Contrast-Induced Nephropathy in Patients Undergoing Intravenous Contrast-Enhanced Computed Tomography in Korea: A Multi-Institutional Study in 101487 Patients

An intravenous administration of iodinated contrast media during contrast-enhanced computed tomography (CECT) can lead to contrast-induced nephropathy (CIN) (1), which is known to increase the length of hospital stay, costs of medical care and the risk of permanent renal impairment (2, 3). Although serious clinical outcomes after CIN have been reported to be rare (4, 5, 6, 7), efforts to reduce CIN caused by CECT are worthwhile (5, 6).

Current guidelines for CIN recommend a pre-contrast assessment of patients to identify risk factors such as impaired renal function, diabetes mellitus (DM) and congestive heart failure (CHF) in order to prompt appropriate actions including measurement of the serum creatinine (SCr) and adequate hydration for high risk patients (4, 8). However, a gap between the evidence-based guidelines and the daily practice of radiologists for CIN prevention in CECT patients has been identified (9), which is likely due to the logistics needed to enable SCr measurement, intravenous hydration and serial follow-up of SCr in the setting of a radiology department (10). The first step should be to measure the size of the population who are at risk of CIN, which would depend on the prevalence of the major risk factors to develop a feasible protocol for the prevention of CIN by CECT. Accordingly, the prevalence data on risk factors of CIN in patients undergoing CECT are essential for designing a control strategy of CIN. Thus far, such data have been collected mostly at a single institution. For this reason, we conducted a multi-center observational study to evaluate the prevalence of known risk factors for CIN and their association with the actual incidence of CIN in patients undergoing CECT in Korea.

Of 140838 CECT examinations was the baseline eGFR greater than or equal to 60 mL/min/1.73 m² in 118424 (84.1%), 45-59 mL/min/1.73 m² in 16630 (11.8%), 30-44 mL/min/1.73 m² in 4421 (3.1%), and 15-29 mL/min/1.73 m² in 1363 (1.0%) (Table 1). The prevalence of risk factors showed an increasing tendency as the renal function worsened. However, the proportion of those in advanced age in the group with the worst renal function was smaller than that in the 3rd eGFR group. The mean age of the group with an eGFR of 15-29 mL/min/1.73 m² was 67.2 (± 13.0), which was smaller than that of the 3rd eGFR group (p < 0.01). The prevalence of risk factors for CIN was 25.1% (95% CI: 24.9-25.3) for the advanced age; 11.9% (11.7-12.1) for DM; 13.7% (13.5-13.8) for HT; 1.7% (1.6-1.8) for CHF; and 0.55% (0.51-0.59) for gout. NSAIDs were prescribed in 25.2% (25.0-25.4), diuretics in 11.6% (11.4-11.7) and ACEI/ARB were prescribed for the day of CECT in 11.8% (11.6-11.9) of cases.

Table 1
Prevalence of Risk Factors for Contrast Induced Nephropathy in Total of 140838 Contrast Enhanced Computed Tomography Examinations

• Click for larger image

Preventive measures were used in 40238 CECT examinations (28.6%), 37448 (26.6%) of which were NS with or without combination with other agents (Table 2). There was a prominent increase in preventive measures as eGFR dropped below 45 mL/min/1.73 m², which was mostly attributed to the increased use of other preventive measures than NS.

Table 2
Use of Preventive Measures

• Click for larger image

We identified 3103 cases of CIN, which accounted for 2.2% of all CECT examinations. All of the risk factors were more prevalent in the CIN group and the difference was statistically significant except for nephrotoxic medications at the univariable analysis (Table 3). In the multivariable analysis, however, decreased eGFR, DM, and CHF only remained statistically significant after the adjustment. The most prominent risk factor was the decreased eGFR with an adjusted relative risk (aRR) of 11.52 (10.43-12.73) for CIN. Gout was with an aRR of 0.78 (0.52-1.18) insignificant after the adjustment. The aRR for advanced age was 0.84 (0.76-0.92). Preventive measures were related to CIN events with an aRR of 2.83 (2.62-3.07). Using the propensity of using preventive measures, we matched the 27369 CECT examinations performed after preventive measures to the same number of CECT examinations performed without preventive measures within the predefined probability difference of using preventive measures. However, the odds ratio (OR) for developing CIN was 2.70 (95% CI: 2.39-3.05) for the group that used preventative measures compared with the cases that did not use preventative measures, similar to the results of the multivariable analysis.

Table 3
Association of Risk Factors of CIN with Actual Occurrence of CIN

• Click for larger image

When we calculated the incidence rate ratio of CIN for each risk factor group stratified by the use of preventive measures, we uniformly found smaller incidence rate ratios for each significant CIN risk factor in the cases where preventive measures were used than in those cases were no preventative measures were used. The aRR for DM were 1.19 (95% CI: 1.02-1.40), for CHF 1.73 (1.24-2.41) and for decreased eGFR 14.17 (12.21-16.44) in cases without use of preventative measures and 1.07 (0.91-1.26), 1.38 (1.09-1.76), and 9.96 (8.79-11.29), respectively, in cases with use of preventive measures. For the decreased eGFR, the 95% CI of the risk estimates was not overlapping, which suggested a significant decrease in CIN risk by preventive measures.

