2.1 Study Design and Participants

This prospective cohort study included a convenience sample represented by all COVID-19 patients admitted to the ICU of a research hospital, which was adapted to become a COVID-19 center between March and June 2020. The exclusion criteria were: age <18 years, transfer to another hospital regardless of the cause (n = 88), a preliminary diagnosis of COVID-19 that was not confirmed after additional examination (n = 66) and nonfatal cardiac arrest before or at the time of admission to the ICU (n = 13). Criteria for ICU admission are specified in the Supplement (Supplementary Table 1).

Health-related quality of life at 1 year follow-up, starting from the day of patient hospital discharge, was designed as the primary endpoint. The secondary endpoints of the study were the components of the physical and mental quality of life (Supplementary material). The SF-36 was used to assess HRQoL [17]. A special electronic form was employed for the survey. The patient filled in the data on his/her own or with the help of a medical specialist (Supplementary material). Specially trained medical staff made up to three attempts to contact patients by telephone (six months and one year after patient discharge). After patients verbal consent to participate, a letter containing a login and password to access the survey was sent. Moreover, the authors attached an informed consent form to the letter, kindly requesting it to be completed and sent back. If it was not possible to contact the patient, it was marked off as “lost to follow-up”.

2.2 Demographics and disease-related characteristics

The collected socio-demographic variables were age, sex, height, weight, and body mass index (BMI). Disease-related characteristics included comorbidities, need of mechanical ventilation, P/F ratio (PaO₂/FIO₂ ratio) and laboratory parameters at the time of ICU admission (Supplementary Table 2). Patient’s condition at admission was assessed using the National Early Warning Score (NEWS) [18] and its latest version NEWS2 [19], as well as the Sequential Organ Failure Assessment (SOFA) score. The following comorbidities were recorded: Charlson comorbidity index [20] (Supplementary material); chronic obstructive pulmonary disease, cerebrovascular insufficiency, peripheral artery disease, diabetes, arterial hypertension, myocardial infarction, atrial fibrillation, coronary heart disease, congestive heart failure, chronic kidney disease, liver failure, diabetic organ damage, peptic ulcer, hemiplegia and cancer. The length of stay (LOS) in the hospital (total LOS and LOS before transfer to the ICU) and the ICU LOS were also assessed. Medical therapy in the ICU was collected: anticoagulant therapy (Enoxaparin or Unfractionated Heparin with the choice at the discretion of the attending physician), anticytokine therapy (Tocilizumab) as well as Hydroxychloroquine, antiviral therapy and antibiotic therapy. Additional information on prescribed regimens is provided in Supplementary Table 3 and ICU complications are listed in Supplementary Table 4. Data were collected by trained research associates.

2.3 HRQoL assessment

HRQoL analysis was performed using the SF-36 Health Status Survey [17]. This is a standardized non-specific questionnaire, representing one of the most widely used indicators of health in Europe and the United States concerning human quality of life [21]. SF-36 consists of 11 questions and 36 items, combining 8 health areas (domains or subscales, see Supplementary material): Physical Functioning (PF), Role Limitations Due to Physical Problems (RP), Role Limitations Due to Emotional Problems (RE), Vitality (VT), Mental Health (MH), Social Functioning (SF), Bodily Pain (BP) and General Health (GH). Physical functioning, role limitations due to physical problems and bodily pain subscales contribute primarily to the physical component of HRQoL (physical health), while social functioning, emotional well-being and role limitations due to emotional problems affect the mental component of HRQoL (mental health) [22]. The two remaining domains, general health and vitality, are associated with both physical and mental components. The scores are expressed as a range from zero, when the respondent has the worst possible health, to 100, when the respondent is in the best possible health. The subscale scores are then converted by using Z-scores and weighted to obtain the values used to calculate the scores for the physical and mental components of health-related quality of life. Physical health and mental health are integral indicators, while each subscale allows for a more accurate characterization of individual elements of the patient’s health-related quality of life. The reference level for all subscales, physical health and mental health is 50 points and is obtained in general population studies in the USA, Italy, Australia and France [21]. Subsequently, patients were divided into two groups depending on the physical and mental components as follows: (1) less than 50 points (reduced HRQoL) and (2) 50 or more points (normal HRQoL).

