Abstract: FR-PO108

Variation in Community-Based AKI Trends Using Administrative Codes versus Serum Creatinine Values Among >5 Million Adults Between 2004-2014

Session Information

Category: Acute Kidney Injury

  • 003 AKI: Clinical and Translational

Authors

  • Go, Alan S., Kaiser Permanente Northern California, Oakland, California, United States
  • Hsu, Chi-yuan, University of California San Francisco, San Francisco, California, United States
  • Tan, Thida C., Kaiser Permanente Northern California, Oakland, California, United States
  • Liu, Kathleen D., University of California San Francisco, San Francisco, California, United States
  • Zheng, Sijie, Kaiser Permanente Northern California, Oakland, California, United States
  • Yang, Jingrong, Kaiser Permanente Northern California, Oakland, California, United States
Background

We evaluated potential variation in community-based temporal trends in acute kidney injury (AKI) incidence using administrative codes vs. serum creatinine (SCr)-based changes.

Methods

In Kaiser Permanente Northern California, a large integrated healthcare delivery system, we identified all hospitalized AKI episodes between 2004-2014 using revised KDIGO criteria (>50% relative rise in SCr from baseline, >0.3 mg/dL SCr increase within 48 hours or receipt of acute dialysis) (KDIGO-AKI) vs. primary or secondary discharge ICD-9 diagnostic codes (DIAG-AKI). We examined age-sex-adjusted incidence and multivariable-adjusted incidence of AKI per year using each AKI definition.

Results

Among 5,253,185 adults, mean age was 48 years, 53% were women and 45% were minorities. Age-sex-adjusted incidence (per 100,000 person-years) of KDIGO-AKI rose from 587 in 2004 to 645 in 2006 but then decreased progressively to 471 by 2014. In contrast, age-sex-adjusted incidence of primary DIAG-AKI remained stable over time, while secondary DIAG-AKI consistently increased from 297 in 2004 to 484 in 2014. After adjustment for potential confounders using Poisson regression, compared with 2004, the relative incidences of KDIGO-AKI and primary DIAG-AKI peaked in 2006 but decreased through 2014; however, the relative incidence of secondary DIAG-AKI steadily increased throughout the study period (Figure).

Conclusion

These data extend prior studies which have reported suboptimal operating characteristics of administrative AKI codes, with increased sensitivity for detecting AKI in later calendar years. Importantly, however, estimates of population-based temporal trends in AKI that rely solely on administrative codes to define AKI are likely to be biased compared with using SCr-based definitions.

Funding

  • NIDDK Support