Predicting acute kidney injury in adults after cardiac surgery with a novel clinical metric: The cardiac diuretic-responsiveness index

Figures

• 

  Multivariate model using the cardiac diuretic responsiveness index predicts kidney injury.

• 

  Figure 1. Urine output in response to diuretic. Representative urine output for single patient. The red area was used to calculate responsiveness of diuretic (1 hour).

• 

  Figure 2. Urine production in 1 hour per weight per milligram of IV diuretic equivalents. A, Mean urine production per hour per kg per mg of IV furosemide equivalents. Error bars represent standard error and ∗∗∗∗ represents P value < .0001 from Wilcoxon rank-sum test. B, Histogram of population urine production per hour per kg per mg of IV furosemide. Red/blue histograms indicate AKI/non-AKI cohorts; dotted lines indicate respective group medians. IV, Intravenous; AKI, acute kidney injury; UOP, urine output production.

• 

  Figure 3. Identifying urine production per hour per kg per mg of IV furosemide equivalent threshold of patients who developed an AKI. A, Patients binned into various urine production per hour per kg per mg of IV furosemide equivalent levels plotted as a percentage of AKI development. The dotted red line represents apparent change in slope of AKI development with percentages of AKI per bin over the bar graphs. B, Segmental regression plot showing optimized threshold where slope of AKI development versus diuretic responsiveness found to be 0.075 represented by the dotted black line. Patients above the threshold are labeled responders and below are labeled nonresponders. Black dots represent percent AKI development in each bin. IV, Intravenous; AKI, acute kidney injury.

• 

  Figure 4. AKI development with clinical threshold. A, Mean percent of AKI development using 0.075 cardiac diuretic responsiveness index (CDRI) expressed as urine volume per weight per hour per milligram of IV diuretic equivalents. Error bars represent 95% CIs and ∗∗∗∗ represents P < .0001 from Wilcoxon rank-sum test. B, OR of AKI development when CDRI is <0.075 compared with those with ≥0.075. Error bars represent 95% CIs. C, Urine production per 1 hour expressed as AKI stage. ∗P value < .05, ∗∗P < .01, and ∗∗∗∗P < .0001. Error bars are standard error of the mean. D, Urine production in 6 hours expressed as AKI stage. ∗P < .05, ∗∗P < .01, and ∗∗∗∗P < .0001. Error bars are standard error of the mean. AKI, Acute kidney injury; IV, intravenous; OR, odds ratio.

• 

  Figure E1. Inclusion criteria for assessing diuretic-responsiveness for urine output production in 6 hours. CPB, Cardiopulmonary bypass; CKD, chronic kidney disease; IV, intravenous; AKI, acute kidney injury.

• 

  Figure E2. A, Mean total urine production in 1 hour of patients who developed an AKI versus those who did not. Error bars represent standard error and ∗∗∗∗ represents P < .0001 from Wilcoxon rank-sum test. B, Mean urine production per hour per kilogram of patients who developed an AKI versus those who did not. Error bars represent standard error and ∗∗∗∗ represents P < .0001 from Wilcoxon rank-sum test. AKI, Acute kidney injury; UOP, urine output production.

• 

  Figure E3. External validation of CDRI versus oliguria in the MIMIC-III database, illustrated by the AUC. CDRI, Cardiac diuretic responsiveness index; MIMIC, Medical Information Mart for Intensive Care; AUC, area under the receiver operating characteristic; KDIGO, Kidney Disease Improving Global Outcomes.

• 

  Figure E4. Forest plot of significant variables found on multivariable logistic regression analysis. Dots represent odds ratio and error bars are 95% CIs. ∗P < .05, ∗∗ is P < .01, and ∗∗∗P < .001. AKI, Acute kidney injury; CDRI, cardiac diuretic responsiveness index; CPB, cardiopulmonary bypass.

• 

  Figure E5. Multivariate analysis of AUC for cardiac diuretic responsiveness index (CDRI) versus positive control of < 0.5 mL/kg/h. AUC, Area under the receiver operating characteristic; UOP, Urine output production.