Timing and Degree of Acute Kidney Injury in Patients Requiring Non-Surgical Intensive Care

Abstract

Background: The degree and timing of acute kidney injury (AKI) on admission and during hospitalization in patients requiring non-surgical intensive care remain unclear.

Methods and Results: In this study, 3,758 patients requiring intensive care were analyzed retrospectively. AKI was defined based on the ratio of serum creatinine concentrations recorded at each time point (i.e., on admission and during the first 5 days in the intensive care unit and during hospitalization) to those measured at baseline. Patients were grouped by combining AKI severity (RIFLE class) and timing (i.e., from admission to 5 days [A-5D]; from 5 days to hospital discharge [5D-HD]) as follows: No-AKI; New-AKI (no AKI to Class R [risk; ≥1.5-fold increase in serum creatinine], I [injury; ≥2.0-fold increase in serum creatinine], and F [failure; ≥3.0-fold increase in serum creatinine or receiving dialysis during hospitalization]); Stable-AKI (Class R to R; Class I to I); and Worsening-AKI (Class R to I or F; Class I to F). Multivariate logistic regression analysis indicated that 730-day mortality was independently associated with Class R, I, and F on admission; Class I and F during the 5D-H period; and New-AKI and Worsening-AKI during A-5D and 5D-HD.

Conclusions: AKI on admission, even Class R, was associated with a poor prognosis. An increase in RIFLE class during hospitalization was identified as an important factor for poor prognosis in patients requiring intensive care.

The concept of acute kidney injury (AKI) was introduced in the past 2 decades. Complex and multifactorial mechanisms underlie the renal dysfunction in patients who require intensive care, including both hemodynamic (renal arterial hypoperfusion, renal venous congestion) and non-hemodynamic (inflammatory mediators [e.g., infection, tissue damage], iatrogenic damage [e.g., contrast media, nephrotoxic medication], elevated intra-abdominal pressure) factors.^1,2 These mechanisms differ between patients and have not been completely elucidated. Therefore, an accurate understanding of this area is essential to improve intensive care.

In 2004, the Acute Dialysis Quality Intensive (ADRI) group introduced the Risk, Injury, Failure, Loss, and End stage (RIFLE) criteria for the diagnosis of AKI.³ In 2007, the Acute Kidney Injury Network (AKIN) criteria were proposed.⁴ In 2012, guidelines for the management and treatment of AKI were finally established based on evidence obtained from several fields.⁵ The definition of baseline creatinine is particularly important in these criteria. Serum creatinine concentrations recorded on admission were traditionally used as the baseline creatinine value for AKI.⁶ Therefore, previous definitions, such as the RIFLE and AKIN criteria, cannot be used to define AKI on admission in many patients. To this end, obtaining data on prehospital kidney function has been recommended whenever possible.⁷ A retrospective study also suggested an alternative approach of using the estimated glomerular filtration rate (eGFR) of 75 mL/min/1.73 m² and the lowest inpatient serum creatinine concentration in patients without prehospital data.⁷ We previously reported using this baseline methodology among patients with acute heart failure (AHF) admitted to the intensive care unit (ICU) and found that AKI was present in 33.2% of patients upon admission, increasing to 73.4% of patients during hospitalization.⁸ The definition used here can define AKI on admission in all patients, and is the rationale for the analyses in the present study.

The clinical significance of AKI upon admission and its timing after admission in patients who require no surgical intensive care has not been investigated thus far. The aim of the present study was to determine the clinical significance and prognostic impact of the degree and timing of AKI on admission and/or during hospitalization in patients requiring non-surgical intensive care.

Methods

Patients

We retrospectively screened 4,172 patients admitted to the non-surgical ICU of Nippon Medical School Chiba Hokusoh Hospital (Inzai, Chiba, Japan) from the emergency room and general wards between May 2011 and October 2020. Patients who were readmitted to the non-surgical ICU during the same hospitalization (n=218) and those who underwent hemodialysis prior to admission (n=196) were excluded from the study. Ultimately, 3,758 patients were enrolled in the present study, admitted to the non-surgical ICU from the emergency room (n=3,307; 88.0%) and general wards (n=451; 12.0%).

The enrolled patients were admitted to the non-surgical ICU and treated by a cardiologist in a “closed ICU”. The 2 ICU departments (i.e., surgical and non-surgical) at Nippon Medical School Chiba Hokusoh Hospital are “closed ICUs”, and all physicians practicing in the non-surgical “closed ICU” are cardiologists. Thus, patients who were admitted to the surgical ICU (e.g., for trauma, burns, drowning, and cerebrovascular disease) were excluded from the present study.

Patients admitted to the non-surgical ICU with the following diseases were included in this study: acute coronary syndrome (ACS); AHF; acute aortic disease (acute aortic dissection and acute aneurysmal rupture); pulmonary embolism; arrhythmia (tachycardia [arterial fibrillation/tachycardia/flutter and ventricular fibrillation/tachycardia including cardiac arrest] and bradycardia [sick sinus syndrome and complete atrioventricular block]); sepsis; pericarditis; coronary spasm angina; Takotsubo cardiomyopathy; respiratory emergency disease (exacerbation of chronic obstructive pulmonary disease, bronchial asthma, institutional pneumonia); neurogenic emergency disease; gastrointestinal disease; allergic disease; or the presence of severe symptoms that required a differential diagnosis.

Evaluation of AKI

Because urine output could not be precisely measured in the general ward and some patients had received diuretics, which influence urine output, AKI was investigated solely based on the creatinine criteria of the RIFLE classification.³ Furthermore, although the guideline recommended that AKI should be evaluated vs. baseline creatinine within 7 days,⁵ we used the original criteria to investigate the timing of AKI.

