|
" Causal Inference Using Cross-sectional Data: "
Shelley, Courtney Diane
Kass, Philip H.
| Document Type
|
:
|
Latin Dissertation
|
| Language of Document
|
:
|
English
|
| Record Number
|
:
|
1105157
|
| Doc. No
|
:
|
TLpq2299780170
|
| Main Entry
|
:
|
Kass, Philip H.
|
|
|
:
|
Shelley, Courtney Diane
|
| Title & Author
|
:
|
Causal Inference Using Cross-sectional Data:\ Shelley, Courtney DianeKass, Philip H.
|
| College
|
:
|
University of California, Davis
|
| Date
|
:
|
2019
|
| student score
|
:
|
2019
|
| Degree
|
:
|
Ph.D.
|
| Page No
|
:
|
115
|
| Abstract
|
:
|
Vaccines are the greatest public health achievement of the 20th century.1 Diligent vaccination programs have led to the global elimination of smallpox2 and the reduction of circulating wild polio to only Pakistan, Afghanistan, and Nigeria.3 Circulating measles was eliminated in the Americas in 20164 after a 22-year long vaccination campaign. Maintaining the high vaccination coverage necessary to suppress outbreaks and endemic circulation of infectious diseases requires diligent efforts on the part of public health officials to maintain public trust and awareness, especially as the diseases vaccines protect against fade into memory and the limitations of the vaccines themselves become more evident.4 Vaccine hesitancy is the “… delay in acceptance or refusal of vaccines despite availability of vaccine services.”5 Vaccine hesitancy is “present when vaccine acceptance in a specific setting is lower than would be expected, given the availability of vaccine services”.5 Vaccine coverage estimates cannot be used as a measure of vaccine hesitancy without consideration of other reasons for underimmunization, including barriers to access or situations where communities lack the opportunity to accept or reject vaccine(s). This dissertation research used the National Immunization Survey – Child 2016, a publicly-available dataset designed to provide current, population-based, state and local area estimates of vaccination coverage.6 To use US national immunization coverage data to estimate vaccine hesitancy, social determinants driving a lack of access must first be assessed. In Chapter 1 of this research, a causal diagram of social factors was created using the Andersen Behavioral Model of Healthcare Utilization,7,8 and the impact of health insurance status on age-appropriate vaccination was assessed using regression-based causal analysis. The causal effect of health insurance coverage was shown to be a complex interaction of coverage type, maternal race, and maternal education levels. In Chapter 2, the mechanism of action for health insurance coverage with regard to the causal model was assessed using mediation analysis. The impact of potential healthcare system interventions to increase vaccination coverage, such as universal healthcare or shifting routine preventive services to primary care settings, were analyzed. If social determinants are driving undervaccination, policy interventions to reduce barriers to access will increase vaccination coverage. If instead an unmeasured confounder such as vaccine hesitancy beliefs is driving undervaccination, the observed effect measure between health insurance coverage and undervaccination will reflect a confounded association rather than a true effect measure. In Chapter 3 of this research, the strength of a potential unmeasured confounder between health insurance coverage and undervaccination outcome is assessed using E-values, which quantify the strength of association between an exposure, confounder, and outcome necessary to nullify an observed effect measure.
|
| Subject
|
:
|
Epidemiology
|
|
|
:
|
Immunization
|
| |