Modeling Area-Level Health Rankings

A two-stage bayesian model for assessing the geography of racialized economic segregation and premature mortality across US counties

Racialized economic segregation, a key metric that simultaneously accounts for spatial, social and income polarization in communities, has been linked to adverse health outcomes, including morbidity and mortality. Due to the spatial nature of this metric, the association between health outcomes and racialized economic segregation could also change with space. Most studies assessing the relationship between racialized economic segregation and health outcomes have always treated racialized economic segregation as a fixed effect and ignored the spatial nature of it. This paper proposes a two-stage Bayesian statistical framework that provides a broad, flexible approach to studying the spatially varying association between premature mortality and racialized economic segregation while accounting for neighborhood-level latent health factors across US counties. The two-stage framework reduces the dimensionality of spatially correlated data and highlights the importance of accounting for spatial autocorrelation in racialized economic segregation measures, in health equity focused settings.

The association of wildfire air pollution with COVID-19 incidence in New South Wales, Australia

The 2020 COVID-19 outbreak in New South Wales (NSW), Australia, followed an unprecedented wildfire season that exposed large populations to wildfire smoke. Wildfires release particulate matter (PM), toxic gases and organic and non-organic chemicals that may be associated with increased incidence of COVID-19. This study estimated the association of wildfire smoke exposure with the incidence of COVID-19 in NSW. A Bayesian mixed-effect regression was used to estimate the association of either the average PM₁₀ level or the proportion of wildfire burned area as proxies of wildfire smoke exposure with COVID-19 incidence in NSW, adjusting for sociodemographic risk factors. The analysis followed an ecological design using the 129 NSW Local Government Areas (LGA) as the ecological units. A random effects model and a model including the LGA spatial distribution (spatial model) were compared. A higher proportional wildfire burned area was associated with higher COVID-19 incidence in both the random effects and spatial models after adjustment for sociodemographic factors (posterior mean = 1.32 (99% credible interval: 1.05-1.67) and 1.31 (99% credible interval: 1.03-1.65), respectively). No evidence of an association between the average PM₁₀ level and the COVID-19 incidence was found. LGAs in the greater Sydney and Hunter regions had the highest increase in the risk of COVID-19. This study identified wildfire smoke exposures were associated with increased risk of COVID-19 in NSW. Research on individual responses to specific wildfire airborne particles and pollutants needs to be conducted to further identify the causal links between SARS-Cov-2 infection and wildfire smoke. The identification of LGAs with the highest risk of COVID-19 associated with wildfire smoke exposure can be useful for public health prevention and or mitigation strategies.

Measuring the spatial distribution of health rankings in the United States

We rank counties in the United States with respect to population health. We utilize the five observable county health variables used to construct the University of Wisconsin Population Health Institute's County Health Rankings (CHRs). Our method relies on a Bayesian factor analysis model that estimates data-driven weights for our rankings, incorporates county population sizes into the level of rank uncertainty, and allows for spillovers of health stock across county lines. We find that demographic and economic variation explains a large portion of the variation in health rankings. We address the importance of uncertainty caused by imputation of missing data and show that there is a substantial quantity of uncertainty in rankings throughout the rank distribution. Analyzing the health of counties both within and across state lines shows notable degrees of disparity in county health. While we find some disagreement between the ranks of our model and the CHRs, we show that there is additional information gained by utilizing the rankings produced by both methods.

Environmental and sociodemographic risk factors associated with environmentally transmitted zoonoses hospitalisations in Queensland, Australia

Zoonoses impart a significant public health burden in Australia particularly in Queensland, a state with increasing environmental stress due to extreme weather events and rapid expansion of agriculture and urban developments. Depending on the organism and the environment, a proportion of zoonotic pathogens may survive from hours to years outside the animal host and contaminate the air, water, food, or inanimate objects facilitating their transmission through the environment (i.e. environmentally transmitted). Although most of these zoonotic infections are asymptomatic, severe cases that require hospitalisation are an important indicator of zoonotic infection risk. To date, no studies have investigated the risk of hospitalisation due to environmentally transmitted zoonotic diseases and its association with proxies of sociodemographic and environmental stress. In this study we analysed hospitalisation data for a group of environmentally transmitted zoonoses during a 15-year period using a Bayesian spatial hierarchical model. The analysis incorporated the longest intercensal-year period of consistent Local Government Area (LGA) boundaries in Queensland (1996-2010). Our results showed an increased risk of environmentally transmitted zoonoses hospitalisation in people in occupations such as animal farming, and hunting and trapping animals in natural habitats. This risk was higher in females, compared to the general population. Spatially, the higher risk was in a discrete set of north-eastern, central and southern LGAs of the state, and a probability of 1.5-fold or more risk was identified in two separate LGA clusters in the northeast and south of the state. The increased risk of environmentally transmitted zoonoses hospitalisations in some LGAs indicates that the morbidity due these diseases can be partly attributed to spatial variations in sociodemographic and occupational risk factors in Queensland. The identified high-risk areas can be prioritised for health support and zoonosis control strategies in Queensland.