AKI on admission was defined based on the ratio of the serum creatinine concentrations recorded on admission to those measured at baseline. AKI during the first 5 days and entire period of hospitalization was also defined based on the ratio of the maximum serum creatinine concentrations recorded at those time points to those measured at baseline. Patients were classified into the following groups: no AKI; Class R (risk); Class I (injury); and Class F (failure) AKI. This classification was based on <1.5-, ≥1.5-, ≥2.0-, and ≥3.0-fold increases, respectively, in serum creatinine concentrations from baseline. Patients who received continuous renal replacement therapy within 24 h of admission, during 5 days in the ICU, and at hospitalization were assigned to Class F on admission, during 5 days and at hospitalization, respectively. In patients with chronic kidney disease (CKD), defined as a syndrome characterized by a >3-month history of low eGFR (<60 mL/min/1.73 m²),⁹ the baseline creatinine concentration was defined as the lowest value recorded during admission. In patients without CKD, the lower of either the lowest creatinine concentration during hospitalization or the Modification of Diet in Renal Disease (MDRD) creatinine concentration was used as the baseline creatinine concentration. The MDRD creatinine concentrations were calculated using the MDRD equation, as recommended by the Acute Dialysis Quality Initiative as follows:

eGFR (mL/min/1.73 m²) = 194 × Age^−0.287 × Creatinine^−1.094 (×0.739 for women)

This calculation was performed assuming an eGFR of 75 mL/min/1.73 m².¹⁰

As noted above, CKD was diagnosed based on a >3-month history of low eGFR before admission, which used the creatinine levels detected within 1 year prior to admission. Furthermore, among patients in whom creatinine concentrations had not been measured during that period, those who had been diagnosed with CKD in the past or at another institution were considered to have CKD. Patients without medical records at Chiba Hokusoh Hospital for 3 months before admission were diagnosed with CKD using data obtained from other institutions to which they had been admitted. Kidney damage, identified based on abnormal findings in the urine and imaging analyses, was used to diagnose CKD in some patients. In the present study, 945 of 3,758 patients (25.1%) were diagnosed with CKD.

AKI Group Comparison and Prognosis

The following parameters were compared among the 4 groups based on the RIFLE classification on admission: age, sex, etiology of HF, medical history (diabetes, hypertension, dyslipidemia, hyperuricemia), vital signs and status (systolic blood pressure [SBP], diastolic blood pressure, heart rate, respiratory rate, body temperature, body mass index, left ventricular ejection fraction [LVEF] upon admission), arterial blood gas (pH, PCO₂, PO₂, HCO₃⁻, SaO₂, and lactate), laboratory data (white blood cell count, hemoglobin, blood urea nitrogen [BUN], creatinine, sodium, potassium, blood glucose, C-reactive protein [CRP], B-type natriuretic peptide [BNP]), and mechanical support during the ICU stay (non-invasive positive pressure ventilation, endotracheal intubation, intra-aortic balloon pumping, percutaneous cardiopulmonary support, continuous hemodiafiltration). The Acute Physiology and Chronic Health Evaluation (APACHE II) score¹¹ was also compared between these groups.

The short-term prognosis was evaluated in the form of the length of ICU stay, length of total hospitalization, and in-hospital mortality rate. The long-term prognosis was evaluated in the form of all-cause death within 2 years. Patients were clinically followed up at a routine outpatient clinic. In patients followed up at other institutions, prognosis was determined by telephone contact. The prognostic value for 730-day mortality was evaluated using the Cox proportional hazards regression model and Kaplan-Meier curve. Age, etiology (ACS yes/no), sodium, hemoglobin, CRP, mean blood pressure, and pulse rate were retrieved on admission for all 3,758 cases. These variables were included in the multivariate logistic regression model. Continuous variables were evaluated by classifying patients into 2 groups using the median as the cut-off value. The analysis was divided into the 2 separate parts of RIFLE classification on admission and comparison of the timing of AKI. Survival and event-free rates were analyzed using Kaplan-Meier curves according to the RIFLE classification on admission and the timing of AKI.

Statistical Analyses

All data were analyzed using SPSS 22.0 J (SPSS Japan Institute, Tokyo, Japan). All numerical data are expressed as the median with interquartile range (IQR). The Mann-Whitney U and Kruskal-Wallis tests were used to compare the No-AKI and AKI groups and for comparisons between the No-AKI, Class R, Class I, and Class F groups, respectively. Comparisons of all proportions were performed using Chi-squared tests. P<0.05 was considered statistically significant.

The prognostic value of RIFLE criteria on admission and the timing of AKI was assessed using a Cox proportional hazards regression model by determining the hazard ratio (HR) for 730-day mortality. This model was developed by simultaneous forced entry. The cumulative survival rates in each of the 4 groups (RIFLE criteria on admission, timing of AKI from admission to 5 days after admission, and timing of AKI from 5 days after admission to the end of hospitalization) were analyzed using Kaplan-Meier curves. The log-rank test was used to calculate the statistical significance of differences.

Ethical Considerations

The study protocol was reviewed and approved by the Research Ethics Committee of Nippon Medical School Chiba Hokusoh Hospital. Due to the retrospective design of this study, the requirement for written informed consent was waived.

Results

Patient Characteristics and Prognosis Associated With RIFLE Criteria at Each Timing Evaluated

The ICU patient cohort consisted of 2,689 (71.6%) male and 1,069 female patients (median age 70 years). Of these patients, 1,674 (44.5%) had ACS, 853 (22.7%) had AHF, 170 (4.5%) had acute aortic disease, 268 (7.1%) had arrhythmia, 292 (7.8%) had infectious diseases, 126 (3.4%) had pulmonary embolism, and 375 (10.0%) had other diseases, including coronary spasms (n=47; 1.3%), Takotsubo cardiomyopathy (n=58; 1.5%) and other diseases requiring intensive care diseases (n=270; 7.2%; Table 1).

