Using Administrative Data to Identify COPD Patients
Author Information
Author(s): Colin R Cooke, Min J Joo, Stephen M Anderson, Todd A Lee, Edmunds M Udris, Eric Johnson, David H Au
Primary Institution: University of Michigan
Hypothesis
Can administrative data accurately identify patients with chronic obstructive pulmonary disease (COPD) when spirometry is the gold standard?
Conclusion
Using multiple diagnostic codes in combination with pharmacy data improves the ability to accurately identify patients with COPD.
Supporting Evidence
- 47.7% of patients had fixed airflow obstruction.
- The best performing model achieved an AUC of 0.79.
- Adding pharmacy data improved model performance.
- Outpatient ICD-9 codes provided better discrimination than inpatient codes.
Takeaway
Doctors can use patient records and prescriptions to find out who has COPD, which helps them treat the right people better.
Methodology
Sequential logistic regression models were constructed using data from patients with postbronchodilator spirometry at two Veterans Affairs medical centers.
Potential Biases
Ascertainment bias may be present due to the inability to assess visits to non-VA facilities.
Limitations
The study did not externally validate the model and was limited to a specific patient population (mostly older white men).
Participant Demographics
The cohort consisted mostly of older white men from the Veterans Affairs system.
Statistical Information
P-Value
<0.001
Confidence Interval
95% CI 0.74-0.76
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website