Abstract 3799: The Currently Accepted Definition of Worsening Renal Function is a Flawed Metric for the Study of Cardio-Renal Interactions
Introduction: A large body of literature supports the prognostic power of worsening renal function (WRF) in acute decompensated heart failure. An increase in creatinine >/=0.3 mg/dL, which has only been validated prognosticly, continues to be the standard endpoint in studies of cardio-renal interactions despite the recent change in focus from prognosis to more mechanistic investigation.
Hypothesis: Given the exponential relationship between creatinine and GFR, we hypothesized that the definition of WRF in current use is biased by baseline renal function. Additionally we hypothesized that the association between WRF and baseline renal insufficiency, reported in prior studies, is also an artifact of the exponential relationship between creatinine and GFR.
Methods: We reviewed consecutive admissions from 2004 to 2008 with a primary discharge diagnosis of congestive heart failure. The performance of the standard definition of WRF (increase in creatinine >/=0.3 mg/dL) was compared to a variable less dependent on baseline renal function (decrease in GFR of >/=20%). Chronic renal insufficiency (CRI) was defined as an admission GFR <60 ml/min.
Results: A total of 993 patients met eligibility criteria. WRF, by the standard definition, occurred in 31.5% of the population and a decrease in GFR >/=20% occurred in 32.7%. CRI predicted 30 day mortality (OR=2.7, p=0.018). Similar to previous reports, CRI was significantly more common in patients developing WRF by the standard definition (OR=1.9, p<0.001). On the contrary, a decrease in GFR >/=20% actually occurred less frequently in those with CRI (OR=0.51, p<0.001). An increase in GFR by 20% had a slightly better ability to predict death at 30 days (OR=2.3, p=0.026) compared to the standard definition (OR=2.1, p=0.048). Using Cox regression analysis, a decrease in GFR >/=20% added incrementally to CRI in predicting 30 day mortality (p=0.009), however, WRF by the standard definition did not independently improve the predictive power over CRI (p=0.11).
Conclusions: Results from this cohort suggest that the definition for WRF in current use is heavily biased by baseline renal function. Alternative definitions for WRF, such as a decrease in GFR >/=20%, should be included in future studies of cardio-renal interactions.