Background: Vulnerability to heat related events is influenced by household income, gender, quality of housing, basic services, and occupation. Low-income groups generally face higher levels of heat-related illnesses (HRIs) and higher levels of ill-health and injury. Individuals with co-existing conditions may be disproportionately affected by heat waves and comorbidity makes treating other illnesses more difficult and more costly. A substantial literature has evolved around the descriptive epidemiology of HRIs, however less is known about the interactions between chronic conditions, injuries, HRIs and socio-economic spatial patterns. There may be unanticipated interactions between HRIs, acute injury, and co-existing medical conditions that remain unclear, and a lack of research on the complexity of these interactions represents a significant gap in the literature. When paired with other epidemiological methods, network analysis may be uniquely suited to the study of structural and relational aspects of these health interactions. The novel use of quantitative network analysis and Geographic Information Services (GIS) can enhance our understanding as well as identification of geographical high risk zones and populations. Objective: Our study aims to identify risk factors for the excess mortality and morbidity during the 2010 heat event in North Carolina, as well as demonstrate a novel method to explore and visualize the complex interactions between chronic disease, injury, socio-economic spatial patterns, and HRIs. Methods: This study explores the interaction between chronic conditions, including HRIs, and socio-economic conditions during 2010 heat event, one of the warmest on record in North Carolina. Data from the National Weather Service describe two periods of excessively hot and humid conditions during 6-8 and 23-25 July, 2010 in central North Carolina. Using data from the 2010 NC State Inpatient Database (H-CUP); network measures were computed to quantify interactions between diagnoses to characterize HRI-diagnosis patterns in counties in North Carolina for several socio-economic and demographic groups. Results: The mean summer temperature in July of 2010 was more than 2°F higher than the previous record set in summer 2007. The percent of hospital admissions of Hispanic males 18-29 & 55 and older and Hispanic females 55 – 79 that died during hospitalization in July were significantly higher than other months. Average length-of-stays were greater in July among Hispanic males and females 55 and older, and older Hispanic males were diagnosed with more conditions during the heat wave. Younger Hispanic males were hospitalized more often for injuries than other groups during this period. We constructed county level, diagnosis networks for various subgroups: sex, race/ethnicity, age group and health insurance coverage. The network analysis reveals unexpected interaction for Hispanics between HRIs, chronic disease and injury. The combination of epidemiology with network analysis constitutes a strategy for enhancing vulnerability assessments, disease burden projections, and intervention planning. Discussion: Preliminary analyses show that working age Hispanic adults have distinct risks associated with HRIs that manifest as an unexpected increase in mortality through the interaction of injury and chronic illness associated with the extreme heat events in July 2010. Mapping and visualization of HRI burdens and interactions between chronic medical conditions, heat injury and illness using network analysis can improve our understanding of the underlying mechanisms driving increased mortality and morbidity. Network analysis and visualization may represent an additional, useful tool of analysis for environmental and occupational health professionals.

To better understand environmental health outcomes, it is essential to have finer resolution health, environmental, and demographic data (i.e., sub-county data). There are, however, fewer indicators available at the sub-county level, with difficulties in analyzing and interpreting such data. CDC’s National Environmental Public Health Tracking Program conducted the Sub-County Data Pilot Project to better understand the challenges in increasing the availability of standardized sub-county data within the National Environmental Public Health Tracking Network (Tracking Network). Florida, Maine, New York State, Washington, and Wisconsin Tracking participated in the project. The team reviewed available sub-county data, developed data standards, and conducted a pilot data submission between grantees and CDC. Four states submitted data on acute myocardial infarction and low birth weight, and one state submitted data on children’s blood lead levels and private well water. The team noted the necessity in balancing the need for fine scale data, stable measures, and confidentiality. Lessons learned included: challenges with geocoding and changing geographies, aggregation, population estimates, and data stability issues. Recommendations included: the creation of two standard aggregated sub-county geographies and guidelines for calculations of measures, uncertainty, and handling difficult records. Collaboration with data stewards and enhanced policies on disseminating sub-county data will be a key part of advancing these efforts. Establishing sub-county data standards will enable comparability between areas, over time, and across different datasets. Expanding the availability of standardized sub-county data is important for enhancing the capability of the Tracking Network, improving our understanding of environmental health, and informing local level decision-making.

