This retrospective, observational cohort study was conducted at National Taiwan University Hospital (NTUH), a 2500-bed tertiary medical center located in Taipei City (population density of approximately 10,000 people/km) with 110,000 annual emergency department (ED) visits [18]. The Institutional Review Board of the hospital approved the study (202203002RINB) and waived participant consent due to the nature of the study. Procedures were followed in accordance with the institutional ethical standards.

The primary dataset was sourced from the hospital medical records and included demographic information, past medical history, cardiac arrest events, postarrest management, laboratory examinations, and outcomes. This study adhered to the Strengthening the Reporting of Observational Studies in Epidemiology reporting guidelines [19].

Patients with cardiac arrest were categorized as either OHCA at a residential or public setting, including transfers from external hospitals or in-hospital cases after triage in the ED. An initial shockable rhythm was defined as the initial recorded rhythm being ventricular fibrillation or ventricular tachycardia. Repeated CPR was characterized as another arrest episode within 1 h after the initial ROSC. Cardiogenic arrest was recorded when the cause of arrest was attributed to ischemic heart disease, structural heart disease, heart failure, or arrhythmia without electrolyte imbalances. The determination of cardiac arrest causes was made by the responsible primary care physicians, who were blinded to the present study.

The lowest mean arterial pressure (MAP) during the initial 24 h following ROSC was categorized as $\ge$65 mmHg or 65 mmHg [20]. Patients with a pre-arrest Cerebral Performance Category (CPC) score of 1 or 2, no active bleeding or intracranial hemorrhage, and comatose consciousness after ROSC were eligible candidates for targeted temperature management (TTM). The TTM protocol at NTUH involved using cold saline and cooling devices with auto feedback to lower patients’ body temperatures to the targeted temperature of 33 °C within 4–6 h after ROSC. This targeted temperature was maintained for 24 h, followed by rewarming of patients at the rate of 0.25 °C per hour until 36 °C was achieved [21]. Temperature management to avoid fever was continued for another 24 h after rewarming. Instances of intra-aortic balloon pump and extracorporeal membrane oxygenation (ECMO) implantation were recorded if they occurred during the initial resuscitation. Emergent coronary angiogram and contrast-enhanced computed tomography scan were performed when indicated in patients within 24 h of ROSC [22]. Baseline laboratory test results at ROSC were documented. The levels of lactic acid (LA) in cardiac arrest survivors were categorized as 5 mmol/L, 5–10 mmol/L, and $>$10 mmol/L [8].

The highest amount of vasopressor use during the first 24 h of ROSC was denoted by VIS${}_{\text{max}}$, which reflects the degree of hypotension and severity of hemodynamic compromise during the early postarrest period. This was calculated as follows: dopamine dose (µg/kg/min) + dobutamine dose (µg/kg/min) + 100 $\times$ epinephrine dose (µg/kg/min) + 10 $\times$ milrinone dose (µg/kg/min) + 10,000 $\times$ vasopressin dose (unit/kg/min) + 100 $\times$ norepinephrine dose (µg/kg/min) using the maximum dosing rates of inotropic medications. The VIS${}_{\text{max}}$ was categorized into 3 groups: no VIS${}_{\text{max}}$, low VIS${}_{\text{max}}$ ($\le$30), and high VIS${}_{\text{max}}$ ($>$30) [23].

The primary outcome was the development of AKI during the early postarrest period. The criteria for diagnosing kidney injury were based on the Kidney Disease: Improving Global Outcomes (KDIGO) definition and AKI staging system characterized as an increase in serum creatinine by $\ge$0.3 mg/dL within 48 h [24]. AKI stage 3 was considered as severe AKI. Baseline creatinine levels were documented using records primarily from the previous 12 months or the admission creatinine, depending on the availability. Urine output volume was not used in this study. Secondary outcomes included the need for any modality of KRT during admission, in-hospital mortality, and poor neurological outcomes at hospital discharge. Poor neurological outcomes were defined as a CPC score of 3 (severe disability) to 5 (brain death).

