Table 2 Statistical results of semantic queries of different complexity (complexity is represented by number of query concepts). All statistics are calculated with MedCalc® [29]. Raw data can be found in Additional file 1: Table S2 in the addendum
From: Integrating terminologies into standard SQL: a new approach for research on routine data
Query | “Vitium cordis” | “Respiration disorder” | “Lethal malformation” |
|---|---|---|---|
Number of query concepts | 1 | 5 | 16 |
Sample size (Percentage of all cards) | 1868 (100%) | 467 (25%) | 1868 (100%) |
Sensitivity | 93.22% (CI 95%: 89.22–96.08%) | 97.76% (CI 95%: 95.79–98.97%) | 92.13% (CI 95%: 89.48–94.29%) |
Specificity | 99.94% (CI 95%: 99.66–100.00%) | 95.45% (CI 95%: 87.29–99.05%) | 96.30% (CI 95%: 95.16–97.24%) |
Positive predictive value | 99.55% (CI 95%: 96.87–99.94%) | 99.24% (CI 95%: 97.75–99.75%) | 90.57% (CI 95%: 87.75–92.92%) |
Disease prevalence | 12.57% (CI 95%: 11.10–14.15%) | 85.90% (CI 95%: 82.41–88.92%) | 27.80% (CI 95%: 25.78–29.89%) |
F score | 0.96 | 0.98 | 0.91 |