This was a single-center prospective longitudinal study which is a part of a registered study (Clinicaltrials.gov NCT02975908). We consecutively included all adult patients (>18 years old) who met the ARDS Berlin definition for the first incidence of ARDS in the medical ICU in China-Japan Friendship Hospital from March 1, 2016, to September 30, 2020 (1). The exclusion criteria were as follows: chronic respiratory failure due to chronic respiratory diseases, such as chronic obstructive pulmonary disease; bronchiectasis or lung fibrosis; AKI prior to the onset of ARDS; and inability or unwillingness to provide informed consent (Figure 1). The study protocol was approved by the China-Japan Friendship Hospital ethics committee. China-Japan Friendship Hospital is the national clinical research center for respiratory disease in China, so most ARDS patients in the MICU are intrapulmonary ARDS. The respiratory support our center can provide includes High-flow Nasal Cannula (HFNC), non-invasive positive pressure ventilation (NPPV), invasive positive pressure ventilation (IPPV) and extracorporeal membrane oxygenation (ECMO) as well as continuous renal replacement therapy (CRRT) renal support. The clinician decided the methods of ventilation and when to start the renal support according to the guidelines (15, 16). Epidemiological, demographic, clinical, laboratory, treatment and outcome data were extracted from electronic medical records using the specialized case report form by the investigators. The data for this study were a post hoc analysis derived from this registry database.

The clinical diagnosis of ARDS, according to the Berlin ARDS definition, includes the following: (1) the presence of acute hypoxemic respiratory failure; (2) onset of respiratory symptoms within 1 week or the presence of new (within 7 days) or worsening respiratory symptoms; (3) bilateral opacities on chest X-ray or computed tomography not fully explained by effusions, lobar or lung collapse, or nodules; and (4) acute hypoxemic respiratory failure not caused by cardiac failure or fluid overload. All ICU patients who had acute hypoxemia with PaO₂/FIO₂ (P/F) ≤ 300 mmHg were screened for ARDS. Patients with ARDS were categorized on the day of ARDS diagnosis based on their PaO2/FIO2 ratios into mild (200 < PaO₂/FIO₂ ≤ 300 mmHg), moderate (100 < PaO2/FIO2 ≤ 200 mmHg), and severe (PaO2/FIO2 ≤ 100 mmHg) according to the Berlin definition. AKI was diagnosed and graded according to the Kidney Disease Improving Global Outcomes (KDIGO) 2012 guidelines using serum creatinine (SCr) and urine output criteria (16). Patient baseline SCr values were obtained between 7 and 365 days before hospitalization. However, if the measured baseline SCr was unavailable, an estimated baseline creatinine, according to the Modification of Diet in Renal Disease (MDRD) formula assuming a GFR of 75 ml/min, was used (17, 18). If a patient had a reported history of CKD, their estimated SCr was back-calculated with the MDRD formula assuming a GFR of 60 ml/min. Renal recovery was defined as alive, free of renal replacement treatment (RRT), improvement in SCr < 1.5 times the baseline SCr, and urine output > 0.5 mL/kg/h for more than 6 h (19). Early AKI was defined as AKI occurring within 48 h of ARDS diagnosis. Patients who developed AKI later than 48 h after ARDS diagnosis were classified into the late AKI group (20). The observation period of this study was from admission to discharge.

Day 1 was defined as the first day that the ARDS criteria were met after ICU admission. Each patient with ARDS was followed up to 28 days after diagnosis, until hospital discharge, or death. The risk factors for ARDS included intrapulmonary (pneumonia, aspiration, and drowning pulmonary contusion) and extrapulmonary factors (trauma, extrapulmonary sepsis, non-cardiogenic shock and blood transfusion) (21). The acute physiology and chronic health evaluation (APACHE) II score and sequential organ failure assessment (SOFA) score were recorded using data from the first 24 h in the ICU. SOFA scores ≥ 3 referring to one organ were defined as failure of the respective organ. Shock was defined as arterial systolic blood pressure less than 90 mmHg, mean arterial pressure less than 60 mmHg, or the use of norepinephrine at any dosage to maintain systolic blood pressure at 90 mmHg or higher or mean arterial pressure at 60 mmHg or higher. Barotrauma refers to pneumothorax, pneumomediastinum, subcutaneous emphysema, pneumopericardium. Invasive ventilator-free days were calculated as the number of days from weaning from invasive ventilation to day 28. Patients who died before weaning were considered to have a ventilator-free-day value of 0. The immunosuppressed state included the following: corticosteroid or immunosuppressant use within 90 days, seropositive for human immunodeficiency virus, malignant tumor, and patients receiving radiation therapy or chemotherapy for an underlying malignancy within 90 days (22). The respiratory support method recorded in this paper is the maximal respiratory support method for the patient during hospitalization. Mechanical ventilation parameters, arterial blood gas, and fluid balance were collected for the first 14 days and on Day 21 and Day 28 after the onset of ARDS. Laboratory examination (blood examinations, coagulation profiles and BNP or NT-proBNP) was recorded for the first 4 days and on Day 7. Serum creatinine, urine volume, and the use of RRT were recorded daily until 28 days after ARDS diagnosis or hospital discharge to determine the highest stage of AKI. Fluid input included all fluids infused by intravenous or enteral routes. Fluid output included urine output, volume of fecal matter and fluid loss from drains. Fluid balance was calculated as the difference between fluid input and fluid output. All data were recorded as close as possible to 8 AM each day. Exposure to nephrotoxic agents included vancomycin, aminoglycoside, sulfamethoxazole-trimethoprim, colistin, amphotericin B, voriconazole, angiotensin converting enzyme inhibitors/angiotensin receptor blockers (ACEIs/ARBs), diuretics, cyclosporine and non-steroid anti-inflammatory drugs (NSAIDs) before AKI was diagnosed.