Table 1. Association Between Patient Characteristics and Degree of AKI on Admission

	All (n=3,758)	RIFLE on admission				P valueA	P valueB
		No AKI (n=2,639)	Class R (n=500)	Class I (n=187)	Class F (n=432)
Age (years)	70 [61–78]	69 [60–78]	72 [64–76]	73 [62–79]	72 [64–79]	<0.001	<0.001
Male sex	2,689 (71.6)	1,940 (73.5)	328 (65.6)	116 (62.0)	305 (70.6)	<0.001	<0.001
Etiology
ACS (yes)	1,674 (44.5)	1,368 (51.8)	171 (34.2)	56 (29.9)	79 (18.3)	<0.001	<0.001
AHF (yes)	853 (22.7)	612 (23.2)	135 (27.0)	35 (18.7)	71 (16.4)	<0.001	0.287
AAD (yes)	170 (4.5)	135 (5.1)	19 (3.8)	3 (1.6)	13 (3.0)	0.033	0.004
PTE (yes)	126 (3.4)	87 (3.3)	23 (4.6)	10 (5.3)	0 (0.0)	0.020	0.418
Arrhythmia (yes)	268 (7.1)	143 (5.4)	49 (9.8)	19 (10.2)	57 (13.2)	<0.001	<0.001
Coronary spasms (yes)	47 (1.3)	44 (1.7)	3 (0.6)	0 (0.0)	0 (0.0)	0.004	<0.001
Takotsubo cardiomyopathy (yes)	58 (1.5)	36 (1.4)	9 (1.8)	9 (4.8)	4 (0.9)	0.002	0.112
Sepsis (yes)	292 (7.8)	112 (4.2)	52 (10.4)	33 (17.6)	95 (22.0)	<0.001	<0.001
Other disease requiring intensive care (yes)	270 (7.2)	102 (3.9)	39 (7.8)	22 (11.7)	107 (24.8)	<0.001	<0.001
Medical history
Hypertension (yes)	2,582 (68.7)	1,823 (69.1)	338 (67.6)	117 (62.6)	304 (70.4)	0.232	0.465
Diabetes (yes)	1,377 (36.6)	935 (35.4)	181 (36.2)	75 (40.1)	186 (43.1)	0.016	0.020
Dyslipidemia (yes)	1,944 (51.7)	1,438 (54.5)	245 (49.0)	67 (35.8)	194 (44.9)	<0.001	<0.001
Hyperuricemia (yes)	1,066 (28.4)	683 (25.9)	157 (31.4)	51 (27.2)	175 (40.5)	<0.001	<0.001
CKD (yes)	945 (25.1)	634 (24.0)	95 (19.0)	25 (13.4)	191 (44.2)	<0.001	<0.001
Vital signs and status
SBP (mmHg)	139 [112–162]	144 [121–166]	128 [100–162]	119 [95–140]	105 [80–136]	0.001	0.001
DBP (mmHg)	79 [61–94]	82 [68–98]	72 [56–97]	66 [54–79]	80 [60–95]	0.001	0.001
Pulse (beats/min)	87 [69–108]	85 [70–105]	93 [70–99]	93 [72–113]	89 [60–112]	0.002	0.005
Respiratory rate (beats/min)	20 [16–28]	20 [16–26]	22 [17–25]	21 [16–30]	20 [15–29]	0.002	0.019
Body temperature (℃)	36.4 [35.8–36.8]	36.4 [35.9–36.8]	36.3 [35.7–36.7]	36.4 [35.7–37.0]	36.4 [35.7–37.0]	0.490	0.356
Body mass index (%)	23.5 [21.2–26.0]	22.6 [21.3–26.3]	23.0 [21.1–26.4]	23.3 [21.0–25.7]	23.1 [20.6–25.6]	0.015	0.001
LVEF (%)	52 [40–63]	52 [40–63]	50 [35–64]	50 [32–65]	46 [34–60]	0.026	0.010
Arterial blood gas
pH	7.41 [7.34–7.45]	7.42 [7.38–7.45]	7.40 [7.30–7.45]	7.37 [7.23–7.43]	7.37 [7.29–7.43]	0.001	0.001
PCO2 (mmHg)	37 [32–42]	38 [33–42]	37 [32–42]	36 [30–44]	33 [37–41]	<0.001	<0.001
PO2 (mmHg)	113 [82–167]	114 [84–165]	118 [78–165]	103 [77–160]	109 [79–160]	0.001	0.255
HCO3− (mmol/L)	22.8 [19.8–25.2]	23.5 [21.3–25.5]	21.9 [18.3–25.7]	20.4 [16.0–23.5]	17.9 [13.5–21.4]	0.001	0.001
SaO2 (%)	98 [96–99]	98 [96–99]	98 [95–99]	98 [94–99]	98 [95–99]	0.001	<0.001
Lactate (mmol/L)	1.6 [1.1–2.9]	1.5 [1.0–2.4]	2.0 [1.2–2.2]	2.9 [1.6–6.9]	2.7 [1.4–7.3]	0.001	0.001
Laboratory data
WBC (U/L)	9,320 [7,100–12,120]	8,900 [6,905–11,425]	10,120 [7,600–10,900]	10,770 [8,400–14,385]	10,900 [7,700–15,675]	0.001	0.001
Hemoglobin (g/dL)	13.3 [11.3–14.8]	13.6 [12.0–15.0]	13.0 [10.8–15.1]	12.6 [10.7–14.2]	11.4 [9.4–13.3]	0.001	0.001
BUN (mg/dL)	18.6 [14.4–27.7]	17.0 [13.6–22.0]	22.3 [16.0–20.5]	27.0 [19.3–40.7]	48.6 [32.0–74.4]	0.001	0.001
Creatinine (mg/dL)	0.93 [0.72–1.34]	0.83 [0.68–1.07]	1.13 [0.89–1.03]	1.24 [1.02–1.65]	2.57 [1.68–4.26]	0.001	0.001
eGFR (mL/min/1.73 m2)	57.8 [38.4–77.0]	66.2 [51.7–83.1]	45.8 [37.0–81.0]	41.1 [31.1–53.5]	18.6 [11.0–30.0]	0.001	0.001
Sodium (mmol/L)	140 [137–142]	140 [138–142]	139 [136–142]	140 [136–142]	138 [134–141]	<0.001	<0.001
Potassium (mmol/L)	4.1 [3.8–4.5]	4.0 [3.7–4.4]	4.2 [3.7–4.3]	4.1 [3.6–4.7]	4.7 [4.0–5.7]	0.001	0.001
BG (mg/dL)	153 [120–221]	149 [118–208]	171 [128–187]	181 [130–278]	152 [117–254]	<0.001	<0.001
CRP (mg/dL)	0.35 [0.09–2.44]	0.25 [0.08–1.08]	0.57 [0.10–0.75]	2.15 [0.24–8.94]	3.12 [0.35–11.35]	0.001	0.001
BNP (pg/mL)	153 [38–554]	115 [31–443]	219 [63–284]	244 [73–640]	437 [116–1,141]	0.001	0.001
Scoring
APACHE II (points)	11 [7–16]	9 [7–13]	13 [9–12]	15 [10–21]	19 [14–26]	0.001	0.001
Mechanical support (cases) during ICU stay
NPPV (yes)	872 (23.2)	596 (22.6)	143 (28.6)	44 (23.5)	89 (20.6)	0.016	0.176
ETI (yes)	712 (18.9)	285 (10.8)	140 (28.0)	80 (42.8)	207 (47.9)	<0.001	<0.001
Pacing (yes)	209 (5.6)	105 (4.0)	39 (7.8)	16 (8.6)	49 (11.3)	<0.001	<0.001
IABP (yes)	512 (13.6)	292 (11.1)	85 (17.0)	31 (16.6)	104 (24.1)	<0.001	<0.001
PCPS (yes)	179 (4.8)	73 (2.8)	18 (3.6)	12 (6.4)	76 (17.6)	<0.001	<0.001
CHDF (yes)	441 (11.7)	47 (1.8)	29 (5.8)	18 (9.6)	347 (80.3)	<0.001	<0.001
In-hospital mortality (death)	477 (12.7)	207 (7.8)	80 (16.0)	39 (20.9)	43 (19.7)	<0.001	<0.001

