Difference in Risk Factors Associated with Acute Respiratory Distress Syndrome and Intubation/Death in Patients with Coronavirus Disease 2019: A Two-hospital Retrospective Cohort Study in Hubei, China

Background: The reviews on the risk factors with ARDS and the worse outcomes concluded lacking robust data of risk factors to prevent COVID-19 and identied an urgent need for large sample and high-quality research in this area, as well as the features of the ARDS. Methods: This retrospective cohort study included 333 COVID-19 inpatients at two hospitals in Hubei of China in 2020. The COVID-19-related ARDS was diagnosed according to the Berlin criteria. The outcomes were ARDS development and the intubation or in-hospital death. The cox proportional hazard ratio (HR) models were employed to determine the signicant risk factors. Results: The median number of days from symptom onset to ARDS diagnosis was 11.0 (IQR, 8.0–13.0). Up to 84.1% COVID-19-related ARDS patients demonstrated multiple organ injuries. The mortality rates were 41.9% and 85.7% in moderate and severe ARDS. The survival patients on invasive mechanical ventilation (IMV) had been intubated earlier since ARDS diagnosis than those who had not survived (5.5 median days, IQR 4.0-7.0 days versus 11.5 median days, IQR 6.0-14.0 days, P < 0.001). Males and all abnormal laboratory indices associated with the higher risk of ARDS (P<0.05) but were not linked with the risk of intubation or death (P>0.05). The sensitivity analyses found that lymphocyte count of < 1000 per mm3 at hospital admission were still signicantly associated with developing ARDS when adjusting for age and male gender (HR, 4.10; 95% CI, 2.40-7.10), and oxygenation index (OI) ratio < 150 were more likely to predict the intubation/death after age adjustment (HR, 2.50; 95% CI, 1.17-5.30). Conclusion: The SARS-CoV-2-caused ARDS was not the typical ARDS according to Berlin criteria. The alive patients with IMV had been intubated earlier since ARDS diagnosis than those who had not survived. We identied male gender and abnormal laboratory indices associated with the ARDS but were not linked with the intubation/death. Sensitivity analysis concluded lymphocyte count of < 1000 per mm3 could predict ARDS while OI ratio less than 150 could predict intubation/death. RRs: risk ratios; IQR: interquartile range; CI: condence intervals; SARS-CoV-2: severe acute respiratory syndrome coronavirus 2; BMI: body mass index; ESR: erythrocyte sedimentation rate; GFR: glomerular ltration rate; WBC: white blood cell; TBIL: total bilirubin; AST: aspartate transaminase; ALT: alanine aminotransaminase; LDH: lactate dehydrogenase; GFR: glomerular ltration rate; PT: prothrombin time; INR: international normalized ratio; ESR: erythrocyte sedimentation rate; SBP: systolic blood pressure; DBP: diastolic blood pressure; MAP: mean arterial pressure


Introduction
Since the COVID-19 outbroke in December 2019 in Wuhan of China, it has rapidly developed into a worldwide pandemic, with more than 129 million patients infected which resulted in over 2 million deaths until March 2021 (1). The data published suggest that 5-20% of COVID-19 patients develop severe illness that is characterized by acute respiratory distress syndrome (ARDS), the mainstay of treatment for which is mechanical ventilation (2)(3)(4). However, the mortality has been reported ranging from 30-80% in the COVID-19 patients who progressed into invasive mechanical ventilation (IMV) (2,4).
The systematic review on risk factors with ARDS reported lacking robust data of risk factors to prevent  and identi ed an urgent need for high quality research in this area (5). The previous studies generally suffered from unclear de nition, small sample sizes, and a lack of adjustment for confounders (5). Until now, the risk factors leading to the intubation or death have not been well delineated (4,6). As those patients with ARDS who progressed on intubation and mechanical ventilation had higher deaths (2,4), identifying the relevant risk factor associated with the intubation or death in the population could change the worst outcomes waiting for global populations fully vaccinated. Additionally, the views that COVID-19 causing an "atypical" ARDS remain different and have been discussed (7,8). Comparably large studies are really required to conclude that if COVID-19-related ARDS has different features.
To make full use of the past valuable medical records and improve the future clinical outcomes for the pandemic particularly in several countries, we retrospectively compared the clinical features between COVID-19 patients with ARDS and those without ARDS, and between subgroups of intubation and never intubation in COVID-19 patients from two hospitals of Hubei of China. We analysed the initial intubation time between survivors and non-survivors who were given IMV. We then studied the risk factors for COVID-19 ARDS and the ARDS patients who progressed into intubation or death.

