Date of Award

Spring 1-3-2018

Degree Type


Degree Name

Master of Public Health (MPH)


Public Health

First Advisor

Matt Hayat, PhD

Second Advisor

Maya Sternberg, PhD


INTRODUCTION: Hemoglobin A1C (A1C) is a common test used in the diagnosis of diabetes mellitus and for long-term glucose management. Within the literature, it has been shown that there may be association between iron deficiency (ID) and A1C. In this study, we analyzed observational data using propensity score analysis to further explore this association.

AIM: The aim of this study is to compare the association of A1C and ID among non-pregnant, non-diabetic women aged 12-49 without a history of chronic renal disease using various statistical methods to control for covariate imbalance.

METHODS: Data on 4,656 women obtained from the National health and Nutrition Examination Survey (NHANES) during the period 2003-2008 was used to compare four different statistical methods to examine the association between A1C and ID: unadjusted and adjusted multivariable logistic regression and two different propensity score analyses to impose covariate balance between those who were ID and not-ID.

RESULTS: The unadjusted crude odds ratio between ID and elevated A1C was 2.32 (95% CI: 1.75, 3.07); while the adjusted odds ratio was slightly attenuated [OR = 2.1 (95% CI: 1.55, 2.85)] after controlling for age, income, education level, race, BMI, smoking status, and 24-hour dietary recall of iron intake. Further adjustment for the propensity score in the logistic model yielded an odds ratio of 1.88 (95% CI: 1.41, 2.49) if the propensity score was treated as a linear variable and 1.92 (95% CI: 1.45, 2.55) if treated as a categorical variable using the quintiles propensity score.

DISCUSSION: This study confirms the presence of a statistically significant association between A1C and ID, and further suggests this may be a causal association. These findings may have implications for diabetes screening if ID causes shifts in A1C.