Table 2

Reliability of different potential definitions of diabetes, using HbA1c ≥6.5% as the gold standard

Potential definitionPatients meeting criteria (%)*Sensitivity (%)Specificity (%)PPV (%)NPV (%)Accuracy† (%)
(A) RPG >140 mg/dL33.870.974.337.692.173.7
(B) RPG >180 mg/dL14.350.493.663.189.685.8
(C) RPG ≥200 mg/dL10.843.696.472.488.786.9
(D) ≥1 FPG levels ≥126 mg/dL36.278.473.038.893.974.0
(E) ≥2 FPG levels ≥126 mg/dL17.058.992.262.291.186.2
(F) ≥1 FPG levels >140 mg/dL23.167.086.551.992.383.0
(G) ≥2 FPG levels >140 mg/dL10.042.297.175.888.587.2
(H) ≥1 FPG ≥126 and RPG (I) ≥200 mg/dL8.640.498.484.488.388.0
(J) ≥1 FPG >140 and RPG (K) ≥200 mg/dL7.737.998.988.488.088.0
  • *Prevalence of diabetes per HbA1c screening=17.9%.

  • †Accuracy=% of true results (ie, (true positives+true negatives)/total).

  • FPG, fasting plasma glucose; HbA1c, glycosylated hemoglobin; NPV, negative-predictive value; PPV, positive-predictive value; RPG, random plasma glucose (collected at admission).