Hierarchical Poisson Model Research Articles

Ecological change increases malaria risk in the Brazilian Amazon

Ecological change in the Brazilian Amazon is closely linked to human mobility and health. Mining, agriculture, logging, and other activities alter highly diverse ecological and demographic contexts and subsequent exposure to diseases such as malaria. Studies that have attempted to quantify the impact of deforestation on malaria in the Brazilian Amazon have produced conflicting results. However, they varied in methodology and data sources. Most importantly, all studies used annual data, neglecting the subannual seasonal dynamics of malaria. Here, we fill the knowledge gap on the subannual relationship between ecological change in the Brazilian Amazon and malaria transmission. Using the highest spatiotemporal resolution available, we estimated the effect of deforestation on malaria cases between 2003 and 2022 using a stratified Bayesian spatiotemporal hierarchical zero-inflated Poisson model fitted with the Integrated Nested Laplace Approximation. The model was also stratified by state. We found that a 1% increase in 1-mo lagged deforestation increased malaria cases in a given month and municipality by 6.3% [95% credible interval (Crl): 6.2, 6.5%]. Based on an interaction term included in the model, the effect of deforestation on malaria was even larger in areas with higher forest cover. We found that the coefficients for deforestation and mobility were highly variable when stratified by state. Our results provide detailed evidence that, on average, deforestation increases malaria transmission, but that the relationship is not spatiotemporally uniform. These results have implications for stratifying malaria control interventions based on ecological dynamics to help Brazil achieve its goal of malaria elimination by 2035.

A Multi-institution Study on the Association of Virtual Reality Skills with Continence Recovery after Robot-assisted Radical Prostatectomy

BackgroundVirtual reality (VR) simulators are increasingly being used for surgical skills training. It is unclear what skills are best improved via VR, translate to live surgical skills, and influence patient outcomes. ObjectiveTo assess surgeons in VR and live surgery using a suturing assessment tool and evaluate the association between technical skills and a clinical outcome. Design, setting, and participantsThis prospective five-center study enrolled participants who completed VR suturing exercises and provided live surgical video. Graders provided skill assessments using the validated End-To-End Assessment of Suturing Expertise (EASE) suturing evaluation tool. Outcome measurements and statistical analysisA hierarchical Poisson model was used to compare skill scores among cohorts and evaluate the association of scores with clinical outcomes. Spearman’s method was used to assess correlation between VR and live skills. Results and limitationsTen novices, ten surgeons with intermediate expertise (median 64 cases, interquartile range [IQR] 6–80), and 26 expert surgeons (median 850 cases, IQR 375–3000) participated in this study. Intermediate and expert surgeons were significantly more likely to have ideal scores in comparison to novices for the subskills needle hold angle, wrist rotation, and wrist rotation needle withdrawal (p < 0.01). For both intermediate and expert surgeons, there was positive correlation between VR and live skills for needle hold angle (p < 0.05). For expert surgeons, there was a positive association between ideal scores for VR needle hold angle and driving smoothness subskills and 3-mo continence recovery (p < 0.05). Limitations include the size of the intermediate surgeon sample and clinical data limited to expert surgeons. ConclusionsEASE can be used in VR to identify skills to improve for trainee surgeons. Technical skills that influence postoperative outcomes may be assessable in VR. Patient summaryThis study provides insights into surgical skills that translate from virtual simulation to live surgery and that have an impact on urinary continence after robot-assisted removal of the prostate. We also highlight the usefulness of virtual reality in surgical education.

Understanding Social Disorganization and the Nonprofit Infrastructure: An Ecological Study of Child Maltreatment Rates

Advocates and researchers have emphasized the role of disorder in neighborhood processes, with serious consequences for families, however, neighborhood structures may also support families and reduce child maltreatment. Nonprofits maintain a range of strategies to support nearby families, including direct services, facilitation of social networks, and formalizing advocacy for increased attention from government. Using agency data on child maltreatment, nonprofit locations, and indicators of social disorganization, this article studies the role of nonprofit organizations in the spatial distribution of child maltreatment among Cuyahoga County Ohio census tracts ( N = 442). Accounting for spatially structured and tract-specific variation with a hierarchical Poisson model implemented through a Bayesian methodology, the results indicate the presence of nonprofits is a protective factor: negatively associated with child maltreatment (posterior mean: −0.18, CI: −0.34, −0.03), with heterogeneity by type. The study highlights neighborhoods as a propitious site of intervention and emphasizes the intra-county distribution of nonprofits.

