The effects of timing onset and progression of AKI on the clinical outcomes in AKI patients with sepsis: a prospective multicenter cohort study

Abstract

Background

Limited studies are available concerning on the earlier identification of AKI with sepsis. The aim of the study was to identify the risk factors of AKI early which depended on the timing onset and progression of AKI and investigate the effects of timing onset and progression of AKI on clinical outcomes.

Methods

Patients who developed sepsis during their first 48-h admission to ICU were included. The primary outcome was major adverse kidney events (MAKE) consisted of all-cause mortality, RRT-dependence, or an inability to recover to 1.5 times of the baseline creatinine value up to 30 days. We determined MAKE and in-hospital mortality by multivariable logistic regression and explored the risk factors of early persistent-AKI. C statistics were used to evaluate model fit.

Results

58.7% sepsis patients developed AKI. According to the timing onset and progression of AKI, Early transient-AKI, early persistent-AKI, late transient-AKI, late persistent-AKI were identified. Clinical outcomes were quite different among subgroups. Early persistent-AKI had 3.0-fold (OR 3.04, 95% CI 1.61 − 4.62) risk of MAKE and 2.6-fold (OR 2.60, 95%CI 1.72 − 3.76) risk of in-hospital mortality increased compared with the late transients-AKI. Older age, underweight, obese, faster heart rate, lower MAP, platelet, hematocrit, pH and energy intake during the first 24 h on ICU admission could well predict the early persistent-AKI in patients with sepsis.

Conclusion

Four AKI subphenotypes were identified based on the timing onset and progression of AKI. Early persistent-AKI showed higher risk of major adverse kidney events and in-hospital mortality.

Trial registration

This study was registered in the Chinese Clinical Trials Registry (www.chictr.org/cn) under registration number ChiCTR-ECH-13003934.

Keywords:

• Sepsis
• acute kidney injury
• major adverse kidney events
• in-hospital mortality
• intensive care unit

Background

Acute kidney injury (AKI) is a common and well-recognized conditions of critical illness with the incidence of up to 50% [1,2]. The association between AKI and increased risk of length of stay (LOS) in hospital, short- or long-term mortality [1], prolonged mechanical ventilation (MV) [3], higher healthcare costs [4,5] and persistent renal dysfunction [6] have been well known. Sepsis and shock are the most common cause of AKI in ICUs, accounting for one-half of cases in the ICU [7,8]. When these conditions occur together, the incidence of AKI is up to 80% [9] with unacceptably mortality as high as 60%–80% [10]. The sepsis patients with AKI also had more changes in vital signs and symptoms, more severe illness and higher rate of nonrenal organ dysfunction or failure [11].

However, the diagnosis of AKI usually based on the urine output and serum creatinine which is often delayed [12,13], and there was no specific treatment are currently available, except for renal replacement therapy (RRT) [14,15]. Therefore, early recognition of high-risk patients, enhanced clinical monitoring are the most effective preventive interventions for AKI. Previous studies have shown that there were significantly different baseline characteristics, risk factors and outcomes between early- (within 48 h) and late-AKI (> 48 h) patients after major surgery [13,16–18], the early-AKI was associated with higher risk of mortality [18], even though renal function returned to baseline level [16]. In addition, the recovery patterns and evolution after AKI were also associated with adverse outcomes [19,20]. According to the Acute Disease Quality Initiative Workgroup (ADQI), Fabrice and colleges classified AKI as ‘transient’ or ‘persistent’ based on the duration less or more than 48 h and demonstrated that the persistent-AKI was an independent risk factor for mortality in patients with sepsis [19]. However, the effects based on the timing onset and evolution of AKI on clinical outcomes in patients with sepsis are still unclear. In addition, few studies based on clinical variables have predicted the risk factors of occurrence and clinical outcomes for AKI.

In this study, we hypothesized that the AKI subphenotypes based on timing onset and progression of AKI had distinct clinical characteristics and outcomes. Furthermore, early clinical variables on admission to ICU could well predict the timing onset and evolution of AKI patients with sepsis.

Methods

Study design and population

The China Critical Care Sepsis Trial (CCCST) was a prospective, multicenter, cohort study, which conducted at 18 ICUs of 16 tertiary hospitals in China. From 1 January 2014 to 31 August 2015, a total of 4910 adult patients consecutively admitted to ICUs for at least 24 h were enrolled in this study. For the multiple ICU admission, only the initial admission was considered. We excluded 2824 patients who did not develop sepsis during their first 48 h admission of ICU and 11 patients with missing data of baseline information. In addition, the patients with end-stage chronic kidney disease, RRT or kidney transplantation were also excluded. Finally, 2075 patients included in the analysis and 1211 patients developed AKI (Figure 1).

