Dissertations & Theses (Open Access)

Date of Award

Fall 12-2018

Degree Name

Master of Public Health (MPH)


Joseph B. Mccormick, Md

Second Advisor

Paul J. Rowan, Phd, Mph


A pilot study was undertaken to evaluate the appropriateness of a previously published diabetes risk stratification tool in a diabetic population. The tool was applied to a sample of 500 prediabetic and diabetic adults receiving primary care services at a Federally Qualified Health Center (FQHC) in Cameron County, Texas. The study population was largely Hispanic and underserved. The National Health and Nutrition Examination Survey (NHANES) 2015-2016 data set was used as a comparison group. The risk assessment tool was applied to separately to the prediabetic and diabetic subset of both study groups. The tool stratified the patients into three risk categories: green (low risk), yellow (moderate risk ) and red (high risk). The tool was applied to both the weighted and unweighted NHANES data; however, unweighted NHANES data was used for most of the comparisons as this was a pilot study. After applying the tool, among the prediabetic clinic patients, 20% were categorized into the red zone, while 1% of the prediabetic comparison group was placed in this zone. For diabetic clinic patients, 56% fell into the red zone, with 42% of the comparison group in this zone. These differences were significant. The utility of the tool was limited by the degree of missing data points, particularly among the clinic patients. The tool uses the values of the Patient Health Questionnaire 9 (PHQ9) in the risk stratification process. At least 64% of the PHQ9 scores were missing in clinic patients. The average PHQ9 score was computed and assigned to those clinic patients with missing PHQ9 scores. Applying the tool to this simulated data reduced the percentage of prediabetic clinic patients in the red zone to 16% and the percentage of diabetic patients in this zone to 44%. After this simulation, the distribution of the risk zones of the diabetic patients was no longer significantly different from the comparison group. This study demonstrates the importance of assessing for missing data in applying a risk stratification tool.