Introduction

The novel pandemic caused by SARS coronavirus-2 (SARS-CoV-2) has been associated with increased burden on healthcare system capacities worldwide1. Patients with COVID-19 have high hospital and intensive care unit (ICU) admission rates2. Mechanical ventilation continues to be the cornerstone of management of COVID-19 patients with life-threatening acute respiratory distress syndrome (ARDS); approximately one-third of these patients require some form of mechanical ventilation3,4,5. Outcome data for patients with COVID-19 receiving invasive mechanical ventilation have varied considerably. Meta-analyses of patients requiring ICU admission and mechanical ventilation referred to mortality rates of 40–45%6. Understanding the factors associated with mortality in patients requiring critical care and mechanical ventilation is highly relevant. Early studies on COVID-19 highlighted the presence of several factors such as patients’ baseline characteristics (hypertension, diabetes mellitus, obesity) and increased inflammatory markers levels as predictors of poor outcomes2,7,8,9,10,11,12,13,14.

Recently published data demonstrated that applying novel risk predictive models via specific machine learning algorithms could provide clinicians with more precise diagnostic tools, identifying patients at higher risk of complications such as health care-associated infections and may play a crucial role in improving patients’ outcome15,16.

However, the mortality of critically ill patients with COVID-19 receiving invasive mechanical ventilation has been reported to be highly variable. Causes of this inconsistency likely include the heterogeneity in the management of these patients and in the presentation of outcome data3,17,18.

It has been assumed that management of acute hypoxaemic respiratory failure and ARDS in COVID-19 patients largely mirrors that for non-COVID-19 etiologies. Recent data have shown that ARDS related to COVID-19 shares common pathophysiological and clinical features with ARDS of other causes; treatment goals of ARDS seem to be similar with goals for non-Covid-19 etiologies19,20. However, patients undergoing invasive mechanical ventilation have significantly longer mechanical ventilation and ICU stays than those with ARDS of other causes, whereas mortality in those patients varies considerably in different reports. Additionally, surging patients volume overwhelms hospitals bed capacity, causes staffing strains, drug and equipment supply shortages; therefore, evolving understanding of the novel pathogen and available treatment options have fueled rapid operational and clinical practice shifts in critical care19.

The COVID-19 pandemic crisis has been a turning point for the achievement of sustainable development goals, with all consequences at the political, economic, and socio-cultural levels. COVID-19 should be taken as an opportunity to learn from lessons taught, plan a more efficient agenda, focus on the development of additional disease-modifying interventions and adapt the actions to the current changing times in the aftermath of the pandemic21.

There is limited information pertaining to outcome of COVID-19 patients admitted to ICU in Greece. Accordingly, the goal of this study was to describe, for the first time, the clinical characteristics and outcomes of a cohort of mechanically ventilated patients with COVID-19 who were admitted to the intensive care in a tertiary hospital in Northern Greece. In parallel, we performed a thorough review of the literature concerning the factors that affect the ICU outcome in this category of patients worldwide.

In this paper, we present the methodology and the study population, then the results from the statistical analyses performed. The next part is the discussion section, with special analysis of the pandemic waves and viral variants of concern, the admission clinical and laboratory data, the treatment regimens and the complications during ICU stay. The limitations of the study are also discussed there, and the paper ends with the conclusion of the study and the references used.

Materials and methods

Patients and study design

The study was carried out through a prospective design in a single center ICU in Regional Hospital “G. Papanikolaou”, Thessaloniki, Greece. The ICU has a capacity of 18 beds and during the study operated as a COVID-19 ICU.

The study was based on the collection of data obtained during routine clinical practice in patients over 18 years old who were tested positive for SARS-CoV-2 infection by polymerase chain reaction (PCR) test between April 2020 and February 2022 (RNA extraction and RT-PCR using Extralab and Amplilab machines, respectively, by Adaltis, Italy). All the patients were intubated due to acute respiratory insufficiency and received Invasive Mechanical Ventilation.

All methods were performed in accordance with the relevant guidelines and regulations for observational studies in human subjects.

