This was a retrospective descriptive case/control monocentric study carried out in the medical ICU of Bizerte hospital (a tertiary teaching hospital in north of Tunisia) over a period of 18 months (September 2020-February 2022). This medical ICU is managed by medical intensivists with a novel unit of six beds created for the COVID-19 outbreak.

The primary endpoint was the incidence of AKI complicating critical forms of COVID-19. The second endpoint was ICU mortality.

All adult patients (>18 years) admitted to the ICU for critical forms of COVID-19 during the study period were included. Patients with a history of chronic kidney disease (CKD) were excluded in order to have a homogeneous group and avoid confounding factors. Those who did not meet the critical COVID-19 criteria were also excluded. Laboratory-confirmation of COVID-19 diagnosis was performed by detection of the SARS-CoV-2 RNA in nasal swabs using reverse transcription-polymerase chain reaction. Patients were divided in two groups: the case group which included the critical COVID-19 patients who developed AKI during their ICU stay according to the Kidney Disease: Improving Global Outcomes (KDIGO) classification: AKI patients, and the control group which included those who didn’t develop AKI during their ICU stay according to the same classification: No AKI patients.

All patients included had at least one creatinine measurement on ICU admission and one or more prior measurement in the department from which they were transferred.

-

Critical form of COVID-19 was considered in all included patients as defined by the WHO: “criteria for acute respiratory distress syndrome (ARDS), sepsis, septic shock, or other conditions that would normally require the provision of life-sustaining therapies such as mechanical ventilation (invasive or non-invasive) or vasopressor therapy”. ¹

-

Oxygen support was: noninvasive including noninvasive ventilation (NIV) and high-flow nasal cannula (HFNC); or invasive for patients requiring mechanical ventilation (MV).

We focused for each patient on demographic (age and gender) and clinical features (comorbidities, initial pleuropulmonary, cardiovascular and neurological examinations data), initial laboratory findings (arterial blood gases, renal function tests, complete blood count, CRP levels, prothrombin time and CPK), initial thoracic computed tomography (CT) scan data, drugs received prior to AKI onset, respiratory support, renal function during hospitalization, need for RRT, ICU length of stay (LOS) and mortality.

The classification of the French “Société d’Imagerie Thoracique” was used to assess lesions extension. It’s based on visual assessment of parenchymal extension. Five stages were considered according to the percentage of lung affected: absent or minimal involvement (<10%), moderate (10-25%), extensive (25-50%), severe (50-75%) or critical (>75%). ¹⁴

Free open Jamovi software was used for data collection and analysis. ¹⁵ For the descriptive study, we calculated means with standard deviations for quantitative variables with a Gaussian distribution and medians with interquartile range for variables with a non-Gaussian distribution. These variables were compared with a nonparametric Mann-Whitney test. We calculated counts and percentages for qualitative variables. Percentages were compared with Pearson’s chi-square test and with Fisher’s exact test, if this test was invalid. For analytic study; univariate logistic regression model then multivariate logistic regression analysis was done to assess AKI risk factors. The receiver operating characteristic (ROC) curve was used to ascertain the cut-off values of the continuous data, which were subsequently converted into dichotomous form. A comparison survival curve was obtained by means of the Log Rank test, and a multivariable Cox regression model was employed to identify independent factors associated with mortality.

In all statistical tests, the significance threshold was set at 0.05.

The Ethics Committee of our hospital (Habib Bougatfa hospital of Bizerte Tunisia) approved the study on July 20, 2023 (Approval number 1/2023) and waived informed consent because of the retrospective and descriptive design of the study. The principles outlined in the Declaration of Helsinki were followed in the protocol study.

With the aim of carrying out this work by the end of 2022, we called all surviving patients and relatives of deceased ones who met the inclusion criteria to obtain their consent to use their data anonymously and confidentially. Unfortunately, we were unable to reach all of them. We therefore obtained consent from 31 surviving patients (51 survivors in total) and consent from 44 suitable legal guardians of deceased patients (93 deceased in total). As we were unable to obtain consent from a significant number of the patients we wished to include, we referred this problem to our hospital's ethics committee. As this was a retrospective, observational study, and it was impossible to contact all the patients or their relatives, the ethics committee members waived informed consent for those we could not reach, and we obtained their agreement to carry out this study.

Among 160 patients who were admitted to the ICU in the study period, 16 didn’t meet the inclusion criteria. Thus, overall, 144 patients were included. Seventy-eight (54%) were transferred from COVID units, 42 (29%) from the emergency department and 24 (17%) from other medical or chirurgical units. Forty-one (28%) patients developed AKI ( Figure 1).

