Introduction

Critically ill COVID-19 patients present greater need for life-sustaining treatments, high mortality, and prolonged length of intensive care unit (ICU) stay [1]. In-hospital mortality can reach 31% and is expected that as much as 76% of patients will require invasive mechanical ventilation (IMV) [2]. Worst long-term sequelae are also expected for ICU COVID-19 survivors, although the one-year mortality following ICU discharge appears to be influenced more by patient’s comorbidities rather than the severity of COVID-19 pneumonia itself [3].

In relation to the recovery from COVID-19, it is not uncommon to witness the persistence of symptoms for several months following the acquisition of SARS-CoV-2 infection [4]. This condition has been recognized as "post-COVID-19 condition" (PCC) in medical literature [5]. However, the understanding and evidence concerning long-term outcomes in individuals who required critical care due to COVID-19 diagnosis are currently limited. A prospective cohort study conducted in Spain in 2021, evaluating the sequelae occurred at the first month after hospital discharge among patients who required ICU admission for severe COVID-19 pneumonia, concluded that these patients need a long-term follow-up due to its expected bad prognosis, and this should be extended beyond 30 days after ICU discharge [6]. However, evidence regarding the optimal duration for follow-up care in individuals who have survived COVID-19 after being admitted to the ICU is still lacking [7]. The present study aimed to assess early and late mortality after ICU discharge among surviving critical COVID-19 patients and describe their most relevant predictors, in order to propose recommendations regarding the most appropriate follow-up strategy for COVID-19 ICU survivors.

Materials and methods

Results

From May 2020 to July 2021, 601 COVID patients admitted to the ICU were screened and 586 met eligibility criteria and were enrolled in the study. 468 patients survived after ICU admission and were included in the 12-month follow-up cohort. Among the 586 patients enrolled, 155 (26.5%) died before completing 1 year of follow-up and of these, 118 patients (20.1%) died in the ICU and 37 (6.4%) patients died after ICU discharge. Amongst the 468 patients eligible for the 1-year follow-up, which was completed on August 16, 2022, all patients were assessed.

The characteristics of all patients enrolled are summarized in Table 1. Median age was 65 years; 32.1% of patients were women. Regarding pre-ICU state of health, 75.9% of patients had high comorbidity and the most common comorbidities were hypertension (61.3%), obesity (35.8%), and diabetes (30.7%). Asthma and chronic obstructive pulmonary disease (COPD) were present in 4.9% and 5.6% of the patients, respectively.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Characteristics of enrolled patients.

https://doi.org/10.1371/journal.pone.0293883.t001

Regarding critical illness, 1.2% of patients were admitted to the ICU due to non-medical (surgical) reasons. At ICU admission, the median SAPS-2 score was 33. According to the highest score on the WHO ordinal severity scale during hospital stay [8], 30 (5.1%) were categorized as score 2, 26 (4.0%) as score 3, 177 (30.2%) as score 4, 208 (35.5%) as score 5, and 145 (24.7%) as score 6. Regarding organ dysfunctions during ICU stay, 30.9% of patients needed non-invasive oxygen therapy, 53.9% of patients needed MV and 51.4% needed vasopressors. Less than 10% needed RRT and delirium was diagnosed in 19.6%. The median length of ICU and hospital stay were, respectively, 9 and 18 days, and the incidence of any ICU-acquired infections was 43,5%, mostly bacterial pneumonia (38.4%). Almost 12% of patients had treatment limitations justified by clinical decision during ICU stay.

Mortality

At the time of ICU discharge, health status was evaluated as “worse than it was before ICU admission” by 88.9% of the patients and 1.3% were directly discharged home after ICU. Overall ICU mortality was 20,1% [95%CI: 17.1% to 23.6%] (118 of 586 critically ill COVID-19 patients) and overall hospital mortality was 25.4% [95%CI: 22.1% to 29.1%] (149 of 586 critically ill COVID-19 patients).

We considered as early mortality all deaths that occurred during the first 30 days after ICU discharge, and we observed a 5.3% [95%CI: 3.6% to 7.8%] (25/468) early post-ICU mortality rate, among ICU survivors. Late mortality was defined as deaths occurring during the first 365 days after ICU discharge. Between days 31st and 365th after ICU discharge, we observed a 2.7% [95%CI: 1.6% to 4.7%] late post-ICU mortality rate, among ICU survivors that were alive at 30 days after ICU discharge (12/439). Thus, we observed an overall mortality rate among ICU survivors during the first year after ICU discharge of 7.9% [95%CI: 5.8% to 10.8%] (37/468) and an overall 1-year mortality among all critically ill COVID-19 patients of 26.5% [95%CI: 23.0% to 30.2%] (155/586). Among the 37 patients who died after ICU discharge, 31 patients died in-hospital (from these, 25 patients died up to 30 days and 6 died between 31 and 365 days after ICU discharge) and 6 patients died after hospital discharge.