Figure 1. Flowchart of participants included in the current study.

Collection of demographical, laboratory variables and outcomes

Standard case report forms were used to collect demographic information, anthropometrics, physiologic, length of stay (LOS) in ICU, total number of days during in the hospital, the ICU and hospital mortality. Vital signs and laboratory values on the first 24 h on admission ICU included temperature, heart rate, systolic blood pressure, diastolic blood pressure, Glasgow coma score (GCS), white blood cell (WBC), hematocrit (HCT), platelet (PLT), bilirubin, pH, PaCO₂, PaO₂, FiO₂, lactate (LAC). Fluid intake and output during the first 24 h on admission ICU were also collected daily. Serum creatinine (SCr) and urine output were recorded daily on the first seven days. The initialed nutrition therapy and nutrition target on the first 24 − 48 h on admission ICU were also reported. For the multiple recorded variables on the first 24 h admission to ICU, the value associated with the highest Acute Physiology and Chronic Health Evaluation II score was employed.

All patients were followed up for 30 days or until discharge from hospital or death, whichever occurred earlier. The primary outcome was major adverse kidney events (MAKE). In hospital mortality, ICU mortality, length of stay in the ICU and hospital were secondary outcomes. Death was confirmed from the participating hospitals or from the local citizen registry.

Definitions

Sepsis was defined as a dysregulated response to infection, which result in life-threatening organ failure (defined as Sequential Organ Failure Assessment (SOFA) equal to or greater than two scores) on admission to ICU or within the first 48 h after admission to ICU [21]. The Kidney Disease Improving Global Outcomes (KDIGO) guideline was used to classify the AKI severity [22]. The baseline SCr was estimated by using the Modification of Diet in Renal Disease (MDRD) equation for an assumed glomerular filtration rate (GFR) of 75 mL/min/1.73 m² or the lowest SCr value during the ICU stay, whichever was lower [23]. Patients who developed AKI after 7-day on admission to ICU were defined as non-AKI. According to the timing onset, AKI was classified as early-AKI (occurring within the first 48 h after admission to ICU) or late-AKI (> 48 h) [13]. Transient-AKI was defined as a reversal of KDIGO criteria within 48 h of AKI onset. If one or more KDIGO criteria remained present beyond 48 h of AKI onset, or they normalized within 48 h but relapsed within the next 48 h, the AKI defined as persistent-AKI [19,24,25]. Fluid balance (FB) was defined as the difference between fluid intake and fluid output. Fluid intake included all intravenous and oral fluids, which included resuscitation and maintenance fluids, blood products, drug infusions, and enteral and parenteral nutrition. Fluid output includes urinary volume, ultrafiltration fluid, drain fluid and estimated gastrointestinal losses. For the difficulty of assessment, the insensible loss was not considered in our study. Major adverse kidney event (MAKE) is a composite outcome, which consisted of all-cause mortality, RRT-dependence, or an inability to recover to 1.5 times of the baseline creatinine value up to 30 days [26,27]. For those discharged with normal serum creatinine or recovered to lower than 1.5 times up to 30 days defined as favorable outcome.

Missing values

The missing variables were shown in Figure S1. Missing variables for height and weight were imputed with mean values. For variables which were recorded on the first 24 h, multiple imputation methods were performed [28]. However, for there were no reliable surrogate markers, the missing data of nutritional therapy target were not performed imputation.

Sensitivity analysis

To examine the robustness of the findings, the Beijing Acute Kidney Injury Trial (BAKIT) database was used to repeat the analysis. Detail of the BAKIT on the rationale, design and major results have been published previously [23,29,30]. Due to the lack of lactate values in the BAKIT, sepsis was defined using sepsis-1 criteria [31]. Sepsis was defined as systemic inflammatory response syndrome (SIRS) caused by suspected or documented infection [31].

Statistical analysis

Categorical variables were expressed as the number of cases and proportions and were compared by using the Mantel–Haenszel χ² test. While continuous variables were expressed as mean (standard deviation, SD) or median (interquartile range, IQR) based on the distribution of variables, which were compared by using Student t test, one-way analysis of variance or Wilcoxon rank-sum test. Sankey diagrams were used to show the association between phenotype of AKI, AKI severity and clinical outcomes.

