Risk factors for post-contrast acute kidney injury in patients sequentially administered iodine- and gadolinium-based contrast media on the same visit to the emergency department: a retrospective study

Study design and population

This was a single-center, retrospective cohort study conducted by reviewing data extracted from electronic medical records of patients who visited the ED of Chungnam National University Hospital; approximately 56,000 patients visit the ED annually. This study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Institutional Review Board of Chungnam National University Hospital (No. 2021-11-024). The extracted data included only clinical data; no personally identifiable information was collected. Therefore, the need for informed consent was waived.

Patients who were admitted from the ED after either a single administration of ICM or sequential administration of both contrast media were included in this study. Among them, pediatric patients (aged <18 years) and patients who had an eGFR of <30 mL/min/1.73 m², as measured in the ED; had a medical history of kidney transplantation; or were missing data for creatinine or eGFR measured before the administration of contrast medium were excluded. The enrolled patients were divided into two groups: ICM alone (single administration of ICM in the ED) and ICM + GBCA (sequential administration of ICM and GBCA).

Data from July 2016 to July 2021 were extracted by an experienced research assistant who underwent rigorous training on our explicit protocol, including clearly defined variables and standardized coding methods. The systematic data abstraction was performed by abstractors who were blinded to the overall goals of the research to ensure unbiased chart reviews. Any conflicting or ambiguous charts were flagged by the abstractors for additional review by two board-certified emergency physicians and nephrologists. We recorded the following data from the index visit: baseline demographics (age, body mass index [BMI], sex), Charlson comorbidity index (CCI) calculated from preexisting illness, ED chief complaint, ED disposition, ED vital signs, laboratory results in ED, type and interval of each enhanced image examination in the cohort (both CT and MRI) performed in the ED on the same day, and mortality rates.

Interventions (process of radiologic examinations)

The enhanced CT and MRI procedures included a chest CT, abdominal CT, three-phase CT, brain CT angiography, head and neck CT, neck CT angiography, extremity CT angiography, head MRI with magnetic resonance angiography, spine MRI, and perfusion brain MRI. All CT and MRI scans were performed using a 64-channel system (Somatom Sensation 64; Siemens Healthineers) and a 3T scanner (Achieva 3T; Philips Healthcare), respectively. According to our institutional policy, preventive hydration with a fixed volume (500 mL) of normal saline was infused into patients at high risk of developing PC-AKI after the administration of contrast medium (serum creatinine level of >1.7 mg/dL or eGFR of <45 mL/min/1.73 m²) at the discretion of the attending physicians.

All administrations of contrast medium (Supplementary Table 1, available online) were performed according to institutional protocols (available online at https://www.ctisus.com/protocols). The ICM used in the ED is a low-osmolality and non-ionic contrast agent that was administered intravenously according to image examination-specific protocols at a volume of 80 to 120 mL. Similarly, GBCA was administered at 0.1 mL/kg according to examination-specific protocols.

Outcomes

The primary outcome of this study was the development of PC-AKI. Traditionally, PC-AKI has been defined as a significant increase in serum creatinine from baseline within 72 hours after the last administration of contrast medium [9]. Recently, however, eGFR has gained attention as a potentially better marker of PC-AKI risk [12,13] because it predicts the true GFR more accurately than serum creatinine [14]. Therefore, we added an eGFR-based criterion and defined PC-AKI in this study as an increase in serum creatinine of ≥25% or 0.5 mg/dL over the baseline value [9] or the reduction of eGFR by ≥25% of the baseline value within 72 hours after the last administration of contrast medium (ICM or GBCA) [15]. eGFR was estimated according to the guidelines of the Korean Society of Nephrology [16]. The secondary outcome was AKI recovery, defined as a return to the baseline serum creatinine level, within 7 days after the ED visit [17].

Statistical analysis

Categorical and continuous variables had a non-normal distribution in this study, so differences between the groups were analyzed using the chi-square test with continuity correction in 2 × 2 tables or Fisher exact test for categorical variables and the Mann-Whitney U test for continuous variables, with the results expressed as a frequency with the percentile and median values with interquartile ranges (IQRs), respectively.