Regarding all COVID-19 patients admitted to the ICU, one-year mortality was 15.9% [95%CI: 11.6% to 20.3%] and 35.4% [95%CI: 30.2% to 40.7%] in non-invasive ventilated patients (43 of 270 patients) and patients submitted to IMV (112 of 316 patients), respectively. Fig 1 shows survival data over time. Median time to death was 11 days after ICU discharge (IQR, 7.0–23.0) for early mortality, and 22 days after ICU discharge (IQR, 8.0–65.5) for all patients dying during the first year after ICU discharge.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. ICU survivors data represented as Kaplan-Meier survival curves.

Caption: A, Survival curve of the 30 days after ICU discharge. B, Survival curve during the one-year follow up.

https://doi.org/10.1371/journal.pone.0293883.g001

Predictors of ICU mortality and early and late post-ICU mortality

Results of univariable analysis for predictors of ICU mortality and multivariable analysis for predictors independently associated with ICU mortality are shown in S1 and S2 Tables, respectively. Severity of critical illness, age greater than or equal to 65 years, delirium, ICU-acquired bacterial pneumonia, need of RRT, oxygen therapy, vasopressor, and the length of ICU stay were associated with ICU mortality.

Results of univariable analysis for predictors of early and late post-ICU mortality are presented in Table 2 and multivariable analysis for predictors independently associated with early and late post-ICU mortality are presented in Table 3. Detailed univariable analysis for predictors of early and late post-ICU mortality can be found in S3 and S4 Tables, respectively. Previously diagnosed conditions (comorbidities) were identified as the most relevant and significant independent predictors associated with early and late mortality after ICU discharge. Pre-ICU hypertension (HR, 5.24; p = 0.007) and pre-ICU heart failure (HR, 4.74; p = 0.002) were independent predictors associated with early post-ICU mortality. Pre-ICU hypertension (HR, 4.96; p < 0.001), pre-ICU chronic kidney disease (CKD) (HR, 1.84; p = 0.007), pre-ICU chronic obstructive pulmonary disease (HR, 4.62; p = 0.018) and pre-ICU cancer (HR, 1.79; p = 0.017) were the independent predictors associated with late post-ICU mortality.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Univariable analysis of predictors associated with early (up to 30 days after ICU discharge) and late mortality (up to 365 days after ICU discharge).

https://doi.org/10.1371/journal.pone.0293883.t002

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Multivariable regression models with independent predictors for early and late post-ICU mortality.

https://doi.org/10.1371/journal.pone.0293883.t003

Discussion

In the present cohort study, performed in adult critically ill patients with COVID-19 admitted to the ICU and followed up during the first 12 months after ICU discharge, we observed an ICU mortality of 20.1%, which is inferior to previous studies where the reported mortality ranged between 30 to 65% [12–14]. Moreover, regarding ICU survivors, early mortality, considering 30 days after ICU discharge, was 5.3% and late mortality (from day 31 to day 365) was 2.6%; thus, the overall 1-year mortality observed among ICU survivors was 7.9% and the overall 1-year mortality among all the COVID-19 patients admitted to the ICU was 26.5%.

There are very few studies reporting the long-term outcomes of critically ill COVID-19 patients. Those studies have shown a long-term mortality (at 1-year follow up) for COVID-19 survivors between 3.5% and 7.5% [3, 15], and these studies rarely report separate estimates for early and late mortality. In the present study, both early and late ICU mortality were higher in patients with more severe COVID-19 illness, defined as need for MV during hospitalisation, which is consistent with the reports from previous studies and a systematic review by Serafim et al. [1]. Despite MV is a necessary treatment for most severely ill COVID-19 patients, it has been associated with adverse long-term outcomes, including hospital readmission and death [16].

We observed that advanced age, SAPS 2 score, delirium during ICU, ICU-acquired bacterial pneumonia and the severity/complexity of management of critical disease (need for RRT, oxygen therapy and vasopressors) were associated with ICU mortality for COVID-19 patients. This is consistent with previous findings from observational studies showing that COVID-19 patients who died at ICU tended to be older, experienced more severe disease, had a greater burden of comorbidities, and developed more complications during their ICU stay [17, 18].

We found a 30-day mortality of all COVID-19 patients admitted to the ICU of 24.4%. This is an estimate higher than those reported in some previous studies, for example, a retrospective cohort study involving 2406 critically ill patients aged 70 years and older, that reported mortality rates ranging from 14% to 17% [19], but it is in agreement with a Swedish cohort study, with 6350 adult COVID ICU patients admitted between March 2020 and April 2021, showing a 30-day mortality of 24.3% [20]. Most of all early mortality studies considered all COVID patients admitted to the ICU and not only ICU survivors. Regarding ICU survivors, it is expected an early mortality rate near to 8% for non-COVID ICU patients [21], a little higher than our COVID-19 ICU survivors early mortality rate (5.3%).