Spatial patterning of occupational stress and its related factors in Iranian critical care nurses using a hierarchical Bayesian technique.

In each community, health problems' patterns and geographical changes are of prime importance to determine high and low-risk areas. This study aimed to investigate the Spatial patterning of occupational stress and its related factors in Iranian critical care nurses using a hierarchical Bayesian techniqueMETHODS:The current research was a cross-sectional descriptive-analytical study. The data includes the number of critical care unit nurses who show a high stress level based on a questionnaire. We used variables such as age, gender, collaboration status, working time, marital status, clinical experience, education, supervisor support, stress score, and working on holiday days for this study. The survey participants had to be at least 18 years old, a registered nurse, and working in the intensive care unit (ICU). OpenBUGS version 3.2.3 was used to implement the Bayesian hierarchical Poisson model and find partial patterning of occupational stress and its related factors. The final sample size was 17414 nurses. The overall prevalence of occupational stress in ICU nurses was estimated at 70%. The lowest and highest prevalence was 65.8% in the North Khorasan province and 75.2% in Golestan province. Occupational stress had a statistically significant association with collaboration status, but with demographic variables, shift work, supportive supervisor, and working on holidays had no statistically significant association. According to the findings, it is necessary to eliminate or reduce job stress and increase efficiency in Iranian nurses, encourage teamwork and collaboration as an essential element of a healthy workplace environment.

BAHAMA: A Bayesian Hierarchical Model for the Detection of MedDRA®-Coded Adverse Events in Randomized Controlled Trials.

IntroductionPatients participating in randomized controlled trials (RCTs) are susceptible to a wide range of different adverse events (AE) during the RCT. MedDRA® is a hierarchical standardization terminology to structure the AEs reported in an RCT. The lowest level in the MedDRA hierarchy is a single medical event, and every higher level is the aggregation of the lower levels.MethodWe propose a multi-stage Bayesian hierarchical Poisson model for estimating MedDRA-coded AE rate ratios (RRs). To deal with rare AEs, we introduce data aggregation at a higher level within the MedDRA structure and based on thresholds on incidence and MedDRA structure.ResultsWith simulations, we showed the effects of this data aggregation process and the method's performance. Furthermore, an application to a real example is provided and compared with other methods.ConclusionWe showed the benefit of using the full MedDRA structure and using aggregated data. The proposed model, as well as the pre-processing, is implemented in an R-package: BAHAMA.

Effectiveness of controlling COVID-19 epidemic by implementing soft lockdown policy and extensive community screening in Taiwan

Strict and repeated lockdowns have caused public fatigue regarding policy compliance and had a large impact on several countries’ economies. We aimed to evaluate the effectiveness of a soft lockdown policy and the strategy of active community screening for controlling COVID-19 in Taiwan. We used village-based daily confirmed COVID-19 statistics in Taipei City and New Taipei City, between May 2, 2021, and July 17, 2021. The temporal Gi* statistic was used to compute the spatiotemporal hotspots. Simple linear regression was used to evaluate the trend of the epidemic, positivity rate from community screening, and mobility changes in COVID-19 cases and incidence before and after a level three alert in both cities. We used a Bayesian hierarchical zero-inflated Poisson model to estimate the daily infection risk. The cities accounted for 11,403 (81.17%) of 14,048 locally confirmed cases. The mean effective reproduction number (Re) surged before the level three alert and peaked on May 16, 2021, the day after the level three alert in Taipei City (Re = 3.66) and New Taipei City (Re = 3.37). Mobility reduction and a lower positive rate were positively associated with a lower number of cases and incidence. In the spatiotemporal view, seven major districts were identified with a radial spreading pattern from one hard-hit district. Villages with a higher inflow degree centrality among people aged ≥ 60 years, having confirmed cases, specific land-use types, and with a higher aging index had higher infection risks than other villages. Early soft lockdown policy and detection of infected patients showed an effective strategy to control COVID-19 in Taiwan.