Study design and participants
This retrospective cohort study included adult COVID-19 inpatients (≥ 18 years old) from 26 January 2020 to 1 April 2020 treated at General Hospital of Central Theatre Command of PLA in Wuhan (a tertiary hospital) and at People's Hospital of Honghu (a community hospital) in Jingzhou, a city close to Wuhan, in Hubei province, China. COVID-19 was diagnosed based on the World Health Organization (WHO)'s technical guidance on COVID-19, including con rmation with reverse transcription polymerase chain reaction (RT-PCR) (9).
The COVID-19-related ARDS was diagnosed according to the Berlin criteria published in JAMA: "development of acute and bilateral pulmonary in ltrates and hypoxemia (PaO 2 /FIO 2 ≤ 300 mmHg or SpO 2 /FiO 2 ≤ 315 as a surrogate if no PaO 2 data are available in the study) not primarily due to heart failure or volume overload" (10).
In the cohort, the ARDS patients were diagnosed not considering their positive end-expiratory pressure (PEEP) in order to identify the risk factors for the intubation or death earlier. Patients were screened and included by reviewing their electronic medical records and laboratory results. We excluded COVID-19 patients who were not con rmed by the RT-PCR testing, or pneumonia caused by other pathogens, such as bacteria, fungi, mycoplasma or chlamydia.
The study was approved by the Chinese Ethics Committee of Registering Clinical Trials (ChiECRCT20200113), and the requirement for informed consent was waived because of the retrospective study design. All methods were performed in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement (11).

Data collection
We reviewed the electronic medical records, laboratory results and radiographic reports of all admitted COVID-19 patients. Using a modi ed version of the standardized case report form developed by the International Severe Acute Respiratory and Emerging Infection Consortium and WHO (12), we extracted data on the demographic characteristics, medical history, comorbidities, illness onset date, initial symptoms, vital signs on admission and other data such as the severity of pneumonia graded by physicians. We also extracted the results of routine hematological, biochemical and in ammation-related indices obtained within 72 hours of hospital admission. treatments (antiviral drugs, antibiotics, glucocorticoids and vasopressor agents) or advanced therapies (intensive care unit (ICU) care, continuous renal-replacement therapy (CRRT), prone position ventilation and extracorporeal membrane oxygenation (ECMO); complications with death and other clinical outcomes. Further, we collected information on ARDS diagnosis, including chest imaging; FiO 2 , PaO 2 or SpO 2 values to calculate oxygenation index (OI); oxygen supplement modalities; intubation date; IMV duration; Reports of chest radiographs were obtained, and any factors associated with ARDS and acute respiratory failure warranting intubation in the cohort or with death were recorded.

Outcomes
The outcomes were ARDS development and intubation or death at hospital.

Statistical analysis
Continuous data are expressed as medians with interquartile ranges (IQRs). Categorical data are presented as counts (n) with percentages (%). Missing data were not imputed. We used Kruskal-Wallis rank test and Fisher's exact test to compare the results between ARDS and non-ARDS patients of COVID-19, and between ARDS patients who were never intubated and these who were intubated. Bivariate cox proportional hazard ratio (HR) models were used to determine HRs and 95% CIs for the risk factors associated with the ARDS or the development from ARDS to intubation/death. Sensitivity analysis yielding adjusted HRs for the popular concerned predictors of lymphopenia and oxygenation index ratio less than 150 were conducted by using multivariate cox proportional HR model. Due to the explorative nature of the study, no formal hypotheses and the sample size calculation were proposed, and we included the maximum number of COVID-19 patients from the two hospitals. P values less than 0.05 were considered signi cant. All analyses were performed in SPSS (version 25