Multivariable logistic regression was used to investigate the association between the timing onset and evolution of AKI and the clinical outcomes. The late transient-AKI was defined as reference group. A stepwise logistic regression model was used to assess the effect of characteristics, symptom and laboratory values on clinical outcomes, timing onset and evolution of AKI. The variables with p values equal to or less than 0.1 in the univariate analysis were set in the multivariate analysis. The likelihood ratio test was used to test the overall statistical significance of the logistic model. The variance inflation factors (VIFs) and tolerance coefficients were computed to test multicollinearity among the covariates. Values of VIF exceeding 10 are regarded as multicollinearity and were removed in the model. In the multivariable analysis, the centers were included as a random effect. Adjusted odds ratios (ORs) and their 95% confidence intervals (CIs) were calculated. Furthermore, to assess the ability of logistic model to discriminate patients who will experiment early persistent-AKI, a concordance statistic (C-statistic; equal to the area under the receiver operating curve) was calculated [8]. A value of C-statistic greater than 0.8 predict a better discrimination.

A two-side value of p < 0.05 was considered to indicate statistically significant. All analyses were performed with Stata version 15 (StataCorp, College Station, Texas, USA) and R package (version 3.6.2).

Results

Baseline characteristics of participants

A total of 2075 eligible patients with sepsis were included in the final analysis. Among them, 1211 (58.4%) patients developed AKI (Figure 1). The baseline characteristics and outcomes of AKI and non-AKI patients were shown in Table S1. Of 1211 patients with AKI, 293 (24.2%) patients experienced early transient-AKI, 522 (43.1%) patients experienced early persistent-AKI, 124 (10.3%) patients experienced late transient-AKI, and 272 (22.4%) patients evolved late persistent-AKI (Figure 1).

Table 1. Baseline characteristics and outcomes of patients indifferent subgroups according to the timing onset and evolution of AKI.

Characteristics	Early transient-AKI	Early persistent-AKI	Late transient-AKI	Late persistent-AKI	p Value
Number	293	522	124	272
Age, Mean (SD)	62.4 (16.5)	65.7(17.9)	62.4 (17.8)	62.9 (18.2)	0.097
Sex, male (%)	181 (61.8)	361 (69.2)	69 (55.6)	185 (68.0)	0.012
Baseline SCr, mean (SD), µmmol/L	83.3 (23.6)	81.4 (24.4)	82.9 (25.3)	82.6 (24.8)	0.164
Body mass index, n (%)
< 18.5	19 (6.5)	64 (12.3)	7 (5.6)	20 (7.4)	0.004
18.5–24.9	202 (68.9)	293 (56.1)	82 (66.1)	183 (67.3)
25.0–27.9	48 (16.4)	81 (15.5)	29 (23.4)	52 (19.0)
≥ 28.0	24 (8.2)	84 (16.1)	6 (4.9)	17 (6.3)
Source of admission, n (%)
Emergency Department	62 (21.2)	124 (23.8)	34 (27.4)	66 (24.3)	<0.001
General Wards	97 (33.1)	210 (40.2)	35 (28.2)	108 (39.7)
Surgery	56 (19.1)	54 (10.3)	47 (37.9)	81 (29.8)
Others	78 (26.6)	134 (25.7)	8 (6.5)	17 (6.3)
Comorbidities, n (%)
Respiratory disease	31 (10.6)	29 (5.6)	8 (6.5)	19 (7.0)	0.049
Cardiovascular disease	51(17.4)	82 (15.7)	25 (20.2)	58 (21.3)	0.225
Hypertension	90 (30.7)	244 (46.7)	39 (31.5)	70 (25.7)	<0.001
Diabetes mellitus	56 (19.1)	110 (21.1)	19 (15.3)	65 (23.4)	0.223
Chronic renal failure	5 (1.7)	75 (14.4)	4 (3.2)	27 (9.9)	<0.001
Cancer	31 (10.6)	45 (8.6)	11 (8.9)	29 (10.7)	0.721
None	64 (21.8)	96 (18.4)	57 (46.0)	102 (37.5)	<0.001
Site of infection^a, n (%)
Lung	150 (51.2)	269 (51.5)	68(54.8)	178 (65.4)	0.001
Abdominal	79 (27.0)	186 (35.6)	27 (21.8)	58 (21.3)	<0.001
Urinary	23 (7.8)	53 (10.2)	5 (4.0)	14 (5.1)	0.028
Bloodstream	21 (7.2)	49 (9.4)	14 (11.3)	22(8.1)	0.506
Others	38 (13.0)	55(10.5)	9 (7.3)	25 (9.2)	0.285
Unclear	31 (10.6)	109 (20.9)	15 (12.1)	33 (12.4)	<0.001
Outcomes
MAKE, n(%)	134 (45.7)	305 (58.4)	32 (25.8)	112 (41.2)	<0.001
Hospital mortality, n(%)	104 (35.5)	235 (45.0)	23 (18.5)	101 (37.1)	<0.001
28-d free ICU LOS, days, median (IQR)	17.0 (13.0 − 23.0)	13.0 (6.0 − 19.0)	20.0 (13.0- 25.0)	16.0 (10.0- 24.0)	0.025
28-d free Hospital LOS, days, median (IQR)	10.0 (4.0- 25.0)	7.0 (4.0- 20.0)	17.0 (9.0- 30.0)	9.0 (5.0 − 24.0)	0.005

