A Bayesian non-stationary heteroskedastic time series model for multivariate critical care data.

We propose a multivariate GARCH model for non-stationary health time series by modifying the observation-level variance of the standard state space model. The proposed model provides an intuitive and novel way of dealing with heteroskedastic data using the conditional nature of state-space models. We follow the Bayesian paradigm to perform the inference procedure. In particular, we use Markov chain Monte Carlo methods to obtain samples from the resultant posterior distribution. We use the forward filtering backward sampling algorithm to efficiently obtain samples from the posterior distribution of the latent state. The proposed model also handles missing data in a fully Bayesian fashion. We validate our model on synthetic data and analyze a data set obtained from an intensive care unit in a Montreal hospital and the MIMIC dataset. We further show that our proposed models offer better performance, in terms of WAIC than standard state space models. The proposed model provides a new way to model multivariate heteroskedastic non-stationary time series data. Model comparison can then be easily performed using the WAIC.

Identifying deprived "slum" neighbourhoods in the Greater Accra Metropolitan Area of Ghana using census and remote sensing data.

Background: Identifying urban deprived areas, including slums, can facilitate more targeted planning and development policies in cities to reduce socio-economic and health inequities, but methods to identify them are often ad-hoc, resource intensive, and cannot keep pace with rapidly urbanizing communities. Objectives: We apply a spatial modelling approach to identify census enumeration areas (EAs) in the Greater Accra Metropolitan Area (GAMA) of Ghana with a high probability of being a deprived area using publicly available census and remote sensing data. Methods: We obtained United Nations (UN) supported field mapping data that identified deprived "slum" areas in Accra's urban core, data on housing and population conditions from the most recent census, and remotely sensed data on environmental conditions in the GAMA. We first fitted a Bayesian logistic regression model on the data in Accra's urban core (n=2,414 EAs) that estimated the relationship between housing, population, and environmental predictors and being a deprived area according to the UN's deprived area assessment. Using these relationships, we predicted the probability of being a deprived area for each of the 4,615 urban EAs in GAMA. Results: 899 (19%) of the 4,615 urban EAs in GAMA, with an estimated 745,714 residents (22% of its urban population), had a high predicted probability (≥80%) of being a deprived area. These deprived EAs were dispersed across GAMA and relatively heterogeneous in their housing and environmental conditions, but shared some common features including a higher population density, lower elevation and vegetation abundance, and less access to indoor piped water and sanitation. Conclusion: Our approach using ubiquitously available administrative and satellite data can be used to identify deprived neighbourhoods where interventions are warranted to improve living conditions, and track progress in achieving the Sustainable Development Goals aiming to reduce the population living in unsafe or vulnerable human settlements.

Within city spatiotemporal variation of pollen concentration in the city of Toronto, Canada.

BACKGROUND: The exacerbation of asthma and respiratory allergies has been associated with exposure to aeroallergens such as pollen. Within an urban area, tree cover, level of urbanization, atmospheric conditions, and the number of source plants can influence spatiotemporal variations in outdoor pollen concentrations. OBJECTIVE: We analyze weekly pollen measurements made between March and October 2018 over 17 sites in Toronto, Canada. The main goals are: to estimate the concentration of different types of pollen across the season; estimate the association, if any, between pollen concentration and environmental variables, and provide a spatiotemporal surface of concentration of different types of pollen across the weeks in the studied period. METHODS: We propose an extension of the land-use regression model to account for the temporal variation of pollen levels and the high number of measurements equal to zero. Inference is performed under the Bayesian framework, and uncertainty of predicted values is naturally obtained through the posterior predictive distribution. RESULTS: Tree pollen was positively associated with commercial areas and tree cover, and negatively associated with grass cover. Both grass and weed pollen were positively associated with industrial areas and TC brightness and negatively associated with the northing coordinate. The total pollen was associated with a combination of these environmental factors. Predicted surfaces of pollen concentration are shown at some sampled weeks for all pollen types. SIGNIFICANCE: The predicted surfaces obtained here can help future epidemiological studies to find possible associations between pollen levels and some health outcome like respiratory allergies at different locations within the study area.

Estimating the lagged effect of price discounting: a time-series study on sugar sweetened beverage purchasing in a supermarket.

