Measurement of adverse events using "incidence flagged" diagnosis codes.
The authors analyzed Australian hospital administrative data and found that ICD-10 codes alone are less accurate in identifying adverse events than in combination with "incidence flags."