Developing a surveillance system of sub-county data: Finding suitable population thresholds for geographic aggregations

The Centers for Disease Control and Prevention's National Environmental Public Health Tracking Program created standardized sub-county geographies that are comparable over time, place, and outcomes. Expected census tract-level counts were calculated for asthma emergency department visits and lung cancer. Census tracts were aggregated for various total population and sub-population thresholds, then suppression and stability were examined. A total of 5,000 persons was recommended for the more common outcome scheme and a total of 20,000 persons was recommended for the rare outcome scheme. Health outcomes with a median case count of 17.0 cases or higher should produce stable estimates at the census tract level. This project generated recommendations for three sub-county geographies that will be useful for surveillance purposes: census tract, a more common outcome aggregation scheme, and a rare outcome aggregation scheme. This methodology can be applied anywhere to aggregate geographic units and produce stable rates at a finer resolution.

Assessment of Area-Level Disease Control and Surveillance Vulnerabilities: An Application to Visceral Leishmaniasis in Brazil

The large number of activities contributing to zoonoses surveillance and control capability, on both human and animal domains, and their likely heterogeneous implementation across administrative units make assessment and comparisons of capability performance between such units a complex task. Such comparisons are important to identify gaps in capability development, which could lead to clusters of vulnerable areas, and to rank and subsequently prioritize resource allocation toward the least capable administrative units. Area-level preparedness is a multidimensional entity and, to the best of our knowledge, there is no consensus on a single comprehensive indicator, or combination of indicators, in a summary metric. We use Bayesian spatial factor analysis models to jointly estimate and rank disease control and surveillance capabilities against visceral leishmaniasis (VL) at the municipality level in Brazil. The latent level of joint capability is informed by four variables at each municipality, three reflecting efforts to monitor and control the disease in humans, and one variable informing surveillance capability on the reservoir, the domestic dog. Because of the large volume of missing data, we applied imputation techniques to allow production of comprehensive rankings. We were able to show the application of these models to this sparse dataset and present a ranked list of municipalities based on their overall VL capability. We discuss improvements to our models, and additional applications.

The Importance of Health and Social Services Spending to Health Outcomes in Texas, 2010-2016

OBJECTIVES

Public health and social services spending have been shown to improve health outcomes at the county level, although there are significant state and regional variations in such spending. Texas offers an important opportunity for examining nuances in the patterns of association between local government health and social services spending and population health outcomes. The primary objectives of this study were to describe local investments in education, health, and social services at the county-area level for all of Texas from 2002 through 2012 and to examine how changes in local investment over time were associated with changes in health outcomes.

METHODS

We used two large secondary data sources for this study. First, US Census Bureau data were used to measure annual spending by all local governments on public hospitals, community health care and public health, and >1 dozen social services. Second, County Health Rankings & Roadmaps data measured county health outcomes. We performed regression models to examine the association between increases in local government spending and a county's health outcomes ranking 4 years later. Multilevel mixed-effects linear regression models accounted for mean spending in each category, county health factors ranking, and county and state random effects.

RESULTS

Local governments in Texas spent an average of $4717 per capita across all health and social services. Although spending was relatively consistent across 2002-2012, there was notable variation in spending across counties and services. Regression models found that changes in four spending categories were associated with significant improvements in health outcomes: fire and ambulance, community health care and public health, housing and community development, and libraries. For each, an additional one-time investment of $15 per capita was associated with a 1-spot improvement in statewide county health rankings within 4 years.

