Identifying work related injuries: comparison of methods for interrogating text fields
2010

Identifying Work-Related Injuries Using Text Analysis

Sample size: 208291 publication 10 minutes Evidence: moderate

Author Information

Author(s): McKenzie Kirsten, Campbell Margaret A, Scott Deborah A, Discoll Tim R, Harrison James E, McClure Roderick J

Primary Institution: National Centre for Health Information Research and Training, Queensland University of Technology

Hypothesis

Can different text interrogation methods effectively identify work-related injuries in emergency department data?

Conclusion

The study supports the development of text searching methods to enhance injury surveillance from emergency department data.

Supporting Evidence

  • The keyword search detected 58% of work-related cases with high specificity.
  • The index search improved detection to 62% but had lower specificity.
  • Content analytic text mining detected 77% of cases with good specificity.
  • Overall, 10.3% of cases were coded as 'Working for an income'.
  • False positives were identified in all methods, highlighting the need for careful review.

Takeaway

This study looked at different ways to find work-related injuries in hospital records by analyzing the words used in injury descriptions.

Methodology

Three methods were used: keyword search, index search, and content analytic text mining to analyze injury description text fields.

Potential Biases

Potential biases may arise from the subjective nature of coding and the accuracy of the Activity codes used as a standard.

Limitations

The study relied on the quality of information recorded in text fields, which can be limited in busy emergency departments.

Participant Demographics

Data was collected from emergency departments in Queensland, Australia, covering a wide range of injury types and demographics.

Statistical Information

P-Value

p<0.01

Statistical Significance

p<0.01

Digital Object Identifier (DOI)

10.1186/1472-6947-10-19

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication