Kidney Week

Abstract: TH-PO882

Diagnostic Validity of Claims Data Diagnoses for CKD With Data From the Berlin Initiative Study (BIS)

Session Information

• CKD: Epidemiology, Risk Factors, Prevention - I
  November 03, 2022 | Location: Exhibit Hall, Orange County Convention Center‚ West Building
  Abstract Time: 10:00 AM - 12:00 PM

Category: CKD (Non-Dialysis)

• 2201 CKD (Non-Dialysis): Epidemiology‚ Risk Factors‚ and Prevention

Authors

• Bothe, Tim, Charite Universitatsmedizin Berlin, Berlin, Berlin, Germany

Tim Bothe,

• Schaeffner, Elke, Charite Universitatsmedizin Berlin, Berlin, Berlin, Germany

Elke Schaeffner,

• Mielke, Nina, Charite Universitatsmedizin Berlin, Berlin, Berlin, Germany

Nina Mielke,

• Schneider, Alice, Charite Universitatsmedizin Berlin, Berlin, Berlin, Germany

Alice Schneider,

• Douros, Antonios, McGill University, Montreal, Quebec, Canada

Antonios Douros,

• Barghouth, Muhammad, Charite Universitatsmedizin Berlin, Berlin, Berlin, Germany

Muhammad Barghouth,

• van der Giet, Markus, Charite Universitatsmedizin Berlin, Berlin, Berlin, Germany

Markus van der Giet,

• Kuhlmann, Martin K., Vivantes Klinikum im Friedrichshain, Berlin, Berlin, Germany

Martin K. Kuhlmann,

• Ebert, Natalie, Charite Universitatsmedizin Berlin, Berlin, Berlin, Germany

Natalie Ebert,

Background

Information on diagnostic validity of kidney disease within claims data is scarce. Thus, we aimed to estimate the validity of diagnostic codes of chronic kidney disease (CKD) in German claims data using estimated glomerular filtration rate (eGFR) as the gold standard.

Methods

We used data of the Berlin Initiative Study (BIS), a community-dwelling cohort of 2,069 persons aged 70 or older, for whom primary clinical data linked with claims data are available. After baseline assessment (2010–2011), 4 follow-up visits were conducted biennially over a total observation period of 8 years. Correspondence of overall and stage-specific CKD ICD-10 diagnoses in claims data with CKD stages assessed by the creatinine-based CKD-EPI equation (eGFR_{CKD–EPI–Crea}) was evaluated over time (cross-sectionally and longitudinally) and for each stage, respectively.

Results

The sensitivity of claims data diagnoses increases with CKD stages (Figure). For stages 3 and 4, stage-specific sensitivity increases from baseline to follow-up 4 (0.29/0.42 vs. 0.55/0.68 for stage 3/4) as well as longitudinally through ±1 year (Δ_% = +3.6 – +12.5 / +3.6 – +17.4 for stage 3/4). Still, false-negative rates are high, as many patients do not receive diagnoses corresponding with the respective CKD stage (e.g., 1-sensitivity_total = 0.91/0.83 for stage 1/2). Over all measures and stages, sensitivity is low (0.38 for overall and 0.22 for stage-specific diagnoses).

Conclusion

Correspondence of claims data diagnoses with eGFR measurements can be interpreted as acceptable for CKD stages 5 and 4, limited for stage 3, and low for stages 2 and 1. Stage-specific sensitivity is lower than overall. Over time (cross-sectionally as well as longitudinally), correspondence seems to increase. Over all measures and stages, diagnostic validity of CKD diagnoses in claims data should be interpreted as restricted.