Data collection

On day 1 at ICU admission, the following data were recorded: patient demographics and characteristics, including age, sex, body mass index (BMI), comorbidities (Diabetes mellitus, Arterial Hypertension, Chronic Ischemic Heart Disease, Chronic Kidney Disease, Chronic Obstructive Pulmonary Disease, Oncology or haematological diseases, Chemotherapy in the previous 6 months, Autoimmune diseases and Hypothyroidism), smoking status, number of days from hospital admission to intubation.

Additional data that were recorded on the same day were:

  • basic laboratory values and inflammatory markers, namely white blood cells count, lymphocyte number, platelets number, urea, creatinine, Lactate Dehydrogenase (LDH), troponin, C-reactive protein (CRP), ferritin, D-Dimer and procalcitonin (PCT).

  • PaO2/FiO2 ratio, SpO2 (from pulse oximetry), Acute Physiology and Chronic Health Evaluation (APACHE II) score, Sequential Organ Failure Assessment (SOFA) score.

During hospitalization the use of specific treatments such as azithromycin, hydroxychloroquine, dexamethasone, remdesivir and tocilizumab was recorded.

Development of important complications such as need for vasopressors, acute heart failure, renal failure, renal replacement therapy, hepatic insufficiency, acute intestinal failure, thromboembolic events, pulmonary barotrauma (pneumothorax, pneumomediastinum) was recorded during the ICU stay.

The allocation of the cases into discrete disease waves was based on the local peaks of disease outbreaks and was defined as follows: the 1st pandemic wave from March to September 2020, the 2nd from October 2020 to January 2021, the 3rd from February 2021 to June 2021 and the 4th from July 2021 to March 2022.

After the patient’s discharge from ICU or death, the length of ICU stay was calculated.

Outcomes

The primary outcome was ICU mortality. Secondary outcomes were 28-day mortality and independent predictors of mortality at 28 days and during ICU hospitalization.

Statistical methodology

Kolmogorov–Smirnov test was used for exploring the normality of distributions. When continuous variables had normal distribution, t-test was used for means comparison between two groups and one-way ANOVA (with Bonferroni post-hoc test) for multiple comparisons. When the variables distribution was not normal, comparisons between two groups were performed using the Mann–Whitney test. Comparisons between discrete variables were made using the x2 test, whereas the binary logistic regression was employed for the definition of factors affecting survival inside the ICU and 28-days survival.

The statistical analysis was performed utilizing the SPSS v21 (IBM Corp) statistical software. Differences were considered statistically significant when p ≤ 0.05.

Ethics approval and consent to participate

The study was approved by the Scientific Council of “G. Papanikolaou” General Hospital of Thessaloniki, Greece. The observational nature of the study and the practical problems posed by the national prohibition of relatives’ visit to the ICU during the COVID-19 pandemic, led to the absence of written consent for patients’ participation to the study, which was also waived by the aforementioned ethics committee.

Results

A total of 375 patients were hospitalized in the 1st ICU of “G. Papanikolaou” Hospital of Thessaloniki, Greece during the Covid-19 pandemic from the beginning of 2020 until March 2022. Of those, 239 (63.7%) were male. Figure 1 shows the general structure of the study population. The patients had a mean age of 64.1 years and had severe ARDS upon admission to the ICU with an average APACHE II of 16.3 and an average length of stay in the unit of 18 days. Table 1 shows the basic descriptive statistics of the continuous variables of the study.

Figure 1
figure 1

Schematic representation of the study population and outcomes.

Full size image
Table 1 Descriptive statistics for the continuous variables.
Full size table

Forty-four subjects were hospitalized during the first wave of the pandemic in Greece, 104 during the 2nd, 87 during the 3rd and 140 during the 4th. They were affected by the following SARS-CoV2 variants: 122 by Alpha, 165 by Beta, 78 by Delta and 10 by Omicron. In total, the ICU survival was 49.6%, whereas the 28-day survival reached 46.9%. The survival rates inside the ICU for the four main viral variants were 54.9%, 50.3%, 39.7% and 50% for the Alpha, Beta, Delta and Omicron variants, respectively. The differences in these rates were not statistically significant. Regarding the survival rates in relation to the different waves of the pandemic in Greece, the following values were observed: 61.4%, 50.96%, 62% and 37.1% for the 1st, 2nd, 3rd and 4th wave, respectively. The x2 test showed significant difference (p = 0.001) between the waves. Only 5 (1.3%) of the patients were fully vaccinated upon admission in the ICU in our cohort.

