Infant product-related injuries: comparing specialised injury surveillance and routine emergency department data

Area of Science:

• Public Health
• Injury Prevention
• Data Science

Background:

• Routine emergency department (ED) data is a potentially rich source for injury surveillance.
• Identifying product-related injuries in infants from ED data requires effective data mining techniques.
• Comparing ED data patterns with specialized injury surveillance data is crucial for validation.

Purpose of the Study:

• To assess the feasibility of using basic text searches in routine ED data to detect infant product-related injuries.
• To compare injury patterns identified in routine ED data with those from specialized injury surveillance units.
• To evaluate the accuracy of text-search methods for injury surveillance.

Main Methods:

• Utilized data from the Emergency Department Information System (EDIS) and Queensland Injury Surveillance Unit (QISU) for injured infants (2009-2011).
• Developed and applied a basic text search to identify top infant products in QISU data.
• Calculated sensitivity, specificity, and positive predictive value, refining the search for EDIS data.
• Manually reviewed results for validity and conducted descriptive analysis to compare datasets.

Main Results:

• Basic text searches demonstrated high sensitivity and specificity for product identification.
• Most searches yielded a high positive predictive value, indicating reliable case identification.
• Injury patterns in EDIS data closely mirrored QISU data, including age-specific injury peaks, admission rates, and injury types.

Conclusions:

• Simple text searching of routine ED data effectively identifies cases of product-related infant injuries.
• Improved text mining capabilities and larger datasets will expand injury surveillance sources.
• Findings support enhanced data utilization for consumer product safety regulators and child safety advocates to target prevention efforts.