Understanding Validation in Electronic Health Records
Author Information
Author(s): Amanda Nicholson, Anne Rosemary Tate, Rob Koeling, Jackie A Cassell
Primary Institution: Brighton & Sussex Medical School
Hypothesis
What does validation of cases in electronic record databases mean?
Conclusion
Natural language processing can improve the validation of diagnoses in electronic health records by accessing free text data.
Supporting Evidence
- Most studies using electronic health records rely on coded data only.
- It is important to separate internal validation (does the code reflect the practitioner's diagnosis?) and external validation (is this diagnosis correct?).
- Existing validation studies often do not look for false negatives, which are diagnosed cases that have not been coded.
- Natural language processing may enable the more widespread use of free text contained in electronic records.
Takeaway
This study talks about how doctors use electronic health records and why it's important to check if the information is correct. New technology can help find mistakes in these records.
Potential Biases
Potential bias due to non-representative samples in validation studies.
Limitations
The study highlights the ambiguity in the term 'validation' and the challenges in identifying false negatives in electronic health records.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website