Validation of a Classification Algorithm for Chronic Kidney Disease Based on Health Information Systems

Area of Science:

• Nephrology
• Public Health
• Health Informatics

Background:

• Chronic kidney disease (CKD) presents a significant global health challenge with high comorbidity, mortality, and healthcare costs.
• Accurate identification of CKD using administrative data is crucial for epidemiological studies and resource allocation.
• Existing methods for CKD diagnosis often rely on clinical data, necessitating the development of administrative data-driven algorithms.

Purpose of the Study:

• To validate a previously developed algorithm for diagnosing chronic kidney disease (CKD) using administrative data.
• To assess the algorithm's performance in identifying both general CKD and advanced CKD stages.
• To determine the feasibility of using administrative data for large-scale CKD prevalence estimation and epidemiological research.

Main Methods:

• The study validated an algorithm utilizing administrative data from the Lazio Region, Italy.
• Serum creatinine measurements from 2012-2015 at Policlinico Gemelli Hospital were used for validation.
• Chronic kidney disease (CKD) and advanced CKD were defined based on estimated glomerular filtration rate (eGFR) thresholds (<60 and <30 mL/min/1.73 m2).
• Key performance metrics including sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated.

Main Results:

• The study included 30,493 adult participants with at least two serum creatinine measurements.
• The prevalence of CKD and advanced CKD in the study population was 11.1% and 2.0%, respectively.
• The algorithm demonstrated high specificity (96.5%) and adequate sensitivity (51.0%) for CKD detection.
• For advanced CKD, the algorithm showed high specificity (98.1%) and a high negative predictive value (99.3%).

Conclusions:

• The algorithm based on administrative data exhibits high specificity and acceptable performance for identifying advanced CKD.
• This validated algorithm can be effectively utilized for estimating CKD prevalence in large populations.
• The findings support the use of administrative data algorithms for robust epidemiological research in nephrology.