Research
Presentations
To view a presentation abstract, click on the presentation title.
2008
International Society for Environmental Epidemiology (ISEE) Annual Meeting
Visualization of Spatial and Temporal Data with Dynamic Maps
Castronovo, D.A.; Naumova E.N.
Recent advances in Geographic Information Systems (GIS) and visualization software allow for the representation of public health data over space and time through dynamic maps. Animation has potential to show the spread of infectious disease in relation to climatic factors at various spatial and temporal scales. However, dynamic maps represent a delicate balance between usefulness and incomprehensible information overload. We develop a conceptual framework for building informative dynamic maps using data on waterborne diseases and extreme weather events.
Designing dynamic maps requires several considerations beyond the traditional static maps including: choice of a temporal scale, cartographic representation of environmental covariates with disease outcomes, appropriate symbolization of disease outcome data, selection of the duration and rate of change for each map frame, and tools to direct the audience’s attention to space-time variations and anomalies.
Monthly waterborne disease hospitalization data for the U.S. elderly were abstracted from the Centers for Medicare and Medicaid Services (CMS) for a 5-year period (1998-2002). Monthly rates were calculated for each county and mapped using graduated dots that represent the entire distribution of the data over the entire time period. Hospitalization rates for each month were overlaid on average monthly maximum precipitation data from the PRISM climate data set. The resulting dynamic map reveals seasonal patterns in waterborne disease hospitalization and suggests that dynamic mapping is informative for displaying the spread of infectious disease through time and space.
The study of spatiotemporal processes such as the spread of infectious disease requires researchers to recognize and track changes in complex systems. Studying the spread of infectious disease poses a large challenge due to the number and scale of interactions that occur within environmental-health systems. Dynamic maps have the capability to help visualize, hypothesize, and decipher change factors in large spatial and temporal data sets, but there are several factors one must consider to make dynamic maps useful. This presentation will teach the audience about key cartographic and visualization elements that contribute to successful dynamic maps of complex environmental-health systems. [ close abstract ]
American Public Health Association (APHA) Annual Meeting
Pneumonia and influenza outcomes in elderly with dementia: The effect of proximity to healthcare facilities
Pandita, M.; Parisi, S.M.; Castronovo, D.A.; Naumova, E.N.
Inadequate or inaccessible health care services increase the vulnerability of elderly patients and elevate the risk of complications associated with viral infections. Elderly with dementia can be at high risk for influenza induced pneumonia, due to communication difficulties that may prevent timely diagnoses. We used geographic mapping technology to investigate the effect of geographic proximity from place of residence to healthcare facilities on P&I outcomes in elderly with dementia. We abstracted 43,812 hospitalization records with an indication of P&I (ICD 9 codes: 480-487) and dementia (ICD 9 codes: 290-294, 331, 797) from the Centers for Medicare and Medicaid Services (CMS) database between 1998-2002 for New England (NE) residents and estimated the mean length of hospital stay by county of the residence. We also abstracted geographic positioning data for Medicare and Medicaid-certified nursing homes. We quantified absolute and average proximity to health care facilities by measuring geo-spatial distance between nursing homes and neighboring hospitals using ESRI's ArcGIS software. Our preliminary results indicate that in Massachusetts, the average distance varied with regards to county rurality. Rural Berkshire and Franklin counties have the highest mileage to neighboring hospitals (10.5-20.7 miles), whereas urban Suffolk County has the lowest average mileage (0.7-2.3 miles). Overall, the mean length of stay was greater for rural counties (ranging from 7.0-8.65 days) than for urban counties (6.1-6.74 days), suggesting rural elderly may have experienced delayed access to health care. Urban versus rural proximity by county and zip code will be further explored throughout the New England region. [ close abstract ]
2007
American Public Health Association (APHA) Annual Meeting
Estimating number of elderly affected by Atlantic storms and post-storm hospitalizations related to infectious gastrointestinal diseases using GIS
Chui, K.H.K.; Castronovo, D.A.; Naumova, E.N.
Natural disasters pose a serious threat to welfare of vulnerable populations. Due to infrastructure disruption and population displacement that typically occur as the consequences of extreme weather events, the assessment of health effects is difficult. Often even simple estimates of affected population are not readily available to public health professionals. In this study, we develop methodology for estimating the number of elderly affected by Atlantic storms in 1998-2002 using Geographical Information System (GIS).