Categorical variables are presented using frequencies (percentages), while continuous variables are presented as medians (interquartile ranges). Comparisons were conducted using Fisher’s exact or Pearson’s chi-square test for categorical variables and the Mann-Whitney U test for continuous variables. Statistical significance was set at a p-value less than 0.05. Multiple imputation was used for missing data. Multiple logistic regression was performed to assess the associations between the predictor variable and outcomes, adjusted for variables with statistical significance and clinical relevance. Odds ratios (ORs) with 95% confidence intervals (CIs) were reported as an estimate of effect size and variability. Survival curves between groups were illustrated and compared using the log-rank test. All statistical analyses were performed using SPSS for Windows, version 16.0 (SPSS Inc., Chicago, IL, USA).

This study included all adult patients with nontraumatic cardiac arrest at the NTUH ED who underwent successful resuscitation and did not have pre-arrest end-stage kidney disease (ESKD) from January 2017 to December 2019. After excluding individuals who did not survive beyond 48 h (n = 9), those with incomplete data (n = 7), and those transferred to another hospital during treatment (n = 1), a total of 411 patients were included for analysis. These patients were further grouped according to their baseline serum creatinine levels at ROSC [16] into creatinine $\le$1.5 mg/dL as the normal kidney function (NKF) group (n = 247) and creatinine $>$1.5 mg/dL as the impaired kidney function (IKF) group (n = 164) (Fig. 1).

Fig. 1.

Flowchart of patient enrollment. Poor neurological outcome is defined as a Cerebral Performance Category score of 3 to 5. ACKD, acute on chronic kidney disease; AKI, acute kidney injury; ESKD, end-stage kidney disease; ROSC, return-of-spontaneous-circulation.

Within the cohort of 411 patients, the median age was 67 years, with a male predominant majority (71.3%). Among them, 181 (44.0%) patients developed acute kidney injury within 48 h after ROSC, with more than half of these cases eventually requiring KRT (n = 108, 59.7%), resulting in a mortality rate of 71.8%. A comparison of the characteristics, cardiac arrest events, postcardiac arrest interventions, and examinations between patients with and without AKI is presented in Table 1. Among patients with AKI, a higher prevalence of preexisting kidney disease (8.7% vs. 16.6%, p = 0.022), anemia (35.4% vs. 54.7%, p 0.001), and malignancy (14.8% vs. 23.2%, p = 0.030) was observed. For cardiac arrest events, instances of repeated CPR (13.5% vs. 24.3%, p = 0.007) were more frequent among patients with AKI, whereas patients without AKI were more likely to exhibit an initial shockable rhythm (43.0% vs. 30.4%, p = 0.010) and receive less than 3 mg of epinephrine (83.9% vs. 66.9%, p 0.001). Patients who underwent intra-aortic balloon pump (p = 0.020) and ECMO placement (p 0.001) showed a stronger association with AKI, while patients with higher Glasgow Coma Scale scores at ROSC (p = 0.022) and subsequently more likely to receive TTM (p = 0.024) were less likely to develop AKI. Patients with AKI had lower hemoglobin concentrations (p 0.001), pH levels (p = 0.006), bicarbonate levels (p = 0.015), and O${}_{\text{2}}$/FiO${}_{\text{2}}$ ratio (p 0.001), along with higher levels of creatinine (p 0.001), sodium (p = 0.011), potassium (p = 0.016), Troponin-T (p 0.001), and LA (p 0.001).

In terms of hemodynamic status, patients with AKI had a significantly lower MAP (p 0.001) and higher VIS${}_{\text{max}}$ (p 0.001), with the majority of these patients receiving a high VIS${}_{\text{max}}$ $>$30. Approximately two-thirds of these patients necessitated KRT during hospitalization. Compared with patients without AKI, those with AKI stages 1–2 had a two-fold higher in-hospital mortality rate and an increased occurrence of poor neurological outcomes. Furthermore, patients with AKI stages 1–2 and AKI stage 3 were shown to have a lower 90-day survival rate than patients without AKI (Fig. 2A).

Fig. 2.

Survival curves between patients with and without AKI. (A) Comparison for overall patients. (B) Comparison according to baseline kidney function. AKI, acute kidney injury; NKF, normal kidney function; IKF, impaired kidney function.

There were 247 patients with NKF and 164 with IKF, of which 87 (35.2%) and 94 (57.3%) developed AKI, respectively. A comparison of the patient characteristics, cardiac arrest events, post-cardiac arrest interventions, and examinations between patients with and without AKI according to NKF and IKF are presented in Supplementary Tables 1,2, respectively.

