Stepwise Screening for Asymptomatic Diabetes Using Opportunistically Available Random Plasma Glucose and HbA1c Open Access

Legvold, Brian (Spring 2020)

Permanent URL:



Oral glucose tolerance tests (OGTTs) are inconvenient but sensitive for identifying diabetes (DM), whereas more convenient HbA1c tests may be inaccurate.


We asked if an alternative two-step strategy, measuring HbA1c only if opportunistically available random plasma glucose (RPG) is ≥100 mg/dl, could improve screening.


The Screening for Impaired Glucose Tolerance (SIGT) dataset, where 1,573 adults without known DM had measurements of RPG, HbA1c, and OGTTs; was used to evaluate the two-step strategy, using Receiver Operating Characteristic (ROC) analysis adjusted for optimism to identify DM per American Diabetes Association (ADA) OGTT criteria.


Participants were 58% female and 58% black, with mean age 47.9 years, BMI 30.3 kg/m2 and HbA1c 5.4%; 4.6% had DM by ADA OGTT criteria.  The ROC area under the curve was 0.82 for HbA1c to identify DM among all 1,573 participants, but 0.86 in those with RPG ≥100 mg/dl (n=576), vs. 0.58 in those with RPG <100 mg/dl (n=997) (modeled interaction p<0.001). DM participants with RPG ≥100 vs <100 mg/dl had mean fasting plasma glucose 131 vs. 116 mg/dl and 2-hour plasma glucose 225 vs. 183 mg/dl, and HbA1c 6.4% vs. 5.6%, respectively, (all p<0.025) – less severe disease in those with RPG <100 mg/dl.  Limiting OGTTs to those with RPG ≥100 mg/dl and HbA1c ≥5.5% would provide 74% sensitivity and 82% specificity overall and reduce the number of OGTTs needed by 80%. The participants with unrecognized DM who were not identified by this method (n=19) had a mean HbA1c of 5.5% (±0.6%), a fasting glucose of 115 mg/dl (±19.4 mg/dl) and an 2hr OGTT of 195 mg/dl (±56.3 mg/dl).


Use of RPG followed by HbA1c improves the accuracy and efficiency of screening, identifying both individuals who should and should not have an OGTT. Such a strategy might improve recognition of diabetes and prediabetes, permitting initiation of preventive management.

Table of Contents

Table of Contents


Research Design and Methods

Study Population




Statistical Analysis


Participant Demographics

Interaction Assessment

Model Fit

Receiver Operating Characteristics



Table 1 Participant demographics

Table 2 Screening accuracy of reduced model by HbA1c cut point

Table 3 Screening accuracy by HbA1c cut point, relative to complete study population


Figure 1 Random Plasma Glucose versus HbA1c, by Dysglycemia

Figure 2 Distribution of HbA1c by RPG ≥100 mg/dl and Type 2 Diabetes Mellitus


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
Last modified

Primary PDF

Supplemental Files