Unless indicated otherwise, data are given as the median [interquartile range] or n (%). ^ADifferences between the AKI and No-AKI groups were tested using the Mann-Whitney U test or the Χ² test. ^BDifferences between the 4 groups (No-AKI, Class R, Class I, and Class F) were tested using the Kruskal-Wallis test or the Χ² test. AAD, acute aortic disease; ACS, acute coronary syndrome; AHF, acute heart failure; AKI, acute kidney injury; APACHE-II, Acute Physiology And Chronic Health Evaluation II; BG, blood glucose; BNP, B-type natriuretic peptide; BUN, blood urea nitrogen; CHDF, continuous hemodiafiltration; CKD, chronic kidney disease; CRP, C-reactive protein; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; ETI, endotracheal intubation; IABP, intra-aortic balloon pumping; ICU, intensive care unit; LVEF, left ventricular ejection fraction measured on echocardiography; NPPV, non-invasive positive pressure ventilation; PCPS, percutaneous cardiopulmonary support; PTE, pulmonary thromboembolism; RIFLE, Risk, Injury, Failure, Loss, and End stage; SBB, systolic blood pressure; WBC, white blood cell.

AKI was present in 1,119 (29.8%) patients on admission based on our criteria: Class R (n=500); Class I (n=187); and Class F (n=432). Notably, this number was increased to 1,684 (44.8%) patients during the first 5 days and to 2,053 (54.6%) patients entire period of hospitalization (Figures 1,2A). AKI was present in 61.6%, 28.3%, 20.6%, and 18.3% of patients with sepsis, AHF, acute aortic disease, and ACS, respectively (Figure 2B–E). The associations between patient characteristics and the RIFLE criteria on admission are presented in Table 1. SBP and LVEF were significantly lower in the AKI groups than in the No-AKI group. Serum concentrations of creatinine, BUN, potassium, CRP, and BNP were significantly higher in the AKI groups than in the No-AKI group. The same was true for APACHE II scores (Table 1). These results suggest that the AKI groups included a higher proportion of critically ill patients, and that the severity gradually increased with increased RIFLE class.

Figure 1.

Change in Risk, Injury, Failure, Loss, and End stage (RIFLE) class during hospitalization. Patients were classified into groups based on this schema, as shown. AKI, acute kidney injury.

Figure 2.

Time-dependent changes in the incidence of acute kidney injury (AKI). The incidence of AKI exhibited different patterns depending on the etiology. ACS, acute coronary syndrome; AHF, acute heart failure; F, failure; I, injury; ICU, intensive care unit; R, risk.

The Kaplan-Meier survival curves for the RIFLE classes, including all-cause death over 730 days, showed that the survival rates were significantly lower in the AKI groups than in the No-AKI group (Figure 3). Interestingly, these rates were significantly lower in the Class R group than in the No-AKI group, in the Class I group than in the No-AKI group, and in the Class F group than in the Class I, Class R, and AKI groups (Figure 3A). Multivariate logistic regression analysis identified Class R (HR 1.343; 95% confidence interval [CI] 1.085–1.661; P=0.007), Class I (HR 1.475; 95% CI 1.093–1.898; P=0.011), and Class F (HR 2.280; 95% CI 1.880–2.765; P<0.001) on admission as being independently associated with 730-day mortality (Table 2). These results suggest that although ICU patients developed only mild AKI (Class R status) on admission, they were associated with poor long-term prognosis. In the groups evaluated during 5 days after admission to the ICU and during hospitalization, long-term prognosis was significantly lower in the Class I group than in the Class R and No-AKI groups, and in the Class F group than in the Class I, Class R, and No-AKI groups (Figure 3B,C). Multivariate logistic regression analysis indicated that only Class I (Day 5: HR 1.876 [95% CI 1.462–2.407; P<0.001]; hospitalization: HR 1.878 [95% CI 1.449–2.434; P<0.001]) and Class F (Day 5: HR 3.146 [95% CI 2.590–3.822; P<0.001]; hospitalization: HR 3.990 [95% CI 3.243–4.910; P<0.001]) were independently associated with 730-day mortality (Table 2). These results suggest that the development of Class I and F status, but not Class R status, during the first 5 days in the ICU and entire period of hospitalization was associated with a poor prognosis for patients requiring intensive care.

Figure 3.

Kaplan-Meier curves of the degree of acute kidney injury (AKI) at admission (Adm), on Day 5, and at the end of hospitalization (Hosp). (A) The all-cause death rate was significantly higher in the Class R_Adm group than in the No-AKI_Adm group; in the Class I_Adm group than in the No-AKI_Adm group; and in the Class F_Adm group than in the Class I_Adm, Class R_Adm, and No-AKI_Adm groups. (B) The all-cause death rate was significantly higher in the Class I_Day5 group than in the Class R_Day5 and No-AKI_Day5 groups; and in the Class F_Day5 group than in the Class I_Day5, Class_Day5, and No-AKI_Day5 groups. (C) The all-cause death rate was significantly higher in the Class I_Hosp group than in the Class R_Hosp and No-AKI_Hosp groups; and in the Class F_Hosp group than in the Class I_Hosp, Class R_Hosp, and No-AKI_Hosp groups.

