Blood Glucose Variability Measures: Going Beyond Traditional Methods Open Access

Kelleman, Michael (2013)

Permanent URL:


BACKGROUND: The establishment of a formal and effective method to characterize glucose variability is a crucial step in the advancement of reducing morbidity and mortality, especially in critically ill patients in the intensive care unit (ICU). Previous studies have used measures that are unable to account for correlation between blood glucose measurements. We sought to provide a standardized method of data analysis from a glucose management protocol in hospitalized patients that is more appropriate for these repeated measures.

METHODOLOGY: Our study population consisted of 153 pediatric patients from two pediatric ICU (PICU): Children's Healthcare of Atlanta (CHOA) in Atlanta, GA and James Whitcomb Riley Hospital for Children in Indianapolis, IN. Baseline severity of illness scores (PELOD) were calculated at the time of admission and on day 6. Blood glucose measurements were collected throughout their hospitalization and PICU (length of stay) LOS and hospital LOS were recorded. We performed GEE analysis on blood glucose variability measures to determine the association with mortality. Additionally, we performed mixed linear model analysis on blood glucose variability measures to determine the association with PICU LOS, hospital LOS and change in PELOD scores.

RESULTS: We observed increased hospital and PICU LOS with increasing maximal glucose levels and decreased hospital and PICU LOS with increasing minimal glucose levels. No glucose variability measures were found to be associated with mortality either independently or in a GEE model. Standard deviation (SD) and glucose variability index were not found to be significantly associated with PICU or hospital LOS while coefficient of variation CV was found to be significantly associated with both PICU of hospital LOS. CV and was also significantly associated with change in PELOD after controlling for baseline severity of illness while SD and glucose variability index were not significantly associated with changed in PELOD.

DISCUSSION: These results are consistent with previous studies while we introduced a novel way to analyze blood glucose measurement from critically ill pediatric patients in the PICU. Further research on glucose variability and repeated measure analysis should consider incorporating other characteristics of variability measured using area under the curves AUC applied to glucose versus time.

Table of Contents

Chapter I: Introduction

Background 1

Problem/Purpose/Significance 2

Assumptions 4

Definition of Terms 4

Chapter II: Review of the Literature

Glucose Variability 8

Heart Rate Variability 10

Chapter III: Methodology

Study Population and Measurements 12

Analysis 14

Chapter IV: Results

Patient Demographics 16

Glucose Variability Measures 16

Hyperglycemia and Hypoglycemia 17

Maximal Glucose, Minimal Glucose, and Glucose Variability Index Quintiles 17

Modeling Results 19

Chapter V: Discussion

Summary of Study/Strengths and Limitations 21

Conclusions 23

Future Research 23

References 25

Appendix: Figures and Tables

Figure 1. Formulas Used in Describing Glucose Variability 29

Table 1. Pediatric Demographics 30

Table 2. Pediatric Glucose Variability Measures by Hospital 31

Table 3. LOS and Mortality Rates According to Glucose Ranges and Glucose Cutoff Values 32

Table 4. LOS and Mortality Rates According to Glucose and Glucose Variability Index Quintiles 33

Figure 2. Example AUC Chart 34

About this Master's Thesis

Rights statement
  • Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.
Subfield / Discipline
  • English
Research field
Committee Chair / Thesis Advisor
Committee Members
Partnering Agencies
Last modified

Primary PDF

Supplemental Files