A Bayesian Approach to Predict Football Matches with Changed Home Advantage in Spectator-Free Matches after the COVID-19 Break

Since the coronavirus disease 2019 (COVID-19) pandemic, most professional sports events have been held without spectators. It is generally believed that home teams deprived of enthusiastic support from their home fans experience reduced benefits of playing on their home fields, thus becoming less likely to win. This study attempts to confirm if this belief is true in four major European football leagues through statistical analysis. This study proposes a Bayesian hierarchical Poisson model to estimate parameters reflecting the home advantage and the change in such advantage. These parameters are used to improve the performance of machine-learning-based prediction models for football matches played after the COVID-19 break. The study describes the statistical analysis on the impact of the COVID-19 pandemic on football match results in terms of the expected score and goal difference. It also shows that estimated parameters from the proposed model reflect the changed home advantage. Finally, the study verifies that these parameters, when included as additional features, enhance the performance of various football match prediction models. The home advantage in European football matches has changed because of the behind-closed-doors policy implemented due to the COVID-19 pandemic. Using parameters reflecting the pandemic’s impact, it is possible to predict more precise results of spectator-free matches after the COVID-19 break.

PEMODELAN KLAIM ASURANSI MENGGUNAKAN PENDEKATAN BAYESIAN DAN MARKOV CHAIN MONTE CARLO

The determination of the correct prediction of claims frequency and claims severity is very important in the insurance business to determine the outstanding claims reserve which should be prepared by an insurance company. One approach which may be used to predict a future value is the Bayesian approach. This approach combines the sample and the prior information The information is used to construct the posterior distribution and to determine the estimate of the parameters. However, in this approach, integrations of functions with high dimensions are often encountered. In this Thesis, a Markov Chain Monte Carlo (MCMC) simulation is used using the Gibbs Sampling algorithm to solve the problem. The MCMC simulation uses ergodic chain property in Markov Chain. In Ergodic Markov Chain, a stationary distribution, which is the target distribution, is obtained. The MCMC simulation is applied in Hierarchical Poisson Model. The OpenBUGS software is used to carry out the tasks. The MCMC simulation in Hierarchical Poisson Model can predict the claims frequency.

Vulnerable road-user deaths in Brazil: a Bayesian hierarchical model for spatial-temporal analysis

Reducing the road traffic injuries burden is relevant to many sustainable development goals (SDG), in particular SDG3 – to establish good health and well-being. To describe the spatial-temporal trends and identify hotspot regions for fatal road traffic injuries, a Bayesian hierarchical Poisson model was used to analyze data on vulnerable road users (bicyclist, motorcyclist and pedestrians) in Brazil from 1999 to 2016. During the study period, mortality rates for bicyclists remained almost unchanged (0.6 per 100,000 people) but rose dramatically for motorcyclists (from 1.0 in 1999 to 6.0 per 100,000 people in 2016) and decreased for pedestrians (from 6.3 to 3.0 per 100,000 people). Spatial analyses accounting for socio-economic factors showed that the central and northeastern microregions of Brazil are hotspot areas for fatal injuries among motorcyclists while the southern areas are for pedestrians.

Fine-scale geographic variations of rates of renal replacement therapy in northeastern France: Association with the socioeconomic context and accessibility to care.

BackgroundThe strong geographic variations in the incidence rates of renal replacement therapy (RRT) for end-stage renal disease are not solely related to variations in the population's needs, such as the prevalence of diabetes or the deprivation level. Inequitable geographic access to health services has been involved in different countries but never in France, a country with a generous supply of health services and where the effect of the variability of medical practices was highlighted in an analysis conducted at the geographic scale of districts. Our ecological study, performed at the finer scale of townships in a French area of 8,370,616 inhabitants, investigated the association between RRT incidence rates, socioeconomic environment and geographic accessibility to healthcare while adjusting for morbidity level and medical practice patterns.MethodsUsing data from the Renal Epidemiology and Information Network registry, we estimated age-adjusted RRT incidence rates during 2010–2014 for the 282 townships of the area. A hierarchical Bayesian Poisson model was used to examine the association between incidence rates and 18 contextual variables describing population health status, socioeconomic level and health services characteristics. Relative risks (RRs) and 95% credible intervals (95% CrIs) for each variable were estimated for a 1-SD increase in incidence rate.ResultsDuring 2010–2014, 6,835 new patients ≥18 years old (4231 men, 2604 women) living in the study area started RRT; the RRT incidence rates by townships ranged from 21 to 499 per million inhabitants. In multivariate analysis, rates were related to the prevalence of diabetes [RR (95% CrI): 1.05 (1.04–1.11)], the median estimated glomerular filtration rate at dialysis initiation [1.14 (1.08–1.20)], and the proportion of incident patients ≥ 85 years old [1.08 (1.03–1.14)]. After adjusting for these factors, rates in townships increased with increasing French deprivation index [1.05 (1.01–1.08)] and decreased with increasing mean travel time to reach the closest nephrologist [0.92 (0.89–0.95]).ConclusionThese data confirm the influence of deprivation level, the prevalence of diabetes and medical practices on RRT incidence rates across a large French area. For the first time, an association was found with the distance to nephrology services. These data suggest possible inequitable geographic access to RRT within the French health system.