CONCLUSIONS

Existing evidence regarding the association between social services spending and health outcomes may not yield sufficiently granular data for policy makers within a single state. Investments in certain social services in Texas were associated with improvements in health outcomes, as measured by improvements in the County Health Rankings, in the years subsequent to spending increases. Similar analyses in other states and regions may yield actionable avenues for policy makers to improve population health.

Revisiting area risk classification of visceral leishmaniasis in Brazil

BACKGROUND

Visceral leishmaniasis (VL) is a neglected tropical disease of public health relevance in Brazil. To prioritize disease control measures, the Secretaria de Vigilância em Saúde of Brazil's Ministry of Health (SVS/MH) uses retrospective human case counts from VL surveillance data to inform a municipality-based risk classification. In this study, we compared the underlying VL risk, using a spatiotemporal explicit Bayesian hierarchical model (BHM), with the risk classification currently in use by the Brazil's Ministry of Health. We aim to assess how well the current risk classes capture the underlying VL risk as modelled by the BHM.

METHODS

Annual counts of human VL cases and the population at risk for all Brazil's 5564 municipalities between 2004 and 2014 were used to fit a relative risk BHM. We then computed the predicted counts and exceedence risk for each municipality and classified them into four categories to allow comparison with the four risk categories by the SVS/MH.

RESULTS

Municipalities identified as high-risk by the model partially agreed with the current risk classification by the SVS/MH. Our results suggest that counts of VL cases may suffice as general indicators of the underlying risk, but can underestimate risks, especially in areas with intense transmission.

CONCLUSION

According to our BHM the SVS/MH risk classification underestimated the risk in several municipalities with moderate to intense VL transmission. Newly identified high-risk areas should be further evaluated to identify potential risk factors and assess the needs for additional surveillance and mitigation efforts.

Diving into the consumer nutrition environment: A Bayesian spatial factor analysis of neighborhood restaurant environment

Neighborhood restaurant environment (NRE) plays a vital role in shaping residents' eating behaviors. While NRE 'healthfulness' is a multi-facet concept, most studies evaluate it based only on restaurant type, thus largely ignoring variations of in-restaurant features. In the few studies that do account for such features, healthfulness scores are simply averaged over accessible restaurants, thereby concealing any uncertainty that attributed to neighborhoods' size or spatial correlation. To address these limitations, this paper presents a Bayesian Spatial Factor Analysis for assessing NRE healthfulness in the city of Kitchener, Canada. Several in-restaurant characteristics are included. By treating NRE healthfulness as a spatially correlated latent variable, the adopted modeling approach can: (i) identify specific indicators most relevant to NRE healthfulness, (ii) provide healthfulness estimates for neighborhoods without accessible restaurants, and (iii) readily quantify uncertainties in the healthfulness index. Implications of the analysis for intervention program development and community food planning are discussed.

Uses of Population Health Rankings in Local Policy Contexts: A Multisite Case Study

Population health rankings are a common strategy to spur evidence-informed health policy making, but little is known about their uses or impacts. The study aims were to (1) understand how and why the County Health Rankings (CH-Rankings) are used in local policy contexts, (2) identify factors that influence CH-Rankings utilization, and (3) explore potentially negative impacts of the CH-Rankings. Forty-four interviews were conducted with health organization officials and public policy makers in 15 purposively selected counties. The CH-Rankings were used instrumentally to inform internal planning decisions, conceptually to educate the public and policy makers about determinants of population health, and politically to advance organizational agendas. Factors related to organizational capacity, county political ideology, and county rank influenced if, how, and why the CH-Rankings were used. The CH-Rankings sometimes had the negative impacts of promoting potentially ineffective interventions in politically conservative counties and prompting negative media coverage in some counties with poor rank.

Eliminating composite bias in treatment effects estimates: Applications to quality of life assessment

Researchers are often interested in composite measures such as overall ratings, indices of physical or mental health, or health-related quality-of-life (HRQoL) outcomes. Such measures are typically composed of two or more underlying discrete variables. In this paper, I investigate conditions where the estimated treatment effect based solely on the composite outcome is biased under non-random treatment assignment, which I refer to as composite bias. I then compare the magnitude of this bias across a variety of estimators, including ordinary least squares, propensity score estimators, and an alternative two-stage approach that first estimates treatment effects on the underlying outcomes and then combines these effects into an overall effect on the composite outcome of interest. The results highlight the presence of composite bias, identify general conditions under which such bias exists, and offer guidance as to how best to minimize this bias in practice.