Using Chart Review as the Gold Standard | |||||||||
---|---|---|---|---|---|---|---|---|---|
Concept | Operational Definition of Preliminary Case-finding Algorithm | No. in Population | No. in Random Sample* | No. True Positives | No. False Positives | Sensitivity (%) | 95% CI | PPV (%) | 95% CI |
Disease confirmation | ≥ 1 diagnosis code from any provider | 1153 | 97 | 67 | 30 | 100 by definition | — | 69 | 59–78 |
≥ 2 diagnosis codes from any provider** | 772 | 64 | 58 | 6 | 87 | 76–93 | 91 | 80–96 | |
≥ 1 diagnosis code from rheumatology | 696 | 60 | 54 | 6 | 81 | 69–89 | 90 | 79–96 | |
New-onset disease | ≥ 1 diagnosis code from any provider | 1153 | 97 | 44 | 53 | 100 by definition | — | 45 | 35–56 |
≥ 1 diagnosis code from rheumatology | 696 | 60 | 42 | 18 | 95 | 84–99 | 70 | 57–81 | |
≥ 1 diagnosis code from rheumatology, ≥ 2 laboratory tests performed (ANA, RF, HLA-B27) | 581 | 52 | 39 | 13 | 93 | 79–98 | 75 | 61–86 | |
≥ 1 diagnosis code from rheumatology, ≥ 2 laboratory tests performed (ANA, RF, HLA-B27), ≥ 12 months of prediagnostic enrollment | 488 | 45 | 37 | 8 | 95 | 81–99 | 82 | 67–91 |
↵* A random sample of about 10% of all preliminary cases was selected by assigning random numbers from 0–1.00 to the 1153 cases and selecting all cases with numbers from 0–0.10 for validation using medical record review. The number in the random sample was 97 of 1153, with subsets of the 97 meeting various operational definitions tested as preliminary case-finding algorithms. PPV: positive predictive value; JIA: juvenile idiopathic arthritis; ANA: anti-nuclear antibody; RF: rheumatoid factor.