BACKGROUND: Price discount is an unregulated obesogenic environmental risk factor for the purchasing of unhealthy food, including Sugar Sweetened Beverages (SSB). Sales of price discounted food items are known to increase during the period of discounting. However, the presence and extent of the lagged effect of discounting, a sustained level of sales after discounting ends, is previously unaccounted for. We investigated the presence of the lagged effect of discounting on the sales of five SSB categories, which are soda, fruit juice, sport and energy drink, sugar-sweetened coffee and tea, and sugar-sweetened drinkable yogurt. METHODS: We fitted distributed lag models to weekly volume-standardized sales and percent discounting generated by a supermarket in Montreal, Canada between January 2008 and December 2013, inclusive (n = 311 weeks). RESULTS: While the sales of SSB increased during the period of discounting, there was no evidence of a prominent lagged effect of discounting in four of the five SSB; the exception was sports and energy drinks, where a posterior mean of 28,459 servings (95% credible interval: 2661 to 67,253) of excess sales can be attributed to the lagged effect in the target store during the 6 years study period. CONCLUSION: Our results indicate that studies that do not account for the lagged effect of promotions may not fully capture the effect of price discounting for some food categories.

Socioeconomic factors and bacillary dysentery risk in Jiangsu Province, China: a spatial investigation using Bayesian hierarchical models.

Bacillary dysentery (BD) is an acute diarrheal disease prevalent in areas affected by socioeconomic disparities. We investigated BD risk and its associations with socioeconomic factors at the county-level in Jiangsu province, China using epidemiological and socioeconomic data from 2011-2014. We fitted four Bayesian hierarchical models with various prior specifications for random effects. As all model comparison criteria values were similar, we presented results from a reparameterized Besag-York-Mollié model, which addressed issues with the identifiability of variance captured by spatial and independent effects. Our model adjusted for year and socioeconomic status showed 18-65% decreased BD risk compared to 2011. We found a high relative risk in the northwestern and southwestern counties. Increasing the percentage of rural households, rural income per capita, health institutions per capita, or hospital beds per capita decreases the relative risk of BD, respectively. Our findings can be used to improve infectious diarrhea surveillance and enhance existing public health interventions.

Extended Bayesian endemic-epidemic models to incorporate mobility data into COVID-19 forecasting.

Forecasting the number of daily COVID-19 cases is critical in the short-term planning of hospital and other public resources. One potentially important piece of information for forecasting COVID-19 cases is mobile device location data that measure the amount of time an individual spends at home. Endemic-epidemic (EE) time series models are recently proposed autoregressive models where the current mean case count is modelled as a weighted average of past case counts multiplied by an autoregressive rate, plus an endemic component. We extend EE models to include a distributed-lag model in order to investigate the association between mobility and the number of reported COVID-19 cases; we additionally include a weekly first-order random walk to capture additional temporal variation. Further, we introduce a shifted negative binomial weighting scheme for the past counts that is more flexible than previously proposed weighting schemes. We perform inference under a Bayesian framework to incorporate parameter uncertainty into model forecasts. We illustrate our methods using data from four US counties.

La prévision du nombre de cas quotidiens de COVID­19 est cruciale pour la planification à court terme de ressources hospitalières et d'autres ressources publiques. Les données de localisation des téléphones mobiles qui mesurent le temps passé à la maison peuvent constituer un élément d'information important pour prédire les cas de COVID­19. Les modèles de séries chronologiques endémiques­épidémiques sont des modèles auto­régressifs récents où le nombre moyen de cas en cours est modélisé comme une moyenne pondérée du nombre de cas antérieurs multipliée par un taux auto­régressif (reproductif), plus une composante endémique. Les auteurs de ce travail généralisent les modèles endémiques­épidémiques pour y inclure un modèle à décalage distribué, et ce, dans le but de tenir compte du lien entre la mobilité et le nombre de cas de COVID­19 enregistrés. Pour saisir les variations de temps supplémentaires, ils y incorporent une marche hebdomadaire aléatoire d'ordre supérieur. De plus, ils proposent un schéma de pondération binomiale négative décalée pour les dénombrements passés, qui est plus flexible que les schémas de pondération existants. Ils utilisent l'inférence bayésienne afin d'intégrer l'incertitude des paramètres aux prédictions du modèle et ils illustrent les méthodes proposées avec des données provenant de quatre comtés américains.

Commentary: "How's It Going?" Training Experiences of Pediatric Postdoctoral Fellows During COVID-19.