This study analyzed the prevalence of CIN risk factors and their relationship with the actual incidents of CIN using large data obtained from 16 tertiary hospitals in Korea. The incidence of CIN from all CECT procedures included to the analysis was 2.2% which was comparable to that found in previous studies. Balemans et al. (5) reported 23 CIN cases out of 944 procedures (2.43%), without the need for hemodialysis in any of the cases. Katzberg and Lamba (15) reviewed studies including 1075 patients with intravenous contrast media showing a 5.1% incidence of CIN without dialysis or mortality. El-Hajjar et al. (16) presented a very low incidence of CIN: only 4 CINs were reported out of 400 patients who underwent CT angiography with preventive measures. Although the prevalence of risk factors and the use of preventive measures varied among studies, the low incidence of CIN among CECT patients was consistent to previous studies.

A decreased renal function (eGFR < 45 mL/min/1.73 m²) was the most important independent risk factor for the development of CIN, although the aRR was larger than that of the previous studies. We found that DM, CHF and decreased renal function were associated with CIN; however, the age was associated with a decrease in the CIN. Many studies reported age as an independent risk factor of CIN (17, 18, 19). As we mentioned in the results section, the mean age of the group with an eGFR 15-29 mL/min/1.73 m² was 67.2 years (± 13.0), which was smaller than that in the eGFR 30-44 mL/min/1.73 m² group. Although a strong association between age and the deterioration of renal function has been identified (20), the mean all-cause mortality rate was the highest in elderly patients with low eGFR (21). The age distribution across eGFR groups in our data reflect the low chance of survival of elderly patients with advanced stage of chronic kidney disease. Furthermore, it was obvious that all the other risk factors increased as the patients became older. For those reasons, the incidence of CIN could be explained by the effect of risk factors other than advanced age in a multivariable model.

The adjusted OR was 2.83 (2.62-3.07) for preventive measures. However, it is well known that those treatments have been beneficial in prevention against CIN in patients who underwent a CECT. Like many other observational studies, our study was susceptible to unmeasured confounding. Regarding the results of the multivariable model, we did not attempt to conclude that preventive measures may increase the risk of CIN. This counterintuitive adjusted OR might have resulted from the fact that the preventive measures were likely performed in patients with a higher risk of CIN while some of the risk factors could not be accounted our statistical adjustment. The relevant data available in the medical records of the patients were likely not comprehensive enough to nullify the effect of all the confounders. Similarly, the results from the propensity score matching analysis also suggested that the input variables used to calculate the propensity score were not sufficient enough to properly evaluate the effect of the preventive measures.

Our data clearly showed that the risk factors of CIN were more prevalent in patients with a lower eGFR (Table 1) and that preventive measures became more prevalent as the eGFR decreased (Table 2). Therefore, it was obvious that patients with risk factors of CIN were more likely to have preventive measures referring to this data. We found this tendency consistently in every stratum of CIN risk factors and in the patient subgroup without any risk factors prior to CECT. Weisbord et al. (7) also reported a similar tendency that patients who were hospitalized or had lower baseline eGFR levels were more likely to undergo preventive measures and were seemingly more likely to derive benefit from such care. The hydration (OR 1.6, 95% CI: 0.6-4.1) and NAC (OR 2.4, 95% CI: 1.9-5.8) were associated with CIN when they defined CIN as a 25% increase in SCr.

Our results should be interpreted with caution due to several limitations. Although we pooled the electronic data of a very large population, this was not a random sample of the target population; thus we cannot say that it is generalizable to all CECT patients. Carstensen et al. (22) reported a prevalence of renal insufficiency that was very similar to this study. However, the study participants could have had more risk factors than ordinary CECT patients, as we selected patients from tertiary hospitals, which might still result in an over-estimation of the prevalence of risk factors. A large number of patients without SCr checked either before or after CECT were excluded, which could have biased the estimates. We found that the patients excluded from the analysis due to the lack of SCr measurements had significantly low rates of risk factors than those who were analyzed in the study (Appendix). Therefore, the excluded patients may have been less likely to develop CIN than the study participants. Although we could not directly assess the incidence of CIN in the patients who were excluded due to the lack of SCr measurements, the prevalence of risk factors and incidence of CIN might have been over-estimated in our study. Another important aspect of the study population was derived from the fact that we did not exclude patients who were on dialysis, but we excluded patients with an eGFR below 15. Considering that there might have been some patients who were on dialysis with an eGFR of at least 15, this study might have included a small number of dialysis patients. However, it is likely that most of the patients on peritoneal dialysis were excluded, so that only a small proportion of patients on hemodialysis were misclassified into the group of study participants.

The incidence of CIN was not directly compared to the population undergoing non-contrast studies, even though a control group would be essential to elucidate the causal relationship between CECT and CIN. In terms of incidence, the interval for SCr measurements was irregular as the data were collected retrospectively, which could have caused a detection bias to underestimate the CIN incidence. The timing and rate of hydration were not available; thus our operational definition of the use of NS as a preventive measure might have inflated the usage rate of the preventive measures.

Some studies have reported the setting of care as an independent risk factor (7) and have noted a higher incidence of CIN in inpatients. We could not identify any difference in the incidence between the inpatients and outpatients since we did not collect information on the setting of care. For the nephrotoxic medications, the medication use a few days before CECT could have had a similar toxic effect that caused an increase in SCr though we defined the exposure as the prescription for the day of CECT. On the other hand, the physician might have given instructions to skip the medication to the patient on the day of CECT, some of which might have not been precisely noted in the medical records. Finally, the use of any other not prescribed nephrotoxic medication such as herbal medicine, if existed, could not be included in our analysis.

In conclusion, we found that DM, CHF and a decreased eGFR were significantly associated with the occurrence of CIN after CECT in Korea and preventive measures were underutilized. Further efforts are needed to develop a system for the identification of susceptible patients and for the facilitation of preventive measures are needed.