Date of Award

January 2016

Document Type

Open Access Thesis

Degree Name

Master of Public Health (MPH)


School of Public Health

First Advisor

Jim Hadler



Background: Shiga toxin-producing E. coli (STEC) is a reportable illness. It is associated with a spectrum of clinical manifestations that range from watery and/or bloody diarrhea to hemolytic uremic syndrome (HUS), a life threatening condition which causes acute renal failure in both children and adults. A study in Connecticut (CT) by Whitney et al. (2015) found a strong association between higher socioeconomic status (SES) and Shiga toxin-producing E. coli incidence. In the same study, there was a weaker association demonstrated between rurality and STEC incidence as well. The primary aim of this study is to investigate the relationship between SES disparities and the incidence of reported STEC-cases in CT between 2012 and 2014. By linking surveillance data with census tract information, populations can be identified that exhibit higher levels of disease burden. The value of geocoding and linking surveillance data to census data in describing the epidemiology of infectious diseases has been demonstrated in many instances. The study will also investigate the relationship of the geographical distribution and densities of cattle/ruminants and related environmental structures, specifically to farm animal-displaying county fairs, with human STEC infection in CT. County fairs have been suspected as an important point source for Shiga toxin- producing E. coli transmission to humans, following the work of Crump (2003) where agricultural fairs exhibiting livestock were implicated in STEC O157:H7 (STEC O157:H7) outbreaks. Data from Keen (2006) suggest that STEC O157 is common and highly transmissible among the livestock displayed at agricultural fairs and can persist in the environment well after the fair is concluded. Thus, fairs where livestock and other animals are displayed and kept for a time in very close contact conditions are worth exploring as sources of the STEC disease.

Methods: A total of 180 incident STEC cases were reported in CT from 2012-2014. For analysis, the cases were linked to neighborhood poverty level which was broken down into four categories based on percent of people living under the federal poverty line: 0–4.99%, 5–9.99%, 10–19.99%, and greater than 20%. County fairs in CT were compiled and geocoded with ArcGIS. Census Tracts were further classified as being in proximity to a county fair if they were located within the five mile buffer zone of a geocoded fair. Using national 2010 census information, three year age- and sex-standardized Shiga STEC incidence rates were calculated for each poverty category and census tract proximate and not proximate to a county fair. High risk cases with onset dates that fell within the ten day period of risk (3 day fair activity and 7 day STEC incubation period) were examined to see if people living in those census tracts at county fair time were at higher risk of STEC infection than they were when the fair was not in operation. County fair operation time was defined as a period of 94 days during summer and fall when significant fair activity occurs. Incidence rates were also analyzed by race/ethnicity and by year. Generalized linear models were employed to investigate the association of STEC risk with neighborhood poverty levels after adjusting for age, sex, race/ethnicity and county fair proximity. A similar model was constructed to see if there was an association of STEC and fair proximity while also adjusting for age, sex, race/ethnicity and neighborhood poverty level.

Results: One hundred and seventy-three of the 180 incident cases of STEC (96%) were geocoded based on the case’s address using ArcGIS. Neighborhood poverty level was found to be significantly correlated with STEC risk. Overall, there was a progressive increase in IR with increasing age, females had 1.4-fold higher incidence rates than males, and non-Hispanic whites had both the most cases (83%) and an IR that was six-fold higher than non-Hispanic blacks. Age- and sex-standardized rates for all STEC infections (all serogroups) revealed an inverse relationship between poverty and STEC incidence. These results were consistent with the previous Whitney et al. (2015) study that found STEC risk highest in areas with less poverty. The STEC incidence in the second wealthiest SES category (5-9.99%) increased in 2012-2014 as compared to 2000-2011. The generalized linear model adjusted for age group, sex, race/ethnicity and proximity to fair, also showed a significant association between neighborhood poverty level and STEC risk. In terms of county fairs, the incidence rate was higher in the sum of census tracts with a fair in proximity (1.79 per 100 000 person years vs 1.42 per 100 000 person years, IRR 1.26), but the association was not statistically significant. The cases identified as being potentially exposed through a county fair had an incidence rate of 3.68 per 100,000 person years. This rate was found to be significantly higher in the 94 day fair activity period versus census tracts not in proximity to an operating county fair. However, again, the linear model found no significant association between STEC risk and operating fair proximity.

Conclusions: This study showed that in Connecticut STEC incidence decreases as neighborhood poverty increases. These results mirror the results of Whitney et al. (2015). These findings can be used to more effectively target education and interventions, starting with higher-income neighborhoods. This data analysis failed to completely elucidate the relationship between the geographical location of county fairs and STEC risk. The generalized linear model showed no association between STEC risk and proximity to a county fair even though the univariate model suggested positive risk association. Looking to the future, studies designed to increase the understanding of the mechanism driving the differences in STEC risk between higher and lower income areas are warranted. Though the relationship between STEC risk and rurality and county fair proximity was not well elucidated, its further exploration appears justified as well. It is likely that the risk presented by county fairs is more complex than what can be represented by simple geographical location data.


This is an Open Access Thesis.

Open Access

This Article is Open Access