Evaluation of administrative case definitions for chronic kidney disease in children

Area of Science:

• Pediatric Nephrology
• Public Health Surveillance
• Health Informatics

Background:

• Administrative data is increasingly used for chronic disease surveillance.
• The validity of administrative data for defining pediatric chronic kidney disease (CKD) cases is unknown.

Purpose of the Study:

• To evaluate the performance of case definitions for CKD in children using administrative data.
• To assess the accuracy of algorithms based on claims and pharmaceutical use compared to laboratory-based definitions.

Main Methods:

• Utilized population-based administrative data from the Manitoba Center for Health Policy.
• Developed and evaluated algorithms combining hospital claims, physician visits, and pharmaceutical use over 1-3 years in children (<18 years).
• Compared algorithms against a laboratory-based definition (eGFR <90 mL/min/1.73 m² and/or proteinuria).

Main Results:

• Algorithms demonstrated very low sensitivity (0.20-0.39) but excellent specificity (0.98-0.99).
• Positive predictive value was moderate (0.52-0.68), while negative predictive value was high (0.96-0.97).
• Sensitivity improved with more years of data; 3-year algorithms with physician claims and prescriptions showed highest sensitivity and ROC accuracy (0.60-0.68).

Conclusions:

• Administrative data algorithms for pediatric CKD exhibit unacceptably low sensitivity for screening purposes.
• High specificity correctly identifies children without CKD, but alternate data sources are necessary for effective population-based surveillance.
• Further research into alternative data sources is required for accurate pediatric CKD surveillance.