Opportunities for weather-related injury surveillance
There are no routinely published national-level data on weather-related injuries in Australia. The most viable existing sources that contain both injury and external cause data are hospitalisations data from the NHMD and deaths data from the NMD and National Coronial Information System (NCIS). However, there are limitations in associating injury and weather events (as outlined in the limitations discussion)
Improve existing data sources
Research to estimate the robustness of the NHMD and NMD would help validate these data sets as reliable sources of information on weather-related injuries. One way to validate NMD data would be to analyse it in conjunction with NCIS data. Access to standardised emergency management system datasets across regions (for example, ambulance call outs) and increased coded (rather than free text) emergency response data collection would provide valuable acute information related to natural hazards and associated health system responses.
Improving national ED data
External cause data is fundamental for understanding the causes and circumstances of over 2 million injury presentations in emergency departments (EDs) each year (AIHW 2022a). Jurisdictions do collect information about the external cause and circumstances of injury, but there is no national standard for this. One option for improving the National Non-Admitted Patient Emergency Department Care Database (NNAPEDCD) would be for the AIHW to create a pilot national best-endeavours (non-mandatory) data set add-on to the NNAPEDCD for injury cases.
Below are two examples, from Victoria and South Australia, on how jurisdictions collect weather-related injury data from EDs. Investment in national ED data improvement is needed to develop a useful national data source for injury reporting not only on weather-related injuries, but for all injury reporting.
The Victorian Emergency Minimum Dataset (VEMD) contains the same data items as the NNAPEDCD as well as information on injury intent (unintentional, intentional, undetermined), cause (for example fall, poisoning), place of occurrence, activity (for example sport, work), body region and description of the injury event (as open text field) (VISU 2022). The inclusion of external cause information provides the potential to identify weather-related injury records in the VEMD.
South Australia’s Non-Admitted Emergency Care (NAEC) data set contains information on patients presenting to EDs in public hospitals. The NAEC data set includes a non-mandatory data item where a heat-related condition can be flagged and defined as ‘heat exhaustion’, ‘heat stroke’ or ‘other’. Staff can also enter further free-text information into the system.
Implementing real time surveillance systems
Real time data on weather events should be incorporated into surveillance systems at local levels to trigger system response alerts within the health system when injury presentations depart from the usual patterns.
Data linkage is another way to improve what we know about weather-related injury, either by using existing data linkage assets such as the Australian Institute of Health and Welfare (AIHW) National Integrated Health Services Information (NIHSI), Australian Bureau of Statistics (ABS) Multi-Agency Data Integration Project (MADIP, to be renamed Person-Level Integrated Data Asset, PLIDA) or by creating new linked data sets. For example, it would be possible to investigate topics such as:
- how many weather-related hospitalisations were immediately preceded by an emergency department presentation.
- how many deaths occur within a certain period following weather-related hospitalisations.
- what were the Medicare Benefits Schedule (MBS) and Pharmaceutical Benefits Schedule (PBS) utilisation patterns like for people who had been hospitalised for a weather-related injury. From this, linked data could be used for generating health service cost estimates.
- Linking Bureau of Meteorology data such as minimum and maximum temperatures and rainfall by location to usual place of residence in the NHMD and NMD would enable easier routine analysis of local weather around the time of a hospitalisation or death, bearing in mind that the area and time of the injury event will not always match up with usual place of residence. This type of analysis would result in potential correlations between some weather conditions and injury rates.
Using linked data can overcome some but not all the technical limitations of the component data sets.
Creating new linked data assets requires time, cooperative effort between data providers, and high levels of governance if there are data privacy issues to manage.