We have 47 cases and 302 controls with genotypic results. Previously we had randomly selected 25 cases and 25 controls for abstraction validation. I compared the electronic case/control status versus our abstraction and here are the results:
Of the 25 cases that we selected, all 25 are still valid cases electronically per the most recent algorithm. 24/25 (96%) validated as true cases per abstraction. The one case that did not validate could not confirm atopic dermatitis per abstraction.
Of the 25 controls that we selected, 21 are still valid controls electronically per the most recent algorithm. 19/21 (90%) validated as true controls per abstraction. The two that did not validate do not have atopic dermatitis, but have exclusion allergies that were found in the chart notes and the subjects were never given the allergy diagnoses. One was allergy to cats/dogs (ICD 9 477.2), and the other was allergy to latex (ICD 9 V15.07).
| Cases (0) | Actual Class (Expectation) | |
|
Predicted Class (Observation) |
True Positive: 24.00 |
False Positive: 1.00 |
|
Case Positive Predictive Value:
96%
|
||
| Control (0) | Actual Class (Expectation) | |
|
Predicted Class (Observation) |
True Positive: 19.00 |
False Positive: 2.00 |
|
Control Positive Predictive Value:
90.48%
|
||