Date of Award

5-1-2024

Language

English

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

College/School/Department

Department of Epidemiology and Biostatistics

Dissertation/Thesis Chair

Shao Lin

Committee Members

Samantha Friedman, Akiko S Hosler, Scott Sheridan, Wangjian Zhang

Keywords

Cold exposure, Diabetes comorbidities, Diabetic complications, Heat exposure, Hospital admissions, Transitional months

Subject Categories

Epidemiology

Abstract

Background: Although total environmental factors have been linked with adverse health outcomes, the impacts of temperature and other total environmental factors on diabetes-related hospitalizations and emergency department (ED) visits remain inconclusive.Objective: This dissertation aims to Aim 1) test the associations between extreme heat and diabetes hospitalizations in the summer and its transitional months, identify populations vulnerable to extreme heat’s effects, and examine the modification effects of greenness on the heat-diabetes associations in warm months (Chapters 3 and 4); Aim 2) investigate the associations between extreme cold and diabetes hospitalizations in the winter and its transitional months, identify populations vulnerable to extreme cold’s effects, and assess the modification effects of ultrafine particles (UFP) on the cold-diabetes associations in cold months (Chapters 5); and Aim 3) develop predictive models for high prevalence of diabetes admissions, determine the effect responses by numbers and combinations of selected risk factors for diabetes hospitalizations, and assess the fact-diabetes associations (Chapter 6). Methods: We conduct a time-stratified case-crossover study at individual-level in Aim 1 and 2. These aims use conditional logistic regressions to estimate the associations of interest while controlling for air pollutants and time variant variables. Multiplicative and additive scales of interactions between extreme cold exposure (ECE) and UFP are evaluated, while only multiplicative interaction between extreme heat exposure (EHE) and greenness is assessed. Aim 3 utilizes machine learning methods to develop predictive models and examine effect responses by numbers and combinations of selected risk factors for diabetes hospitalizations in an ecological study conducted at census-tract level. We utilize univariate and multivariate regression models to estimate the factor-diabetes associations. Results: Aim 1) A significantly increased risk of diabetes hospitalizations is associated with per interquartile increase of temperature in May (ranges of excess risk: 3.1%-4.8%) and in July (ranges of excessive risk: (4.1%-4.2%). Females, patients aged 65 years and older, urban residents, and patients with heart diseases are vulnerable to heat’s effects. Additionally, the immediate risks of diabetes hospitalizations associated with EHE significantly increase in low greenness areas with a dose-response effect related to the levels of greenness; Aim 2) ECE significantly increases diabetes hospitalizations in winter months (January and February, ranges of odds ratios (ORs): 1.099-1.242, P < 0.05) and in March (ranges of ORs: 1.068-1.177, P < 0.05), but the risks increase more consistently in March. Females, patients aged 65 years and older, urban residents, and patients with hyperlipidemia and myocardial infarction are susceptible to cold’s effects. Moreover, significant ECE-UFP interactions on diabetes are at both multiplicative (P < 0.05 for product terms) and additive scales; and Aim 3) The prevalence ratios (PRs) range from 1.4 (95% CI: 1.15-1.7) to 3.87 (95% CI: 3.18-4.71) for the associations between risk factors and diabetes. We found effect responses of numbers and combinations of selected risk factors on predictability of diabetes hospitalizations. Having 20 or more risk factors and two high-risk combinations (non-Whites/low tree canopy coverage/no interest dividends or net rental income/never married/live alone and public health insurance/chromium/using public transportation to work/live alone) show 100% predictability for diabetes hospitalizations. Conclusions: EHE and ECE significantly increase the burden of hospitalizations for diabetes in the transitional months of May and March, respectively. This study finds synergistic interactions between ECE and UFP on diabetes. In addition, vulnerability to heat/cold varies by demographics and comorbidities. Furthermore, the census tracts with 20 or more risk factors or having certain high-risk combinations are particularly vulnerable to diabetes hospitalizations.

Download

Included in

Epidemiology Commons

Share

COinS