O b jecti ve: To describe the nature and frequency of adverse events reported in routinely compiled national mortality and hospital morbidity data collections, and to gauge the usefulness of these data collections (using ICD to classify injury and poisoning, and external causes) as sources of data on adverse events in Australia. Adverse events reported in a national survey of general practitioner activity will also be described.

Design: Retrospective analysis of data in the AIHW National Mortality Database, the AIHW National Hospital Morbidity Database and the BEACH survey of general practitioner activity.

Subjects: Deaths registered in Australia in 1997 and 1998, admitted patient episodes (separations) in almost all Australian hospitals in 1997-98, and 201,757 weighted patient encounters with randomly selected general practitioners in 1998-99 and 1999-00.

Main outcome measures: Numbers and proportions of deaths registered with an adverse event as a cause of death, numbers and proportions of separations from hospital with an adverse event reported as a diagnosis or as an external cause of injury or poisoning, and proportions of general practitioner-patient encounters with an adverse event reported as a problem managed.

R esults: A total of 2,594 deaths registered in 1997 (2.0% of all deaths) and 2,939 registered in 1998 (2.3%) had an adverse event reported as a cause of death, an average of 14.9 deaths per 100,000 population per year. The adverse event was reported as the underlying cause of death for 177 of these deaths in the two-year period. Complications of surgical and medical care were the reported adverse events for 73.3% of the 5,533 deaths, adverse drug effects for 26.6%, iatrogenic diseases for 62.4% and misadventures to patients for 1.3%.

An adverse event was reported as a diagnosis or an external cause of an injury or poisoning for 264,347 separations from Australian hospitals in 1997-98 (4.75% of all separations). Complications of surgical or medical care were reported for 72.2% of these, drug adverse effects for 20.2%, iatrogenic diseases for 75.8% and misadventures to patients for 1.8%.

An adverse event was a reported problem for about 0.9% (95% CI 0.8%-1.0%) of general practice encounters in both 1998-99 and 1999-00. The problems were mainly adverse effects of medical agents and complications of treatment.

Conclusions : The mortality and hospital morbidity databases have the advantages of being routinely collected, fully covering deaths and hospital separations in Australia, and being supported by substantial national data collection infrastructures. They appear to usefully record a range of adverse events. However, some features of the ICD-9(-CM) (and ICD-10(-AM)) classifications and of the source material (death certificates and hospital medical records), and uncertain data quality mean that their use in routine monitoring of the occurrence of adverse events may not be possible without further assessment. For hospital data, in particular, it could be important to undertake studies of the validity of the coded data against medical records (for example, to assess the capacity of hospital morbidity data to record various types of adverse events) and enhancements to the ICD-10-AM classification and other changes to data collection practices that would improve the sensitivity of analyses of the data for adverse events.