The sex had no significant difference in survival in the ICU or the mortality at 28 days (male 28-days survival = 46.4%, female 28-days survival = 47.8%).

Figure 2 presents the duration of ICU stay of patients in the four discrete pandemic waves.

Figure 2
figure 2

Bar chart showing the mean duration of ICU stay in relation to the Covid-19 wave in the Greek healthcare system.

Full size image

The Kolmogorov–Smirnov test marked only the IL-6 and the PLT value as variables having normal distribution. The Mann–Whitney test for the comparison of continuous variables between patients who died in the ICU and those who survived yielded the following statistically significant results (Table 2):

Table 2 Mann–Whitney test for differences in various variables between non-survivors and survivors in the ICU.
Full size table

Moreover, patients who were treated with remdesivir had shorter ICU stay (4.74 vs 5.84 days, p = 0.014) and had better survival rates during the ICU stay (62.5% vs 41.15%, p < 0.001) as well as at 28 days.

The ANOVA tool with Bonferroni post-hoc tests among the four virus variants showed significant differences in:

  • Length of ICU stay: Beta–Alpha = 4.7 days & Delta–Alpha = 13.2 days

  • Age: Beta–Delta = 4.54 years

  • d-dimer: Delta–Alpha = 13.7 & Delta–Beta = 13.2

Concerning the x2 tests for comparisons between discrete variables, the following noteworthy results were obtained:

  • Patients with acute kidney injury (AKI), coronary disease, enteral insufficiency, heart failure, liver insufficiency and barotrauma had worse outcome in the ICU.

  • The same conditions had similar effect to the 28-days survival.

  • Smokers had better outcomes in terms of ICU and 28-days survival.

  • The wave of the pandemic played an important role in survival (both the ICU and 28-days) with the 1st wave having a survival rate of 61.4%, the 2nd 51%, the 3rd 62% and the 4th 37.1%.

  • Tocilizumab and T-lymph therapies did not show significant impact on outcome.

Tables 3 and 4 show the statistically significant differences in discrete variables between survivors and non-survivors in the ICU and at 28 days, based on the x2 tests.

Table 3 Differences in discrete variables in relation to ICU survival (x2 tests).
Full size table
Table 4 Differences in discrete variables in relation to 28 days-survival (x2 tests).
Full size table

It is of interest to note that 50 patients (13.3%) developed barotrauma during the ICU stay. One additional parameter approached statistical significance in 28-days survival: hypothyroidism (31% vs 48.4%, p = 0.053).

Logistic regressions for outcome revealed that the following parameters were independently associated with ICU survival: wave, SOFA day1, remdesivir use, AKI, enteral insufficiency, sepsis, duration of ICU stay and WBC.

Similarly, by applying binary logistic regression, we found that the parameters affecting the 28-days survival were: duration of stay in ICU, SOFA day1, WBC, Wave, AKI and Enteral Insufficiency. Two additional parameters were close to achieving statistical significance: remdesivir and development of barotrauma.

Concerning the statistical power of the sample size, it was discovered that for a difference of average mortality of approximately 6% (which was the case in our population compared to the literature) and an equal distribution between death and survival, a sample size of 380 subjects would yield a statistical power of 80%.

Discussion

This study identified the association of the characteristics of pandemic (waves, virus strains responsible for infection), baseline patient characteristics, clinical and laboratory data on admission and during ICU stay with 28-day mortality and mortality at ICU discharge in patients with SARS COV-2 admitted to the ICU of a tertiary hospital. This is the first study designed to describe clinical characteristics and outcomes of COVID-19 patients with ARDS receiving prolonged mechanical ventilation from a large Northern Greece area.