Table 1 shows the characteristics and the evolution of all patients and both groups: AKI and No AKI patients. We have summarized the epidemiological and clinical features, in addition to the laboratory and CT scan findings at ICU admission. Predisposing conditions to AKI, therapeutics and evolution were also assessed. In fact, AKI patients were older and had more comorbidities (notably diabetes and hypertension). Their heart rate, mean arterial pressure (MAP) and severity scores on admission were also higher compared to No AKI patients. Initial laboratory findings showed higher levels of white blood cells count (WBC) and C reactive protein (CRP). In addition, their baseline serum urea and creatinine rates on admission were higher. Nephrotoxic antibiotics, shock and MV requirement were the main predisposing conditions to AKI.

According to the KDIGO criteria the AKI patients (41 cases) were staged class 1 (5 cases: 12%), class 2 (16 cases: 39%) or class 3 (20 cases: 49%). AKI occurred within a median of 4 days (3, 8.5) and extremes between 1 and 32 days. The mean creatinine level at the onset of the AKI was 285±185 μmol/L (extremes between 106 and 955 μmol/L). Thirteen patients (32%) underwent RRT. Renal function improved in seven cases (17%). As shown in Table 1: age, diabetes, hypertension, APACHE II, SAPS II, heart rate, MAP, serum baseline urea and creatinine, WBC count, CRP, shock, MV and nephrotoxic antibiotics were all predictors of AKI in univariate analysis. However, diabetes, nephrotoxic antibiotics, and shock were the three independent risk factors in the multivariate analysis ( Table 2).

Mean ICU length of stay (LOS) was longer in AKI patients without a significant difference (p: 0.454) but mortality was significantly higher (88% versus 55%, p< 10 ⁻³) ( Table 1). Only five patients of the AKI group survived (three were classified KDIGO 1 and two KDIGO 2). All AKI KDIGO 3 patients had fatal outcome.

Figure 2 shows the cumulative survival rates of the total population and the two groups according to the occurrence of AKI.

In univariate analysis, age > 53 years, severity scores, CT scan lesion extension > 67.5%, septic shock complicating hospital acquired infection (HAI), AKI, MV were all predictive of poor outcome. Besides, age > 53, septic shock complicating HAI and MV were the three independent factors of mortality ( Table 3).

In this study among the 144 patients enrolled, 41 (28%) developed AKI during ICU-hospitalization within a median of 4 days (3, 8.5). It was staged KDIGO 3 in about half of the cases. Thirteen patients underwent RRT and renal function improved in only seven cases. Diabetes, nephrotoxic antibiotics and shock were the three independent risk factors of AKI. Mortality was significantly higher in AKI group, but AKI didn’t appear to be an independent risk factor of poor outcome in multivariate analysis.

In patients undergoing conventional hospitalization, the incidence of AKI ranged from 0.5% to 5,3%. ^{3 , 16 – 18}

The prevalence of AKI increases in parallel with the COVID-19 severity. In the study by Hu et al, AKI occurred in 1.3% (2 of 151), 3.4% (5 of 146), and 38.5% (10 of 26) of non-severe patients, severe, and critical patients respectively. ¹⁹ Similar findings were reported by Zheng et al, who found an incidence of AKI of 1.0% (3 of 297), 6.8% (13 of 190), and 39.4% (13 of 33) in non-severe, severe, and critical patients, respectively. ²⁰ In a systematic review and meta-analysis of 58 studies focused in AKI and RRT in COVID-19 patients, 13 studies reported on AKI incidence among critical patients. Overall, AKI occurred in 312/565 ICU patients with a pooled incidence rate of 39.0%. ²¹

There is also a difference in the prevalence of AKI depending on the patients’ geographical distribution. Data from Chinese studies estimated the AKI prevalence between 8.3% and 50.6% in ICU COVID-19 patients. ^{16 , 22 – 25} More recent studies, from the United States, have found a higher prevalence ranging from 19% to 76%. ^{26 – 29} A total of 61/215 (28.4%) patients admitted to a Sub-Saharan African ICU developed AKI. ³⁰ This rate seems to be more important in European ICUs reaching levels above 80%. ^{4 , 5}

AKI is also variable in severity. KDIGO is the most commonly used classification, and the kidney damage was staged KDIGO 1, 2 and 3 in 25-39%; 3.5-35% and 30-63% respectively in several previous series. ^{25 , 27 , 28 , 30} AKI is usually diagnosed within 5 to 9 days of hospital admission and a median of 12 to 21 days after the onset of symptoms. ^{23 , 25 , 31} However, Hirsch et al. reported a high frequency of AKI occurrence (37%) within 24 hours of admission. ²⁸ Depending on the study, the use of RRT in ICU is variable from 16% to 73% of patients with AKI. ^{4 , 5 , 16 , 23 , 25 , 28 – 30}

These discrepancies between studies concerning the incidence of AKI, its severity, its time of onset and the use of RRT could be explained by: variation of the definition of “severe” disease and AKI, heterogeneity of the studied populations, genetic predisposition to kidney involvement and RRT resource limitations.