The COVID-19 pandemic has impacted the lives and workplaces of individuals across the world substantially, in ways that are yet largely unknown. This commentary aims to provide an early snapshot of the experiences of pediatric postdoctoral fellows in academic medical settings; specifically, we will explore the impact of the pandemic on developing mastery within several competencies (e.g., research, professional development, clinical, interdisciplinary). These competencies are critical elements to fellowship to prepare for independent practice. Several models of training competencies for professional psychology and pediatric psychology exist, which focus on trainee skill development. Measures taken to minimize the spread of COVID-19 have directly impacted hospital systems and training, requiring programs to adapt competencies in various domains, such as increased familiarity with telehealth and virtual supervision. Additionally, fellows experienced an impact of the pandemic on securing employment following fellowship, conducting research and program development activities, and on cognitive flexibility and self-care. Governing bodies, such as the APA and Council of Chairs of Training Councils, have released statements and guidelines on addressing training of postdoctoral fellows including increasing flexibility of training methods, limiting in-person contact, and adjusting educational and licensing requirements. This paper offers informed commentary and diverse perspectives from current postdoctoral fellows engaged in a variety of clinical and research responsibilities regarding how the COVID-19 pandemic has impacted their training. We hope this paper will provide important insight into the unique experiences of postdoctoral fellows during the capstone year(s) of training prior to independent work and inform recommendations for postdoctoral training programs.

Generating community measures of food purchasing activities using store-level electronic grocery transaction records: an ecological study in Montreal, Canada.

OBJECTIVE: Geographic measurement of diets is generally not available at areas smaller than a national or provincial (state) scale, as existing nutrition surveys cannot achieve sample sizes needed for an acceptable statistical precision for small geographic units such as city subdivisions. DESIGN: Using geocoded Nielsen grocery transaction data collected from supermarket, supercentre and pharmacy chains combined with a gravity model that transforms store-level sales into area-level purchasing, we developed small-area public health indicators of food purchasing for neighbourhood districts. We generated the area-level indicators measuring per-resident purchasing quantity for soda, diet soda, flavoured (sugar-added) yogurt and plain yogurt purchasing. We then provided an illustrative public health application of these indicators as covariates for an ecological spatial regression model to estimate spatially correlated small-area risk of type 2 diabetes mellitus (T2D) obtained from the public health administrative data. SETTING: Greater Montreal, Canada in 2012. PARTICIPANTS: Neighbourhood districts (n 193). RESULTS: The indicator of flavoured yogurt had a positive association with neighbourhood-level risk of T2D (1·08, 95 % credible interval (CI) 1·02, 1·14), while that of plain yogurt had a negative association (0·93, 95 % CI 0·89, 0·96). The indicator of soda had an inconclusive association, and that of diet soda was excluded due to collinearity with soda. The addition of the indicators also improved model fit of the T2D spatial regression (Watanabe-Akaike information criterion = 1765 with the indicators, 1772 without). CONCLUSION: Store-level grocery sales data can be used to reveal micro-scale geographic disparities and trends of food selections that would be masked by traditional survey-based estimation.

Fluoroquinolone Use and Seasonal Patterns of Ciprofloxacin Resistance in Community-Acquired Urinary Escherichia coli Infection in a Large Urban Center.

Urinary tract infections caused by the bacterium Escherichia coli are among the most frequently encountered infections and are a common reason for antimicrobial prescriptions. Resistance to fluoroquinolone antimicrobial agents, particularly ciprofloxacin, has increased in recent decades. It is intuitive that variation in fluoroquinolone resistance is driven by changes in antimicrobial use, but careful study of this association requires the use of time-series methods. Between April 2010 and December 2014, we studied seasonal variation in resistance to ciprofloxacin, trimethoprim-sulfamethoxazole, and ampicillin in community-acquired urinary E. coli isolates in Montreal, Quebec, Canada. Using dynamic linear models, we investigated whether seasonal variation in resistance could be explained by seasonal variation in community antimicrobial use. We found a positive association between total fluoroquinolone use lagged by 1 and 2 months and the proportion of isolates resistant to ciprofloxacin. Our results suggest that resistance to ciprofloxacin is responsive to short-term variation in antimicrobial use. Thus, antimicrobial stewardship campaigns to reduce fluoroquinolone use, particularly in the winter when use is highest, are likely to be a valuable tool in the struggle against antimicrobial resistance.

