Medication-related critical incidents (CIs) comprise harmful and potentially harmful events. The aim of CI monitoring is quality improvement through system changes. In a prospective survey, we analysed our drug-related CIs of the year 2001 with an emphasis on how they contributed to system changes. A voluntary, anonymous, non-punitive CI reporting was used. The study was performed in a multidisciplinary, 23-bed, neonatal-paediatric intensive care unit (ICU). CI severity was graded: minor (no interventions required), moderate (requiring routine therapy, available outside the ICU), major (need for therapeutic interventions specific to the ICU). There were 284 drug-related CIs, 76% (95% confidence interval 71%-81%) of minor, 19% of moderate and 5% of major severity. A total of 24 CIs were potentially life threatening (if not detected). Some 27% of CIs were intercepted, 17% before preparation and 10% before administration of the drug to the patient. There was a negative correlation between median delay (from CI to detection) and mean severity of the different drug classes involved (P = 0.027). As to the impact on quality, 46 CIs were followed by system changes and 63% (95% confidence interval 49%-77%) of these CIs were of minor severity. Examples of system changes are: double checking for potentially harmful drugs, standardised prescription form and contact to the national drug control agency regarding misleading drug labels.
Conclusion: most of the system changes were based on minor critical incidents which were often detected only after a longer period of time. This shows the value of our "low-threshold" critical incident monitoring. Repeated checks along the drug delivery process (prescription, preparation, administration) are an important means to reduce adverse drug events.