^apercentage is not equal to 100 for patients may present one or more sites of infection. SD: standard deviation; Scr: serum creatinine; MAKE: major adverse kidney events; ICU, intensive care unit; LOS, length of stay; IQR: interquartile range.

Compared with other three groups, the patients with early persistent-AKI were more likely to be male, underweight and overweight or obesity, have more history of comorbidities and unclear infection (Table 1). They were likely to have more unstable hemodynamics (heart rate > 100/min: 70.3% vs 78.2% vs 58.9% vs 68.8%, MAP < 65 mmHg: 36.2% vs 56.7% vs 40.3% vs 21.0%. p < 0.001), conscious disturbance (Glasgow Coma Scale < 10: 31.4% vs 50.8% vs 41.3% vs 25.0%, p < 0.001) and more hypoxemia (PO₂/FIO₂ mmHg: 207.2 vs 202.0 vs 233.6 vs 214.6, p < 0.001) and elevated serum lactate (Lactate > 4.0 mmol/L, 15.4% vs 25.9% vs 7.3% vs 20.2%, p < 0.001), higher WBC (×10⁹/L, 12.1 vs 14.0 vs 12.1 vs 12.3, p < 0.001), lower Platelet count (< 100 × 10⁹/L, 23.2% vs 38.1% vs 10.5% vs 24.3%, p < 0.001) and lower HCT (< 30%, 38.9% vs 60.7% vs 16.2% vs 42.6%, p < 0.001) and received more fluid (L, 2.1 vs 3.9 vs 1.1 vs 1.8, p < 0.001). More than half (161, 52.1%) patients with early persistent-AKI experienced higher energy intake. However, only 11.7% patients with late and transient-AKI received >20 kcal/h of energy intake (Table 2).

Table 2. Clinical symptom and library values in subgroup patients according to the timing onset and evolution of AKI.