Recognizing the variables that independently predict death in COVID-19 is of great importance. Logistic regressions for outcome revealed that the following parameters were independently associated with 28-day and ICU survival: wave sequence, SOFA score on admission, the use of Remdesivir, presence of AKI, presence of gastrointestinal failure, and WBC values/levels. In addition, sepsis influenced ICU survival but not 28-day survival, a finding in agreement with recent literature22. The use of remdesivir and development of barotrauma were close to achieving statistical significance in affecting 28-day outcome, however no association was observed in relation with the ICU mortality rate.

Waves, viral variants of concern (VOCs) and outcome

The 28-day survival rate of 46.9% was demonstrated, whereas the overall survival rate upon ICU discharge reached 49.6%. Statistically significant differences in ICU mortality were observed among patients admitted during the different waves of the pandemic: the mortality rate was significantly higher in patients admitted to the ICU during the first and third waves of the pandemic.

Diverse COVID-19 associated mortality rates have been reported from different studies among critically ill patient populations during the beginning of the pandemic. During the first wave of the pandemic, the mortality rates of patients with COVID-19 treated in ICUs ranged broadly (23.3–81%)23,24,25,26,27,28.

A multicentre retrospective cohort study of Carbonella et al. referred to overall ICU mortality of 30.7%, without significant differences between study periods (first wave 31.7% vs second/third waves 28.8%, p = 0.06). After adjusting for confounding factors through a multivariable analysis, no significant association was found between the COVID-19 waves and mortality (OR 0.81, 95% CI 0.64–1.03; p = 0.09)29.

The impact of waves sequence on patients’ outcomes during the course of the pandemic around the world could be difficult to analyze and interpret due to the fact that in different regions pandemic waves occurred in different periods of time, patients included in studies had diverse underlying medical conditions and diverse severity of illness and need for invasive mechanical ventilation30,31,32.

Recently published studies have demonstrated the association between increased mortality and hospital load on critical care capacity30,33. Possible explanations for the higher mortality during the first waves in comparison with the fourth wave may have to do with the insufficient knowledge of the disease during the early phases of pandemic, as well as the claimed inability of the health systems to control the unprecedented ICU capacity strain resulting from sharp increase in patient volume during the peaks of the initial waves29. ICU capacity strain can be characterized as a mismatch between supply of recourses including beds, staff, and/or other resources and demand (the need to admit and provide care for critically ill patients34). ICU capacity strain has adverse consequences for patient outcomes but, additionally, may has an adverse effect on the well-being of the healthcare providers by increasing stress levels35,36. It is imperative to investigate how excessive ICU demand may negatively affect patients’ outcome and to identify regulatory strategies to prevent healthcare system overload associated with high mortality.

The lack of some effective antiviral treatment could be considered as an additional factor affecting the mortality rate in our patients, as our results revealed significant increase on the use of antivirus therapy (remdesivir) during the fourth wave in comparison with the previous periods of pandemic.

Our data demonstrated that the ICU survival rate did not differ significantly when infection with the four main VOCs was considered. However, The Bonferroni post-hoc tests among the four virus variants showed significant differences in the length of ICU stay among different VOCs: patients with Beta and Delta variants had longer duration of stay in comparison with patients with Alpha variants.

Owing to a possible risk of increasing the transmissibility of the virus, severity of the infected individuals, and the ability to escape the antibody produced by the vaccines, the four SARS-CoV-2 variants of Alpha (B.1.1.7), Beta (B.1.351), Gamma (P.1), and Delta (B.1.617.2) have attracted the most widespread attention. However, the conclusions regarding disease severity and impact on outcome of these variant viruses are not consistent in the literature. A recent review showed that variants Alpha, Beta and Gamma all had a higher risk of hospitalization and ICU admission compared with the wild-type virus, with the higher risk for Beta variant37. A systematic review and meta- analysis study of Lin et al. aimed to determine the effects of SARS-CoV-2 variants of concern on disease severity and clinical outcomes. The analysis showed that all SARS-COV-2 VOCs have a higher risk of disease severity than the wild-type virus. By comparing with the wild-type virus, in terms of the risk of hospitalization, ICU admission, and mortality, the variants Beta and Delta have a higher risk than the variants Alpha and Gamma, however, Alpha variant had a higher risk of disease severity than the wild-type virus38. The results showed that in the risk of hospitalization, ICU admission, and mortality, all the SARS-CoV-2 VOCs had different degrees of increase compared with wild-type virus; Delta variant had the highest risk of ICU admission and mortality, and Beta variant had the highest risk of hospitalization. Contrastingly, Funk et al. found that Alpha variant was more threatening than the wild-type virus and associated with significantly higher risk of hospitalization and ICU admission39.

