Monitoring caustic injuries from emergency department databases using automatic keyword recognition software.

Area of Science:

• Public Health
• Toxicology
• Medical Informatics

Background:

• Chemical products are implicated in 0.9% of home and leisure accidents in Italy, with 90% of toxic substance exposures occurring domestically.
• Hospital admissions due to chemical agents are double the average for domestic injuries, indicating a high risk of severe harm.
• Existing European data may underestimate the incidence of home-based caustic agent injuries.

Purpose of the Study:

• To monitor injuries caused by caustic agents in Italy.
• To evaluate the effectiveness of an automated free-text recognition system for identifying these injuries in Emergency Department (ED) medical databases.

Main Methods:

• Development of a Stata software program using an agent-specific keyword list to automatically identify caustic or corrosive injury cases.
• Deployment across ten hospitals in six Italian regions.
• Assessment of the program's sensitivity and specificity through quality control via manual reading of ED reports.

Main Results:

• The automated system identified 112 potential cases, with 99 confirmed as true positives (88.4% recognition rate).
• This represents 0.59% of all home injuries, a figure significantly higher (p < 0.0001) than the 0.9% reported by the European Union Injury Database.
• The system demonstrated excellent recognition capacity with acceptable excess sensitivity (11.6% false positives).

Conclusions:

• Automated free-text recognition is a valuable tool for identifying caustic agent injuries in ED databases.
• The actual incidence of injuries from caustic substances in Italian homes is significantly underestimated by current European codified data.
• Findings underscore the need for improved public health surveillance and prevention strategies for household chemical exposures.