Table 2. Cox Proportional Hazards Regression Model Associated With 730-Day All-Cause Death Based on the RIFLE Criteria at Each Time Point

	RIFLE on admission			RIFLE during the first 5 days			RIFLE during hospitalization
	HR	95% CI	P value	HR	95% CI	P value	HR	95% CI	P value
RIFLE classes
No AKI	1.000			1.000			1.000
Class R	1.343	1.085–1.661	0.007	1.229	0.993–1.521	0.058	1.102	0.870–1.395	0.423
Class I	1.475	1.093–1.898	0.011	1.876	1.462–2.407	<0.001	1.878	1.449–2.434	<0.001
Class F	2.280	1.880–2.765	<0.001	3.146	2.590–3.822	<0.001	3.990	3.243–4.910	<0.001

The model was adjusted using the following factors: ACS (yes), age (≥70 years), mean blood pressure (>98 mmHg), pulse (≥88 beats/min), hemoglobin (≥13.3 g/dL), sodium (≥140 mmol/L), CRP (≥0.35 mg/dL), and LVEF (≥53%). CI, confidence interval; HR, hazard ratio. Other abbreviations as in Table 1.

Time-Dependent Changes in RIFLE Criteria During Hospitalization and Prognosis

As noted above, the number of patients with AKI increased time dependently during hospitalization. We identified AKI in 1,119 patients on admission (Class R: n=500; Class I: n=187; Class F: n=432). Although 2,639 patients did not have AKI on admission, AKI developed in an additional 565 patients during the first 5 days (Class R: n=415; Class I: n=85; Class F: n=65); these patients were defined as New-AKI_Day5. Of the 500 patients in Class R on admission, 391 remained in Class R, whereas 79 and 30 were assigned to Class I and Class F, respectively, during the first 5 days in the ICU. Of the 187 patients in Class I on admission, 156 remained in Class I, whereas 31 were assigned to Class F. Based on these results, patients were assigned to 2 other groups according to the timing of AKI during the first 5 days: patients who had a stable RIFLE classification (Class R to R or Class I to I; n=547 [Stable-AKI_Day5]) and patients who were assigned to another RIFLE class (Class R to I and F or and Class I to F; n=140 [Worsening-AKI_Day5]). Patients were also classified into 4 groups based on how their RIFLE classification changed from Day 5 to the end of hospitalization, using the same method: No-AKI_Hosp (n=1,705); New-AKI_Hosp (n=369); Stable-AKI_Hosp (n=932); and Worsening-AKI_Hosp (n=194; Figure 1).

Kaplan-Meier survival curves, including all-cause death, for the groups based on the timing of AKI are shown in Figure 4. The survival rates were significantly lower in the Worsening-AKI_Day5 group than in the New-AKI_Day5, Stable-AKI_Day5, and No-AKI_Day5 groups (Figure 4A). A similar tendency was observed in the evaluation of Day 5 to the end of hospitalization; the survival rates were significantly lower in the Worsening-AKI_Hosp group than in the New-AKI_Hosp, Stable-AKI_Hosp, and No-AKI_Hosp groups (Figure 3B).

Figure 4.

Kaplan-Meier curves of the timing of acute kidney injury (AKI) (A) from admission to Day 5 and (B) from Day 5 to the end of hospitalization (Hosp). (A) The all-cause death rate was significantly higher in the Worsening-AKI_Day5 group than in the New-AKI_Day5, Stable-AKI_Day5, and No-AKI_Day5 groups. (B) The all-cause death rate was significantly higher in the Worsening-AKI_Hosp group than in the New-AKI_Hosp, Stable-AKI_Hosp, and No-AKI_Hosp groups. Patients were classified into groups according to the schema shown in Figure 1.

Factors associated with 730-day mortality were assessed using multivariate logistic regression analysis. The results indicate that New-AKI (admission to Day 5: HR 1.714 [95% CI 1.376–2.134; P<0.001]; Day 5 to end of hospitalization: HR 1.505 [95% CI 1.134–1.999; P=0.005]) and Worsening-AKI (admission to Day 5: HR 2.431 [95% CI 1.797–3.288; P<0.001]; Day 5 to end of hospitalization: HR 2.944 [95% CI 2.228–3.891; P<0.001]) were independent predictors of 730-day mortality (Table 3). These findings suggest that increases in RIFLE class in specific groups of AKI on admission are an important factor for poor prognosis in patients requiring intensive care.

Table 3. Cox Proportional Hazards Regression Model Associated With 730-Day All-Cause Death Based on the Timing of AKI

	Admission to Day 5			Day 5 to end of hospitalization
	HR	95% CI	P value	HR	95% CI	P value
Timing of AKI
No AKI	1.000			1.000
New AKI	1.714	1.376–2.134	<0.001	1.505	1.134–1.999	0.005
Stable AKI	1.328	1.055–1.673	0.016	1.217	0.970–1.528	0.090
Worsening AKI	2.431	1.797–3.288	<0.001	2.944	2.228–3.891	<0.001

The model was adjusted using the following factors: ACS (yes), age (≥70 years), mean blood pressure (>98 mmHg), pulse (≥88 beats/min), hemoglobin (≥13.3 g/dL), sodium (≥140 mmol/L), CRP (≥0.35 mg/dL), and LVEF (≥53%). Abbreviations as in Tables 1,2.

Association Between Diagnosis and AKI in Patients Requiring Intensive Care

Time-dependent changes in AKI by etiologies and the ratio of etiologies in severe AKI (e.g., Class F status on admission, and worsening AKI during the first 5 days and hospitalization) were also notable. On admission, AKI was present in 18.3%, 28.3%, and 20.5% of patients with ACS, AHF, and acute aortic disease, respectively. Notably AKI was present in a substantial proportion of patients with sepsis (61.6%). During the first 5 days, the rate of AKI gradually increased to 31.2%, 48.7%, and 51.2% in patients with ACS, AHF, and acute aortic disease, respectively. During hospitalization, these rates increased to 40.7%, 63.2%, and 62.3%, respectively. Nonetheless, the greatest increase was observed in patients with sepsis during the first 5 days and during hospitalization (76.0% and 82.2%, respectively; Figure 2).