We performed propensity score matching (PSM) between the ICM alone and ICM + GBCA groups to balance potential covariables. A binary logistic regression model was used to determine the propensity scores for the ICM + GBCA group using baseline characteristics and clinical status in the ED. For the PSM analysis, each patient in the ICM + GBCA group was matched to one patient in the ICM alone group to the nearest fifth decimal point using a nearest-neighbor algorithm. A caliper setting of 0.2 was used. Standardized differences (SDs) were used to confirm a balanced matching result. The matching result was considered balanced when the SD was <0.1. There was no overlapping of non-exposure cases in the final models. Many subjects received ICM alone; therefore, we also used many-to-one PSM (1:1, 2:1, and 3:1) to minimize the standard error between the groups. After PSM, multivariable logistic regression analyses were performed for each PSM cohort. Adjusted odds ratios (aORs) with 95% confidence intervals (CIs) for the exposure variable were calculated in each analysis. Backward selection was used to develop the final adjusted model. The goodness of fit of the final model was evaluated using the Hosmer-Lemeshow test. The results of the logistic regression analysis are expressed as aORs with 95% CIs.

A subgroup multivariable logistic regression analysis was performed to identify the independent risk factors for PC-AKI in the ICM + GBCA group. All variables with a p-value of <0.1 in the univariable analyses were included using the same multivariable logistic regression method just described. Potential multicollinearity was assessed using tolerance and the variance inflation factor (VIF) to verify that multicollinearity did not significantly influence the model’s coefficients. Multicollinearity between variables was defined as a tolerance of <0.1 or a VIF of >10 [18]. Receiver-operating characteristic (ROC) curves were constructed to determine the predictive value of factors independently associated with the development of PC-AKI, and the area under the ROC curve (AUROC) was obtained for single effective variables and their combination. The combination was divided into two steps: first, a probability value was obtained using a binary logistic regression analysis; second, an ROC curve analysis was performed using this probability value as a test variable. The analysis was performed using R software (version 4.1.0; R Foundation for Statistical Computing) and MedCalc version 15.2.2 (MedCalc Software). Results were considered statistically significant at a p-value of <0.05.

Study population

Of the 13,839 patients who received ICM for enhanced CT in the ED, 349 pediatric patients, 64 patients for whom the data for estimating renal function before the use of contrast medium were unavailable, 6,974 patients whose eGFR was <30 mL/min/1.73 m² at baseline, and 134 patients who had undergone kidney transplantation were excluded (Fig. 1). Of the remaining 6,318 patients, 6,179 (97.8%) and 139 (2.2%) were in the ICM alone and ICM + GBCA groups, respectively.

Figure 1.

Flow diagram of the patients included in this study.

CKD, chronic kidney disease; ED, emergency department; eGFR, estimated glomerular filtration rate; ESRD, end-stage renal disease; GBCA, gadolinium-based contrast agents; HD, hemodialysis; ICM, iodine-based contrast medium; PD, peritoneal dialysis.

The demographic and baseline characteristics of the total cohort in this study are provided in Table 1. The groups differed significantly in liver disease (335 [5.4%] vs. 18 [12.9%], p < 0.001) and ischemic heart disease (417 [6.7%] vs. 22 [15.8%], p < 0.001) as past medical history; however, the CCI did not differ between the groups (3 [IQR, 1–4] vs. 3 [IQR, 1–5], p = 0.14) (Table 1). Neurologic symptoms (675 [10.9%] vs. 47 [33.8%], p < 0.001) and cardiac arrest (54 [0.9%] vs. 7 [5.0%], p < 0.001) as the chief complaint during the ED visit were significantly higher in the ICM + GBCA group than in the ICM alone group (Table 1). After performing PSM to adjust the balance between the groups, the baseline characteristics and clinical status of the two groups were as displayed in Table 2, and the adjustment status after PSM that we estimated using the standard mean differences between the groups is shown in Supplementary Fig. 1 (available online). None of the baseline characteristics or clinical status in the ED differed significantly between the groups after PSM (Table 2).

Table 1.

Baseline demographics, characteristics, and outcomes of the study population

Table 2.

Comparison of baseline characteristics and clinical status between two groups in each propensity score matched cohorts

Incidence of post-contrast acute kidney injury and the renal recovery rate within 7 days after the emergency department visit

The ICM + GBCA group developed PC-AKI more frequently than the ICM alone group (676 [10.9%] vs. 49 [35.3%], p < 0.001) (Table 1), but the rate of renal recovery did not differ significantly between the groups (501 [74.1%] vs. 30 [78.9%], p = 0.57) (Table 1). The incidence of PC-AKI remained significantly higher in the ICM + GBCA group than the ICM alone group in all PSM cohorts (Table 2).