	Early transient-AKI	Early persistent-AKI	Late transient-AKI	Late persistent-AKI	P value
No.	293	522	124	272
Symptom
Temperature ≥38 °C	160 (54.6)	270 (51.7)	66 (53.2)	131 (48.3)	0.669
Heart rate/min, mean (SD)	111.2 (17.4)	117.0 (21.6)	106.5 (15.8)	109.2 (19.9)	<0.001
>100	206 (70.3)	408 (78.2)	73 (58.9)	187 (68.8)	<0.001
Respiratory rate/min, mean (SD)	26.8 (6.6)	26.7 (7.3)	24.2 (6.6)	24.8 (7.5)	<0.001
>20	215 (73.4)	373 (71.5)	80 (64.5)	174 (64.0)	<0.001
MAP < 65 mmHg, n (%)	106 (36.2)	296 (56.7)	50(40.3)	57 (21.0)	<0.001
Glasgow Coma Scale < 10, n (%)	92 (31.4)	265 (50.8)	51 (41.3)	68 (25.0)	<0.001
Laboratory findings, n (%)
WBC (×10^9/L), median (IQR)	12.1 (8.2-17.3)	14.0 (9.1-18.7)	12.1(8.0-16.0)	12.3 (8.5-16.6)	0.012
HCT < 30%	114(38.9)	317 (60.7)	19 (16.2)	116 (42.6)	<0.001
Platelet < 100 (×10^9/L), n (%)	68 (23.2)	199 (38.1)	13 (10.5)	66 (24.3)	<0.001
Bilirubin >23.3 mmol/L, n (%)	50 (17.1)	93 (17.8)	19 (15.3)	40 (14.7)	0.827
pH, mean (SD)	7.39 (0.10)	7.32 (0.12)	7.40 (0.11)	7.37 (0.09)	<0.001
PO2/FIO2, mmHg, median (IQR)	207.2 (138.0-307.0)	202.0 (124.0-317.0)	233.6(158.7-341.0)	214.6 (140.0-323.8)	<0.001
PCO2, median (IQR)	36.2 (11.8)	35.7 (10.2)	39.7 (11.7)	38.3(10.8)	<0.001
Lactate >4.0 mmol/L, n (%)	45 (15.4)	135 (25.9)	9 (7.3)	55 (20.2)	<0.001
Fluid balance, day1, L,	2.1 (0.8 − 3.6)	3.9 (1.8 − 5.1)	1.1 (0.6 − 2.6)	1.8 (0.7 − 3.1)	<0.001
Target energy within first 24–48 hours, n (%)					0.081
10–20 kcal/hr	76 (43.7)	89 (28.8)	65 (84.4)	112 (52.8)
<10 kcal/hr	16 (9.2)	59 (19.1)	3 (3.9)	19 (9.0)
>20 kcal/hr	82 (47.1)	161 (52.1)	9 (11.7)	81 (38.2)
Duration of ICU stay, n (%)
Mechanical ventilation	228 (77.8)	415 (79.5)	96 (77.4)	228 (83.8)	0.271
Renal replace therapy	41 (14.0)	228 (55.2)	17 (13.7)	53 (19.5)	<0.001
Vasopressor	204 (69.6)	345 (66.1)	61 (49.2)	129 (47.4)	<0.001

SD: standard deviation; IQR: interquartile range; ICU: intensive care unit; MAP, mean arterial pressure; WBC: white blood cell; HCT: hematocrit; PaO2/FiO2: partial pressure of oxygen and inhaled oxygen concentration ratio.

Severity and outcomes of AKI subgroups

Timing onset and evolution of renal dysfunction were diverse, although there were not significantly difference between baseline SCr (Table 1). The distribution of the severity among the four groups was shown in the Figure 2. Only 216 (26.3%) patients with early-AKI and 221 (27.8%) patients with persistent-AKI were mild KDIGO criteria (stage1), more than half patients with early-AKI (418, 51.3%) or persistent-AKI (424, 53.4%) were severity KDIGO criteria (stage 3). When combined timing onset and evolution of AKI, 354 (67.8%) patients with early persistent-AKI presented severity KDIGO criteria (stage 3) and more than half (228, 55.2%) of those patients received RRT during their ICU stay (Table 2).

Figure 2. Sankey diagrams showing the severity of AKI by subgroups. (A)AKI onset; (B) AKI progression; (C) Combine AKI onset and evolution. AKI, acute kidney injury.

Crude MAKE and in-hospital mortality in patients with early persistent-AKI were 58.4% (305) and 45.0% (235), respectively (Figure 2(C)). Which were significantly higher than other three subgroups. However, the LOS in ICU or hospital were likely to be shorter in early transient-AKI patients (Table 1).

The ORs of clinical outcomes based on the timing onset and evolution of AKI are shown in Figure 3. Compared with the patients with late transients-AKI, the patients who developed early persistent-AKI had worse clinical outcomes, with multivariable-adjusted ORs for the ‘MAKE’ and ‘in hospital mortality’ of 3.04 (95% CI, 1.61 − 4.62) and 2.60 (95%CI, 1.72 − 3.76), respectively.

Figure 3. The association between AKI onset, evolution and clinical outcomes.

