Skip to main content
. Author manuscript; available in PMC: 2020 May 1.
Published in final edited form as: Pharmacoepidemiol Drug Saf. 2019 Jan 24;28(5):632–639. doi: 10.1002/pds.4718

Table 2.

Random sample of patients with possible T2DM by computable phenotype (CP) and site at selected health care systems in PCORnet, 2012-2016

Total Reviewed Patients with confirmed T2DM (n=1,512) Patients without confirmed T2DM (n=60)

N PPV1 (95% CI2) Did not meet ≥ 2 Diabetes criteria3 Type 1 diabetes, gestational diabetes or prediabetes
Overall 1572 1512 96.2 (95.1, 97.0) 52 8

By computable phenotype
 CP1 - Diabetes diagnosis and medication use 1124 1084 96.4 (95.2, 97.4) 37 3
 CP2 - Diabetes diagnosis and HbA1c ≥6.5% 432 415 96.1 (93.8, 97.5) 13 4
 CP3 - HbA1c ≥6.5% and medication use 16 13 81.3 (57.0, 93.4) 2 1

By site
 Vanderbilt University Health System 398 379 95.2 (92.7, 96.9) 18 1
 OneFlorida 391 382 97.7 (95.7, 98.8) 8 1
 University of North Carolina 396 380 96.0 (93.5, 97.5) 15 1
 Tulane 387 371 95.9 (93.4, 97.4) 11 5
1

PPV: Positive predictive value

2

Confidence interval using Wilson’s formula

3

Criteria included evidence of a diabetes diagnosis, diabetes medication use or elevated HbA1c ≥6.5%