The incidence of AKI was 28% in our study, which is a low rate compared to previous series. This may be explained by the fact that all patients included didn’t have a history of CKD. Moreover, as this population had critical clinical presentation with several AKI risk factors, AKI was rather severe (only 12% were classified KDIGO 1).

-Demographic risk factors

In our study AKI patients were older than no AKI ones with a significant difference in univariate analysis, however age was not considered as an independent predictor of AKI in multivariate analysis. Older age was considered as a risk factor for AKI and RRT in an Italian cohort of 99 invasively ventilated COVID-19 patients. ³² Likewise, in a large Chinese study by Hirsh et al including 5449 COVID-19 patients, 1993 (36.6%) developed AKI and older age was an independent predictor of AKI (OR: 1.03; 95% CI: (1.03–1.04); p<0.001). ²⁸ Similar findings were reported by Dereli et al. ²

Lin et al analyzed the data of 79 research articles: 8 studies investigated the risk factors of COVID-19 induced AKI and also showed that age ≥ 60 years and severe infection were independent factors predicting AKI with ORs: 3.53 (95% CI: (2.92-4.5); p<0.001), and 6.07 (95% CI (2.53-14.58); p<0.001) respectively. ³³

While male gender was much more associated with AKI, as reported by Hirsh JS et al. ²⁸ and Ng JH et al., ³⁴ sex ratio was comparable in our cohort and other previous studies. ^{2 , 4 , 32}

-Comorbidities

Most of the critical COVID-19 patients have pre-existing comorbidities which were also associated to AKI. The most common are hypertension and other cardiovascular disorders, diabetes and obesity. Diabetes was an independent factor in our study as well as in several series. ^{28 , 34} Hypertension was also significantly much more frequent in AKI patients in our study as well as in previous studies. ^{2 , 28} In addition, cardiomyopathy, chronic respiratory failure and BMI were also reported as risk factors of AKI. ^{2 , 28} According to these findings, in a recent meta-analysis of forty-four studies with a total number of 114 COVID-19 patients with AKI, Sabaghian et al found that factors including older age, hypertension, cardiovascular disease, diabetes, high BMI, chronic kidney disease, immunosuppression, and smoking are the potential risk factors of AKI. ⁷

These comorbidities are well-known factors of renal vulnerability causing histological lesions of nephroangiosclerosis or diabetic glomerulosclerosis which are considered as underlying renal fragility factors in COVID-19 patients. ^{35 – 38} Moreover, due to these conditions, patients are frequently treated with drugs that interfere with regulation of renal flow, such as ACE inhibitors. ⁹ Besides, AKI patients had higher baseline serum creatinine with a significant difference in our cohort and similar findings were reported in several studies. ^{17 , 28 , 34 , 39} This could be explained by the premorbid kidney disease potentially related to the frequent comorbidities especially diabetes and hypertension.

-Acute disease severity and therapeutics

In addition to these non-modifiable demographic factors, the severity of the COVID-19 on admission was the major predictor of AKI. In fact, severity scores were significantly higher in the AKI patients in our study and in several previous series. ^{2 , 4} In addition, ARDS requiring MV, shock and vasopressor support were reported as predictive of AKI. ^{2 , 4 , 28 , 40 , 41}

Since AKI patients had more serious forms of COVID-19, they require much more MV which was predictive of AKI in our univariate analysis but was not considered an independent factor. In fact, critical COVID-19 patients are at a high risk of AKI as a complication of MV. Specifically, high positive end-expiratory pressure used for COVID-19 associated ARDS leads to increased intrathoracic pressure and can ultimately result in increased renal venous pressure and reduced filtration. ⁴² Besides, positive pressure ventilation can increase sympathetic tone, leading to secondary activation of the renin–angiotensin system. ⁴³ Furthermore, upregulation of proinflammatory mediators associated to biotrauma, may subsequently induce multiple system organ failure including the kidney. the kidney-lung crosstalk theory is due to the increased release of cytokines in the blood, which is promoted by lung injury. Elevated levels of cytokines, especially IL-6, increase alveolar-capillary permeability and pulmonary hemorrhage. It even may lead to distant organs dysfunction, notably damage of the kidney vascular endothelium. ⁴⁴

Moreover, restrictive fluid strategy recommended for ARDS patients, who may initially present with relative volume depletion due to fever and gastrointestinal losses, may worsen hypovolemia and compromise renal perfusion. ⁴⁵ Thus, hypovolemia and hemodynamic instability cause renal hypoperfusion and, consequently, AKI. Moreover, shock is associated to lactic acidosis, hyperkalemia and rhabdomyolysis which all had a negative impact on kidney function. ⁴⁵ Therefore, careful attention to volume status is needed to avoid AKI.

