Date of Award

Summer 2018

Degree Type

Thesis

Degree Name

Master of Public Health (MPH)

Department

Public Health

First Advisor

Matthew Hayat

Second Advisor

Ashli Owen-Smith

Abstract

INTRODUCTION: Diabetes Mellitus (DM) is one of the fastest growing and most costly chronic diseases in the United States. DM is severely underdiagnosed, resulting in increased complications, costs, and mortality. Primary goals of the Affordable Care Act (ACA) were to increase health insurance coverage and access to care, and to improve chronic disease outcomes. However, the effects of the legislation have not been widely studied, particularly the relationship between proper diabetes diagnosis and a variety of health related factors.

AIM: Determine the relationship between DM prevalence and under-diagnosis, to healthcare utilization, usual source of care, insurance, type of insurance, and population characteristics have changed since the implementation of the ACA.

METHODS: Data collected between 2005 and 2016 in the National Health and Nutrition Examination Survey were used for this work. The Andersen behavioral health model was used as a theoretical framework and selection of study variables. Descriptive statistics and advanced statistical modeling techniques were applied. Distinct multilevel models were used to model the logit of the probability of DM and the logit of the probability of a proper DM diagnosis each as a function of study variables with an indicator of pre- or post-ACA included as a fixed effect. Marginal models are multilevel models that apply population averaged estimates for parameters. Marginal models were specified to account for clustering by time, and generalized estimating equations used to estimate model parameters. The quasi-likelihood under the null (QIC) statistic was estimated for model comparisons. The SAS Software System was used for data analysis and the level of significance set at .05.

RESULTS: The sample consisted of 31,225 participants, with half pre-ACA (n=15,612) and half post-ACA. Females comprised 51.64% of the study sample with 43.50% White, 25.82% Hispanic, and 20.65% Black and a mean (standard deviation) age of 49.3 (17.9) years. About 11.45% of those in the Pre ACA period had a diagnosis of DM, while 13.5% of those in the Post ACA period had a diagnosis of DM. The percentage of uninsured was 23.95% in the Pre ACA period and 20.69% in the Post ACA time period. The prevalence of undiagnosed DM patients was 26.7% before the ACA, and 21.3% after. A multilevel model with DM status as the dependent outcome showed that sex (females vs males: OR=0.83, 95%CI=0.78,0.89,p =.02), USC (yes vs no: OR=1.28, 95%CI=1.03,1.59, p=.03), health insurance (yes vs no: OR=1.21, 95%CI=1.17,1.26, p =.02), and education level(college graduate vs less than high school: OR=0.79, 95%CI=0.64,0.97, p=.05, high school graduate vs less than high school: OR=0.97, 95%CI=0.93,1.03, p=.05) were significantly associated with presence of DM. Participants were more likely to have their DM properly diagnosed after the ACA: in the final multivariable multilevel model, only ACA time period had a significant effect on correct DM diagnosis (OR=1.51, 95%CI=1.24,1.85, p=.04).

CONCLUSIONS: Although prevalence of DM has increased in recent years, under-diagnosis is less of an issue after the ACA. In the multivariable model comparing DM status (having the disease) to selected covariates, sex, health insurance, education, and USC were related to DM status. The ACA time period had no significant relationship with DM status in the multivariable model. However, in the multivariable model for correctly diagnosed DM, ACA time period was the only independent variable that had a significant association with correct DM diagnosis.

DOI

https://doi.org/10.57709/12248171

Download

Share

COinS