Risk factors for developing post-contrast acute kidney injury in all propensity-matched cohorts

The multivariable logistic regression analysis for each PSM cohort is provided in Table 3. Sequential administration of ICM and GBCA, compared with a single administration of ICM, was independently associated with the development of PC-AKI in all PSM cohorts (Table 3).

Table 3.

Multivariable logistic regression analyses for development of post-contrast acute kidney injury in each propensity score matched cohorts

Subgroup analyses in the iodine contrast medium + gadolinium-based contrast medium group to investigate the risk factors for developing post-contrast acute kidney injury

The demographic and baseline characteristics of subgroups of patients who received sequential administrations of ICM and GBCA are provided in Supplementary Table 2 (available online). The PC-AKI group showed significantly higher osmolality (303.0 mOsmol/kg [IQR, 295.0–316.5] vs. 291.0 mOsmol/kg [IQR, 283.0–305.1], p < 0.01), sodium (142.1 mEq/L [IQR, 138.9–146.3] vs. 138.8 mEq/L [IQR, 136.2–140.8], p < 0.01), chloride (106.2 mEq/L [IQR, 98.0–110.5] vs. 103.3 mEq/L [IQR, 99.0–108.1], p = 0.04), and glucose (152.0 mg/dL [IQR, 110.8–227.0] vs. 114.0 mg/dL [IQR, 92.0–154.5], p < 0.01) and lower platelet counts (180.0 × 103/μL [IQR, 130.8–247.5] vs. 226.0 × 103/μL [IQR, 167.5–281.5], p = 0.03) than the non-PC-AKI group. In addition, the time interval between the administrations of ICM and GBCA (<4 or ≥4 hours) in the PC-AKI group was significantly shorter than in the non-PC-AKI group (4.6 hours [IQR, 2.2–8.0] vs. 1.8 hours [IQR, 1.0–3.4], p < 0.01).

In our assessment for potential multicollinearity, sodium and glucose did not show significant collinearity with serum osmolality (sodium [tolerance, 0.818; VIF, 1.223] and glucose [tolerance, 0.919; VIF, 1.089]); however, they were excluded from the multivariable analysis because unrecognized collinearity was strongly expected between them and serum osmolality. In the multivariable analysis, the following variables were adjusted: serum osmolality, whether or not the time between the administration of the contrast media was >4 hours, platelet count, creatinine level, eGFR, and chloride level. The multivariable analysis revealed that osmolality (aOR, 1.05 [95% CI, 1.01–1.10]; p = 0.02) and eGFR (aOR, 0.93 [95% CI, 0.88–0.98]; p = 0.01) were independently associated with the development of PC-AKI, whereas the time interval between contrast administrations was not (Table 4).

Table 4.

Multivariable analysis for development of PC-AKI in the patients sequentially administered ICM and GBCA

Subgroup analysis to find the predictive performance of osmolality and estimated glomerular filtration rate for post-contrast acute kidney injury development

Fig. 2 shows the predictive performance of serum osmolality and eGFR when examined independently and in combination. The AUROC values for serum osmolality and eGFR were 0.701 (95% CI, 0.592–0.796) and 0.613 (95% CI, 0.503–0.716), respectively (Fig. 2). Their corresponding cut-off values, sensitivity, and specificity were as follows: 292 mOsmol/kg, 87.1% (95% CI, 70.2–96.4), and 50.0 (95% CI, 36.1–63.9), respectively, for osmolality; 78 mL/min/1.73 m², 58.3% (95% CI, 36.6–77.9), and 66.7% (95% CI, 53.7–78.0), respectively, for eGFR (Fig. 2). The AUROC values for the development of PC-AKI were numerically higher when the two variables (osmolality and eGFR) were used in combination than when they were used alone (AUROC, 0.714 [95% CI, 0.578–0.825]) (Fig. 2).

Figure 2.

Predictive performance of serum osmolality (A), eGFR (B), and their combination (C) on the development of PC-AKI in patients who received sequential administration of ICM and GBCA. (A)

AUROC, area under the receiver-operating characteristic curve; CI, confidence interval; eGFR, estimated glomerular filtration rate; GBCA, gadolinium-based contrast agents; ICM, iodine-based contrast media; NA, not applicable; PC-AKI, post-contrast acute kidney injury.