BACKGROUND: Screening for non-insulin-dependent diabetes mellitus (NIDDM) can be useful in clinical practice and in epidemiologic and genetic studies, but the available information for choosing between screening methods is limited. In this study, characteristics of several screening tests for NIDDM were compared. METHODS: Among Pima Indians participating in an epidemiologic study, the sensitivity and specificity for detecting NIDDM of fasting plasma glucose (FPG) levels and two measures of glycated hemoglobin (HbA1 or HbA1c) were compared in 2092 fasting subjects. Glycated hemoglobin, quantitative glycosuria, and dipstick glycosuria were compared in 237 nonfasting subjects. Diabetes was diagnosed using an oral glucose tolerance test if the 2-hour postload venous plasma glucose concentration was 11.1 mmol/L (200 mg/dL) or greater. The area under the relative operating characteristic curve was used to compare tests. RESULTS: In fasting subjects, the sensitivity for detecting diabetes with 98% specificity was 78.8% for HbA1 level of 7.5% or greater, 80.3% for HbA1c level of 6.3% or greater, and 88.0% for FPG level of 6.83 mmol/L (123 mg/dL) or greater. By relative operating characteristic analysis, there were no significant differences between FPG and HbA1c, but FPG was significantly more sensitive than HbA1. In nonfasting subjects the sensitivity at 98% specificity was 92.9% for HbA1 level of 7.3% or greater, 80.6% for quantitative urine glucose level of 1.94 mmol/L (35 mg/dL) or greater, and 64.3% for trace or greater of dipstick glycosuria. The area under the relative operating characteristic curve was significantly greater for glycated hemoglobin than for either measure of glycosuria. CONCLUSIONS: Although FPG has the best screening properties, HbA1c, HbA1, and quantitative urine glucose also provide high specificity and approximately 80% sensitivity in detecting NIDDM. The choice of a particular method could depend on cost, convenience, and availability.
BACKGROUND: Screening for non-insulin-dependent diabetes mellitus (NIDDM) can be useful in clinical practice and in epidemiologic and genetic studies, but the available information for choosing between screening methods is limited. In this study, characteristics of several screening tests for NIDDM were compared. METHODS: Among Pima Indians participating in an epidemiologic study, the sensitivity and specificity for detecting NIDDM of fasting plasma glucose (FPG) levels and two measures of glycated hemoglobin (HbA1 or HbA1c) were compared in 2092 fasting subjects. Glycated hemoglobin, quantitative glycosuria, and dipstick glycosuria were compared in 237 nonfasting subjects. Diabetes was diagnosed using an oral glucose tolerance test if the 2-hour postload venous plasma glucose concentration was 11.1 mmol/L (200 mg/dL) or greater. The area under the relative operating characteristic curve was used to compare tests. RESULTS: In fasting subjects, the sensitivity for detecting diabetes with 98% specificity was 78.8% for HbA1 level of 7.5% or greater, 80.3% for HbA1c level of 6.3% or greater, and 88.0% for FPG level of 6.83 mmol/L (123 mg/dL) or greater. By relative operating characteristic analysis, there were no significant differences between FPG and HbA1c, but FPG was significantly more sensitive than HbA1. In nonfasting subjects the sensitivity at 98% specificity was 92.9% for HbA1 level of 7.3% or greater, 80.6% for quantitative urine glucose level of 1.94 mmol/L (35 mg/dL) or greater, and 64.3% for trace or greater of dipstick glycosuria. The area under the relative operating characteristic curve was significantly greater for glycated hemoglobin than for either measure of glycosuria. CONCLUSIONS: Although FPG has the best screening properties, HbA1c, HbA1, and quantitative urine glucose also provide high specificity and approximately 80% sensitivity in detecting NIDDM. The choice of a particular method could depend on cost, convenience, and availability.
Authors: Arch G Mainous; Richard Baker; Azeem Majeed; Richelle J Koopman; Charles J Everett; Sonia Saxena; Barbara C Tilley Journal: Public Health Rep Date: 2006 May-Jun Impact factor: 2.792
Authors: Arch G Mainous; Azeem Majeed; Richelle J Koopman; Richard Baker; Charles J Everett; Barbara C Tilley; Vanessa A Diaz Journal: Public Health Rep Date: 2006 Jan-Feb Impact factor: 2.792
Authors: Caroline West; David Ploth; Virginia Fonner; Jessie Mbwambo; Francis Fredrick; Michael Sweat Journal: Am J Med Sci Date: 2016-02-10 Impact factor: 2.378