Results
Demographics, characteristics, and clinical outcomes From 26 January to 1 April 2020, 338 patients with COVID-19 con rmed by laboratory ndings or otherwise clinical diagnosis were treated at the two hospitals in Hubei province of China. We included 333 patients, of which 65 subjects developed ARDS, and we excluded 5 cases due to negative/no laboratory testing or insu cient data. The details of enrolment are provided in Fig. 1.
Compared with patients without ARDS, patients with ARDS were older (P < 0.001) and trended higher BMI (P = 0.055), and more patients with ARDS were over 65 years (P = 0.015) and males (P = 0.008) ( Table 1). Median days from symptoms onset to hospital admission were shorter in ARDS than non-ARDS patients (P = 0.012).
However, more patients with ARDS presented with initial symptoms of fever (P = 0.028), cough (P = 0.009), fatigue (P = 0.050), expectoration (P < 0.001), palpitation and chest distress (P < 0.001), and at least three symptoms (P < 0.001) compared with those who did not develop ARDS. Patients who progressed into ARDS had higher breath rate (P = 0.008), more comorbidities (P = 0.008), and severest pneumonia graded by physicians (P = 0.008) on admission than patients not progressing into ARDS. All subjects were found to have bilateral in ltrate in lung in the severe illness of COVID-19 compared to 15.2% in the non-severe patients (P < 0.001). For treatment, the signi cantly higher percentages of patients with ARDS received general or advanced therapies, including antivirus therapy (P = 0.002), antibiotic (P < 0.001), glucocorticoid (P < 0.001), ICU-level care (P < 0.001), prone positioning (P < 0.001), CRRT (P < 0.001), and ECMO (P < 0.001). At last, when compared with patients who did not happen ARDS, patients with the severe illness further progressed more on IMV due to failure of noninvasive respiratory support (P < 0.001), stayed longer in hospital (P = 0.012), and died more (P < 0.001). No patient in non-ARDS group occurred severe complications compared to the 13 in ARDS group had acute respiratory failure (P < 0.001) and acute circulatory failure (P < 0.001), and the 7 occurred Shock/MODS (P < 0.001).       In the subgroup of COVID-19 patients with ARDS, more patients who progressed on IMV were over 65 years (P = 0.015), reported fatigue (P = 0.009) and at least three symptoms at admission (P = 0.011), and had higher SBP (P = 0.004) or MAP levels (P = 0.021) with admission fever compared with patients who were never intubated ( Table 1) intubated. Nevertheless, ARDS patients on IMV trended lower cell counts of WBC, neutrophil, and lymphocyte, and elevated markers related to the damages of liver, pancreas, and kidney than the patients never intubated.    * Data regarding the cTnI were missing for 2 ARDS patients who were never intubated. † Data regarding the Glucose were missing for 1 ARDS patient who were never intubated. ‡ Data regarding both the Potassium and Sodium were missing for 6 ARDS patients, with 4 at never intubation group, 1 at early intubation group, and 1 at late intubation group. § Data regarding the PTs were missing for 1 ARDS patient who were never intubated.
|| Data regarding the ESR were missing for 6 ARDS patients who were never intubated.
High mortality of COVID-19 ARDS patients with intubation and the initial time of intubation COVID-19 ARDS patients requiring intubation showed a high mortality at 70.0% (Table 1). However, we found the survivors had been intubated earlier since ARDS diagnosis than the non-survivors, with a median of 5.5 and IQR of 4.0-7.0 days versus a median of 11.5 and IQR 6.0-14.0 days (P < 0.001) (Fig. 2).

