In this cohort of 150 critically ill patients with COVID-19, about half (46.7%) patients developed AKI during hospitalization, and most of whom (62.9%) were severe AKI. Patients with severe AKI had extremely high in-hospital mortality (79.5%). After adjustment for confounders, severe AKI was independently associated with in-hospital mortality. In addition, patients with high a level of serum cytokines were more likely to develop AKI during hospitalization.
AKI occurred frequently in critically ill patients with COVID-19. However, previous studies reported that the incidence varies from 8.9% ~ 28.8% [12–14], which was much lower than 46.7% reported in our study. This variation probably results from differences in the case number, illness severity, and coexisting conditions. Another major reason was that we retrospectively recorded the AKI occurrence during the entire hospitalization, and the previous studies were cross-sectional  or had short follow-up duration [12, 13].Our result demonstrated that the median of hospital length of stay was 19 days and the median duration from admission to AKI occurrence was 11 days, which suggested that some AKI cases were not able to be detected in the early stage of hospitalization. In addition, it is reported that the incidence of AKI in critically ill patients was 58.0% in MERS  and 61% in septic shock patients . Hence, the actual incidence might be higher considering that some AKI might be missed in our study. In accordance with the recent reports on the outcomes of COVID-19 patients, 5 ~ 26% patients will be admitted to the intensive care unit during hospitalization [4–6]. Given that the number of COVID-19 cases is enormous and growing rapidly, there would be a substantial burden of medical care from AKI during the treatment of COVID-19. Therefore, plans should be made at local and regional levels for how to prepare and manage the potential surge in the need for critical care resources for AKI, including renal replacement treatment.
In our study, patients with AKI demonstrated an extremely poor prognosis. Of note, the mortality of severe AKI was 79.5%, which was much higher than those who developed ARDS . Likewise, the reported mortality of AKI in patients with SARS was up to 91.7% . One possible explanation is the development of AKI may indicate the involvement of multiple organs dysfunction. In addition, AKI may worsen lung injury via several mechanisms including impaired fluid excretion, direct capillary endothelial injury, and exacerbating inflammatory response . Hence, in order to detect the AKI timely and reduce the deaths, closer monitoring of serum creatinine and urine output is needed in the management of critically ill patients with COVID-19. Moreover, our findings showed that only severe AKI but not mild AKI was associated with in-hospital mortality. Therefore, a clear escalation plan on individualized treatment should be made depending on the severity of AKI.
The increased cytokines might be a predictor of AKI development in critically ill patients with COVID-19. Accumulating evidence suggests that cytokine release syndrome plays a role in severe COVID-19 . It is reported that an inflammation-related cytokine profile, characterized by IL2, IL2R, IL-8, IL-10 and tumor necrosis factor-α (TNFα), were associated with COVID-19 disease severity [21, 22]. Our study was the first to confirm the high level of IL6 and IL10 were associated with high risk of AKI development in critically ill patients with COVID-19. In addition, we found that cytokines were increased remarkably when AKI occurred. Although kidney tissues from postmortems showed that AKI might be attributed to reduced renal perfusion and widespread tubular necrosis , our results suggested more complex and subtle mechanisms of immune-mediated microvascular and tubular dysfunction was involved in AKI in critically ill patients with COVID-19. Moreover, timely inhibiting the excessive inflammatory response in its early stage through such means as immunomodulators and cytokine antagonists, as well as continuous renal replacement therapy, might be breakthroughs in the treatment of critically ill patients with COVID-19.
Some conventional risk factors for AKI in critically ill patients, including age and severity of disease assessed by SOFA score, were not identified in our study. One possible explanation is that Tongji Hospital was one of the designated hospitals for COVID-19 patients and the study population was relatively homogeneous with respect to age and disease severity. Most patients included in our study were the elderly, with median age of 70. And almost all patients were admitted to intensive care unit due to the respiratory failure, thus, the SOFA score on admission was lower than studies included patients with different causes and multiple organs dysfunction . In addition, our study was performed in single center and the generalizability of the study is limited by the small sample size. Further study in a larger cohort is needed to gain a better understanding of risk factors for AKI in critically ill COVID-19 patients.
We also summarized the medications during the hospitalization in critically ill patients with COVID-19. Our result showed that most patients have been treated with antibiotics and antivirus, which might induce and worsen AKI. Thus, the appropriate dosage of antibiotics and antivirus should be considered on basis of kidney function in clinical practice. In addition, we found that patients with AKI had high proportion of treatment with diuretics, which are often frequently applied to control hypervolemia and prevent pulmonary edema in critically ill patients . Previous studies showed the use of diuretics in critically ill patients with acute kidney injury was associated with an increased risk of death and nonrecovery of kidney function . Therefore, diuretics should be used with caution in management of critically ill patients with COVID-19. However, due to the small number of patients and the bias in different therapy of COVID-19 patients, causal relationship between medications and AKI in critically ill COVID-19 patients remains undetermined.
Our study has several limitations. First, an accurate baseline serum creatinine and urine output was not available, which may have led to an underestimation of AKI or erroneous associations. Second, our study has a retrospective observational design with its inherent biases and potential for unmeasured confounding variables. Finally, this investigation was conducted in a medical intensive care unit of a tertiary referral center, potentially limiting the generalizability of our study results.