Information on 25 storms was abstracted from the U.S. National Oceanic & Atmospheric Administration including date, time, and epicenter coordinates in a six-hour interval. To reflect degree of exposure the storm tracks were mapped with three buffers: 50, 75, and 100 miles from storm trajectory. Number of elderly older than 65 affected by storms were estimated by overlaying the storm buffer areas with county maps linked to Census 2000 demographic data. Counties with centroid inside in the buffer area were considered affected.
Of 25 Atlantic storms made landfall between 1998 and 2002, 5 were category 4 or higher, 6 were category 2 and 3. In the 31 states from South Atlantic, Middle Atlantic, New England, East North Central, and West South Central divisions, the recorded storms have accumulatively affected 3,531 counties with 390.3 million people out of whom 49.7 million (12.7%) were elderly. Proportion of elderly affected to total population varies storm by storm, from 0.2% to 33.4%.
The obtained estimates provide important insights on the magnitude of exposure to extreme weather events among the US elderly. [ close abstract ]
32nd International Symposium of Remote Sensing of Environment (SRSE)
The Use of Remote Sensing to Assess Global Trends in Seasonality of Cryptosporidiosis.
Jagai, J.S.; Castronovo D.A.; Naumova E.N.
Cryptosporidiosis is a diarrheal illness caused by protozoan, which is spread though water contaminated with human or animal feces. Cryptosporidiosis typically manifests through a low endemic level and well-pronounced seasonal outbreaks, indicating strong effect of environmental factors. Though cryptosporidiosis is self-limiting in immuno-competent people, it can be life threatening to immuno-compromised individuals, such as those with AIDS and the malnourished. Studies conducted in various locations have noted an increase in cryptosporidiosis during the warm and rainy season. We conducted a meta-analysis to examine how an increase in cryptosporidiosis relates to precipitation and ambient temperature worldwide. We also investigated the potential of using remote sensing data as a proxy for exposure to cryptosporidiosis globally. [ close abstract ]
2006
American Public Health Association (APHA) Annual Meeting
Preexisting rate of gastroenteric infections among elderly residing in areas affected by Hurricane Katrina.
Chui, K.H.K.; Kosheleva A.; Jagai, J.S.; Castronovo D.A.; Naumova E.N.
An increase in gastroenteric infections (GI) associated with microbial contaminations in water supply has been noted during warm and rainy seasons worldwide. Vulnerable populations residing in areas prone to severe rainfall are likely to be exposed to contaminated water and display higher incidence of GI rates. Attempting to document historical status of GI among elderly (≥65yo) in the areas affected by Hurricane Katrina in Louisiana and Mississippi, we gathered information from various sources and described GI rates for the two states from 1998 to 2002.
We abstracted and aggregated GI counts from the Centers for Medicare and Medicaid Services database at the zip code level. Results were geo-merged with Census 2000 data on Zip Code Tabulation Area level (ZCTA) using zip code centroids for annual rate calculation (cases/1000 population) and demographic comparisons. Both states provided information on areas flooded by Katrina. Geographic distribution of GI rates are shown using maps created with ArcInfo 8.3.
In the 989 ZCTAs, elderly composed of 13 ± 6% of the population, ranging from 0% to 70%. Seventeen ZCTAs flooded by Katrina in Louisiana had similar percentage of elderly (11% ± 3%, p = 0.15) but higher population density (2600 vs. 200 person/km2, p < 0.001) and larger proportion of African Americans (66% vs. 26%, p < 0.001). Overall GI rates (ICD9: 001 to 009) among ≥65yo were similar between the flooded and non-flooded areas (0.17 vs. 0.25 per 1000 of population ≥65yo, p = 0.10). [ close abstract ]
Environmental Indicators for Clostridium Difficile in the US Elderly.
Jagai, J.S.; Kosheleva A.A; Castronovo D.A.; Naumova E.N.