2.4 Statistical analysis

All quantitative values were presented as medians (interquartile range—IQR) for continuous variables and reported as absolute values along with percentages for categorical variables. Normality of data distribution was assessed using the Shapiro-Wilk test. To compare demographics, clinical characteristics and health-related quality of life parameters, the Wilcoxon rank-sum test (Mann-Whitney U-test for continuous variables, independent samples) or Wilcoxon signed-rank test (for continuous variables, related samples-comparison of HRQoL after 6 months and 1 year after discharge), as well as Fisher’s exact test (for nominal data, independent samples) and McNemar’s test (for nominal data, related samples) were adopted. The participants were divided into groups according to their health-related quality of life 1 year after discharge (reduced health-related quality of life on the SF-36 scale-physical or mental component <50; normal health-related quality of life—component values ≥50). Boxplot diagrams were used to graph the median and quartiles of HRQoL parameters for different groups. Spearman’s rank correlation coefficient was used to assess the strength of the relationship between quantitative predictors (correlations were plotted and a “heatmap” diagram was created—the “Seaborn 0.11.1” library for “Python 3.7” in the “PyCharm” environment was used for development).

Multivariable analysis based on binary logistic regression with an estimate of the adjusted odds ratio (adjusted OR) and its 95% confidence interval (CI) was used to determine independent predictors of patients’ health-related quality of life in view of the presence of potential bias variables (severity of the condition, comorbidity, co-administration of drugs, etc.). A stepwise inverse method (likelihood ratio) was used to include predictors in the regression model. This method begins by fitting a model with all predictors and discarding the least significant factors at each stage until only significant independent predictors of outcome remain in the model. To assess the quality of the parameters as predictors of HRQoL, we used the receiver operating characteristic (ROC) analysis with the estimation of the area under curve (AUC) parameter and its 95% CI. For sensitivity investigation, we performed an analysis on outcomes of different cut-off values chosen to achieve the optimal sensitivity/specificity ratio according to the results of the ROC analysis (Youden’s J statistic). Comparison of the predictive ability of quantitative parameters (comparison of AUCs) at 6 months and 1 year after discharge was carried out using the DeLong method with the assessment of the Z-test [23]. Positive and negative predictive values were also assessed for the selected cut-off point. If more than 10% of laboratory parameters were missing, an additional note was included in the data table.

All statistical analyses were performed using IBM SPSS Statistics for Windows® Software version 25 (IBM Corp., Armonk, NY, USA) and MedCalc® Statistical Software version 19.5.6 (MedCalc Software Ltd, Ostend, Belgium). The significance level was set at 0.05.

3.1 Baseline characteristics

Overall, 222 out of 403 (55.1%) COVID-19 patients previously admitted to the ICU from March to June 2020 were discharged from the hospital: they were 62 ± 15.3 years old (range from 21 to 97) and 57% were men. Eighteen patients (18/403 = 4.5%) died after hospital discharge and within one year. 103 patients (65 men, 63.1%) agreed to take part in this research project and signed the informed consent 1 year after discharge, while the other 101 discharged patients either refused the survey (n = 47) or were lost for follow-up (n = 54) (Fig. 1).

Fig. 1.Flow chart of the study.

To assess population representativeness, we performed a comparison of baseline characteristics and prescribed therapy in 103 patients who completed the SF-36 forms versus those 119 who did not (died, refused the survey or were lost to follow-up). In general, these groups were comparable in terms of baseline characteristics, severity of the condition upon admission to the ICU and length of stay (all p > 0.05), however, patients who did not complete the SF-36 formless frequently received Tocilizumab (19.3% vs. 40.8%, p = 0.001) and antiviral therapy (20.2% vs. 33.0%, p = 0.03) (Table 1).

