Epidemiology of acute kidney injury in intensive care units in Beijing: the multi-center BAKIT study

Acute kidney injury (AKI) is a life-threatening disease and global health burdens with increasing incidence in both developed and developing countries [1, 2]. AKI commonly occurs in the intensive care unit (ICU), and is caused by multiple risk factors, leading to adverse clinical outcomes, increasing costs, and the development of chronic kidney disease (CKD) [3‐8]. The definition of AKI has evolved from the Risk, Injury, Failure, Loss, End-stage (RIFLE) criteria and the AKI Network (AKIN) classification to the Kidney Disease Improving Global Outcomes (KDIGO) classification [9‐11]. A multinational epidemiological study using KDIGO criteria showed that the incidence of AKI in the ICU was 57.3% [12], which implies concern regarding AKI the in ICU globally with extremely high morbidity reported. It is essential for physicians, researchers, and health policy makers to establish an accurate incidence of AKI [12]. There have been were large epidemiological studies of patients hospitalized with AKI in the Chinese population [13, 14]. However, there have been limited epidemiological data on AKI in the ICU in mainland China reported [15]. Therefore we performed a cohort study of all the adult patients in 30 ICUs of 28 hospitals in Beijing to determine the incidence, risk factors, renal replacement therapy (RRT) practice, and the outcome of patients with AKI.

A total of 3107 participants were included in the final analysis among the 9049 patients admitted to the participating ICUs during the study period. The flow chart is presented in Fig. 1. In comparison of the characteristics between patients with and without AKI, patients with AKI have a higher median age, baseline serum creatinine, APACHE II score, SAPS II score, SOFA score, non-renal SOFA score, more comorbidity, a higher rate of mechanical ventilation, use of inotropic agents and diuretics, and WH/WD. (Table 1).

Our results indicated a high incidence of AKI in the ICU. Approximately half of our ICU patients experienced AKI. The incidence in the present study was significantly higher than those in an international epidemiological study in 2005 [20] and the FINNAKI study [21], which might be attributed to the discrepancy in diagnostic criteria. A study comparing the three diagnostic criteria indicated that the KDIGO criteria identified more patients with AKI and were more predictive of short-term mortality [22]. The incidence in the present study was similar to the Acute Kidney Injury- Epidemiological Prospective Investigation (AKI-EPI) study [12]. The AKI-EPI study [12] was the first multinational epidemiological study in the ICU using the KDIGO criteria. Both AKI-EPI and our study demonstrated an unsatisfactory prevention and increasing burden of AKI. In our study, the top three possible causes reported by physicians were hypovolemia, sepsis on ICU admission and low cardiac output, which was similar to the result of the FINNAKI study (preceding AKI severe sepsis, pre-ICU hypovolemia and pre-ICU hypotension) [21]. Our logistic regression showed comprehensible risk factors. The RRT ratio in our study was similar to that of the FINNAKI study (8.6% vs 9.4%) [21], but lower than that of the AKI-EPI study (13.5%) [12]. The higher RRT ratio in the AKI-EPI study could be attributable to the higher ratio (KDIGO stage 3: 30.0%) of patients with more severe AKI in comparison with the other two studies (KDIGO stage 3: 15.7%; AKIN stage 3: 14.1%). The RRT pattern results indicated that CRRT was mainly chosen (97.9%) and that IRRT was seldom used. A previous observational study in French ICUs showed that 40% of RRT patients had CRRT and 60% had IRRT [23], which varied greatly from our results. Although the previous studies have shown no difference in clinical outcomes comparing CRRT and IRRT in the ICU [24‐26], our results revealed a strong inclination to CRRT in our ICUs. The advantages of CRRT include better hemodynamic tolerance, accurate balance control and better clearance of the middle molecule [11, 27]. Besides our ICU physicians and nurses were familiar with CRRT and unfamiliar with IRRT. Our results showed that the femoral vein (74.6%) was the most common catheter location, differing from the recommendations in the guideline (the right internal jugular vein) [28]. Studies have suggested that the internal jugular vein might be preferable to the femoral vein to minimize dialysis catheter dysfunction and blood recirculation to improve RRT provision and reduce the risk of infection [29‐32]. The reasons for our physicians preferring femoral access might be the advantages of convenience, efficiency and fewer complications in the operating procedure. The results indicated that citrate was the most commonly used for anticoagulant. Meta-analyses [33‐35] suggested that citrate is preferable to heparin in anticoagulation for CRRT to prolong circuit life span, reduce the risk of bleeding [34, 35], and increase the delivered RRT dose [33]. Citrate is novel and has not been used for long in CRRT; nevertheless, our results indicated that it has already been widely used in Beijing. However, considering the advantages of better medical resources in Beijing, the capital city of China, the result may not be representative of the whole country. The distinction of guideline compliance with respect to vascular access and anticoagulation might imply that changes in medication are more acceptable for physicians than are changes in techniques. The mortality rate was comparable to the FINNAKI study [21]. The regression analysis indicated that a greater severity of AKI, and comorbidity of sepsis were associated with an increased risk of mortality, which was in accordance with the previous studies [12, 13, 21]. Our results showed a significantly higher ICU costs for RRT patients, which is reasonable and comparable to other studies [26, 35]. RRT patients had a higher AKI stage, more complications, higher disease severity score and longer ICU-LOS, all of which lead to higher costs [35]. Our results showed that increased AKI stage and septic AKI were associated with higher mortality, and those trends were also found in a study that evaluated patients with AKI with and without sepsis [36]. More than 1/4 of survivors were RRT dependent, which would lead to heavy burdens. There are strengths in our study. This is the first large, prospective, multicenter cohort study of AKI in the ICU in Beijing. We used the KDIGO criteria to evaluate the prevalence of AKI. Previous studies [22, 37, 38] indicated that the KDIGO criteria defined more patients with AKI in comparison with RIFLE and AKIN. Further analysis showed that the patients missed by RIFLE had higher mortality rate and longer hospital-LOS than the patients missed by KDIGO [22]. For the first time the incidence of AKI in the ICU for ten consecutive days was reported, to the best of our knowledge. The results revealed an obvious downtrend of AKI onset over time. We consider this result to be meaningful epidemiological data that might imply the necessity of key vigilance against the risk of AKI in the first 4 days after ICU admission. Furthermore, we investigated the comprehensive situation in RRT practice. Thus, we understood our insufficiencies and underlying causes, which are important areas for policy makers and physicians to make improvements. There were limitations in our study. First, our participating hospitals were all located in Beijing. As the capital city of China, Beijing enjoys better medical resources. Thus, the results of our study might not be representative of hospitals nationwide. Second, we used the MDRD equation to estimate the baseline serum creatinine for missing values, as recommended [18, 39]. However, the MDRD method may result in under- or overestimation of baseline creatinine [39, 40]. Third, we screened patients for AKI for the first 10 days after admission to the ICU. Thus, we were unable to analyze later-onset AKI. However, based on our results, the majority of patients with AKI (87.6%) had AKI onset in the first 4 days after admission. Our data on the occurrence of AKI during the first 10 days may imply that the onset of AKI in the ICU decreases over time, and new onset after 10 days would be minimal.

There was a high incidence of AKI in the ICU. Approximately half of our ICU patients experienced AKI. The majority of patients with AKI developed AKI during the first four ICU days. For RRT patterns, continuous RRT, predilution, citrate, and femoral vein were the most commonly used RRT procedure, dilution mode, anticoagulant and vascular access, respectively. AKI was associated with increased mortality and costs, incomplete kidney recovery and a series of adverse outcomes. Higher AKI stage, septic-AKI and the need for RRT were associated with increased mortality.