In patients with Class F status on admission, sepsis was a major diagnosis, accounting for 22.0% of the total number of cases. In patients with worsening AKI during the first 5 days, those with ACS and AHF accounted for 33.6% and 23.4% of cases, respectively, whereas during hospitalization they accounted for 25.2% and 37.1% of cases, respectively (Figure 5). These results suggest that patients with sepsis were already complicated with severe AKI at the time of admission. Surprisingly, AKI was also present on admission in some patients with cardiovascular disease, and increasingly occurred during the course of hospitalization; ACS and AHF were major diagnoses of worsening AKI.

Figure 5.

Etiological differences depending on the degree and timing of acute kidney injury (AKI): (A) etiology in patients with Class F status on admission; (B) etiology in patients with worsening AKI from admission to Day 5; (C) etiology in patients with worsening AKI from Day 5 to the end of hospitalization. ACS, acute coronary syndrome; AHF, acute heart failure; CM, cardiomyopathy; PE, pulmonary edema.

Discussion

In the present study, AKI on admission was associated with a worse outcome in non-surgical ICU patients, even if they developed Class R status (“mild AKI”). Furthermore, new onset of AKI after admission and exacerbation of AKI during hospitalization, in particular the presence of AKI on admission, was associated with poor prognosis in patients with AHF. Diagnostic differences in the degree and/or timing of AKI were also indicated. AKI on admission was frequently present in patients with ACS, AHF, and sepsis, and the exacerbation of AKI during hospitalization was increased in patients with ACS and AHF.

AKI on Admission

The most common diagnosis of AKI on admission was ACS (n=306), followed by AHF (n=241) and sepsis (n=180).

Although ACS made up the highest proportion among the diagnoses, the rate of AKI on admission in patients with ACS was relatively low because a large number of ACS patients (n=1,674) was admitted during the present cohort. Type 1 cardiorenal syndrome manifests as an acute cardiac event that results in kidney injury and renal dysfunction; this condition is termed AKI in cardiogenic shock.¹² The complication of AKI by cardiovascular shock is thus a well-described phenomenon. Substantial evidence supports the conclusion that complex and multifactorial mechanisms underlie renal dysfunction in patients with cardiogenic shock, involving both hemodynamic (renal arterial hypoperfusion, renal venous congestion) and non-hemodynamic factors.^12–14 Almost all AKI cases in patients with ACS were due to these mechanisms, and the incidence of cardiovascular shock is approximately 6–13% in ACS.^15,16 This is consistent with the results of the present study, which show that the incidence of AKI on admission in patients with ACS was relatively low (18.3%). SBP on admission was significantly lower in the AKI than No-AKI group in the limited ACS cohort (median [IQR] 141 [120–160] vs. 112 [87–137] mmHg, P<0.001; data not shown). Within the overall cohort, the same tendency (low blood pressure and LVEF on admission) was observed. The mechanisms of cardiogenic shock were the main driving factors in the non-surgical intensive care cohort examined.

The incidence of cardiovascular shock in patients with AHF is only approximately 4%,^17–19 and the mechanisms underlying AKI on admission in AHF (relatively high at 28.3%) could thus not be explained by cardiovascular shock alone. Recently, research attention on the mechanism underlying AKI has shifted from cardiac output (cardiogenic shock, “forward failure”) to venous congestion (congestive kidney, “backward failure”) as the most important hemodynamic determinant.²⁰ The development of “congestive kidney failure” induced by the increased renal venous pressure arising from venous congestion (increased renal afterload) and increased renal interstitial pressure (intrinsic renal compromise) may play an important role in the development of AKI in patients with cardiovascular disease.²¹ A recent study reported that persistent venous congestion, as well as arterial and organ hypoperfusion, reflected by a lower arterial pressure and cardiac index, are associated with both the incidence and severity of AKI.²² Pressure-induced reduction in renal blood flow, renal hypoxia, and increased interstitial fibrosis directly lead to AKI through renal congestion. The combination of these mechanisms made AHF the second most frequent etiology of AKI on admission in the non-surgical ICU cohort.

According to previous reports involving a general ICU cohort, one of the major diagnoses of AKI was sepsis, which accounts for 45–70% of all cases of AKI in critically ill patients.²³ Although sepsis was not a common diagnosis (n=292) in the present study, the rate of AKI on admission among those with sepsis was relatively high (61.6%) on admission and greater (82.2%) during hospitalization. These rates were affected by the difference in definition of the baseline creatine value. Therefore, our results suggest that AKI developed with the same timing as the onset of sepsis, and thus already complicated AKI on admission.

AKI During Hospitalization

Worsening renal function (new-onset AKI after admission and AKI with an increase in RIFLE class during hospitalization) is common in patients with cardiovascular disease (mainly ACS and AHF).^24,25 In the present study, 934 patients among those who were not diagnosed with AKI on admission experienced AKI complications during hospitalization. A multifactorial pathogenesis of AKI has been considered in these patients, suggesting that medication is also associated with the occurrence of “late-onset” AKI. In general, treatment with intravenous diuretics is required in patients with decompensated HF, and potent diuretics are necessary during the acute phase of the disease.²⁶ Therefore, diuretics remain the mainstay of treatment for patients with AHF and HF complicated by ACS. Numerous studies have also shown that potent diuretics are associated with worsening renal function,²⁷ particularly in patients with pre-existing renal dysfunction.²⁸ Furthermore, bolus infusions do not promote gradual diuresis; therefore, they do not allow time for the fluid in the periphery to move from the extravascular space to the intravascular space. This leads to a significant decrease in renal perfusion and, as a result, worsening renal function during hospitalization.²⁹ The use of angiotensin-converting enzyme inhibitors (ACEI) and angiotensin II receptor blockers (ARBs) is recommended depending on the associated components of the pathogenesis.¹⁷ Patients with renal impairment may derive greater benefit from ACEI and ARBs than other patients because they are at a higher absolute risk of mortality. Excessive circulating and tissue angiotensin II and aldosterone concentrations cause remodeling and dysfunction in cardiovascular and renal tissues. Current guidelines recommend the same treatment for renal impairment patients as for patients with normal kidney function with minor dose adjustments.³⁰ However, the levels of creatinine often increase after the initiation of ACEI or ARBs in patients with AHF and ACS, especially those with pre-existing renal impairment.³¹ The clinical importance of each mechanism is likely to vary depending on the patient and clinical situation.

