The Changing Definition of Contrast-Induced Nephropathy
The Changing Definition of Contrast-Induced Nephropathy
A total of 58,957 patients underwent PCI during the study period. After excluding patients already on renal replacement therapy (1,208) and those with missing Cr values (10,365), the study cohort included 47,480 patients. A total of 1,601 (3.4%) patients developed CINTraditional, whereas 4,308 (9.1%) patients developed CINNew. There were 2,707 patients who met the definition of CINIncremental; that is, they met the definition of CINNew but not CINTraditional. There were 55 patients who met the definition of CINTraditional but not CINNew. The mean preprocedural Cr in these patients was 3.6 ± 1.75 mg/dL and rose postprocedurally to 4.3 ± 1.94 mg/dL.
The baseline characteristics of patients who developed CINTraditional, CINNew, and CINIncremental are listed in Table 1. As compared to patients who met the definition of CINTraditional, patients who developed CINIncremental were more likely to be younger and have preserved renal function on presentation. They were less likely to have a history of diabetes and congestive heart failure or present with an acute myocardial infarction.
When we evaluated the occurrence of CINTraditional vs CINNew in patients categorized by baseline renal function, CINNew was more likely to be seen in patients with preserved renal function, whereas CINTraditional was more likely to occur in patients with abnormal renal function (Figure 1).
(Enlarge Image)
Figure 1.
Distribution of baseline Cr clearance among patients who developed CINTraditional and CINNew.
In our study group, 177 (0.37%) patients experienced NRD and 627 (1.3%) experienced in-hospital death. Compared with patients who developed CINNew, patients developing CINTraditional were more likely to require dialysis (9.81% vs 3.64%) and experience in-hospital death (16.74% vs 7.29%). Patients with CINIncremental had a low mortality rate (1.7%), and none developed a need for dialysis (Figure 2).
(Enlarge Image)
Figure 2.
Frequencies of in-hospital death and NRD among patients meeting the definitions of CINTraditional, CINNew, and CINIncremental.
The area under the receiver operating curve was highest for absolute change in Cr, followed by relative change in Cr and change in Cr clearance for both NRD (0.93 vs 0.85 vs 0.76, respectively) and in-hospital death (0.80 vs 0.78 vs 0.76, respectively). Because absolute change in serum Cr had higher discrimination for identifying patients at risk for dialysis, optimization of sensitivity and specificity was sought using the Youden index (j). This suggested that a rise in Cr between 0.35 and 0.45 had similar accuracy, but moving the definition from 0.5 to 0.4 resulted in a loss of specificity and a gain in sensitivity.
We then validated the traditional definition, the new definition, and a proposed CIN>0.4 (rise in serum Cr >0.4 mg/dL) based on the Youden index in a cohort of 23,804 patients who underwent PCI in 2009 and did not meet any of our exclusion criteria, already on renal replacement therapy (n = 559), or with missing Cr values (n = 4,073). Thus, this study cohort included 23,804 patients. A switch from CINTraditional to CIN>0.4 would increase the incidence of CIN in our population from 3.18% to 4.57% without any increase in the sensitivity for detecting NRD (0.93 vs 0.93) and with a trivial increase in specificity (0.97 vs 0.99). A slight increase in the sensitivity for predicting death would, however, be observed (0.51 vs 0.57), with a similar small change in specificity (0.97 vs 0.99) (Table 2).
Results
A total of 58,957 patients underwent PCI during the study period. After excluding patients already on renal replacement therapy (1,208) and those with missing Cr values (10,365), the study cohort included 47,480 patients. A total of 1,601 (3.4%) patients developed CINTraditional, whereas 4,308 (9.1%) patients developed CINNew. There were 2,707 patients who met the definition of CINIncremental; that is, they met the definition of CINNew but not CINTraditional. There were 55 patients who met the definition of CINTraditional but not CINNew. The mean preprocedural Cr in these patients was 3.6 ± 1.75 mg/dL and rose postprocedurally to 4.3 ± 1.94 mg/dL.
The baseline characteristics of patients who developed CINTraditional, CINNew, and CINIncremental are listed in Table 1. As compared to patients who met the definition of CINTraditional, patients who developed CINIncremental were more likely to be younger and have preserved renal function on presentation. They were less likely to have a history of diabetes and congestive heart failure or present with an acute myocardial infarction.
When we evaluated the occurrence of CINTraditional vs CINNew in patients categorized by baseline renal function, CINNew was more likely to be seen in patients with preserved renal function, whereas CINTraditional was more likely to occur in patients with abnormal renal function (Figure 1).
(Enlarge Image)
Figure 1.
Distribution of baseline Cr clearance among patients who developed CINTraditional and CINNew.
In our study group, 177 (0.37%) patients experienced NRD and 627 (1.3%) experienced in-hospital death. Compared with patients who developed CINNew, patients developing CINTraditional were more likely to require dialysis (9.81% vs 3.64%) and experience in-hospital death (16.74% vs 7.29%). Patients with CINIncremental had a low mortality rate (1.7%), and none developed a need for dialysis (Figure 2).
(Enlarge Image)
Figure 2.
Frequencies of in-hospital death and NRD among patients meeting the definitions of CINTraditional, CINNew, and CINIncremental.
The area under the receiver operating curve was highest for absolute change in Cr, followed by relative change in Cr and change in Cr clearance for both NRD (0.93 vs 0.85 vs 0.76, respectively) and in-hospital death (0.80 vs 0.78 vs 0.76, respectively). Because absolute change in serum Cr had higher discrimination for identifying patients at risk for dialysis, optimization of sensitivity and specificity was sought using the Youden index (j). This suggested that a rise in Cr between 0.35 and 0.45 had similar accuracy, but moving the definition from 0.5 to 0.4 resulted in a loss of specificity and a gain in sensitivity.
We then validated the traditional definition, the new definition, and a proposed CIN>0.4 (rise in serum Cr >0.4 mg/dL) based on the Youden index in a cohort of 23,804 patients who underwent PCI in 2009 and did not meet any of our exclusion criteria, already on renal replacement therapy (n = 559), or with missing Cr values (n = 4,073). Thus, this study cohort included 23,804 patients. A switch from CINTraditional to CIN>0.4 would increase the incidence of CIN in our population from 3.18% to 4.57% without any increase in the sensitivity for detecting NRD (0.93 vs 0.93) and with a trivial increase in specificity (0.97 vs 0.99). A slight increase in the sensitivity for predicting death would, however, be observed (0.51 vs 0.57), with a similar small change in specificity (0.97 vs 0.99) (Table 2).
Source...