Related Experiment Videos
A reliable method to retrieve accident & emergency data stored on a free-text basis
U N Premaratne1, G B Marks, E J Austin
1Department of Public Health Medicine, UMDS, St Thomas' Hospital, London, U.K.
Respiratory Medicine
|February 1, 1997
Summary
Searching free-text fields for "asth" can reliably identify asthma patients in emergency departments. However, sensitivity may decrease during asthma epidemics, requiring careful data interpretation.
Area of Science:
- Emergency Medicine
- Respiratory Medicine
- Health Informatics
Background:
- Accident & Emergency (A & E) data is crucial for asthma research.
- Final A & E diagnoses are often uncodated in UK hospital databases.
- Free-text presenting complaint fields are a potential data source.
Purpose of the Study:
- To validate the use of the text string "asth" in A & E free-text fields for identifying asthma patients.
- To assess the accuracy of this method for retrieving asthma-related attendances.
Main Methods:
- Retrieved A & E attendance data from two UK hospitals.
- Evaluated records containing "asth", "wheeze", "breathing problems", or GP referrals without diagnosis.
- Calculated sensitivity and specificity of the "asth" string for asthma diagnosis.
Main Results:
- The string "asth" demonstrated 80.3% sensitivity and 96.7% specificity for identifying asthma diagnoses.
- This method provides reliable data for asthma attendance tracking.
- Sensitivity was notably lower during a reported asthma epidemic.
Conclusions:
- Free-text fields in hospital databases can be effectively searched for reliable asthma data.
- The "asth" string is a valid tool for identifying asthma attendances.
- Reduced sensitivity during epidemics highlights the need for nuanced data analysis.