An Area-Level Indicator of Latent Soda Demand: Spatial Statistical Modeling of Grocery Store Transaction Data to Characterize the Nutritional Landscape in Montreal, Canada.

Measurement of neighborhood dietary patterns at high spatial resolution allows public health agencies to identify and monitor communities with an elevated risk of nutrition-related chronic diseases. Currently, data on diet are obtained primarily through nutrition surveys, which produce measurements at low spatial resolutions. The availability of store-level grocery transaction data provides an opportunity to refine the measurement of neighborhood dietary patterns. We used these data to develop an indicator of area-level latent demand for soda in the Census Metropolitan Area of Montreal in 2012 by applying a hierarchical Bayesian spatial model to data on soda sales from 1,097 chain retail food outlets. The utility of the indicator of latent soda demand was evaluated by assessing its association with the neighborhood relative risk of prevalent type 2 diabetes mellitus. The indicator improved the fit of the disease-mapping model (deviance information criterion: 2,140 with the indicator and 2,148 without) and enables a novel approach to nutrition surveillance.

Joint Modeling of Resistance to Six Antimicrobials in Urinary Escherichia coli Isolates in Quebec, Canada.

Empirical treatment of urinary tract infections should be based on susceptibility profiles specific to the locale and patient population. Additionally, these susceptibility profiles should account for correlations between resistance to different types of antimicrobials. We used hierarchical logistic regression models to investigate geographic, temporal, and demographic trends in resistance to six antimicrobials in community-acquired and nosocomial urinary E. coli isolates from three communities in the province of Quebec, Canada, procured between April 2010 and December 2017. A total of 74,986 community-acquired (patient age, ≥18 years) and 4,384 nosocomial isolates (patient age, ≥65 years) were analyzed. In both community-acquired and nosocomial isolates, we found geographic variation in the prevalence of resistance. Male sex (community-acquired hierarchical mean odds ratio [OR], 1.24; 95% credible interval [CI], 1.02 to 1.50; nosocomial hierarchical mean OR, 1.16, 95% CI, 0.92 to 1.41) and recent hospitalization (community-acquired hierarchical mean OR, 1.49; 95% CI, 1.33 to 1.66; nosocomial hierarchical mean OR, 1.31; 95% CI, 0.99 to 1.78) were associated with a higher risk of resistance to most types of antimicrobials. We found distinct seasonal trends in both community-acquired and nosocomial isolates, but only community-acquired isolates showed a consistent annual pattern. Ciprofloxacin resistance increased sharply with patient age. We found clinically relevant differences in antimicrobial resistance in urinary E. coli isolates between locales and patient populations in the province of Quebec. These results could help inform empirical treatment decisions for urinary tract infections. In the future, similar models integrating local, provincial, and national resistance data could be incorporated into decision support systems for clinicians.

Bayesian estimation of the average treatment effect on the treated using inverse weighting.

We develop a Bayesian approach to estimate the average treatment effect on the treated in the presence of confounding. The approach builds on developments proposed by Saarela et al in the context of marginal structural models, using importance sampling weights to adjust for confounding and estimate a causal effect. The Bayesian bootstrap is adopted to approximate posterior distributions of interest and avoid the issue of feedback that arises in Bayesian causal estimation relying on a joint likelihood. We present results from simulation studies to estimate the average treatment effect on the treated, evaluating the impact of sample size and the strength of confounding on estimation. We illustrate our approach using the classic Right Heart Catheterization data set and find a negative causal effect of the exposure on 30-day survival, in accordance with previous analyses of these data. We also apply our approach to the data set of the National Center for Health Statistics Birth Data and obtain a negative effect of maternal smoking during pregnancy on birth weight.

Longitudinal evaluation of a household energy package on blood pressure, central hemodynamics, and arterial stiffness in China.

