Demographics and clinical characteristics of COVID ICU patients

A total of 145 COVID ICU patients were enrolled in the study. The average age was 67.6 ± 14.3 years, range [18–96] years, and among them, 103 (71.0%) patients were males and 59 (43.7%) were smokers. The participants’ top three most common comorbid conditions were hypertension (n = 91, 62.8%), diabetes mellitus (n = 62, 42.8%), and cardiovascular diseases (n = 62, 42.8%) (Table 1). Most COVID ICU patients had low oxygen saturation levels < 95% at triage (n = 101, 70.1%). The average ED boarding time was 6.8 days ± 4.9. In total, 95 (65.5%) died, among them, 52 (55.0%) died in ICU and the remaining 42 (45.0%) died in ED (Table 1). A cutoff age value of 65.5 years (AUC = 0.665, 95% CI 0.569–0.761, sensitivity = 0.705, 1-specificity = 0.46) was used for analysis. To note, an average of ED length of stay of around 7 days was found (p value 0.03).

Table 1 Association of demographic and clinical characteristics of COVID ICU patients with in-hospital mortalityTreatments and health-related complications of COVID ICU patients

The patients’ hospital treatment included steroids (n = 139, 95.9%), antibiotics (n = 137, 94.5%), anticoagulants and antiplatelets (n = 115, 79.3%), Remdesivir (n = 82, 56.6%), Actemra (n = 50, 34.5%), convalescent plasma (n = 50, 34.5%), Ivermectin (n = 48, 33.1%), Baricitinib (n = 23, 15.9%), and hydroxychloroquine (n = 3, 2.1%). About 58 patients (40.0%) received vasopressors (Table 2, in the tables’ section).

Table 2 Association of treatments and health-related complications of COVID ICU patients with in-hospital mortality

About 88.3% of COVID ICU patients developed health-related complications (n = 128). Renal failure (n = 85, 57.9%) was the most commonly reported complication followed by cardiovascular complications (n = 71, 49.0%), acute respiratory distress syndrome (n = 58, 40.0%), septic shock and DIC (n = 49, 33.8%), infection (n = 34, 23.4%), and pneumothorax (n = 19, 13.1%) (Table 2). About 23 patients (15.9%) underwent dialysis and only 5 patients had a tracheostomy inserted. Oxygen was provided to COVID ICU patients through different techniques; BiPAP (n = 56, 38.6%) high-flow nasal cannula (n = 22, 15.2%) or endotracheal intubation (n = 87, 60.0%) (Table 2). Interestingly, out of patients who developed pneumothorax, a trend showed that 16 cases (17.2%) required intubation whereas 3 cases (5.8%) did not, with an exact p value of 0.05.

Predictors of In-Hospital Mortality in COVID ICU patients

As summarized in Table 1, 71.6% of patients were males (n = 68, p = 0.842). It also showed that they were significantly older (70.5% vs. 46.0%, p = 0.004). They were more likely to have had chemotherapy in the previous year (50% (n = 9) vs 28.6% (n = 2), p = 0.407). Mortality was significantly associated with ED LOS, as our analysis showed that patients who died stayed longer in the ED compared to those who did not die (7.45 days versus 5.58 days, p = 0.03) (Table 1). However, there was no significant difference in terms of gender, smoking status, and types of comorbidities between COVID ICU patients who died or survived. Patients who died were more likely to have cardiovascular diseases (46.3% vs. 36.0%) and 2.9 times more likely to have chronic kidney diseases (20.0% vs. 8.0%), but it was not statistically significant (p > 0.05) (Table 1). Patients who died were twice more likely to have low oxygen saturation levels < 95% at triage (75.5% vs. 60.0%, p = 0.053) and 1.9 times more likely to have severe COVID disease on CT (60% vs. 44%, p = 0.066) but was not significant. They significantly required 4.9 times more vasopressors (54.7% vs. 12.0%, p < 0.001). Intubated patients were 6.7 times significantly more prone to death (78.9% vs. 36.0%, p < 0.001) (Table 2).

