Reference Model | New Model | cfNRI_{events}^{a} | cfNRI_{nonevents}^{a} | cfNRI (95% CI)^{a} | P Value | IDI^{b} | P Value |
---|---|---|---|---|---|---|---|

Clinical model^{c} | Clinical model^{c} + uAnCR | 0.28 | 0.39 | 0.67 (0.26 to 1.09) | 0.001 | 0.06 | 0.09 |

Clinical model^{c} + uAnCR | Clinical model^{c} +uAnCR +uRenCR | 0.44 | 0.11 | 0.55 (0.14 to 0.96) | <0.01 | 0.01 | 0.38 |

cfNRI, category free net reclassification improvement; CI, confidence interval; IDI, integrated discrimination improvement; uAnCR, urinary angiotensinogen-to-creatinine ratio; uRenCR, urinary renin-to-creatinine ratio.

↵a cfNRI is a means of calculating the effect of adding a new variable to a predictive model on the overall accuracy of the model. cfNRI is the sum of cfNRI

_{events}and cfNRI_{nonevents}. cfNRI_{events}and cfNRI_{nonevents}are the proportion of patients who met the outcome (events) or those who did not, respectively, which are correctly reclassified by the new model minus the proportion of patients who are incorrectly reclassified. Correct reclassification is defined as a calculated risk of meeting the outcome that is higher for events and lower for nonevents compared with the reference model. If all events and nonevents were correctly reclassified, the cfNR_{Ievents}and cfNRI_{nonevents}would be +1, and the cfNRI would be 2.↵b IDI is a means of quantifying the effect of addition of a new variable to a predictive model on the magnitude of the change in the difference between the average calculated risk of patients who met the outcome compared with those who did not. The mean risk of the two groups is calculated using the reference model and the new model, and IDI is simply the difference between the discrimination slopes of the two models.

↵c Clinical model is a multivariate logistic regression model including the Cleveland Clinic score and the percentage increase in serum creatinine from baseline at the time of sample collection.