Risk factors for the early persistent-AKI

We used multivariable logistic regression to explore the risk factors for the early persistent-AKI. The results are shown in Table 3. Older age (OR 1.02, 95%CI 1.00 − 1.03), underweight (OR 2.04, 95%CI 1.02 − 4.07), obese (OR 2.09, 95%CI 1.12 − 3.92), faster heart rate (OR 1.02, 95%CI 1.01 − 1.06), lower MAP (OR 1.61, 95%CI 1.16 − 2.45), lower platelet (OR 2.37, 95%CI 1.51–3.74), hematocrit (OR 2.33, 95%CI 1.54 − 3.54), pH (OR 1.91, 95%CI 1.24 − 2.94) were associated with early persistent-AKI. For one liter of fluid balance increased, the risk of early persistent AKI increased 1.78-fold (OR 1.78, 95%CI 1.26 − 2.89). In additional, compared with patients of 10–20 kcal/hrs of energy intake during the first 24–48 h, both lower (< 10 kcal/h, OR 2.10, 95%CI 1.11–3.99) and higher (> 20 kcal/h, OR 1.26, 95%CI 1.07–2.01) energy intake were increased the risk of early persistent-AKI.

Table 3. Multivariate logistic regression analysis evaluating the risk factors on early persistent-AKI.

Characteristics	OR(95%CI)	p Value
Age, per year older	1.02 (1.00 − 1.03)	<0.001
Body mass index, kg/m^2
<18.5	2.04 (1.02 − 4.07)	0.035
18.5–24.9	Ref
25.0–27.9	1.44 (0.83 − 2.28)	0.196
≥28.0	2.09 (1.12 − 3.92)	0.021
Heart rate, per 5/min	1.02 (1.01 − 1.06)	0.036
Mean arterial pressure < 65 mmHg	1.61 (1.16 − 2.45)	0.007
Platelet< 100000/cm^3	2.37 (1.51–3.74)	<0.001
Hematocrit < 30%	2.33 (1.54 − 3.54)	<0.001
pH <7.32	1.91 (1.24 − 2.94)	0.003
Fluid balance day 1, per L/24 h	1.78 (1.26 − 2.89)	<0.001
Nutritional strategy during the first 24-48 hour
10–20 kcal/h	Ref
<10 kcal/h	2.10 (1.11–3.99)	0.023
>20 kcal/h	1.26 (1.07 − 2.01)	0.017

C-statistic was 0.794. AKI: acute kidney injury.

Sensitivity analysis

We used data from the BAKIT to examine the robustness of the findings, and similar results were yielded. The participant included and baseline characteristics, clinical values and outcomes were shown in Tables S2 and S3. There were no significant differences in age and sex between subgroup patients according to the timing onset and evolution of AKI. Compared with late transient-AKI patients, the patients with early persistent-AKI had 2.7-fold increased risk of MAKE (OR 3.65, 95% CI 1.41–6.27) and 1.8-fold risk of hospital mortality (OR 2.84, 95% CI 1.44–5.60) after adjusting for potential confounds (Figure S2). The demographic characteristics, symptom and laboratory values could better predict the timing onset and evolution for AKI patients with the C statistic were 0.815 (Table S4).

Discussion

In this large prospective multicenter cohort study, we observed that more than half of sepsis patients developed AKI. The timing onset and evolution of AKI are quite different from in terms of clinical outcomes. Patients with early persistent-AKI had increased risk of MAKE and in-hospital mortality when compared with late transient-AKI. Furthermore, the demographic characteristics, symptom and laboratory values during the first 24 h on admission to ICU could predict the clinical outcomes of AKI in patients with sepsis. These results suggest that we should treat these patients separately when designing therapy and help to study improvement quality process.

Similar to previous studies, our study demonstrated a different profile of AKI. However, the population, definition and timing onset among them varied a lot [13,16,32]. Moriyama et al. used Acute Kidney Injury Network (AKIN) criteria and defined early-AKI in patients with acute myocardial infarction as AKI occurring within 48 h after hospital admission [16]. In a study of 774 critically ill patients, AKI defined according to KDIGO criteria and early-AKI defined as any AKI stage diagnosed within 24 h upon ICU admission [32]. In our study, early-AKI was defined as any AKI stage within 48 h after ICU admission because there was a delay in reaching SCr criteria and this also meet the KDIGO definition of AKI[22], just as Li et al. showed [13].