Treatment with convalescent plasma, steroids, anticoagulants, Remdesivir, or Ivermectin did not significantly affect mortality (p > 0.05) (Table 2). However, patients in the hospital who received Actemra were significantly less likely to die than those who did not receive Actemra (27.4% vs. 48%, p = 0.013).

Patients who died developed multiple complications including acute respiratory distress syndrome (ARDS) (49.5% vs. 22.0%, p = 0.001), pneumothorax (17.9% vs. 4.0%, p = 0.018), acute kidney injury (44.2% vs. 14.0%, p < 0.001), respiratory failure (52.6% vs. 20.0%, p < 0.001), septic shock and DIC (45.3% vs. 12.0%, p > 0.001), and cardiovascular complications (65.3% vs. 18.0%, p < 0.001) (Table 2).

Additionally, 41.1% and 18.9% of patients who were on BiPAP, and high-flow nasal cannula died, respectively. There was no significant difference between patients who were on BiPAP and died versus those who survived. Similarly, there was no significant difference between patients who were on high-flow nasal cannula and died versus those who survived (Table 2).

However, almost 78.9% of patients who got intubated have died (78.9% vs. 36.0%, p < 0.001) (Table 2). There was no significant difference in the ED boarding time between patients who died or survived. Half of the patients who died stayed more than 6 days in the ED (51.9%).

Additionally, 41.1% and 18.9% of patients who were on BiPAP and high-flow nasal cannula died, respectively. There was no significant difference between patients who survived or not and were on BiPAP or high-flow nasal cannula (Table 2).

For the laboratory results, troponin (0.058 ± 0.095 vs 0.026 ± 0.029, p = 0.008) and procalcitonin (1.4 ± 2.8 vs. 0.5 ± 0.8, p = 0.01) were significantly higher in patients who died than those who survived. Hemoglobin level (12.1 ± 2.3 vs. 13.0 ± 2.0, p = 0.021) was significantly lower in patients who died. Other laboratory values including D-dimer, Fibrinogen, ANC, CRP, PT, WBC, and creatinine levels did not show a significant difference (Table 2).

Predictors of mortality using logistic regression

After adjusting for confounding variables using logistic regression, patients > 65 years old were more likely to die (aOR = 11.66, 95%CI = 1.59–85.82). Patients who died were more likely to be on vasopressors (aOR = 11.56, 95%CI = 1.44–92.84) and to have a severe COVID-19 on CT (aOR = 7.07, 95%CI = 1.34–37.42). The patients who received Actemra in the ED had lower odds of dying (aOR = 0.08, 95%CI = 0.02–0.43). Patients who died had significantly more complications including pneumothorax (aOR = 20.7, 95%CI = 1.18–364.28), or cardiovascular complications (aOR = 10.33, 95%CI = 2.05–52.11). Mortality was more likely in those who received high-flow oxygen (aOR = 10.43, 95%CI = 1.14–95.66) and in patients who had endotracheal intubation (aOR = 5.79, 95%CI = 1.23–27.17) (Table 3).

Table 3 Logistic regression: factors associated with mortality in COVID ICU patientsPredictors of intubation in COVID ICU patients

Of the total of 93 patients who were intubated, 70 (75.3%) were males, 61 (65.6%) patients were older than 65 years, and 76 (81.7%) patients had cardiovascular diseases. However, there was no significant difference in gender, age, smoking status, or types of comorbidities between COVID ICU patients who were intubated and those patients who were not intubated. Most patients who were intubated had low oxygen saturation at triage compared to patients who were not intubated (77.4% vs. 56.9%, p = 0.01) (Table 4, in the tables’ section).

