High rates of impaired quality of life and social and economic problems at 6 months after COVID-19-related ARDS

Consecutive patients admitted to the ICUs of two distinct referral hospitals in the northeast Italian region Trentino-Alto Adige (S. Chiara Hospital-Trento and S. Maria del Carmine Hospital-Rovereto) between March 5 and April 30, 2020, were screened for enrollment in this prospective, cohort study investigating the impact of COVID-19 on the quality of life and the socio-economic status. Adult (> 18 years), Italian-speaking patients admitted to ICU (> 24 h) for COVID-19 ARDS, according to the Berlin definition [21], were included. Those who did not respond at follow-up or denied participating were excluded. A flow chart of the study is shown in Fig. 1.

Bild vergrößern

The local ethics committee approved the study, as part of a follow-up study on COVID-19 survivors (APSS; protocol number 1/2021), including a waiver of consent for hospital data collected during the pandemic surge (APSS; protocol number 3/2020). Six months after ICU discharge, COVID-19 ICU survivors were firstly contacted by experienced ICU physicians (AS, SM) with an informative phone call, followed by an email with a brief description of the study. Patients were asked by email for their consent to participate to the study and appointed for a telephone interview. Since restrictions due to the pandemic prevented the possibility to perform face-to-face interviews, the follow-up was performed on the telephone, by an experienced ICU physician previously involved in patient clinical management.

At the time of the structured telephone interview, patients were administered a set of questionnaires including the five-dimension, five-level EuroQoL questionnaire (EQ-5D-5L), the EuroQol Visual Analogue Scale (EQ VAS), and the Short Form Health Survey 36, version 2 (SF-36v2) to assess the HR-QoL, a novel question set designed to determine changes in family circumstances, socio-economic status, and care requirements. The duration of the interview was approximately 45 min.

The EQ-5D-5L scale describes 5 dimensions: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression [22]. Each dimension has 5 levels: 1 no problems, 2 slight problems, 3 moderate problems, 4 severe problems, and 5 extreme problems. The respondent is asked to indicate his/her health state in each of the 5 dimensions. Based on the five answers, a summary EQ-5D-5L index was calculated using a recently published value set for the Italian people, with values ranging from −0.571 for the worst health state (55,555) to 1 for the best health state (11,111). As a reference value, we used mean 0.91 (SD 0.14) for an age-matched population [23].

The EQ VAS records the respondent’s self-rated health on a vertical, visual analogue scale where the endpoints are labeled “Best imaginable health state” and “Worst imaginable health state.” This information can be used as a quantitative measure of health outcome as judged by the individual respondents.

The SF-36 is an 11-question, 36-item questionnaire frequently used to describe HR-QoL after ICU [24, 25]. It investigates 8 health domains: physical functioning, the extent to which health limits physical activity; physical role, the extent to which physical health interferes with work or limits activity; pain, the intensity of pain and the effect of pain on patient’s ability to work; general health, patient’s perception of his or her health; vitality, the degree of energy the patient has; social functioning, the extent to which health or emotional problems interfere with social activities; emotional role, the extent to which emotional problems interfere with work or activities; and mental health, general mental health. Estimate scores for the 8 domains were calculated as recommended in the specific manual and interpretation guide. The overall score on each SF-36 subscale ranges from 0 to 100 with higher scores indicating a better HR-QoL. We also calculated the SF-36 physical (PCS) and mental (MCS) component summary measures by combining the subscale scores physical and mental components, as recommended in the specific manual [26]. A reference level of 50 (SD 10) points was adopted for both components, as calculated for the US population, and validated in Australia, France, and Italy [26]. Patients were then classified into two groups based on physical and mental components as follows: (1) a reduced HR-QoL if the score was lower than 1 SD below the mean reference value (i.e., < 40) and (2) normal HR-QoL if the score was higher than 1 SD below the mean reference value (i.e., > 40).

The socio-economic questionnaire consists of a multiple-choice questionnaire to evaluate the socio-economic-relational impact on patients and families, previously used in general intensive care survivors in the UK and subsequently adapted to the Italian context with the help of a health sociologist (prof. Annamaria Perino, Department of Sociology and Social Research, University of Trento, Italy), who validated the socio-economic question set [14]. In detail, we collected the following: demographic and socio-economic data about patients and family, economic and socio-relational impact of the disease and the ICU stay, information on health service utilization and need for care after ICU discharge, and observed possible stigmatization processes of the survivor and of their families.