Decline of the boreal willow grouse (Lagopus\xa0lagopus) has been accelerated by more frequent snow-free springs

Climate change has influenced a range of species across the globe. Yet, to state a noted decline in the abundance of a given species as a consequence of a specific environmental change, for instance, spatially explicit long-term data are a prerequisite. This study assessed the extent to which prolonged snow-free periods in autumn and spring have contributed to the decline of the willow grouse, the only forest grouse changing into a white winter plumage. Time-series data of willow grouse numbers from summer surveys across the study area were integrated with local data on weather (snow cover), mammalian predator abundance and hunting intensity. Modelling was conducted with a hierarchical Bayesian Poisson model, acknowledging year-, area- and location-specific variability. The results show that while willow grouse numbers had decreased continuously across the study landscapes, the decrease was accelerated at the sites where, and during the years when the preceding April was the most snow-free. This indicates a mismatch between the change into a white winter plumage and the presence of snow, turning the bird into an ill-camouflaged prey. The results thus also confirm past hypotheses where local declines of the species have been attributed to prolonged snow-free periods. Across our study area, autumns and springs have become more snow-free, and the trend has been predicted to continue. Thus, in addition to conservation actions, the future of a species such as the willow grouse is also dependent on its ability to adapt to the changed environmental conditions.

Do Coarser Gap Mosaics in Conifer Plantations Induce More Seed Dispersal by Birds? Temporal Changes during 12 Years after Gap Creation

The creation of canopy gaps is thought to be an efficient silvicultural operation to diversify species composition of monoculture conifer plantations; however, the shortage of regeneration materials in overclosed plantations is one of the concerns related to this operation. Seed dispersal by frugivorous birds may play an important role in inducing the spread of native broad-leaved trees in canopy gaps in plantations in the warm temperate zone of central Japan because bird-dispersed woody species are abundant in this area. We monitored the dynamics of the abundance and species composition of bird-dispersed seeds over 12 years after gap creation in the canopy of a Japanese cypress (Chamaecyparis obtusa (Siebold et Zucc.) Endl.) plantation. We also studied the effects of gap-mosaic patterns (from many small gaps to fewer large gaps) on dispersal. We used a hierarchical Bayesian zero-inflated Poisson model to analyze the factors affecting seed dispersal by frugivorous birds. Seed dispersal by birds increased with gap age except for just after gap creation. Dispersal in coarser gap mosaics was more abundant than that in finer gap mosaics. The species diversity of dispersed seeds in each seed trap (α-diversity) and plot (γ-diversity) showed similar trends in terms of temporal changes and differences between plots related to seed dispersal abundance; β-diversity did not. These trends might have been caused by shrub-vegetation development after gap creation. The effects of shrub vegetation were classified into the attraction by fruits borne within the vegetation, as well as other effects related to vegetation, such as functions of perch availability and insect presence as a food source. The presence of bird-dispersed seeds was strongly promoted by vegetation in all seasons, but only marginally by the presence of fruit-bearing vegetation. However, fruits attracted seed dispersal by frugivorous birds in the winter season. Our results suggest that both vegetation development and fruiting are important for promoting seed dispersal by frugivorous birds, and those effects are different in different seasons according to vegetation conditions and shifting food resources.

The empirical Bayes estimators of the parameter of the Poisson distribution with a conjugate gamma prior under Stein's loss function

For the hierarchical Poisson and gamma model, we calculate the Bayes posterior estimator of the parameter of the Poisson distribution under Stein's loss function which penalizes gross overestimation and gross underestimation equally and the corresponding Posterior Expected Stein's Loss (PESL). We also obtain the Bayes posterior estimator of the parameter under the squared error loss and the corresponding PESL. Moreover, we obtain the empirical Bayes estimators of the parameter of the Poisson distribution with a conjugate gamma prior by two methods. In numerical simulations, we have illustrated: The two inequalities of the Bayes posterior estimators and the PESLs; the moment estimators and the Maximum Likelihood Estimators (MLEs) are consistent estimators of the hyperparameters; the goodness-of-fit of the model to the simulated data. The numerical results indicate that the MLEs are better than the moment estimators when estimating the hyperparameters. Finally, we exploit the attendance data on 314 high school juniors from two urban high schools to illustrate our theoretical studies.