Definition of AKI

The most important factor for defining renal dysfunction is the concentration of creatinine at baseline. It is difficult to define the baseline creatinine concentrations for each individual. Most studies have determined that the levels detected on admission can be used to define worsening renal function. Our definition, as used in the present study, may not be strictly appropriate, which may constitute one of the limitations of this study. Routinely monitored creatinine concentrations would be the best option for baseline levels, because it is unclear whether high creatinine concentrations on admission indicate AKI or CKD. This may be a major drawback of the choice of creatinine value on admission as baseline. Although obtaining data on prehospital kidney function wherever possible has been recommended, other suggestions include using the MDRD method (eGFR 75 mL/min/1.73 m² approach) and the lowest inpatient serum creatinine concentration in patients without pre-hospital data.⁷ Determination of baseline kidney function remains controversial. Because a more rapid diagnosis of AKI is required for better prognosis in intensive care patients, the determination of the presence or absence of AKI on admission is crucial. We previously reported that when using the RIFLE criteria in patients with AHF upon admission to the ICU, AKI was present in 33.2% of patients, which increased to 73.4% during hospitalization.⁸ Efforts to develop reliable early diagnosis metrics were made from various perspectives. In the present retrospective study, we could not evaluate the prehospital creatinine in all cases, and the baseline value was therefore evaluated using the eGFR 75 mL/min/1.73 m² approach or the lowest inpatients serum creatine concentration. We were able to demonstrate a prognostic impact of AKI on admission in the non-surgical ICU cohort. The traditionally used worsening renal function criterion based on the baseline creatinine value on admission may have included patients to whom better prognoses would apply. Therefore, a redefinition of the AKI criteria to indicate AKI on admission may be necessary in patients who require intensive care.

Study Limitations

Several limitations of the present study should be mentioned. The most important consideration is the definition of the baseline creatinine value. Because of the retrospective nature of this study, we could not obtain previous medical records in all cases, obliging us to define the baseline creatinine value by either using the lowest creatinine concentration during hospitalization or a calculated value. We were unable to precisely determine the baseline creatinine concentrations in clinical situations, and the RIFLE classification status could thus not be rapidly evaluated in this study. It is therefore impossible to predict death from the presence or absence of AKI in a real-world clinical setting, because the lowest creatine value cannot be defined until the patient is discharged. We suggest from present study that attention must be paid to AKI on admission in an emergency clinical setting and that the acquisition of the creatinine value before admission be attempted. In addition, we did not establish clear high-flow oxygen, inotropes, and diuretics criteria for admission to the ICU, and, moreover, these admission criteria may have varied over time. Increasing competition for limited ICU beds may force physicians to admit more severe cases to general wards in some cases. The physician ultimately decided where each patient should be admitted in the hospital (i.e., ICU or general ward), and patient bias may have affected this decision. Furthermore, we did not use urine volume criteria for the definition of AKI, as the definition used included the presence of AKI on admission. Another limitation is that no data regarding medication (e.g., prescribed medication on admission, intravenous medication during ICU stay, and medications at discharge from hospital) were obtained, and no statistical associations between AKI and medications were tested in the present study. This is essential to investigate the mechanisms of AKI on admission and during hospitalization for critically ill patients. Finally, the present investigation was a single-center study, which may have led to patient bias. Therefore, further investigations into the detection of the occurrence of AKI at an earlier stage of hospitalization are warranted.

Conclusions

The absence of AKI on admission was associated with good prognosis, whereas the presence of AKI on admission, even if this was only assessed as Class R status, was associated with poor prognosis. An increase in RIFLE classes (i.e., Class R to I or F, and Class I to F) during hospitalization and new AKI (i.e., No-AKI to Class I, R, or F) were also important factors for poor prognosis in patients requiring non-surgical intensive care. This indicates that it is important to understand both the degree and timing of AKI in patients requiring intensive care.

Acknowledgments

The authors are grateful to the staff at the ICU and Medical Records Office of Chiba Hokusoh Hospital, Nippon Medical School, for collecting the medical data.

Sources of Funding

This study did not receive any specific funding.

Disclosures

The authors have no conflicts of interest to declare in association with the present study.

IRB Information

The present study was approved by the Research Ethics Committee of Nippon Medical School Chiba Hokusoh Hospital (Reference no. 841).

