regency. Accuracy varied between health facilities and decreased in less populated areas with fewer prominent landmarks.

METHODOLOGY/PRINCIPAL FINDINGS: Using in silico tools, we designed and expressed four novel P. knowlesi protein products to address the distinct lack of suitable serosurveillance tools: PkSERA3 antigens 1 and 2, PkSSP2/TRAP and PkTSERA2 antigen 1. Antibody prevalence to these antigens was determined by ELISA for three time-points post-treatment from a hospital-based clinical treatment trial in Sabah, East Malaysia (n = 97 individuals; 241 total samples for all time points). Higher responses were observed for the PkSERA3 antigen 2 (67%, 65/97) across all time-points (day 0: 36.9% 34/92; day 7: 63.8% 46/72; day 28: 58.4% 45/77) with significant differences between the clinical cases and controls (n = 55, mean plus 3 SD) (day 0 p<0.0001; day 7 p<0.0001; day 28 p<0.0001). Using boosted regression trees, we developed models to classify P. knowlesi exposure (cross-validated AUC 88.9%; IQR 86.1-91.3%) and identified the most predictive antibody responses.