Risk factors associated with ARDS and intubation/death in COVID-19 patients
Univariate cox proportion hazard models showed that age and age over 65 years were associated with the development of ARDS

Discussion
In this study, we observed that the ARDS caused by COVID-19 was not the typical ARDS de ned by the Berlin criteria due to more than 1-week onset time, multiple organs injuries, and higher mortality. We found that the survivors who were on IMV had been intubated earlier since ARDS diagnosis than the non-survivors. We reported the different risk factors between COVID-19 patients who developed ARDS after admission and who progressed from ARDS to intubation or death.
Older age or age over 65 years was the most consistently reported risk factors for patients who develop ARDS, as concluded by previous investigations (13). Interestingly, we found the hazards ratio of age over 65 years associated with the intubation or death was larger than the ratio of this factor with ARDS, which suggests that the risk for worse clinical outcomes markedly increased once the elder patients developed ARDS. The reason could be the accelerating decline of immune response (14,15). To prevent the ARDS development for the elder patients shall reduce the worse outcomes. Another different nding compared with the studies published was that males faced a signi cantly greater risk of developing ARDS. However, for the typical ARDS caused by other factors, females were found to be more likely to develop the disease than males (16). One possible explanation was the protection of X chromosome and sex hormones which played an important role in innate and adaptive immunity made the females less susceptible to coronavirus infection than men, resulting in the less development of the critical disease (17).
The two-hospital cohort found that the COVID-19 subjects with abnormal laboratory indices at admission were more possibly to progress into ARDS. The lymphopenia, which is de ned as the counts less than 1000 counts/mm3, strongly predicted the ARDS for COVID-19 patients. Even adjusting for age and male gender, the lymphopenia was associated with fourfold risk of ARDS. Lymphocytes express the ACE2 receptor and may be a direct target of SARS-CoV-2 infection, and an increase of pro-in ammatory cytokines in COVID-19, especially IL-6, could induce further lymphocyte reduction (18). When the patient's laboratory indicator showed PCT > 0.1 ng/mL, ESR > 20 mm/h or C-Reactive Protein > 10 mg/L, there was a higher risk of progression to the critical illness. That may support the corticosteroids use to prevent the risk of ARDS, particularly for the elder patients, due to alleviating the systemic in ammation (19). The positive associations between albumin < 32 g/L, LDH > 280U/L, cTnI, ng/mL > 0.04, creatinine > 110 µmol/L or glucose > 7.1 mmol/L and the ARDS indicated that the liver, heart, kidney or pancreas dysfunctions had been involved, and corresponding therapies should be taken in time to prevent the further deterioration of the COVID-19. When the D-dimer > 0.5mg/L meant the high risk of the progression of the COVID-19 to ARDS, the approaches to alter the hypercoagulability of blood, reduce the risk of thrombosis, and improve the ischemia and hypoxia shall be implemented early (18).
No laboratory indicators predicting the need of intubation or death suggests different pathophysiological pathways between ARDS development and hypoxemic respiratory failure/death in COVID-19. The ARDS patients presenting with symptoms ≥ 3 were more likely to be intubated or die at hospital. The clinicians might monitor these ARDS patients more closely to implement therapies in time even though the patients did not suffer any coexisting disease. Additionally, the analysis revealed that COVID-19 patients with lower OI ratios at ARDS diagnosis were more likely to develop hypoxemic respiratory failure necessitating IMV and had higher risk to die.
The COVID-19 Guidelines of Europe (20), USA (21), and Canada (22) have not yet provided any recommendation related to intubation. Our ndings that the decreasing OI ratio in ARDS patients was highly associated with hypoxemic respiratory failure necessitating the intubation and the risk of in-hospital death may supported the early intubation criteria of the COVID-19 Management ---patients with an SpO 2 value of less than 93% of the room air and a OI ratio of less than 300 mmHg should be prepared for intubation if there are indications (23).
The sensitivity analysis showed that the OI ratio less than 150 well predicted the need of intubation and death when adjusting for age. The threshold might be considered as a caution to the deterioration of COVID-19 patients to facilitate the early treatment.