Comparing spatio-temporal methods of non-communicable disease surveillance.

ObjectiveTo determine the merits of different surveillance methods for cluster detection, in particular when used in conjuction with small area data. This will be investigated using a simulated framework. This is with a view to support further surviellance work using real small area data.IntroductionHealth surveillance is well established for infectious diseases, but less so for non-communicable diseases. When spatio-temporal methods are used, selection often appears to be driven by arbitrary criteria, rather than optimal detection capabilities. Our aim is to use a theoretical simulation framework with known spatio-temporal clusters to investigate the sensitivity and specificity of several traditional (e.g. SatScan and Cusum) and Bayesian (incl. BaySTDetect and Dcluster) statistical methods for spatio-temporal cluster detection of non-communicable disease.MethodsCount data were generated using various random effects (RE). A subset of areas was randomly given an increased relative risk (RR) to simulate disease clusters. Simulations were conducted in R using a grid of 625 areas. We used 12 times= nteps within a hierarchical Poisson model. Multiple values of model parameters, including REs and the RR within clusters, were then tested. The range of RE (values) was derived from real-world data from England on common and rare diseases. RR ranging between 1.2 and 1.8 were tested to reflect both low and high exposures to pollutants and other risk factors. ROC analysis, based on 50 simulations, was used to assess the performance of each statistical method for each combination of parameter values.ResultsOur ROC analysis suggested that SaTScan usually had the highest specificity at low sensitivities (&lt;0.5), although its maximum sensitivity was often lower than when using the Bayesian methods. In scenarios where the RR within clusters was lower, all methods had less sensitivity at a given specificity. Cusum usually performed quite similarly to SatScan, while the two Bayesian methods considered often misidentified a high proportion of disease clusters. P-values generated by SaTScan need to be considered with caution as they did not relate closely with the sensitivity or specificity of the ROC curves from our simulations.ConclusionsReal-world investigations of spatio-temporal signals (e.g. disease clusters) are often complex and time consuming. Identifying the best method to reduce the risks of identifying false positives and of missing real clusters is therefore essential. Despite the inherent constraints of theoretical simulations, such a framework allows to objectively assess the performance of different methods. Overall, our simulation framework suggested that SatScan would usually be the easiest, most user-friendly and best performing space-time methods for non-communicable disease surveillance.

Measuring and forecasting mobile game app engagement

Monetization strategies in the large and strongly growing mobile app space must go beyond traditional purchase revenue as most mobile apps are now free to download. One key marketing innovation allows mobile app publishers to monetize ongoing engagement — in-app advertising.We study ongoing user engagement with a mobile app after the initial download decision using the $40 billion mobile gaming industry as an example. Our study investigates and forecasts user engagement after the initial download aiming to help publishers to monetize their engagement via in-app advertising. We leverage a novel dataset containing user-level engagement for 193 mobile games and propose a hierarchical Poisson model on a mobile-game level. We find significant usage heterogeneity across the mobile games studied and generate forecasts publishers can use when trying to monetize engagement via pre-sold contracts.

Combining historical data and bookmakers’ odds in modelling football scores

Modelling football outcomes has gained increasing attention, in large part due to the potential for making substantial profits. Despite the strong connection existing between football models and the bookmakers’ betting odds, no authors have used the latter for improving the fit and the predictive accuracy of these models. We have developed a hierarchical Bayesian Poisson model in which the scoring rates of the teams are convex combinations of parameters estimated from historical data and the additional source of the betting odds. We apply our analysis to a nine-year dataset of the most popular European leagues in order to predict match outcomes for their tenth seasons. In this article, we provide numerical and graphical checks for our model.

Saying Sayonara to the Farm: Hierarchical Bayesian Modeling of Farm Exits in Japan

AbstractOff‐farm employment opportunities are thought to have an effect on farm exit rates, though evidence on the sign of this effect has been mixed. Examining this issue in the context of Japanese agriculture, we find that farm exits are related to off‐farm income as a share of household income, and more specifically to the nature of off‐farm work. Two econometric models are developed: a hierarchical Bayesian linear model and a hierarchical Bayesian Poisson model. Both models perform well in predicting exit rates across the towns and prefectures of Japan.