In the present cohort, considering all COVID-19 patients admitted to the ICU, one-year mortality (26.5%) was similar to other studies, close to 30% [3, 22]. Late mortality observed in our study for ICU survivors (7.9%) was higher than previous studies from Europe and Latin America, that showed late mortality rates among COVID-19 ICU survivors lower than 5% [3, 22, 23]. Further studies focusing on long-term health status of ICU-survivors are warranted to better explore long-term mortality among ICU COVID-19 survivors, but these results should draw our attention to the need to maintain a closer monitoring during the initial phase after ICU discharge. This same recommendation emerged from previous studies on the general ICU population, showing an early phase after ICU discharge characterized by high mortality rates, followed by a gradual decrease in the risk of death in a 60-day period post-ICU discharge [21].

In our study, 35.4% of COVID-19 patients that received IMV at the ICU died during the first year after ICU admission, data not much lower than that reported [22]. Several potential explanations can be explored to elucidate the correlation between the requirement of MV and unfavorable long-term prognosis among individuals who have survived COVID-19. As previously observed, comparing with other causes of pneumonia, it can be inferred that COVID-19 pneumonia has a lower late mortality rate among ICU survivors [24]. The cause behind reduced long-term mortality rates for COVID patients remains unclear, as the incidence of pulmonary anatomical abnormalities and compromised physiological function is significantly high among individuals diagnosed with ARDS as a consequence of COVID-19 [6]. Additionally, research findings indicate that COVID-19 patients necessitating MV exhibit heightened levels of inflammatory markers, multiple organ dysfunction, and increased rates of in-hospital mortality [24–26]. These findings may underscore the influence of pre-existing health conditions on adverse outcomes among survivors of severe COVID-19 cases.

Regarding predictors of post-ICU mortality, pre-ICU comorbidities were the only factors associated with mortality after ICU discharge. Similarly, in our study, previous heart failure increased early mortality and previous severe CKD, COPD and cancer, increased late mortality. Our findings are aligned with prior studies, indicating that the presence of pre-existing comorbidities has a greater impact on post-discharge mortality than the severity of COVID-19 pneumonia itself [3, 15, 27, 28]. Other cohort studies also showed that previous hypertension and CKD increased both early and late mortality [29–31]. Previous studies have consistently shown a significantly higher risk of in-hospital deaths, ICU admissions, and the need for invasive ventilation among individuals with a combination of COVID-19 infection and hypertension [32]. It has been shown that COVID-19 patients admitted to the ICU had a higher comorbidity burden than those not admitted to the ICU [33].

Our data support that the burden directly attributable to COVID-19 in terms of hospitalisation and mortality is notably elevated, surpassing the impact attributed to acute illness alone. Clinicians should develop an effective post-discharge follow-up plan for these patients, considering the factors that may contribute to mortality after discharge.

Our study has several strengths, including the examination of both the ICU period and the post-discharge follow-up period, which provides valuable insights into the planning and management of post-ICU care [34]. In addition, it is a prospective cohort with the assessment of patient-centered outcomes contributing to the standardization of eligibility criteria and data collection methods. This approach not only facilitated consistent data collection but also minimized the occurrence of missing values for fundamental variables. A multicentre design and interviews conducted by trained personnel unrelated to the patient’s care appeared to be successful to avoid potential bias.

Some limitations must be considered. First, the inclusion of patients during the first wave of the pandemic, when the healthcare system was overwhelmed and still in the process of organization may have resulted in some missing data; however, it has been shown that there were no significant differences in ICU mortality observed between waves [17]. Second, COVID-19 can have different effects on long-term outcomes in different contexts as a consequence of differences in post-discharge access to health services and we have not considered these aspects in the present study. Particularly, our study did not evaluate treatment limitations imposed following ICU discharge, which could potentially be significant factors contributing to post-ICU mortality. Third, it is important to note that the present study includes only patients enrolled during the first 14 months of the COVID-19 pandemic, which may limit the generalizability of the results, especially considering the substantial changes in the course of the pandemic over time, mostly after the Omicron variant became dominant [35]. However, we should also highlight the importance of the rigorous analysis of the initial phase of the pandemic and the many relevant lessons learned from this, as most of the predictors of long-term outcomes in critical care described and discussed are applicable and generalizable in broader contexts. Finally, we did not assess variables that have the potential to modify the association between the acute severity of COVID-19 and long-term outcomes, such as vaccination status and infection by different variants of the virus.

Conclusion

By reporting on the early and late mortality post-ICU discharge in critically ill COVID-19 patients, we have shown that there is a period immediately following ICU discharge, specifically in the first 30 days, with a higher mortality rate, but this risk decreases over time up to 1 year. Therefore, it is crucial to closely monitor critically ill COVID-19 patients who have survived intensive care, particularly those with pre-existing comorbidities, to prevent adverse outcomes and mortality, as our findings indicate that pre-ICU comorbidities were the sole factors associated with mortality after ICU discharge. Since we identified a limited number of previous studies, further research is urgently needed on post-ICU mortality rates and to establish an appropriate post-ICU follow-up strategy for surviving critically ill COVID-19 patients.

Supporting information

Acknowledgments

The authors would like to thank the medical staff of each participating ICU for their support and cooperation in data collection.