Moreover, consistently with our results, a meta-analysis study of Kow et al. demonstrated significantly increased hazard of mortality among patients with COVID-19 infected with Alpha variant relative to those infected with the wild-type virus40. Although, the phenotypic effects of SARS-CoV-2 VOCs are still uncertain and the emergence of advanced vaccines may reduce the threat posed by SARS-CoV-2 variants, the impact of VOCs on outcome of patients seems to be important. Of course, other than VOCs factors, including the use of health-care resources, demographic changes, strain posed on healthcare system possibly influenced clinical outcomes in our patients and lead to a higher mortality rate in patients with the Alpha variant.

Admission clinical and laboratory data and outcome

Our results showed that non survivors were older, had shorter duration of ICU stay, higher APACHE II score and SOFA scores on admission, higher levels of laboratory markers, including ferritin, d-dimer, urea and creatinine levels, white blood cells (WBC), troponin (TPI) and lactate dehydrogenase (LDH).

Nicholson et al. evaluated the risk factors on admission (including comorbidities, vital signs, and initial laboratory assessment) associated with need for mechanical ventilation and in-hospital mortality in COVID-19 patients. Older age, male sex, coronary artery disease, diabetes mellitus, chronic statin use, SpO2/FiO2 ratio and body mass index, high levels of lactate dehydrogenase and high levels of inflammatory and infectious biomarkers (neutrophil to lymphocyte ratio, platelet count, procalcitonin, C-reactive protein) were identified as important risk factors for mechanical ventilation requirement and in-hospital mortality. Using these factors, the authors constructed specific risk scores for healthcare providers and researchers41.

The higher severity of illness and severity of organ failure and the excessive levels of inflammatory and infectious markers in our patients confirms the association of aggressive inflammatory response leading to organ dysfunction and the higher mortality rate42. Observational research published at the beginning of the pandemic suggests that the risk of mortality increases with the presence of comorbidities: obesity, hypertension, diabetes mellitus, and chronic lung disease70,71. A large UK cohort study demonstrated a high AKI incidence; 487 (39%) out of 1248 patients experienced AKI. Acute kidney injury was a strong predictor of 30-day mortality with an increasing risk across AKI stages72. Data from a multicenter retrospective cohort analysis of patients with critical COVID-19 in seven large hospitals in Belgium confirm this association and demonstrated even higher incidence of AKI in critically ill patients approaching 80%; additionally, after multivariable adjustment, all AKI stages were associated with ICU mortality73. In our study the presence of acute kidney injury was associated with higher ICU and 28-day mortality rate; the rate of AKI in non survivors was 71.3% compared to 21.4% in survivors. Moreover, AKI diagnosis was independently associated with ICU survival.

Logistic regressions for outcome revealed that the following parameters were independently associated with ICU survival: wave, SOFA day1, Remdesivir use, AKI, enteral insufficiency, ICU length of stay and WBC.