BACKGROUND: Cardiovascular diseases are the leading contributors to disease burden in China and globally, and household air pollution exposure is associated with risk of cardiovascular disease. OBJECTIVES: We evaluated whether subclinical cardiovascular outcomes in adult Chinese women would improve after distribution of an energy package comprised of a semi-gasifier cookstove, water heater, chimney, and supply of processed biomass fuel. METHODS: We enrolled 204 households (n = 205 women) from 12 villages into a controlled before- and after-intervention study on cardiovascular health and air pollution in Sichuan Province. The intervention was distributed to 124 households during a government-sponsored rural energy demonstration program. The remaining 80 households received the package 18 months later at the end of the study, forming a comparison group. One woman from each household had their blood pressure (BP), central hemodynamics, and arterial stiffness measured along with exposures to air pollution and demographic and household characteristics, on up to five visits. We used a difference-in-differences mixed-effects regression approach with Bayesian inference to assess the impact of the energy package on sub-clinical cardiovascular outcomes. RESULTS: Women who did not receive the energy package had greater mean decreases in brachial systolic (-4.1 mmHg, 95% credible interval (95%CIe) -7.3, -0.9) and diastolic BP (-2.0 mmHg, 95%CIe -3.6, -0.5) compared with women who received the package (systolic: -2.7, 95%CIe -5.0, -0.4; diastolic: -0.3, 95%CIe -1.4, 0.8) resulting in slightly positive but not statistically significant difference-in-differences effect estimates of 1.3 mmHg (95%CIe -2.5, 5.2) and 1.7 mmHg (95%CIe -0.3, 3.6), respectively. Similar trends were found for central BP, central pulse pressure, and arterial stiffness. Air pollution exposures decreased on average for both treatment groups, with a greater range of reductions among women who did not receive the package (with package: -30% to -50%; without package: +2% to -69%), likely as a result of increased use of gas fuel and electric stoves among this group. Outdoor air quality changed very little over time. CONCLUSIONS: Gasifier stoves have been widely promoted as the next generation of 'clean-cooking' technologies, however their effectiveness in improving health in real-world settings should be carefully evaluated and communicated before scaling up their implementation.

Zika emergence, persistence, and transmission rate in Colombia: a nationwide application of a space-time Markov switching model.

Zika, a viral disease transmitted to humans by Aedes mosquitoes, emerged in the Americas in 2015, causing large-scale epidemics. Colombia alone reported over 72,000 Zika cases between 2015 and 2016. Using national surveillance data from 1121 municipalities over 70 weeks, we identified sociodemographic and environmental factors associated with Zika's emergence, re-emergence, persistence, and transmission intensity in Colombia. We fitted a zero-state Markov-switching model under the Bayesian framework, assuming Zika switched between periods of presence and absence according to spatially and temporally varying probabilities of emergence/re-emergence (from absence to presence) and persistence (from presence to presence). These probabilities were assumed to follow a series of mixed multiple logistic regressions. When Zika was present, assuming that the cases follow a negative binomial distribution, we estimated the transmission intensity rate. Our results indicate that Zika emerged/re-emerged sooner and that transmission was intensified in municipalities that were more densely populated, at lower altitudes and/or with less vegetation cover. Warmer temperatures and less weekly-accumulated rain were also associated with Zika emergence. Zika cases persisted for longer in more densely populated areas with more cases reported in the previous week. Overall, population density, elevation, and temperature were identified as the main contributors to the first Zika epidemic in Colombia. We also estimated the probability of Zika presence by municipality and week, and the results suggest that the disease circulated undetected by the surveillance system on many occasions. Our results offer insights into priority areas for public health interventions against emerging and re-emerging Aedes-borne diseases.

Bayesian modeling of dynamic behavioral change during an epidemic.

For many infectious disease outbreaks, the at-risk population changes their behavior in response to the outbreak severity, causing the transmission dynamics to change in real-time. Behavioral change is often ignored in epidemic modeling efforts, making these models less useful than they could be. We address this by introducing a novel class of data-driven epidemic models which characterize and accurately estimate behavioral change. Our proposed model allows time-varying transmission to be captured by the level of "alarm" in the population, with alarm specified as a function of the past epidemic trajectory. We investigate the estimability of the population alarm across a wide range of scenarios, applying both parametric functions and non-parametric functions using splines and Gaussian processes. The model is set in the data-augmented Bayesian framework to allow estimation on partially observed epidemic data. The benefit and utility of the proposed approach is illustrated through applications to data from real epidemics.

A joint hierarchical model for the number of cases and deaths due to COVID-19 across the boroughs of Montreal.