Epidemiological and clinical information about ICU stay, severity scores, and laboratory data was manually or digitally extracted and recorded from our electronic system. All data were anonymized and saved in Microsoft Excel files. Education levels were classified according to the International Standard Classification of Education (ISCED): ISCED 0, early childhood education (“less than primary” for educational attainment); ISCED 1, primary education; ISCED 2, lower secondary education; ISCED 3, upper secondary education; ISCED 4, postsecondary non-tertiary education; ISCED 5, short-cycle tertiary education; ISCED 6, bachelor’s or equivalent level; ISCED 7, master’s or equivalent level; and ISCED 8, doctoral or equivalent level.

Descriptive results for continuous variables are reported as mean (SD) if normally distributed, or median (25th–75th percentiles) otherwise, and as count and percentages when indicated. Variable distribution was tested by the Kolmogorov–Smirnov and Shapiro–Wilk tests.

As social and disease-related risk factors, eight clinically and epidemiologically relevant variables (i.e., age, education, income, comorbidities, worst PaO₂/FiO₂ ratio on the first ICU day), SAPSII (simplified Acute Physiology Score), days of artificial ventilation with PEEP (either invasive or noninvasive), and length of hospital stay (LOS) were selected a priori and dichotomized using meaningful reference values, as follows: age ≤ or >65 years (as relevant to COVID-19 [27]), education ISCED ≤ or > 2 (identifying low or medium/high education), income ≤ or >1500 € (as the median value), comorbidities ≤ or > 1 (as relevant to COVID-19 [27]), PaO₂/FiO₂ ratio ≤ or > 150 mmHg (as performed in previous ARDS studies [28]), SAPS II ≤ or > 25 (as the median value), and artificial ventilation and hospital stay, respectively, ≤ or > 15 days and 4 weeks (as relevant to COVID-19 [29]). Each of the EQ-5D-5L dimensions were grouped in two categories: “no problems” and “some problems,” this latter including slight, moderate, severe, and extreme problems, if applicable. Mann Whitney U-test was used for comparisons of non-normally distributed continuous variables, one-sample t-test was used to compare SF-36 dimensions and EQ-5D-5L summary index of survivors with normative data from the general population, and Fisher test was used to analyze categorical variables between groups.

Univariate binomial logistic regression analysis was used to explore risk factors (covariates) associated with each of the EQ-5D-5L domains (dependent variable): pain, mobility, usual activity, self-care problems, and anxiety and depression. Reference value for the dependent variables was set as follows: no problems vs. some problems. Continuous variables were transformed as follows: age/10, PaO₂/FiO₂/100, SAPSII/5, duration of ventilation (days)/10, and hospital LOS (days)/10. Categorical variables (education, income, and comorbidities) were dichotomized as previously described. Wald test was used for regression analysis.

Multivariable binomial logistic regression analysis was used to explore associated factors with reduced HR-QoL, assessed by SF-36 PCS and MCS < 40. For model selection, we used a stepwise selection approach based on the Akaike information criterion (AIC) and calculated the adjusted odds ratios with 95% confidence intervals.

To account for the presence of both patients who only received noninvasive mechanical ventilation, as opposed to patients who were invasively ventilated, two sensitivity analyses were performed; we compared the demographic, socio-economic, and clinical characteristics of the two subgroups of patients, and we repeated all the analyses only in the subgroup of invasively ventilated patients.

Of the 93 potentially recruitable patients, 79 were included in the study (Fig. 1). Clinical and demographic data are described in Table 1. The median age of the patients was 63 years, and 84% were male, 54% were pensioners, and 65% had an ISCED level > 2; most of the patients lived in small towns (< 15,000 inhabitants) as a reflection of the geographical composition of a rural territory. A total of > 75% of the patients had no or only one preexisting comorbidity, mostly hypertension (46%). All patients had ARDS, 69 (87%) required invasive mechanical ventilation, and 10 (13%) received noninvasive ventilation with PEEP ≥ 5 cmH₂0. Laboratory data and information about ICU treatments are summarized in Table 1. The characteristics of the noninvasively ventilated C-ARDS patients and the comparison with the mechanically ventilated patients are shown in Supplementary Table S1. Despite similar gas exchange derangement on ICU admission, noninvasively ventilated patients had a higher BMI, received no or less rescue therapies, and had a shorter ICU and hospital stay.

