#5673 ESTIMATING AND VALIDATING MEASURES OF CKD PREVALENCE AND INCIDENCE USING OBSERVATIONAL AND CLAIMS DATA IN PERSONS AGED 70+

Abstract Background and Aims Estimating prevalence incidence of chronic kidney disease (CKD) based on claims data often underlies bias limitations due to uncertainty diagnostic validity. We investigated CKD over time using from a community-dwelling cohort individuals aged 70+ linked individual data. assessed the validity claims-based compared with eGFR-based definition. Method in two sources: 1) eGFR values participants (n = 2,069) Berlin Initiative Study (BIS), including five biennial study visits 2009-2019, 2) BIS matched person-level. Using eGFR, we defined as <60 ml/min/1.73m² calculated creatinine-based CKD-EPI (2009) equation. For each visit, prevalent cases were considered incident if was ≥60 respective previous visit. In data, in- outpatient ICD-10 diagnoses (N18.3, N18.4, N18.5, N18.8, N18.9, N19) year preceding and/or following visit date used determine prevalence. Incidence without diagnosis first diagnosis. The denominator for all persons at risk, i.e., who not still under observation. reference. Analyses stratified by sex. Results Between 2009 2019, increased 0.42 0.51 0.30 0.43 varied between 0.16–0.22 0.07–0.13 There no clear trend any source. Diagnostic detection is displayed Table 1. Sensitivity while other indicators remained stable time. Detection showed lower sensitivity (0.18–0.30 vs. 0.58–0.70) PPVs (0.40–0.60 0.79–0.83) prevalence, respectively. Prevalence slightly higher males females (Figure 1). Conclusion or definition found an increase 2019. This might be introduction KDIGO guidelines 2012. During same time, rates stable. Comparing definition, 0.58-0.70 0.18-0.30 incidence, indicating that there risk underdetection when solely particular cases.