As of July 2021, Montreal is the epicentre of the COVID-19 pandemic in Canada with highest number of deaths. We aim to investigate the spatial distribution of the number of cases and deaths due to COVID-19 across the boroughs of Montreal. To this end, we propose that the cumulative numbers of cases and deaths in the 33 boroughs of Montreal are modelled through a bivariate hierarchical Bayesian model using Poisson distributions. The Poisson means are decomposed in the log scale as the sums of fixed effects and latent effects. The areal median age, the educational level, and the number of beds in long-term care homes are included in the fixed effects. To explore the correlation between cases and deaths inside and across areas, three different bivariate models are considered for the latent effects, namely an independent one, a conditional autoregressive model, and one that allows for both spatially structured and unstructured sources of variability. As the inclusion of spatial effects change some of the fixed effects, we extend the Spatial+ approach to a Bayesian areal set up to investigate the presence of spatial confounding. We find that the model which includes independent latent effects across boroughs performs the best among the ones considered, there appears to be spatial confounding with the diploma and median age variables, and the correlation between the cases and deaths across and within boroughs is always negative.

A process convolution model for crash count data on a network.

Crash data observed on a road network often exhibit spatial correlation due to unobserved effects with inherent spatial correlation following the structure of the road network. It is important to model this spatial correlation while accounting for the road network structure. In this study, we introduce the network process convolution (NPC) model. In this model, the spatial correlation among crash data is captured by a Gaussian Process (GP) approximated through a kernel convolution approach. The GP's covariance function is based on path distance computed between a limited set of knots and crash data points on the road network. The proposed model offers a straightforward approach for predicting crash frequency at unobserved locations where covariates are available, and for interpolating the GP values anywhere on the network. Inference procedure is performed following the Bayesian paradigm and is implemented in R-INLA, which offers an estimation procedure that is very efficient compared to Markov Chain Monte Carlo sampling algorithms. We fitted our model to synthetic data and to crash data from Ottawa, Canada. We compared the proposed approach with a proper Conditional Autoregressive (pCAR) model, and with Poisson Regression (PR) and Negative Binomial (NB) models without latent effects. The results of the study indicated that although the pCAR model has comparable fitting performance, the NPC model outperforms pCAR when the main goal is to predict unobserved locations of interest. The proposed model also offers lower mean absolute error rates for cross validated crash counts, latent variable values, fixed-effect coefficients, as well as shorter interval scores for singletons. The NPC provides a natural way to account for the road network structure when considering the inclusion of spatially structured latent random effects in the modelling of crash data. It also offers an improved predictive capability for crash data on a road network.

A joint spatial marked point process model for dengue and severe dengue in Medellin, Colombia.

The spatial distribution of surveillance-reported dengue cases and severity are usually analyzed separately, assuming independence between the spatial distribution of non-severe and severe cases. Given the availability of data for the individual geo-location of surveillance-notified dengue cases, we conducted a spatial analysis to model non-severe and severe dengue simultaneously, using a hierarchical Bayesian model. We fit a joint model to the spatial pattern formed by dengue cases as well as to the severity status of the cases. Results showed that age and socioeconomic status were associated with dengue presence, and there was evidence of clustering for overall cases but not for severity. Our findings inform decision making to address the preparedness or implementation of dengue control strategies at the local level.

A Poisson-multinomial spatial model for simultaneous outbreaks with application to arboviral diseases.

Dengue, Zika, and chikungunya are arboviral diseases (AVD) transmitted mainly by Aedes aegypti. Rio de Janeiro city, Brazil, has been endemic for dengue for over 30 years, and experienced the first joint epidemic of the three diseases between 2015-2016. They present similar symptoms and only a small proportion of cases are laboratory-confirmed. These facts lead to potential misdiagnosis and, consequently, uncertainty in the registration of the cases. We have available the number of cases of each disease for the n=160 neighborhoods of Rio de Janeiro. We propose a Poisson model for the total number of cases of Aedes-borne diseases and, conditioned on the total, we assume a multinomial model for the allocation of the number of cases of each of the diseases across the neighborhoods. This provides simultaneously the estimation of the associations of the relative risk of the total cases of AVD with environmental and socioeconomic variables; and the estimation of the probability of presence of each disease as a function of available covariates. Our findings suggest that a one standard deviation increase in the social development index decreases the relative risk of the total cases of AVD by 28%. Neighborhoods with smaller proportion of green area had greater odds of having chikungunya in comparison to dengue and Zika. A one standard deviation increase in population density decreases the odds of a neighborhood having Zika instead of dengue by 18% but increases the odds of chikungunya in comparison to dengue by 18% and by 43% in comparison to Zika.