Many hospitals rely on incident reporting as their key quality and safety measure, despite widespread acknowledgement that many errors go unreported. Medication errors are among the most frequent adverse events in hospitals, as well as the most dangerous. However, it has been estimated that only some 13 per cent are reported by staff. Little is known about whether the types of medication errors reported by staff represent the full spectrum of errors which occur. Thus interpreting and taking action in response to incident data can be difficult.
Around half of all adverse medication events are preventable, thus better identification and reporting of errors can allow for the design of interventions to more effectively reduce harm to patients.
Encouraging staff to report incidents is viewed as an important element in creating a positive safety culture and attention has focused on understanding barriers to reporting by staff. However, an under explored issue is whether under-reporting is also due to a failure of staff to detect medication errors.
We studied medication errors occurring at two large Sydney teaching hospitals by:
- auditing patient records and observing nurses administering drugs to patients to find out how many and what kinds of errors were being made, if staff detected these errors and then if errors were reported to the hospitals’ incident reporting systems
- assessing how the two hospitals differed in terms of the medication error rates observed versus the errors actually reported by staff
Researchers reviewed 3291 patient records to identify prescribing errors (e.g. wrong drug, dose or strength) and evidence of their detection by staff. Errors during the administration of medications to patients were identified from a direct observational study of 180 nurses administering 7451 medications to 1397 patients across the two hospitals. Severity of errors was classified and those likely to lead to patient harm were categorised as ‘clinically important’.
Of the 12,567 prescribing errors identified, 539 or 4.3% were clinically important. There was evidence that staff had detected 21.9% (118) of these clinically important errors, but very few (7, 1.3%) were reported to the hospitals’ incident systems. The remaining 78.1% (n=421) failed to be detected, although it is possible that some of these errors were detected by staff but no information to this effect was recorded in patients’ records.
Of the medication administration errors, most were (79%) procedural (eg failing to check a patient’s identification before administering a drug). One or more clinical errors (e.g. wrong dose) occurred in 27.4% of drug administrations, and in 10.2% the errors were rated as clinically important, with the potential to cause patient harm. None was reported to the incident systems. This matches overseas experience, and may be explained partly by the difficulty of identifying such errors once a drug has been administered.
Comparing the two hospitals, we found no relationship between the number of reported medication incidents and the ‘actual’ rate of prescribing errors. The hospital with the higher number of incident reports had lower ‘actual’ prescribing errors and vice versa. Thus in this instance the higher number of medication incidents reported reflected a lower patient risk. These results support the notion that encouraging the reporting of incidents is an element in creating a safety culture likely to support improved care.
Our study suggests that hospitals’ incident data have significant shortcomings, especially as the basis for new quality and safety procedures. As many clinically important prescribing errors go undetected they also go unreported. Currently, the reporting of incidents does not accurately reflect the profile of medication errors in our hospitals, or real error rates. This means using incident frequency of errors to compare patient risk or performance quality within or between hospitals is unreliable. New approaches including data mining of electronic clinical information systems are required to support more effective medication error detection and to provide the data needed to develop safer practices.