Hierarchical Bayesian modeling of spatio-temporal patterns of scrub typhus incidence for 2009–2013 in South Korea

This study aims to analyze spatio-temporal patterns of scrub typhus incidence and to identify environmental risk factors in South Korea from 2009 to 2013 using hierarchical Bayesian Poisson model. We classified 244 municipal areas into nine categories based on the posterior relative risks and their temporal trends and these patterns were visualized. Environmental covariates (i.e. land surface temperature, precipitation and vegetation) were compiled by season to reflect the seasonal behavior of trombiculid mites and rodents and by year to explain year-based environmental changes. A total of six covariates were incorporated in the regression model and environmental effects were evaluated. We found areas in the south, southwest, and southeast of South Korea that present relatively high risk, and the north and southeast parts show rapid increases during the study period. The spatio-temporal maps present similar patterns as well. A positive correlation is found between the scrub typhus risk and precipitation in summer, and normalized difference vegetation index in fall. In contrast, the scrub typhus risk is negatively correlated with land surface temperature in fall. This information has implications for prioritizing disease control resources on high-risk areas. Furthermore, the regression results might be utilized for predicting scrub typhus incidence and contribute to the design of disease control programs in the future.

Year-round monitoring reveals prevalence of fatal bird-window collisions at the Virginia Tech Corporate Research Center.

Collisions with glass are a serious threat to avian life and are estimated to kill hundreds of millions of birds per year in the United States. We monitored 22 buildings at the Virginia Tech Corporate Research Center (VTCRC) in Blacksburg, Virginia, for collision fatalities from October 2013 through May 2015 and explored possible effects exerted by glass area and surrounding land cover on avian mortality. We documented 240 individuals representing 55 identifiable species that died due to collisions with windows at the VTCRC. The relative risk of fatal collisions at all buildings over the study period were estimated using a Bayesian hierarchical zero-inflated Poisson model adjusting for percentage of tree and lawn cover within 50 m of buildings, as well as for glass area. We found significant relationships between fatalities and surrounding lawn area (relative risk: 0.96, 95% credible interval: 0.93, 0.98) as well as glass area on buildings (RR: 1.30, 95% CI [1.05–1.65]). The model also found a moderately significant relationship between fatal collisions and the percent land cover of ornamental trees surrounding buildings (RR = 1.02, 95% CI [1.00–1.05]). Every building surveyed had at least one recorded collision death. Our findings indicate that birds collide with VTCRC windows during the summer breeding season in addition to spring and fall migration. The Ruby-throated Hummingbird (Archilochus colubris) was the most common window collision species and accounted for 10% of deaths. Though research has identified various correlates with fatal bird-window collisions, such studies rarely culminate in mitigation. We hope our study brings attention, and ultimately action, to address this significant threat to birds at the VTCRC and elsewhere.

Spatial Determinants of Ebola Virus Disease Risk for the West African Epidemic.

Introduction: Although many studies have investigated the probability of Ebola virus disease (EVD) outbreaks while other studies have simulated the size and speed of EVD outbreaks, few have investigated the environmental and population-level predictors of Ebola transmission once an outbreak is underway. Identifying strong predictors of transmission could help guide and target limited public health resources during an EVD outbreak. We examined several environmental and population-level demographic predictors of EVD risk from the West African epidemic. Methods: We obtained district-level estimates from the World Health Organization EVD case data, demographic indicators obtained from the Demographic and Health surveys, and satellite-derived temperature, rainfall, and land cover estimates. A Bayesian hierarchical Poisson model was used to estimate EVD risk and to evaluate the spatial variability explained by the selected predictors. Results: We found that districts had greater risk of EVD with increasing proportion of households not possessing a radio (RR 2.79, 0.90-8.78; RR 4.23, 1.16-15.93), increasing rainfall (RR 2.18; 0.66-7.20; 5.34, 1.20-23.90), and urban land cover (RR 4.87, 1.56-15.40; RR 5.74, 1.68-19.67). Discussion: The finding of radio ownership and reduced EVD transmission risk suggests that the use of radio messaging for control and prevention purposes may have been crucial in reducing the EVD transmission risk in certain districts, although this association requires further study. Future research should examine the etiologic relationships between the identified risk factors and human-to-human transmission of EVD with a focus on factors related to population mobility and healthcare accessibility, which are critical features of epidemic propagation and control.