Beyond shock and diabetes, nephrotoxic antibiotics use was also an independent factor of AKI in our study. In fact, critical COVID-19 patients might be exposed to nephrotoxins as part of their clinical care, in particular, antibiotics, which can result in tubular injury or acute interstitial nephritis.

In a large Chinese study including 210 ICU COVID-19 patients, Sang et al proved that the use of nephrotoxic drug was an independent factor of AKI (OR: 2.67; 95% CI: (1.09–6.55); p: 0.0316). ⁴⁶

Similarly, a Portuguese study including 192 COVID-19 patients (20% of whom needed ICU management), confirmed that the exposure to nephrotoxins during the first week of admission (vancomycin, aminoglycosides, nonsteroidal anti-inflammatory drug and iodine contrast agents) was an independent factor of AKI (OR 3.60 95% CI (1.30–9.94) p=0.014). ³⁹

In a most recent study carried in Argentina including 162 ICU COVID-19 patients, exposure to nephrotoxic drugs (particularly polymyxins and aminoglycosides) was markedly higher in the AKI group (p<0.001). ⁴⁰

The use of iodine contrast agents was not considered as an AKI risk factor in our cohort. This could be explained by the fact that, on the one hand, all patients included didn’t have previous CKD and on the other, they received hydro-electrolytic supplements according to the daily fluid balance calculated by subtracting the total fluid output from the total intake.

Mean ICU LOS was longer in AKI patients without a significant difference but mortality rate was significantly higher in this group and all patients staged KDIGO 3 deceased.

In univariate analysis AKI was a poor prognostic factor but only age >53 years, septic shock complicating HAI and MV were the three independent factors of mortality.

Mortality was also significantly higher in AKI patients in the most reported studies and it increases in parallel with the AKI severity. ^{30 , 39 – 41}

In fact, Nlandu Y et al. found that the death rate of AKI patients was more than 2 time higher than all patients. Besides, this rate was more than 3 times higher, in patients requiring RRT than those classified AKI stage 1. Thus, AKI was an independent prognostic factor in this study (OR: 2.96; 95% CI (1.23-4.65); p: 0.013). ³⁰

Likewise, AKI stage 3 (OR: 5.33; 95% CI (1.15-24.65); p: 0.0321) was independently associated with death in the study by Sang L et al. in addition to critical disease (OR: 69.16; 95% CI (5.86-815.79); p: 0.0008), older age (OR: 1.06; 95% CI (1.02-1.11); p:0.0035) and P/F < 150 (OR: 15.21; 95% CI (4.72-49.07); p<10 ⁻³). ⁴⁶

Beyond older age (OR: 1.07; 95% CI (1.02–1.11); p: 0.004), lower Hb level (OR: 0.78; 95% CI (0.60–0.98); p: 0.035), persistent AKI (OR: 7.34; 95% CI (2.37–22.72); p: 0.001) and severe AKI (OR: 2.65 per increase in KDIGO stage; 95% CI (1.32–5.33); p: 0.006) were also considered independent factors of mortality in the study by Gameiro et al. ³⁹

Thus, most of the studies agree on the negative prognostic impact of AKI on critical COVID-19 patients and this is not surprising. In fact, as AKI most often occurs in elderly patients with multiple comorbidities, severe forms of COVID-19, and requiring life-sustaining therapies (particularly MV and vasopressor therapy), they are expected to have a poor prognosis. Although, our results showed that AKI was associated to mortality in univariate analysis, it wasn’t considered as an independent factor in multivariate analysis. This could be due to the fact that some factors were mutually dependent as shock, MV and AKI.

This study is one of the few works that have focused on the AKI in critical forms of COVID-19 managed in the ICU with a large number of patients which represent its strength. Although some limitations must be noted. The retrospective design of our study was constrained due to the paucity of data on the previous treatments of patients enrolled, notably, prior use of angiotensin converting enzyme inhibitor or angiotensin II receptor blocker. In addition, some laboratory tests were lacking in our hospital such us ferritin and D-dimers. Thus, these missing data considered as a risk factor for AKI in several studies could not be evaluated in our patients.