Categorical variables were analyzed using the χ² test or Fisher’s exact test, and continuous variables were analyzed using Student’s t-test or the Mann–Whitney U test. After testing the distribution of continuous variables, normally distributed variables were presented as the mean ± SD, and non-normally distributed variables were presented as the median (interquartile range). Multivariate analysis was performed to evaluate potential explanatory variables (i.e., variables found to be both clinically significant and statistically significant in univariate analyses) for AKI and clinical outcomes, which were expressed as the odds ratio (OR) and its 95% confidence interval (CI). Survival was assessed by the Kaplan–Meier method and compared between groups using the log-rank test. The hazard ratios of variables related to 28-day mortality were estimated with a Cox proportional hazards regression model. A two-sided p value less than 0.05 was considered statistically significant. SPSS 24.0 (SPSS, Chicago, IL, USA) was used for all statistical analyses (22).

During the observation period, a total of 311 patients were diagnosed with ARDS and met the inclusion criteria. Of these patients, the mean age was 56.6 years, and 216 (69.5%) were male. One hundred twenty-two (39.2%) patients had a history of hypertension, 68 (21.9%) had diabetes mellitus, 32 (10.3%) had coronary heart disease, 26 (8.4%) had CKD, 23 (7.4%) had chronic respiratory disease, 3 (0.9%) had chronic liver disease, 28 (9%) had cerebrovascular disease, and 27 (8.7%) had connective tissue disease. Seven (2.3%) patients had a history of drinking, 124 (39.9%) patients had a history of smoking, 10 (3.2%) patients had surgery in the past 3 months, and 124 (39.9%) patients were immunosuppressed. AKI developed in 161 (51.8%) of the 311 patients with ARDS: 91 (56.5%) patients developed AKI within 48 h of ARDS diagnosis, whereas 70 (43.5%) developed AKI later than 48 h. Basal creatinine values were available for 28 patients, and the remaining 283 patient’s baseline creatinine were estimated by back calculation with the MDRD equation. There were no missing results for other parameters in the result section and also no lost-to-follow-up populations in this study.

Patients in the AKI group were significantly older (59.7 ± 16.5 vs. 53.2 ± 15.8 years, p = 0.00) and had a greater proportion of hypertension (44.7% vs. 33.3%, p = 0.04), diabetes mellitus (27.3% vs. 16%, p = 0.02), coronary heart disease (13.7% vs. 6.7%, p = 0.04) and CKD (14.3% vs. 2%, p = 0.00) than patients in the non-AKI group. Patients with AKI had a greater proportion of multiple comorbidities (Table 1). Table 2 shows the physiology and organ dysfunction parameters on admission, laboratory results and treatments. Admission APACHE II (21 vs. 16, p = 0.00) and SOFA scores (8 vs. 5, p = 0.00) were significantly higher in patients with AKI than in those without AKI. Shock was observed in 109 (67.7%) of 161 patients with AKI and in 48 (32%) of 150 patients without AKI (p = 0.00). There was no significant difference in the cause or severity of ARDS between the two groups. Regarding the laboratory indicators, patients with AKI had a higher prevalence of metabolic acidosis in the first 3 days and a higher level of lactic acid. In addition, AKI patients’ white blood cell and BNP/NT-proBNP levels were significantly higher than those of patients without AKI. However, there was no association between ventilator parameter settings, airway plateau pressure, peak airway pressure, airway resistance, compliance and the development of AKI among ARDS patients. The number of patients in the AKI group who needed NPPV was significantly lower than that in the non-AKI group (12 vs. 24, p = 0.02), whereas the number of patients who needed ECMO support in the AKI group was significantly higher than that in the non-AKI group (41 vs. 21, p = 0.01). AKI patients had more fluid accumulation than non-AKI patients on Day 1 (285 vs. 0, p = 0.00) and Day 6 (153 vs. –32, p = 0.03).

After adjustment in a multivariable analysis model, the development of AKI was independently associated with age [OR 1.027, 95% CI 1.009–1.045], a history of diabetes mellitus [OR 2.110, 95% CI 1.100–4.046], a history of CKD [OR 9.328, 95% CI 2.393–36.363], APACHE II score [OR 1.049, 95% CI 1.008–1.092], the average lactate level in the first 3 days [OR 1.965, 95% CI 1.287–3.020] and the use of ECMO support [OR 2.359, 95% CI 1.154–4.824].

