Abstract: TH-PO537
Limitations of ICD Codes in Detection, Staging, and Assessing Progression of CKD
Session Information
- CKD: Health Services, Disparities, Prevention
November 02, 2017 | Location: Hall H, Morial Convention Center
Abstract Time: 10:00 AM - 10:00 AM
Category: Chronic Kidney Disease (Non-Dialysis)
- 307 CKD: Health Services, Disparities, Prevention
Authors
- Jalal, Kabir, University at Buffalo, Hamburg, New York, United States
- Anand, Edwin J., None, Getzville, New York, United States
- Venuto, Rocco C., Erie County Medical Center, Buffalo, New York, United States
- Arora, Pradeep, VAMC, Buffalo, New York, United States
- Eberle, Joseph A., Intelligent Care Management, East Amherst, New York, United States
Background
The International Classification of Diseases (ICD) coding system is the industry standard tool for disease classification and epidemiology purposes. However, ICD codes in practice unreliably reflect the true diagnosis for a given disease/patient. This study seeks to quantify that inaccuracy among patients with Chronic Kidney disease (CKD).
Methods
Using insurance data consisting of 222,664 insured individuals with serum creatinine measurements over seven years, diagnoses based on a set of CKD-related ICD codes were compared against gold standard Kidney Disease Outcomes Quality Initiative (KDOQI) clinical guidelines to evaluate accuracy of ICD codes to detect CKD-positive patients. Patient serum creatinine levels were used to estimate progression of disease course using a longitudinal mixed model to identify advanced progressors, or those patients with loss of eGFR (estimated glomerular filtration rate) greater than 1 ml/year, and to assess accuracy of ICD codes in detecting advanced progressors.
Results
ICD codes correctly identified only 10,101 of 33,159 individuals as CKD, for a sensitivity of 30.46% with positive predictive value (PPV) of 65.05%; codes correctly identified 184,078 individuals as CKD-negative, for a specificity of 97.14% with negative predictive value (NPV) of 88.87%. In identifying rapid progressors, ICD codes achieved a sensitivity of 11.95% with PPV of 8.46%, and a specificity of 94.73% with NPV of 96.35%.
Conclusion
The use of ICD codes alone is insufficient in identifying patients with CKD. This study is the first to attempt the use of ICD codes in identifying rapidly progressing patients, revealing poor coding performance when compared to gold standard KDOQI guidelines. Use of ICD codes to identify CKD patients, assess disease severity, or to evaluate disease progression for either clinical or epidemiological purposes is not recommended.
Progression Performance Measures
| Measure | Mean | Lower 95% | Upper 95% |
| Sensitvity | 11.95 | 8.020 | 16.90 |
| Specificity | 94.73 | 94.11 | 95.30 |
| Positive Predictive Value | 8.460 | 5.410 | 11.52 |
| Negative Predictive Value | 96.35 | 95.85 | 96.84 |
Funding
- Other U.S. Government Support