Among patients with NKF, those without AKI had a higher proportion of patients with OHCA, initial shockable rhythm, and epinephrine use less than 3 mg. Conversely, patients with AKI had a higher frequency of repeated CPR. Furthermore, compared with patients without AKI, those with AKI received higher VIS${}_{\text{max}}$, had an increased incidence of ECMO placement, and exhibited elevated Troponin-T and LA levels, as well as lower MAP, hemoglobin concentration, pH value, and O${}_{\text{2}}$/FiO${}_{\text{2}}$ ratio. Among patients with IKF, the proportion of anemia was higher in patients with AKI. Similar patterns were noted in patients with NKF concerning epinephrine doses, ECMO placement, and laboratory results, including a lower MAP and higher VIS${}_{\text{max}}$ in patients with AKI. Regardless of baseline kidney function, patients with AKI in both the NKF and IKF groups showed higher mortality and poorer 90-day survival outcomes than those without AKI (Fig. 2B).

Compared with patients without AKI, those who developed AKI had significantly higher mortality rates (34.8% vs. 71.8%, adjusted OR [aOR] 5.40, 95% CI 3.36–8.69, p 0.001) and poor neurological outcomes (43.9% vs. 78.5%, aOR 5.70, 95% CI 3.45–9.43, p 0.001) upon discharge. Moreover, when stratified by AKI severity, patints with both AKI stages 1–2 and AKI stage 3 demonstrated higher mortality rates (aOR 3.85, 95% CI 2.00–7.42, p 0.001 and aOR 6.54, 95% CI 3.76–11.38, p 0.001, respectively) and poor neurological outcomes (aOR 3.90, 95% CI 1.96–7.76, p 0.001 and aOR 7.17, 95% CI 3.93–13.05, p 0.001, respectively), indicating a rising risk of adverse outcomes with increasing severity of kidney injury. These findings were consistently observed in both the NKF and IKF groups (Table 2).

Patients with a high VIS${}_{\text{max}}$ exhibited a nearly three-fold increased risk of developing AKI compared to those who did not receive any vasopressor support (aOR 2.96, 95% CI 1.61–5.45, p 0.001) (Table 3). When stratified by illness severity, a low VIS${}_{\text{max}}$ was associated with the development of AKI stages 1–2 (aOR 2.51, 95% CI 1.20–5.24, p 0.05), whereas a high VIS${}_{\text{max}}$ was linked to the occurrence of AKI stage 3 (aOR 2.46, 95% CI 1.28–4.75, p 0.05). Additionally, a low VIS${}_{\text{max}}$ was correlated with a reduced likelihood of requiring KRT (aOR 0.39, 95% CI 0.20–0.78, p 0.05), while a higher VIS${}_{\text{max}}$ was associated with higher rates of KRT (aOR 1.86, 95% CI 0.99–3.50, p = 0.056).

Regardless of baseline kidney function, a higher VIS${}_{\text{max}}$ was associated with a higher AKI risk (aOR 2.67, 95% CI 1.21–5.88, p 0.05 and aOR 5.23, 95% CI 1.92–14.25, p 0.05, respectively). Among patients with NKF, both low and high VIS${}_{\text{max}}$ increased the risk of AKI stage 1–2 (aOR 4.81, 95% CI 1.74–13.29, p 0.05 and aOR 3.20, 95% CI 1.11–9.20, p 0.05, respectively). On the other hand, among patients with IKF, a high VIS${}_{\text{max}}$ increased the risk of AKI stage 3 (aOR 5.18, 95% CI 2.01–13.35, p 0.05).

This study had several limitations. Firstly, due to its retrospective nature, selection bias was unavoidable, and unidentified confounding factors may be present. As a result of the study design, there is a lack of a strict protocol for the use of vasoactive and inotropic medications, which may be dependent on the attending physicians’ decisions at the time. Additionally, certain data may not be available for all patients; hence, not all possible factors that may contribute to AKI development were taken into account for analysis. Secondly, the definition of AKI in this study was based on serum creatinine alone, and not including urine output may not reflect the true classification of this population. Thirdly, our findings may have limited generalizability depending on the different hospital protocols in terms of postarrest care strategies, which may influence AKI development and overall outcomes. Lastly, only short-term mortality and neurological recovery at hospital discharge were assessed. It would be prudent for future well-designed studies to explore long-term outcomes after discharge.