Two-thirds of sepsis patients (815/1211) presented AKI upon ICU admission, of which the majority was moderate/severe (KDIGO stage 2 or 3), and persistent in three out of five patients. For those delayed AKI, nearly eighty percent had mild/moderate AKI, and one quarter of which were persistent. The clinical outcomes were quite difference between them. The persistent-AKI had higher rate of MAKE and in-hospital mortality. A prospective observational study showed a majority persistent-AKI had severity of renal dysfunction and affects patients survival [19]. Approximately 40% AKI patients unrecovered and associated with worse clinical outcomes [33]. The patients who were not resolve at hospital discharge were 2.0-fold likely to be dead at one year compared to those late recovery [33]. While a large multicenter study indicated that not only sustained-AKI but also transient-AKI were associated with higher risk of readmission hospital in patients with acute decompensated heart failure [34]. These different results maybe contribute to the previous studies only considered the timing onset or the duration of AKI. Even though AKI occurs at the same time, the outcomes may be varied with different duration of AKI.

Combined with the timing onset and evolution of AKI, we classified AKI patients into four subgroups and found that the characteristics, symptoms and laboratory values were quite different among them, clinical outcomes were also varied a lot. Persistent-AKI (no matter timing onset) was associated with higher risk of MAKE and in-hospital mortality, even though transient-AKI with early timing onset was also had higher risk of unfavorable clinical outcome, when in contrast to the late transient-AKI. This study suggests that for those patients with the same timing onset of AKI, the different duration of AKI had various severity of AKI and clinical outcomes.

It generally is well accepted that there is no effective therapy of AKI except RRT [14,15]. A major study based on kidney function (elevated SCr or decreased urine output) to identify AKI, which usually delayed 24–48 h. Although new kidney biomarkers may improve early identification [35], they still do not cause patients to seek earlier medical attention. For AKI, usually it was a silent process and does not present early symptoms. Therefore, based on demographic characteristics, earlier symptoms and laboratory valuables to identify the timing onset and evolution of AKI, enhanced clinical monitoring, avoiding nephrotoxic exposure are the most effective preventive interventions for AKI, especially for those had poor clinical outcomes.

Our results reveal that except for timing onset and evolution of AKI were associated with clinical outcomes in critically ill patient with sepsis. The early persistent-AKI had 3.0-fold increased risk of MAKE and 2.6-fold increased risk of in-hospital mortality, compares with late transient-AKI patients. Furthermore, older age, underweight/obesity, faster heart rate, lower Platelet count and hematocrit were also independent risk factors for early persistent-AKI [13,18,33]. For every liter of fluid balance increased, the risk of early persistent-AKI increased 78%, but higher fluid balance was the cause or a result of severity of AKI remains unclear. Our previous studies demonstrated that fluid overload was an independent risk factor of AKI and increased the severity of AKI, and higher fluid balance was associated with 28-day mortality [29]. However, for patients with AKI always showed oliguria or anuria, cumulated more fluid and lead to fluid overload. In our study, two thirds of AKI patients with sepsis received nutritional therapy during the first 24-48 h upon admission to ICU, less than half of which achieved the targets of trophic/hypocaloric feeding ranged 10-20 kcal/hrs or up to 70% of target goal within 48 h [21]. Both lower and higher caloric was associated with unfavorable clinical outcomes [36–39].

There were several limitations to this analysis. First, the observational study design could not include all the potential confounders. Second, we failed to record diuretic use, which may influence fluid management and outcomes [40]. Third, we used the lowest creatinine level during the ICU stay or the MDRD formula to estimate baseline Scr, which may not reflect the true baseline renal function. Furthermore, we only analyzed the first 24 h of symptoms and laboratory values, which couldn’t illustrate the status of ICU duration and those AKI patients at ICU admission. Although we used two databases to explain our study, the results still need to be further validated in the future studies with larger sample sizes.

Conclusion

In conclusion, based on the timing onset and progression of AKI, four AKI subphenotypes were identified: early transients-AKI, early persistent-AKI, late transients-AKI and late persistent-AKI. Compared with late transients-AKI, the patients with early persistent-AKI had high risk of major adverse kidney events and in-hospital mortality. The demographic characteristics, symptoms and laboratory values could be used to for earlier identification of onset and progression of AKI.

Additional information

Additional file 1: Table S1. Baseline characteristics and outcomes in AKI and non-AKI patients with sepsis. Table S2. Baseline characteristics and outcomes for subgroup patients in BAKIT according to the timing onset and evolution of AKI. Table S3. Symptoms, clinical characteristics for subgroup patients in BAKIT according to the timing onset and evolution of AKI. Table S4. Multivariate logistic regression analysis evaluating the risk factors on early persistent AKI in BAKIT. Figure S1. Missing rate for clinical and laboratory variables extracted from the database. Variables with missing rate greater than 30% were excluded from analysis. Figure S2. The association between AKI onset, evolution and clinical outcomes in BAKIT.