Table 4 Association of COVID ICU patients’ demographic and clinical characteristics with intubation

Treatment with steroids, anticoagulants, Baricitinib, Ivermectin, or Actemra had no significant effect on intubation. About 57.1% and 63.6% of patients who were on BIPAP and high-flow nasal cannula, respectively, were intubated, but it was not significant (p > 0.05) (Table 4).

Compared to non-intubated patients, intubated patients had significantly more respiratory complications (74.2% vs. 38.5%, p < 0.001), as well as septic shock (46.2% vs. 11.5%, p < 0.001) or cardiovascular complications (57% vs. 34.6%, p = 0.01). Additionally, 95.6% of patients on dialysis were intubated (23.7% vs. 1.9%, p = 0.001). Patients who were on antibiotics, Remdesivir, and convalescent plasma were significantly more intubated (Table 4).

After adjusting for confounding variables using logistic regression, SaO2 < 95% at triage (aOR = 5.2, 95%CI = 1.58–17.17), convalescent plasma (aOR = 4.09, 95%CI = 1.28–13.14), and respiratory complications (aOR = 7.89, 95%CI = 2.53–24.62) remained significantly associated with intubation. Patients on BiPAP were significantly less intubated (aOR = 0.27, 95%CI = 0.09–0.795). Patients on vasopressors (aOR = 8.27, 95%CI = 2.43–28.15) or dialysis (aOR = 31.87, 95%CI = 2.73–372.08) were significantly more intubated (Table 5).

Table 5 Logistic regression: predictors of intubation in COVID ICU patientsPredictors of complications in COVID ICU patients

Forty-nine patients developed septic shock (33.8%) and 89 patients had respiratory complications (61.4%) including ARDS, pneumothorax, respiratory failure, or pneumomediastinum. Most of them were males (73.5%, 69.7%, respectively) and older than 65 years old (59.2%, 61.8%, respectively). However, none of the gender, age, or type of comorbidities were significantly associated as predictors for septic shock or respiratory complications (p > 0.05). Patients who developed any complication stayed longer in the ED than those who had no complications (4.55 days ± 6.20 versus 7.12 days ± 4.60, p = 0.04). Treatment with steroids, antibiotics, anticoagulants, Baricitinib, Actemra, or Remdesivir had no significant association with respiratory complications, or septic shock (p > 0.05). Patients on convalescent plasma significantly developed more septic shock (46.9% vs. 28.1%, p = 0.024) and respiratory complications (41.6% vs. 23.2%, p = 0.024). Similarly, patients on Ivermectin significantly developed more respiratory complications (44.9% vs. 14.3%, p < 0.001).

Of the 17 patients who had a pulmonary embolism (PE) (11.7%), 10 were males (58.8%). Interestingly, patients who developed pulmonary embolism were significantly younger than 65 years old (64.7% vs. 34.4%, p = 0.015). However, there was no significant difference in gender or type of comorbidities between patients with or without pulmonary embolism. Treatment with steroids, anticoagulants, convalescent plasma, Actemra, or Remdesivir had no significant effects on PE (p > 0.05). Patients on Ivermectin (64.7% vs. 28.9%, p = 0.003) or Baricitinib (35.3% vs. 13.3%, p = 0.031) had significantly more PEs than those not on Ivermectin or Baricitinib.

After adjusting for confounding variables using logistic regression, vasopressor use (aOR = 3.28, 95%CI = 5.35–32.94) and Ivermectin (aOR = 3.03, 95%CI = 1.17–7.86) remained significantly associated with septic shock. O2 high flow (aOR = 0.048, 95%CI = 1.02–22.28), Ivermectin (aOR = 3.57, 95%CI = 1.35–9.49), and intubation (aOR = 8.73, 95%CI = 3.36–22.695) remained significantly associated with respiratory complications. Ivermectin (aOR = 4.04, 95%CI = 1.3–12.55), Baricitinib (aOR = 3.59, 95%CI = 1.04–12.37), and severe CT (aOR = 4.84, 95%CI = 1.19–19.61) remained significantly associated with pulmonary embolism.