Skip to main content
. 2013 Mar 26;20(e1):e147–e154. doi: 10.1136/amiajnl-2012-000896

Table 3.

Positive predictive value for phenotype case and control algorithms, and for phenotype eligibility algorithms across Electronic Medical Records and Genomics sites

Phenotype Number validated Positive predictive value
Group Health (%) Marshfield Clinic (%) Mayo Clinic (%) Northwestern University (%) Vanderbilt University (%)
Validated for case/control status and eligibility
 Cataract
  Case* 3234 97.7 96.0
  Control* 3184 97.7
 Dementia
  Case* 3778 73.0 89.7 84.0
  Control 505 96.7
 Type 2 diabetes
  Case 300 99.0 98.2 100
  Control 143 98.0 100 100
 Diabetic retinopathy
  Case 229 80.0 67.6
  Control 80 98.0 100
 Resistant hypertension
  Case 354 90.0 100 84.4
  Control 144 91.0 100 93.8 84.0
 Peripheral arterial disease
  Case* 11504 87.5 90.7 95.0
  Control 100 100
 Chronic autoimmune hypothyroidism
  Case 389 92.0 91.3 82.0 98.1 98.0
  Control 290 100 100 96.0 100 100
Validated for eligibility
 Low level of high  density lipoprotein 440 81.6
 Lipids* 1054 78.8 92.3
 Red blood cell indices 391 96.4 98.0 98.0
 White blood cell indices 365 89.6 85.0
 QRS duration 245 100 96.9 97.0
 Height 579 86.9 95.1

Blank cells if did not participate in validation of that phenotype.

*Number large due to pre-existing study with validation.