Michael P. Stern, MD; Ken Williams, MS; Steven M. Haffner, MD, MPH
Acknowledgment: The authors thank Dr. Agustin Escalante for providing helpful advice on the evaluation of ROC curves.
Grant Support: By the National Heart, Lung, and Blood Institute (grants RO1 HL24799 and RO1 HL36820).
Requests for Single Reprints: Michael P. Stern, MD, Division of Clinical Epidemiology, Department of Medicine, University of Texas Health Science Center at San Antonio, 7703 Floyd Curl Drive, San Antonio, TX 78229-3900; e-mail, firstname.lastname@example.org.
Current Author Addresses: Drs. Stern and Haffner and Mr. Williams: Division of Clinical Epidemiology, Department of Medicine, University of Texas Health Science Center at San Antonio, 7703 Floyd Curl Drive, San Antonio, TX 78229-3900.
Author Contributions: Conception and design: M.P. Stern, S.M. Haffner.
Analysis and interpretation of the data: M.P. Stern, K. Williams, S.M. Haffner.
Drafting of the article: M.P. Stern, K. Williams.
Critical revision of the article for important intellectual content: M.P. Stern, K. Williams, S.M. Haffner.
Final approval of the article: M.P. Stern, K. Williams, S.M. Haffner.
Statistical expertise: K. Williams.
Obtaining of funding: M.P. Stern.
Administrative, technical, or logistic support: M.P. Stern.
Collection and assembly of data: M.P. Stern, S.M. Haffner.
The standard method of identifying persons at high risk for type 2 diabetes mellitus involves detection of impaired glucose tolerance, which requires a costly and inconvenient 2-hour oral glucose tolerance test. Because clinical trials have indicated that diabetes is preventable by using behavioral or pharmacologic interventions, less expensive methods of identifying high-risk persons are needed.
To determine whether multivariable models are superior to glucose tolerance tests for identifying persons at high risk for diabetes mellitus.
Prospective cohort study.
San Antonio, Texas.
1791 Mexican Americans and 1112 non-Hispanic whites without diabetes at baseline who were randomly selected from census tracts.
Medical history; body mass index; blood pressure; fasting and 2-hour plasma glucose levels; fasting serum total, low-density lipoprotein, and high-density lipoprotein cholesterol levels; and triglyceride level.
For prediction of 7.5-year incidence of type 2 diabetes, the area under the receiver-operating characteristic (ROC) curve for a multivariable model involving readily available clinical variables was significantly (P < 0.001) greater than the area under the ROC curve for the 2-hour glucose value alone (84.3% vs. 77.5%). Impaired glucose tolerance represents a single point on the latter curve. Adding the 2-hour glucose measurement to the multivariable model increased the area under its ROC curve, but only from 84.3% to 85.7%.
Persons at high risk for diabetes mellitus are better identified by using a simple prediction model than by relying exclusively on the results of a 2-hour oral glucose tolerance test. Although adding the 2-hour glucose variable to the model enhanced prediction, the resulting slight improvement entails greater cost and inconvenience.
Stern MP, Williams K, Haffner SM. Identification of Persons at High Risk for Type 2 Diabetes Mellitus: Do We Need the Oral Glucose Tolerance Test?. Ann Intern Med. 2002;136:575–581. doi: 10.7326/0003-4819-136-8-200204160-00006
Download citation file:
Published: Ann Intern Med. 2002;136(8):575-581.
Cardiology, Coronary Risk Factors, Diabetes, Endocrine and Metabolism.
Results provided by:
Copyright © 2018 American College of Physicians. All Rights Reserved.
Print ISSN: 0003-4819 | Online ISSN: 1539-3704
Conditions of Use