Social and relational life were affected by C-ARDS in a significant proportion of subjects. Table 4 shows that 38% of the patients were negatively affected in their relationships, 58% of the active workers had returned to full-time work after 6 months, whereas 42% had a change in their employment status. A total of 23% of subjects were requiring support in personal care and in activities of daily living, for 20 h a week or less; only two subjects required more assistance. The description of met and unmet health professional needs is summarized in Fig. 5 and indicates physiotherapy as the most prevalent unmet need.

Bild vergrößern

In the last decade, an increased awareness towards the long-term consequences of critical illness has been developed, as ensuring good long-term quality of life, rather than simply survive the acute event, has become the main aim of intensive care medicine. Indeed, up to 30% of ICU survivors develop cognitive, physical, and psychological sequelae, independently of the reason for admission [31]. Follow-up of ICU survivors has been suggested to facilitate prompt recognition and treatment of such symptoms and improve long-term physical, psychological, and cognitive outcomes [32]. A total of 80% ARDS survivors experience some degree of cognitive impairment [33‐35], as well as a high rate of persistent psychological and physical disability [36] and a reduced HR-QoL [37]. Moreover, viral infections are also characterized by several long-term manifestations [38]. Long-term follow-up of SARS and MERS survivors showed a 30–40% prevalence of post-traumatic stress disorder, depression, anxiety, chronic fatigue, and reduced HR-QoL [39‐41]. Like other post-acute viral syndromes, emerging data suggest persistent and prolonged effects after acute COVID-19 [42]. In our cohort, problems in mobility and activities of daily life were reported in about a half of survivors, namely two to three times higher compared to the reference Italian population; compared to norm, impairment in self-care was even more profound (i.e., four times higher). Impressively, only < 25% survivors reported full health. A survey on > 3500 respondents from 56 countries found that systemic and neurological symptoms such as fatigue, cognitive dysfunction, post-exertional malaise, headache, memory troubles, sleep disorders, dizziness, and chest tightness were common and persisting at 6 months after recovery [43]. While still not entirely understood, the underlying mechanism of the long-term neurocognitive consequences of COVID-19 is likely to be multifactorial and might include the direct effects of viral infection, the immunological response, corticosteroid therapy, ICU stay, social isolation, and stigma [44]. Quite interestingly, our results suggest that while previous health state and the severity of the acute illness have a significant impact on the long-term consequences of COVID-19 on health-related quality of life, the level of education and the premorbid income do not.

We found that C-ARDS survivors reported significantly lower HR-QoL domain scores compared to population norms, with the exception of pain (i.e., the most frequently reported problem by normal people [30]) and some of the emotional problems, as assessed by SF-36. Anxiety and depression symptoms were also not significantly different compared to population norms, when using EQ-5D-5L. While it is debated if COVID-19-related acute respiratory failure could be defined as a classic form of ARDS from a pathophysiological point of view [45], in terms of supportive therapy, the disease seems to be similar to the severe forms of classic ARDS, requiring endotracheal intubation, deep sedation, and prolonged mechanical ventilation in the majority of patients. Indeed, the HR-QoL profile observed in our cohort of patients is similar to those observed in non-COVID-19 ARDS survivors [46, 47] or in survivors of ARDS due to other viral infections [48].

As a matter of fact, the physical and psychological sequelae of COVID-19 may even be more frequent than classic ARDS, because of the restriction of visitation and the limitations to social and rehabilitation supports. Of note, only few other studies assessed the HR-QoL in follow-up studies after severe COVID-19 disease. A preliminary report of a 3-month, French follow-up study on 54 ICU survivors found that all patients reported an impaired quality of life, and up to 80% described pain or discomfort in their daily life; almost half complained about mental health disturbances and worsened mobility due to muscular weakness [49]. In a 2-month follow-up in a small cohort of invasively ventilated COVID-19 patients, the overall HR-QoL was reduced to a similar extent as we found, with cognitive and psychological scales showing no impairment, and the most affected dimensions being self-care, usual activities, and pain [50]. In a multicenter cohort mechanically ventilated COVID-19 patients, the HR-QoL at 90 days after ICU discharge was significantly lower than sex- and age-matched random sample population in both physical (mobility, ability to speak, ability to eat) and mental (discomfort, depression, vitality, sexual activity) dimensions [51]. In a 3-month follow-up study, about 75% C-ARDS survivors showed pulmonary functional and structural abnormalities, together with symptoms of dyspnea and cough and a reduced self-reported physical and mental quality of life [10]. Unlike other studies [52], anxiety, depression, and mental problems were far less frequently reported in our cohort of survivors. We wonder whether this might depend on the relatively younger age or prevalent male sex of the patients we enrolled [53], the different tools used for the assessment, or the peculiar characteristic of the national and regional social support system.