The incidence of Clostridium difficile associated diarrhea has risen over the past two decades and in the US more than 300,000 cases per year occur in hospitals or long-term care facilities. C. difficile is associated with antibiotic use and up to 75% of cases of antibiotic associated diarrhea are attributable to C. difficile. It is plausible that spatio-temproal patterns could be driven by pathogens for which antibiotics are prescribed. The presence of pathogens and their pathogenicity can be driven by environmental factors and may differ by climate zones. Therefore, we hypothesize that rates of C. difficile are associated with meteorological characteristics in different climate zones. All hospitalization records for C. difficile, as well as non-specific gastroenteritis infections and symptoms from the Centers for Medicare and Medicaid Services (CMS) MedPAR data file for a 5 year period (1998-2002) were abstracted. Annual rates were calculated for each county (3108) and each county was classified based on the Koppen Climate classification scheme and assigned to a watershed region. We compared the rates across climate categories and watersheds using ANOVA. The average annual county-specific hospitalization rate of C. difficile per 100,000 elderly was 234.59 ± 160.40 cases. The highest rates were seen in the humid climate region which covers the US Midwest (301.75 ± 182.34) and in the Ohio Region watershed (355.48 ± 167.94). Rates were also correlated with non-specific gastroenteritis (Pearson r = 0.345). These preliminary results suggest that higher rates of C. difficile are associated with climate-sensitive environmental characteristics. [ close abstract ]
International Society for Environmental Epidemiology (ISEE)
Annual Meeting
Gastroenteritis infections in the U.S. elderly and extreme weather events: exposures to Atlantic tropical storms of 1998-2002.
K.H. Chui, D.A. Castronovo, J.S. Jagai, A.A. Kosheleva, E.N. Naumova.
An increase in potentially waterborne gastroenteric infections (GI) associated with microbial contaminations in water supply has been noted during warm, rainy seasons. Populations in areas prone to severe rainfall are likely to be exposed to contaminated water and display higher incidence of GI rates. Rainfall and temperature are common exposures of interest in time-series epidemiologic analyses, while historical exposure to storms and hurricanes was usually not controlled for. We developed a storm frequency index along the East Coast of the U.S. at county level, using Atlantic storm track data of 1991-2004, and examined the relations between hospitalization for GI among elderly (≥65yo) and the storm frequency index. [ close abstract ]
Climate and Water Indicators for Enteric Infections in the US Elderly.
Jagai, J.S.; Kosheleva A.A.; Castronovo D.A.; Chui, K.C.; Naumova E.N.
Gastroenteric infections (GI) of waterborne or foodborne etiology are shown to be associated with contaminated water sources and warmer temperatures. Several studies have noted increased rates of GI during warm and rainy seasons. Climate characteristics and drinking water source can affect the rates of GI in an area. We conducted a nationwide analysis of hospitalization rates for five GI among the elderly on the county level to examine associations with climate characteristics and watersheds. We hypothesize that increased rates will be seen in counties with warm, moist climates or watersheds. [ close abstract ]
Spatio-temporal patterns for pneumonia and influenza hospitalization in the frail subpopulations of USA elderly.
Kosheleva, A.; Castronovo, D.A.; Jagai, J.S.; Chui, K.C.; Naumov, Y.N.; Gorski, J., Naumova E.N.
Hospitalization due to pneumonia and influenza (P&I) has a well defined seasonal pattern, that is relatively stable across various climatic zones of the continental USA. Peak of incidence is typically observed in winter months. In the elderly, the P&I incidence increases with age and in presence of co-morbid conditions. The goal of this project is to provide detailed examination for the spatio-temporal distribution of hospitalization rates of P&I in the US elderly (≥65 years old) with respect to aging, preexisting medical conditions, and meteorological stressors. [ close abstract ]
2005
International Society for Environmental Epidemiology (ISEE)
Annual Meeting
Seasonality of Cryptosporidiosis in Africa Compared to Global Trends:
A Meta-Analysis Approach.
Jagai, J.S.; Castronovo, D.A.; Naumova, E.N.
To understand how increases in incidence of cryptosporidiosis relate to precipitation and ambient temperature on a global scale and to determine whether Normalized Difference Vegetation Index (NDVI) can be used as a proxy for the combined effect of precipitation and temperature as a predictor for incidence of cryptosporidiosis. [ close abstract ]