The study complements the existing small studies on the clinical features of ARDS caused by COVID-19 with a comparably large sample. The longer median onset days (11.0, IQR, 8.0-13.0), which exceeds 1-week limit of typical ARDS (10), should alert physicians to monitor the COVID-19 patients longer for adequate treatment. The mortality rates of moderate and severe COIVID-19 ARDS were higher than the corresponding data of 32% and 45% of the typical ARDS (24). One possible explanation we thought is the silent hypoxemia that the patient had severe hypoxemia but was asymptomatic, which would lead to delay treatment (25,26). Therefore, from the perspective of treatment, an adjusted classi cation of ARDS severity that can correctly identify COVID-19 patients who require a timely therapy for has been always needed (7). Most importantly, the ARDS caused by COVID-19 demonstrated injuries in multiple organs, including the heart, the liver, the kidney, and the pancreas, which was supported by the fundamental research that this new coronavirus can attack the tissues of other organs in addition to the respiratory tract (15). At the two-hospital cohort, 78.5% patients with ARDS were given systemic corticosteroids to alleviate hyperin ammation. However, guidance regarding corticosteroids use for COVID-19 is still mixed (27). Some issued a weak recommendation (28), but others issued a weak recommendation against the use of corticosteroids for COVID-19 and ARDS (29). Nevertheless, the latest publication of three randomized trials of corticosteroids and the prospective meta-analysis have strengthened con dence and shifted usual care of COVID-19-related ARDS to include corticosteroids (27).
An interesting result in this cohort was that the survivors with IMV were intubated earlier than those with IMV who did not survive. We noticed the non-survivor group had the median intubation days at 11.5, which exceed one week after ARDS diagnosis or hospital admission. By contrast, we observed the patients intubated within one week after ICU admission or ARDS diagnosis demonstrated much lower mortalities in our cohort, in other hospitals of China (6), and some places of USA (30). To intubate the COVID-19 patients within one week since critical illness development seemed to result in a lower mortality rate. However, because we were not sure the two hospital used the same intubation criteria at that time, and we did not have the speci c values when they were progressed into intubation, we could not say these ARDS patients had survived due to comparable early intubation. Even nowadays, the de nition of what constitutes "early" is still a challenge and requires necessary investigations for COVID-19. As Shoemaker et al.'s study (23) highlights the importance of timely stopping an enlarging oxygen debt using effective oxygenation and ventilation therapies because there is a close association between the oxygen debt accumulated over 48 hour and the chance of survival in patients undergoing high-risk surgery and ICU admission afterward.
The strengths of the study are as follows: the study cohort included a comparably large sample size from two hospitals in two cities of Hubei province, which increases the robustness and generalizability of the ndings; reported HRs for the risk factors and did the sensitivity analysis for the popular concerned risk factors associated with ARDS and ARDS progressing on IMV or death. However, the study has some limitations. First, owing to limited medical resources, only patients with relatively severe COVID-19 were hospitalized during this period. As such, this study may have included disproportionately more patients with poor outcomes. There may also be a selection bias when identifying factors that in uence the clinical outcomes. Second, it is a retrospective cohort conducted at two hospitals. As is known, it was hard to perform a perspective cohort when pandemic happened and developed everywhere. As we have sparse new COVID-19 cases now, we hope our new ndings can encourage a larger prospective cohort study or even randomly controlled trials in other regions who are experiencing the outbreak and in the urgent need to adjust the strategy.

Conclusions
In conclusion, the SARS-CoV-2-caused ARDS was not the typical ARDS for which the management strategies should be modulated accordingly. The survival patients had been intubated earlier since ARDS diagnosis than those who had not survived, which may propose the issue of timing of intubation. Lymphopenia predicted the ARDS and OI ratio less than 150 predicted the intubation/death. The male gender and all the abnormal laboratory indices associated with the ARDS but were not linked with the intubation or death, suggesting different pathophysiological processes at the two stages. The study was approved by the Chinese Ethics Committee of Registering Clinical Trials (ChiECRCT20200113), and the requirement for informed consent was waived.

Consent for publication
Not applicable Availability of data and materials The datasets generated during and analyzed during the current study are not publicly available due to privacy or ethical restrictions but are available from the corresponding author on reasonable request. A proposal with detailed description of study objectives and statistical analysis plan will be needed for evaluation of the reasonability of requests.

Competing interests
We declare no competing interests.  Flowchart of patient's enrolment

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.