The association of acute gastrointestinal injury and mortality among COVID-19 patients was not sufficiently elucidated. The development of high-grade acute gastrointestinal injury (AGI) and feeding intolerance (FI) during ICU stay can serve as a prognostic tool to predict outcomes in critically ill COVID-19 patients, however there is a relative scarcity of data regarding the impact of AGI on outcome COVID-19 patients. Initial data from Wuhan, China reported a high incidence of AGI in patients with COVID-19 disease; diagnosis of AGI was associated with a higher incidence of septic shock and 28-d mortality74. Drakos et al. found extremely high incidence of AGI and feeding intolerance among critically ill COVID-19 patients. The overall incidence of AGI was 95% (45% AGI I/II, 50% AGI III/IV), and FI incidence was 63%. Patients with AGI III/IV were more likely to have prolonged mechanical ventilation (22 days vs 16 days, p-value < 0.002) and higher mortality rate (58% vs 28%, p-value < 0.001) compared to patients with AGI 0/I/II75. Our findings are consistent with conclusions from the previous publications. We categorized AGI into four grades based on collected clinical and imaging data. Logistic regression analysis performed in our study demonstrated that the presence of AGI grades III and IV was associated with higher 28-day and ICU mortality rates. Consequently, a conclusion could be made that presence of AGI III/IV may reflect the degree of inflammatory response and multiple organ damage as reflected by higher WBC levels and higher SOFA score in this patient population, which has been shown to correlate with disease mortality.

A high rate of barotraumas was observed in our patients’ population: 50 patients (13.3%) developed barotrauma during the ICU stay; however, the logistic regression analysis failed to show a strong relationship between presence of barotraumas and outcome. An ever-increasing number of studies have reported an increased incidence of spontaneous pulmonary barotrauma such as pneumothorax, pneumomediastinum, and subcutaneous emphysema in patients with COVID-1976,77. A systematic review and meta-analysis aimed to evaluate the incidence of barotraumas and impact on outcome showed a high incidence 18.4% (13–25.3%) in patients receiving invasive mechanical ventilation. In addition, barotrauma was associated with a longer length of ICU and hospital stay, and higher in-hospital mortality. When compared with non-COVID-19 ARDS, a slightly higher odds of barotrauma were seen in COVID-19 ARDS compared with non-COVID19 cases77. Further studies are required to unravel the underlying pathophysiology and develop safer management strategies.

Limitations

This study has several limitations. First, the study was conducted in the Northern Greece area, therefore, results from our study may not be representative of other regions in Greece and may not be extrapolated to other world regions. Second, the observational nature of the study implies that a number of possible unmeasured confounders could affect the outcomes. Further study into possible differences in the provision of care and outcome for COVID-19 and non-COVID-19 patients is needed. Third, longer follow up such as hospital mortality were not evaluated. Reporting short-term endpoints could be not enough and the investigation of long-term outcomes can be highly important in critical care settings. Fourth, we focused on all-cause ICU mortality. To be able to better interpret mortality rates, more data of the specific causes of death in COVID-19 would be helpful.

Conclusion

Recognizing the variables that independently predict death in COVID-19 is of great importance. In this observational cohort study of 375 COVID-19 patients we report an association between clinical and laboratory data with outcome. Strength of this study is the large number of patients with severe form of coronavirus disease who were all under invasive mechanical ventilation. Overall, 90-day mortality was 49.6%, the wave of the pandemic played an important role in survival which was constantly decreasing over time during the study period. Comparison of the adjusted mortality rates between pandemic waves within a two year-study period adds additional understanding and additional potential and tools for pandemics’ management. Recognizing the variables that independently predict death in COVID-19 is of great importance for pandemic management. Logistic regressions for outcome revealed that the following parameters were independently associated with ICU survival: wave, SOFA on day1, remdesivir use, acute kidney injury, sepsis, presence of gastrointestinal failure, duration of ICU stay and WBC. Similarly, the parameters affecting the 28-days survival were: duration of stay in ICU, SOFA on day1, WBC, wave, acute kidney injury and presence of gastrointestinal failure. Our results did not confirm data from the previous publications regarding the association of comorbidities (diabetes mellitus, arterial hypertension, COPD) with poor outcome in patients with COVID-19. Our results revealed a marginal association of hypothyroidism with patient outcome at 28 days. Future research into the potential complications of COVID-19 on the thyroid gland and function is warranted. One of the strengths of the study include the analysis of relationship between immunotherapy treatment and patients’ outcome. Despite many references that advancing novel therapeutic developments becomes crucial to minimize the number of deaths from COVID-19, the use of Tocilizumab and T-lymphocyte therapy did not demonstrate any advantage regarding outcome parameters in our study population. Future investigations may explore potential biomarkers to optimize patient selection for tocilizumab treatment and combination therapy with various cell therapies.