3.2 HRQoL of overall group and its comparison to the general population

Overall, at 1 year follow-up, 62 patients (60.2%) had a reduced health-related quality of life with regards to the physical component (reduced physical health) while 36 patients (35.0%) had a reduced health-related quality of life as to the mental component (reduced mental health). Female gender was associated with a worse physical health status—the proportion of women in the cohort of patients with a normal physical component of HRQoL was only 12.2% (5/41) (Table 1). Patients that perceived reduced levels of the physical HRQoL component were significantly older—the percentage of patients in the 50+ category was 83.9% (52/62) in the group of patients with reduced physical health and 53.7% (22/41) in the group of patients with a normal health-related quality of life, Fisher’s exact test: p = 0.001 (Table 1). Patients with normal and reduced physical and mental health were comparable in terms of other medical and demographic parameters, which included illness severity during hospitalization, length of stay in the hospital and in the ICU (Table 1), laboratory test results at the time of hospitalization (Supplementary Table 3) and incidence of complications in the ICU (Supplementary Table 4).

According to the results of univariate analysis, the prescription of low molecular weight heparin (LMWH), unfractionated heparin (UFH), interleukin-6 (IL-6) blockers, hydroxychloroquine, antiviral drugs and antibiotics was not associated with changes in physical and mental health 1 year after discharge (Table 1).

Patients with chronic obstructive pulmonary disease (COPD) (Fisher’s exact test: p = 0.025), peripheral artery disease (Fisher’s exact test: p = 0.01), and a history of hypertension (Fisher’s exact test: p = 0.014) were more likely to perceive a reduced level of quality of life related to the physical component HRQoL at 1-year follow-up (Table 2). In addition, a higher Charlson comorbidity index was observed in patients with reduced physical health-2 (IQR: 1 to 4) versus 1 (IQR: 0 to 3), Mann-Whitney U test: p = 0.024. Comorbidities had no effect on the mental component of health (Table 2).

According to the results of the ROC analysis, the age of patients at the time of hospitalization was a significant predictor of reduced physical health (AUC: 0.72, 95% CI: 0.62 to 0.82, p < 0.001). The optimal cut-off point was set at 50.0 years (Sensitivity: 46.3%, 95% CI: 30.4% to 62.6%; Specificity: 83.9%, 95% CI: 72.3% to 92.0%, positive predictive value (PPV): 65.5%, 95% CI: 50.0% to 78.6%; negative predictive value (NPV): 70.3%, 95% CI: 63.5% to 76.2%) (Fig. 2).

Fig. 2.Roc-curve: age as a predictor of physical health in COVID-19 ICU survivors at 1 year after discharge.

3.3 HRQoL from six months to 1 year

The proportion of patients with a normal mental health increased 1 year after discharge, compared to the six-month period outcomes (McNemar’s test: p = 0.033) (Table 3). At the same time, the proportion of patients with a normal physical health component did not change after 1 year (McNemar’s test: p = 0.3).

One year after discharge, there was an increase in both physical health (46.0 (IQR: 38.0 to 53.0) versus 43.7 (IQR: 31.7 to 52.7), Wilcoxon signed-rank test: p = 0.007, Fig. 3A) and mental health (53.0 (IQR: 47.0 to 56.0) versus 50.6 (IQR: 42.0 to 55.8), Wilcoxon signed-rank test: p = 0.035, Fig. 3B) (Table 3).

Fig. 3.Comparative analysis of the physical. (A) and mental (B) Health-Related Quality of Life (SF-36 survey) in COVID-19 ICU survivors at 6 months and 1 year after discharge. Normative value is highlighted in green.