References

• 1.   Verbrugge FH, Dupont M, Steels P, Grieten L, Malbrain M, Tang WH, et al. Abdominal contributions to cardiorenal dysfunction in congestive heart failure. J Am Coll Cardiol 2013; 62: 485–495.
• 2.   Pickkers P, Darmon M, Hoste E, Joannidis M, Legrand M, Ostermann M, et al. Acute kidney injury in the critically ill: An updated review on pathophysiology and management. Intensive Care Med 2021; 47: 835–850.
• 3.   Bellomo R, Ronco C, Kellum JA, Mehta RL, Palevsky P; Acute Dialysis Quality Initiative Workgroup. Acute renal failure – definition, outcome measures, animal models, fluid therapy and information technology needs: The Second International Consensus Conference of the Acute Dialysis Quality Initiative (ADQI) Group. Crit Care 2004; 8: R204–R212.
• 4.   Mehta RL, Kellum JA, Shah SV, Molitoris BA, Ronco C, Warnock DG, et al. Acute Kidney Injury Network: Report of an initiative to improve outcomes in acute kidney injury. Crit Care 2007; 11: R31.
• 5.   Khwaja A. KDIGO clinical practice guidelines for acute kidney injury. Nephron Clin Pract 2012; 120: c179–c184.
• 6.   Ronco C, Bellomo R, Kellum JA. Acute kidney injury. Lancet 2019; 394: 1949–1964.
• 7.   Siew ED, Matheny ME. Choice of reference serum creatinine in defining acute kidney injury. Nephron 2015; 131: 107–112.
• 8.   Hata N, Yokoyama S, Shinada T, Kobayashi N, Shirakabe A, Tomita K, et al. Acute kidney injury and outcomes in acute decompensated heart failure: Evaluation of the RIFLE criteria in an acutely ill heart failure population. Eur J Heart Fail 2010; 12: 32–37.
• 9.   Iseki K. Chronic kidney disease in Japan. Intern Med 2008; 47: 681–689.
• 10.   Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D. A more accurate method to estimate glomerular filtration rate from serum creatinine: A new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med 1999; 130: 461–470.
• 11.   Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II: A severity of disease classification system. Crit Care Med 1985; 13: 818–829.
• 12.   Ronco C, Haapio M, House AA, Anavekar N, Bellomo R. Cardiorenal syndrome. J Am Coll Cardiol 2008; 52: 1527–1539.
• 13.   Shirakabe A, Hata N, Kobayashi N, Shinada T, Tomita K, Tsurumi M, et al. Prognostic impact of acute kidney injury in patients with acute decompensated heart failure. Circ J 2013; 77: 687–696.
• 14.   Shirakabe A, Hata N, Kobayashi N, Okazaki H, Matsushita M, Shibata Y, et al. Worsening renal failure in patients with acute heart failure: The importance of cardiac biomarkers. ESC Heart Fail 2019; 6: 416–427.
• 15.   Vallabhajosyula S, Dewaswala N, Sundaragiri PR, Bhopalwala HM, Cheungpasitporn W, Doshi R, et al. Cardiogenic shock complicating ST-segment elevation myocardial infarction: An 18-year analysis of temporal trends, epidemiology, management, and outcomes. Shock 2022; 57: 360–369.
• 16.   Kolte D, Khera S, Aronow WS, Mujib M, Palaniswamy C, Sule S, et al. Trends in incidence, management, and outcomes of cardiogenic shock complicating ST-elevation myocardial infarction in the United States. J Am Heart Assoc 2014; 3: e000590.
• 17.   McDonagh TA, Metra M, Adamo M, Gardner RS, Baumbach A, Böhm M, et al. 2021 ESC guidelines for the diagnosis and treatment of acute and chronic heart failure. Eur Heart J 2021; 42: 3599–3726.
• 18.   Berg DD, Bohula EA, van Diepen S, Katz JN, Alviar CL, Baird-Zars VM, et al. Epidemiology of shock in contemporary cardiac intensive care units. Circ Cardiovasc Qual Outcomes 2019; 12: e005618.
• 19.   Spinar J, Parenica J, Vitovec J, Widimsky P, Linhart A, Fedorco M, et al. Baseline characteristics and hospital mortality in the Acute Heart Failure Database (AHEAD) main registry. Crit Care 2011; 15: R291.
• 20.   Mullens W, Abrahams Z, Francis GS, Sokos G, Taylor DO, Starling RC, et al. Importance of venous congestion for worsening of renal function in advanced decompensated heart failure. J Am Coll Cardiol 2009; 53: 589–596.
• 21.   Legrand M, Mebazaa A, Ronco C, Januzzi JL. When cardiac failure, kidney dysfunction, and kidney injury intersect in acute conditions: The case of cardiorenal syndrome. Crit Care Med 2014; 42: 2109–2117.
• 22.   Tarvasmäki T, Haapio M, Mebazaa A, Sionis A, Silva-Cardoso J, Tolppanen H, et al. Acute kidney injury in cardiogenic shock: Definitions, incidence, haemodynamic alterations, and mortality. Eur J Heart Fail 2018; 20: 572–581.
• 23.   Zarbock A, Nadim MK, Pickkers P, Gomez H, Bell S, Joannidis M, et al. Sepsis-associated acute kidney injury: Consensus report of the 28th Acute Disease Quality Initiative Workgroup. Nat Rev Nephrol 2023; 19: 401–417.
• 24.   Shirakabe A, Hata N, Kobayashi N, Okazaki H, Matsushita M, Shibata Y, et al. Worsening renal function definition is insufficient for evaluating acute renal failure in acute heart failure. ESC Heart Fail 2018; 5: 322–331.
• 25.   Kadooka K, Miyachi H, Kimura T, Asano K, Onodera K, Masunaga N, et al. Non-cardiovascular disorders in a contemporary cardiovascular intensive care unit in Japan. J Cardiol 2021; 78: 166–171.
• 26.   Shirakabe A, Hata N, Yamamoto M, Kobayashi N, Shinada T, Tomita K, et al. Immediate administration of tolvaptan prevents the exacerbation of acute kidney injury and improves the mid-term prognosis of patients with severely decompensated acute heart failure. Circ J 2014; 78: 911–921.
• 27.   Gottlieb SS, Brater DC, Thomas I, Havranek E, Bourge R, Goldman S, et al. BG9719 (CVT-124), an A₁ adenosine receptor antagonist, protects against the decline in renal function observed with diuretic therapy. Circulation 2002; 105: 1348–1353.
• 28.   Forman DE, Butler J, Wang Y, Abraham WT, O’Connor CM, Gottlieb SS, et al. Incidence, predictors at admission, and impact of worsening renal function among patients hospitalized with heart failure. J Am Coll Cardiol 2004; 43: 61–67.
• 29.   Weng H, Li Y, Nie X, He C, Feng P, Zhao F, et al. Comparative effectiveness and safety of bolus vs. continuous infusion of loop diuretics: Results from the MIMIC-III database. Am J Med Sci 2023; 365: 353–360.
• 30.   Tsutsui H, Ide T, Ito H, Kihara Y, Kinugawa K, Kinugawa S, et al. JCS/JHFS 2021 guideline focused update on diagnosis and treatment of acute and chronic heart failure. Circ J 2021; 85: 2252–2291.
• 31.   Pontremoli R, Borghi C, Perrone Filardi P. Renal protection in chronic heart failure: Focus on sacubitril/valsartan. Eur Heart J Cardiovasc Pharmacother 2021; 7: 445–452.