Background. The use of Electronic Nicotine Delivery Systems (ENDS; e.g., e-cigs) has grown in popularity and recent efforts have begun to examine the proliferation of ENDS retailers. Yelp, an online business listing and rating platform, has demonstrated utility in this regard in Florida and New Jersey. However, little is known about how regional sociodemographic characteristics are associated with prevalence and characteristics of ENDS retailers. We focused on Pennsylvania, a politically and demographically diverse microcosm of the US, which has an evolving policy landscape related to ENDS taxation and public use restrictions. Objectives. To examine sociodemographic characteristics associated with prevalence and characteristics of ENDS retailers in Pennsylvania. Methods. We developed a Python script that leverages the Yelp Application Programming Interface (API) to conduct comprehensive searches in each of Pennsylvania’s 37 metropolitan and micropolitan census regions. The search strategy canvassed for any instance of “vape”, “vapor”, “electronic cigarette”, and “e-cig” within Yelp business listings. Two coders working independently then reviewed each of the business listings and websites for key information such as whether traditional tobacco products are sold in addition to ENDS products. Coders called each establishment on the telephone to confirm operating status, base price of e-liquid, and whether retailers mix their own e-liquids. ENDS retail locations were then mapped and examined in the context of regional sociodemographic characteristics according to recent US Census data. Results. In January 2017, our search resulted in 914 unique business listings throughout Pennsylvania. Of these, 253 were natively identified as “Vape Shops” on the Yelp platform. Coders identified a substantial number of ENDS retailers that were not captured by Yelp’s native categorization. In other cases, retailers that were correctly categorized had closed but were still marked as open on the Yelp platform. However, including these additional retailers and removing closed businesses did not significantly alter regional rankings of ENDS retailer prevalence. Comparisons across regions indicate that vape shop prevalence is positively associated with proportion of black residents, proportion of adults with post-high school education, higher median income, and urban region status. The base cost of e-liquid and prevalence of ENDS retailers that mix their own e-liquids varied by specific region within Pennsylvania. Conclusions. While our findings raise some concern about identifying ENDS retailers based solely on Yelp’s native Vape Shop category, they also support previous work demonstrating the value of using Yelp to estimate ENDS retailer prevalence. Although associations between race and ENDS prevalence conflicted with previous research conducted in New Jersey, this is likely because there are many rural regions with predominately white racial makeup in Pennsylvania. Additional research and validation will allow for investigation of trends in ENDS proliferation nationwide and may help to inform well-tailored health policy at state and local levels.

Background: Frequent readmissions decrease health care efficiency and increase medical costs. Geospatial approaches are of potential value in determining whether high-frequency readmissions are related to area socioeconomic status, e.g., ZIP code-level mean household income. Objectives: Evaluate spatial clustering of hospital readmissions in a Midwest city and assess its association with ZIP code-level mean household income. Methods: We analyzed electronic health records (EHR) from 36 adult inpatients with multiple readmissions between November 1, 2016 and January 31, 2017, and who did not have a primary care provider (PCP). We mapped the number of patients in each ZIP code area and utilized the American Community Survey 5-year estimates (ACS; 2011-2015) to identify ZIP code-level characteristics for each cluster. The ZIP codes were grouped based on high vs. low proportion of frequently admitted patients and general linear models were used to compare the mean household income for the two ZIP code groups in SAS v9.4. Results: A larger number of high frequency readmission patients were located in the Northern part of the city. The mean household income for the ZIP code area with a high proportion of frequently admitted patients was $12,000 lower than that of the ZIP code area with a low proportion of those patients (p=0.07). Conclusion: Most patients with frequent readmissions and no PCP were located in the Northern part of the city, which had ZIP codes with lower mean household income. Linkage of EHR and population-level data through geospatial analyses may be utilized to identify potential areas for future interventions to reduce hospital readmissions.

Xin Qi, Shaobai Zhang, Weiqiu Li, Guihua Zhuang Background: Geographical information system have widely been used in detecting infectious disease incidence. However, most of them did not detect the disease distribution under the local administration level (e.g., township level) in a relative large area (e.g., provincial level), especially in developing countries. Objectives: To describe the spatial distribution of Japanese Encephalitis (JE) (2005-2014) at the township level in Shaanxi, China and to identify the high-risk areas over time. Methods: JE data at the township level were supplied by Shaanxi Center for Disease Control and Prevention. Population data were provided by China Population Census. Descriptive and spatiotemporal cluster analyses were used to display the geographical distribution of JE and detect high-risk clusters of JE incidence. Results: A total of 654 JE cases were included in this study. Two significant clusters (RR=5.97 in the central east, RR=13.06 in the north) were detected in 2013. Other two clusters (RR: 8.81 and 11.69 for each) were detected in the south in 2009. Clusters were also discovered in 2009 (RR=11.69 in the south) and 2010 (RR=11.75 in the central west). Conclusion: High risk cluster varied across the study period, but usually existed in rural areas, especially in the south of Shaanxi. JE control and prevention strategies should focus on high risk areas.

Introduction: Myocardial Infarction (MI) or heart attack rates have been declining in Florida in the past few decades, but MI remains a leading cause of mortality in the state. Further, the declines in MI risks have not occurred uniformly across the state, but the spatial patterns of MI mortality have not been completely characterized at the County level using statistical techniques. Identifying geographic areas with significantly high risks of MI, and how these risks might be changing over time would lead to improved targeting of preventive and control strategies, and contribute to improved population health and reduced health disparities. Objectives: The objectives of this study were to investigate spatial patterns of MI mortality risks in Florida, to identify clusters of MI mortality risks, and to identify trends in MI mortality risks in clusters with persistently high and persistently low risks for the period 2000 to 2014. Methodology: County-level mortality risk data for MI were obtained from the Florida Department of Health. Spatial scan statistics were used to identify the locations of high and low risk clusters, and the spatial patterns were displayed in ArcGIS. Results: Myocardial infarction mortality risks decreased substantially throughout the state during the study period. However, geographical disparities were identified, and they showed a clear north to south gradient. Statistically significant clusters of high mortality risks were identified in the more rural, northern parts of the state, while low mortality risk clusters were observed in the south. The high risk clusters in the north showed more dramatic decreases in mortality risks, with consequent reduction in disparities between the north and the south. However, the counties in the north still lagged behind those in the south by almost 15 years where mortality risks were concerned. Conclusion: MI mortality risks have decreased in Florida over the past 1.5 decades, but significant disparities still exist. Efforts are needed to identify the determinants of these disparities, so as to reduce or eliminate them.