Although the length of ICU stay did not differ between the patients with ARDS with and without AKI, the invasive ventilation-free days to Day 28 were significantly shorter. Moreover, the 28 days mortality and ICU mortality were apparently higher in patients with AKI (Table 3). Kaplan–Meier analysis revealed a significantly higher 28-day mortality rate for patients with AKI (Log-Rank chi-square = 21.313, p = 0.00, Figure 2). The hazard ratio for AKI in 28-day mortality of ARDS patients, estimated using a Cox proportional hazards regression model, was 1.488 (95% CI 1.006–2.201, p = 0.046).

Acute kidney injury was significantly more prevalent in patients with ECMO support. But the onset of AKI, CRRT and 28-day mortality was no different in ECMO patients (Table 4).

Overall, the in-hospital mortality rate for ARDS patients was 42.4%. However, the mortality rate was 57.1% in patients with ARDS with AKI. In univariate analysis, the patients who died with ARDS and AKI were older (62.9 vs. 55.5, p = 0.00), had a lower body mass index (23.9 vs. 25.4, p = 0.03) and were less likely to be hypertensive (35.9% vs. 56.9%, p = 0.01). The incidence of shock was significantly higher among deceased patients (93.5% vs. 33.3%, p = 0.00). Mild ARDS was more common in survivors (21.7% vs. 6.5%, p = 0.01). The P_aO₂/FiO₂ was higher (143 vs. 112.9, p = 0.00) and the FiO₂ (0.6 vs. 0.7, p = 0.00) was lower in this group. Due to the higher shock incidence, a lower PaO₂/FiO₂ and more severe ARDS status, the deceased patients had significantly higher cLac levels (1.8 vs. 1.6, p = 0.00). For AKI severity, the majority of deceased patients were in stage 2, whereas most survivors were in stage 3, and the baseline Cr was significantly higher in the survivor group (143.3 vs. 86.6, p = 0.00). Ventilator-specific variables and nephrotoxic drugs were not correlated with the mortality of ARDS combined with AKI patients. However, regarding fluid accumulation, the deceased patients had significantly greater fluid accumulation than the survivors, especially on D2, D3, D4, and D7 (Table 5).

After adjustment in a multivariable analysis model, the mortality of patients with ARDS combined with AKI was independently associated with shock (OR 54.943, 95% CI 9.751–309.573).

Early AKI was found in 91 (56.5%) patients, and late AKI was found in 70 (43.5%) patients. Patients in the early AKI group had more hypertension (54.9% vs. 31.4%, p = 0.00) and CKD (24.2% vs. 1.4%, p = 0.00) than those in the late AKI group. Compared with patients in the late AKI group, patients in the early AKI group were more severely ill when they were admitted to the ICU based on the higher APACHE II and SOFA scores. In addition, patients in the early AKI group had a higher prevalence of metabolic acidosis and higher BNP/NT-proBNP levels. Regarding the stage of AKI between the two groups, the majority of patients in the early AKI group were in stage 3 (n = 47, 51.6%), whereas half of the patients in the late AKI group were in stage 1 and stage 2 (n = 47, 67.1%). Therefore, the peak serum creatinine in the early-onset group was significantly higher than that in the late-onset group (259.8 vs. 197.3, p = 0.00), and the use of CRRT was significantly higher in the early-onset group (30.8% vs. 14.3%, p = 0.02). The use of nephrotoxic agents, especially aminoglycosides (17.1% vs. 3.3%, p = 0.01), colistin (17.1% vs. 1.1%, p = 0.00), amphotericin B (20% vs. 5.5%, p = 0.01), NSAIDs (17.1% vs. 3.3%, p = 0.01), diuretics (44.3% vs. 18.7%, p = 0.00) and voriconazole (34.3% vs. 12.1%, p = 0.00), was significantly associated with the development of late-onset AKI. The fluid accumulation on Day 7 was significantly higher in the late-onset group than in the early-onset group. Regarding the prognostic indicators, the number of invasive ventilation-free days up to Day 28 in the early-onset group was significantly longer than that in the late-onset group, and the mortality rate was significantly lower in the early-onset group (Table 6).

After adjusting for confounding factors, history of hypertension (p = 0.01), history of CKD (p = 0.03), and the SOFA score upon admission (p = 0.02) were significantly higher in the early-onset group than in the late-onset group. However, the use of nephrotoxic agents, especially colistin (p = 0.01), NSAIDs (p = 0.04) and diuretics (p = 0.03), was significantly associated with the development of late-onset AKI.

In the subgroup analysis, both the ICU length of stay and hospital length of stay were shorter in the early AKI group, but the difference was not statistically significant. The number of invasive ventilation-free days to Day 28 was significantly longer and the mortality rate was lower in the early AKI group than in the late AKI group. However, in Kaplan–Meier analysis, the mortality rate between the two groups was not significantly different (p = 0.39).