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

Abstract

Background: Chronic kidney disease (CKD) is a common condition, characterized by high burden of comorbidities, mortality and costs. There is a need for developing and validating algorithm for the diagnosis of CKD based on administrative data. Methods: We validated our previously developed algorithm that used administrative data of the Lazio Region (central Italy) to define the presence of CKD on the basis of serum creatinine measurements performed between 2012 and 2015 at the Policlinico Gemelli Hospital. CKD and advanced CKD were defined according to eGFR (2, respectively). Sensitivity, specificity, positive and negative predictive values (PPV/NPV) were computed. Results: During the time span of the study, 30,493 adult participants residing in the Lazio Region had undergone at least 2 serum creatinine measurements separated by at least 3 months. CKD and advanced CKD were present in 11.1% and 2.0% of the study population, respectively. The performance of the algorithm in the identification of CKD was high, with a sensitivity of 51.0%, specificity of 96.5%, PPV of 64.5% and NPV of 94.0%. Using advanced CKD, sensitivity was 62.9% (95% CI 59.0, 66.8), specificity 98.1%, PPV 40.4% and NPV 99.3%. Conclusion: The algorithm based on administrative data has high specificity and adequate performance for more advanced CKD; it can be used to obtain estimates of prevalence of CKD and to perform epidemiological research.

Keywords

• chronic kidney disease
• administrative data
• serum creatinine measurements
• validation
• algorithm
• eGFR