We compared the median values of all eight SF-36 subscales at 6 months and 1 year follow-up to determine the main factors contributing to the positive changes in the patients’ health-related quality of life 1 year after discharge. The increase in COVID-19 patients’ health-related quality of life was associated with: (1) improvement in the physical functioning of patients (PF), from 70 (IQR: 35 to 85) to 75 (IQR: 55 to 95; Wilcoxon signed-rank test: p < 0.001); (2) increased role functioning due to physical condition (RP), from 50 (IQR: 0 to 100) to 100 (IQR: 50 to 100; p < 0.001); (3) improved mental health (MH), from 64 (IQR: 48 to 72) to 68 (IQR: 64 to 76; p < 0.001); (4) improved social functioning (SF), from 75 (IQR : 56 to 100) to 88 (IQR: 63 to 100; p = 0.007) (Table 3).

Compared to females, males showed greater improvement in the physical performance and better general health, resulting in increased role functioning at 1 year follow-up. (Table 3).

3.4 Regression analysis

In multivariable regression analysis, age 50+ (adjusted OR: 0.270, 95% CI: 0.098 to 0.739, p = 0.011) and female sex (adjusted OR: 0.144, 95% CI: 0.048 to 0.437, p = 0.001) were independently associated with reduced physical health at the 1-year follow-up. The administration of anticoagulants and other drugs in the ICU did not affect 1-year HRQoL. Moreover, there were no significant predictors of mental health alterations 1 year after discharge.

3.5 Correlation analysis

Patients with an already high physical component of HRQoL 6 months after discharge presented with a further improvement in physical health after 1 year (Spearman r = 0.421, p < 0.001) (Fig. 4A). A similar but weaker relationship was observed for mental HRQoL (Spearman r = 0.222, p = 0.024) (Fig. 4B). A moderately strong negative correlation was found between age and physical functioning (PF, Spearman r = −0.52, p < 0.001), while a weak negative interaction was observed between age and role functioning due to physical condition (RP, Spearman r = −0.37, p = 0.015), age and physical pain (BP, Spearman r = −0.30, p = 0.011), age and general health (GH, Spearman r = −0.35, p = 0.021) (Fig. 5). Other correlations between health-related quality of life, medical-demographic characteristics, values of laboratory parameters and length of hospital and ICU stay are shown on the heat map (Fig. 5).

Fig. 4.Scatter plot and linear regression: relationship of Health-Related Quality of Life at 6 months and 1 year after discharge. Fig. 4A: physical Health-Related Quality of Life. Fig. 4B: mental Health-Related Quality of Life.

Fig. 5.“Heatmap” diagram of the correlation matrix (spearman’s correlation coefficient): the relationship between health-related quality of life, medical and demographic parameters, characteristics of the condition upon admission, laboratory parameters and length of stay in hospital and ICU. Positive relationships are highlighted in red, negative ones in blue, the intensity of the color is related to the strength of the relationship.

4.1 Strengths and limitations of the study

To our knowledge, this is the first study to present data on 1-year follow-up health-related quality of life of patients discharged from the ICU after COVID-19. The study, however, has some limitations: we acknowledge that this is a single center study with obvious external validity limitations and with some lack of follow-up data; moreover, we observed a relatively high hospital mortality rate, even considering that the cohort was limited to ICU patients; furthermore, we failed to specifically identify the causes of death. Because of the significant number of patients who were lost to follow-up, the representativeness of the study population seems to be limited. We observed a difference in the number of patients receiving Tocilizumab and antiviral drugs in patients completing and not-completing the SF-36 forms.

4.2 Future studies and prospects

Future studies should be performed on larger samples and with higher adherence to confirm our findings that COVID-19 ICU survivors experience a health-related quality-of-life improvement between the 6-month and the 1-year follow-up after discharge.

In addition, clear guidelines should be developed for the assessment of the health status of patients with COVID-19 after discharge from the ICU.

Importantly, there is a much-needed comparison to the healthy population QoL standard, which would answer the question whether the changes after 1 year are only related to the ICU stay or other conditions (e.g. aging, comorbidities).