Additional file 2: All other ethical bodies that approved our study in the various centers involved.

Ethics approval and consent participants

This study was approved by the ethics committees of Fu Xing Hospital, Capital Medical University, Beijing China (approval notice number 2013FXHEC-KY018) and all other participating hospitals (Additional file 2). The Institutional Review Board approved the informed consent waiver because of the anonymous and purely observational nature of this study.

Consent for publication

Not applicable.

Author contributions

XMX, YH and LJ conceived, designed and led the study. MPW, XW were involved in the design of the study and analysis of the data. The CCCST group performed the recruitment and data collection. WL, YTZ and JW finalized the analysis and interpreted the findings. MPW wrote the drafts of the manuscript. XMX, YH and LJ commented on and helped revise drafts of the manuscript. All authors contributed to this article, and all have approved the final manuscript.

Abbreviations
AKI	=	acute kidney injury
MAKE	=	major adverse kidney events
OR	=	odds ratio
CI	=	confidence interval
MV	=	mechanical ventilation
ICU	=	intensive care unit
RRT	=	renal replace therapy
ADQI	=	Acute Disease Quality Initiative Workgroup
GCS	=	Glasgow coma scale
WBC	=	white blood cell
HCT	=	hematocrit
PLT	=	platelet
LAC	=	lactate
Scr	=	serum creatinine
FB	=	fluid balance
SIRS	=	systemic inflammatory response syndrome
SD	=	standard deviation
IQR	=	interquartile range

Acknowledgements

The authors gratefully acknowledge the National Science and Technology Supporting Plan of the Ministry of Science and Technology of the People’s Republic of China and the Beijing Municipal Science & Technology Commission, a government fund used to improve health-care quality and data collection, for providing financial support. The authors also acknowledge all the following members of the CCCST workgroup who contributed data and samples and have enabled this study to be conducted: Bin Du, Li Weng, Medical Intensive Care Unit, Peking Union Medical College Hospital, Beijing, China; Tong Li, Department of Critical Care Medicine, Beijing Tongren Hospital, Capital Medical University, Beijing, China; Meili Duan, Department of Critical Care Medicine, Beijing Friendship Hospital, Capital Medical University, Beijing China; Wenxiong Li, Surgical Intensive Care Unit, Beijing Chaoyang Hospital, Capital Medical University, Beijing, China; Bing Sun, Department of Respiratory and Critical Care Medicine, Beijing Institute of Respiratory Medicine, Beijing Chaoyang Hospital, Capital Medical University, Beijing, China; Jianxin Zhou, Department of Critical Care Medicine, Beijing Tiantan Hospital, Capital Medical University, Beijing, China; Jianguo Jia, Surgical Intensive Care Unit, Xuanwu Hospital, Capital Medical University, Beijing, China; Xi Zhu, Department of Critical Care Medicine, Peking University Third Hospital, Beijing, China; Qingyuan Zhan, Department of Critical Care Medicine, China-Japan Friendship Hospital, Beijing, China; Xiaochun Ma, Department of Critical Care Medicine, The First Affiliated Hospital of China Medical University, Shenyang, China; Tiehe Qin and Shouhong Wang, Department of Critical Care Medicine, Guangdong Geriatric Institute, Guangdong General Hospital, Guangdong, China; Yuhang Ai, Department of Critical Care Medicine, Xiangya Hospital, Central South University, Changsha, China; Yan Kang and Xuelian Liao, Department of Critical Care Medicine, West China Hospital, Sichuan University, Sichuan, China; Xiangyuan Cao, Department of Critical Care Medicine, General Hospital of Ningxia Medical University, Ningxia, China; Yushan Wang, Intensive Care Unit, The First Hospital of Jilin University, Changchun, China; and Duming Zhu, Surgical Intensive Care Unit, Department of Anaesthesiology, ZhongShan Hospital, FuDan University, Shanghai, China.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The datasets generated and or analyzed during the current study are available from the corresponding author on reasonable request.

Additional information

Funding

This study was supported by the National Science and Technology Supporting Plan of the Ministry of Science and Technology of the People’s Republic of China [2012BAI11B05], Beijing Municipal Science & Technology Commission [D101100050010058]. The funder had no role in the design and conduct of the study; management, analysis and interpretation of the data; preparation of the manuscript or the decision to submit it for publication.