Table 1. Coding accuracy unassisted and assisted by the DST.
Data fields | Control data (unassisted by the DST) n (%)/Sensitivity (95% CI) |
Trial data (assisted by the DST)
a
n (%)/Sensitivity (95% CI) |
---|---|---|
Intent of injury (grouped) | n = 1,668 | n = 610 |
Accurate coding | 1,653 (99.1) | 605 (99.2) |
Inaccurate coding | 20 (1.2) | 7 (0.8) |
UAvg sensitivity | 0.941 (0.94–0.95) | 0.960 (0.95–0.96) |
Kappa value | 0.889 (0.83–0.94) | 0.945 (0.90–0.99) |
p -Value | <0.001 | <0.001 |
External cause | n = 1,668 | n = 820 |
Accurate coding | 1516 (90.9) | 762 (92.9) |
Inaccurate coding | 152 (9.1) | 58 (7.1) |
UAvg sensitivity | 0.939 (0.898–0.980) | 0.942 (0.89–0.99) |
Kappa value | 0.888 (0.87–0.9) | 0.918 (0.90–0.94) |
p -Value | <0.001 | <0.001 |
Major injury factor | n = 1,668 | n = 1,476 |
Accurate coding | 1,502 (90) | 1,378 (93.4) |
Inaccurate coding | 166 (10) | 98 (6.6) |
UAvg sensitivity | 0.886 (0.850–0.923) | 0.939 (0.909–0.971) |
Kappa value | 0.893 (0.88–0.91) | 0.929 (0.91–0.94) |
p -Value | <0.001 | <0.001 |
Abbreviations: CI, confidence interval; DST, decision support tool; UAvg, unweighted average.
Number of records from the trial dataset included in the analysis varies depending on the number of records classified by the classifier.