Predicting COVID-19 Infections in Eswatini Using the Maximum Likelihood Estimation Method.

COVID-19 country spikes have been reported at varying temporal scales as a result of differences in the disease-driving factors. Factors affecting case load and mortality rates have varied between countries and regions. We investigated the association between socio-economic, weather, demographic and health variables with the reported cases of COVID-19 in Eswatini using the maximum likelihood estimation method for count data. A generalized Poisson regression (GPR) model was fitted with the data comprising 15 covariates to predict COVID-19 risk in the whole of Eswatini. The results show that the variables that were key determinants in the spread of the disease were those that included the proportion of elderly above 55 years at 98% (95% CI: 97-99%) and the proportion of youth below the age of 35 years at 8% (95% CI: 1.7-38%) with a pseudo R-square of 0.72. However, in the early phase of the virus when cases were fewer, results from the Poisson regression showed that household size, household density and poverty index were associated with reported COVID-19 cases in the country. We then produced a disease-risk map of predicted COVID-19 in Eswatini using variables that were selected by the regression model at a 5% significance level. The map could be used by the country to plan and prioritize health interventions against COVID-19. The identified areas of high risk may be further investigated to find out the risk amplifiers and assess what could be done to prevent them.

Bayesian Geostatistical Modeling to Assess Malaria Seasonality and Monthly Incidence Risk in Eswatini.

Eswatini is on the brink of malaria elimination and had however, had to shift its target year to eliminate malaria on several occasions since 2015 as the country struggled to achieve its zero malaria goal. We conducted a Bayesian geostatistical modeling study using malaria case data. A Bayesian distributed lags model (DLM) was implemented to assess the effects of seasonality on cases. A second Bayesian model based on polynomial distributed lags was implemented on the dataset to improve understanding of the lag effect of environmental factors on cases. Results showed that malaria increased during the dry season with proportion 0.051 compared to the rainy season with proportion 0.047 while rainfall of the preceding month (Lag2) had negative effect on malaria as it decreased by proportion - 0.25 (BCI: - 0.46, - 0.05). Night temperatures of the preceding first and second month were significantly associated with increased malaria in the following proportions: at Lag1 0.53 (BCI: 0.23, 0.84) and at Lag2 0.26 (BCI: 0.01, 0.51). Seasonality was an important predictor of malaria with proportion 0.72 (BCI: 0.40, 0.98). High malaria rates were identified for the months of July to October, moderate rates in the months of November to February and low rates in the months of March to June. The maps produced support-targeted malaria control interventions. The Bayesian geostatistical models could be extended for short-term and long-term forecasting of malaria supporting-targeted response both in space and time for effective elimination.

Assessing the relationship between environmental factors and malaria vector breeding sites in Swaziland using multi-scale remotely sensed data.

Many entomological studies have analyzed remotely sensed data to assess the relationship between malaria vector distribution and the associated environmental factors. However, the high cost of remotely sensed products with high spatial resolution has often resulted in analyses being conducted at coarse scales using open-source, archived remotely sensed data. In the present study, spatial prediction of potential breeding sites based on multi-scale remotely sensed information in conjunction with entomological data with special reference to presence or absence of larvae was realized. Selected water bodies were tested for mosquito larvae using the larva scooping method, and the results were compared with data on land cover, rainfall, land surface temperature (LST) and altitude presented with high spatial resolution. To assess which environmental factors best predict larval presence or absence, Decision Tree methodology and logistic regression techniques were applied. Both approaches showed that some environmental predictors can reliably distinguish between the two alternatives (existence and non-existence of larvae). For example, the results suggest that larvae are mainly present in very small water pools related to human activities, such as subsistence farming that were also found to be the major determinant for vector breeding. Rainfall, LST and altitude, on the other hand, were less useful as a basis for mapping the distribution of breeding sites. In conclusion, we found that models linking presence of larvae with high-resolution land use have good predictive ability of identifying potential breeding sites.