To the best of our knowledge, this is the first study to investigate the socio-economic impact of COVID-19-related critical illness. More than 30% of survivors report a worsening of their social relations after returning home, and about 20% require some form of health support during the first 6 months after discharge. About 40% of the patients had not completely returned to their previous employment during the follow-up. This is not unexpected, as a recent systematic review found that about two-thirds of previously employed intensive care unit survivors are jobless up to 3 months following hospital discharge, while those returning to work often experience job loss, occupation change, or worse employment status [54]. As for care received by COVID-19 survivors, this was mainly provided by family members; moreover, about one out of four caregivers reported a change in their employment status. In a UK study, Griffiths and coll. found that one quarter of patients followed up after ICU discharge reported themselves in need of assistance with care at 6 months; reduction in employment and disruption in lifestyle were common in the caregivers, and a negative impact on employment and on reported family earning sources and income was experienced by about 20% of respondents [14].

The length of hospital stay, and the number of comorbidities, was independently associated with a lower perceived quality of life in ICU survivors. Prolonged mechanical ventilation, ARDS class, and comorbidities are known to be associated with worse HR-QoL in “classic” ARDS survivors [46, 47]. In a recent study, the HR-QoL of severe COVID-19 ICU survivors was significantly associated with age, sex, number of comorbidities, the severity of ARDS, duration of invasive mechanical ventilation, and occupational status, while the marital status and degree of instruction were not [51]. A recent prospective cross-sectional, global, social media online survey assessed > 700 COVID-19 survivors and their caregivers’ health-related quality of life and found how survivors with pre-existing health conditions reported significantly more problems on their physical and psychosocial health [55]. Hypertension and diabetes mellitus are among the most prevalent comorbidities in COVID-19 patients and often coexist as multiple comorbidities, as in our cohort. Age was not associated with worse ICU outcome in our study; an explanation may be the young age of the patients or the small sample size.

Quite unexpectedly, neither education nor the monthly income was found to be associated with any HR-QoL domain. However, these social variables are known to play a significant role in determining the HR-QoL [56], and a recent French investigation found that long-term physical recovery appeared to be poorer in socially deprived ICU survivors [17]. On the one hand, several studies have identified material living conditions (housing and neighborhood), health behaviors, early childhood conditions, and psychosocial stress as important factors underlying the association between social status and health [57]. On the other side, it is also possible that, at 6 months from ICU discharge, the entity of disability related to clinical factors still outweighs the social aspects in determining HR-QoL. Another possible explanation to these apparently conflicting results is that the Italian context (and even more the Autonomous Province of Trento), as compared to that of the other studies, despite worse job indicators and lower average incomes, is characterized by a higher degree of social and family support, a better work-life balance, and an easier access to the health care system, with a lower influence of the education level on life expectancy, which might have counterbalanced the baseline inequality in patient income or education level [58].

These results suggest that patients with severe disease and prolonged hospital stay or more comorbidities may need post-discharge care. While we wait for longer follow-up studies in a larger population, which are necessary to understand the full spectrum and the duration of health consequences from COVID-19, we highlight the need for a multidisciplinary follow-up, involving different healthcare (such as physiotherapists) and social sector professionals (in particular social workers, who played an active role in the management of social fragilities highlighted by the health emergency) [59], able to give integrated answers that are fundamental for an adequate recovery and rehabilitation after COVID-19 critical illness. SF-36 and EQ-5D-5L questionnaires are both valid and complementary tools to identify COVID-19 ICU survivors with persistent symptoms and reduced quality of life.

Several limitations of this study need to be acknowledged; as with any follow-up study, the loss-to-follow-up rate may limit the generalizability of the results; this phenomenon was however comparable to other studies [8, 14, 49, 50]. The patient cohort was relatively small and from only two nearby regional centers, which again may limit the generalizability of the results.

In the current investigation, we included in the HR-QoL assessment of survivors both patients who received invasive mechanical ventilation and patients who only received noninvasive ventilation. Whether this might lead to biased results is still an unresolved issue; in a recent follow-up study on critically ill, C-ARDS patients, 23% of the patients were noninvasively ventilated patients [60]. Latronico et al. assessed 1-year physical, cognitive, and mental health outcomes in a group of critically ill, C-ARDS survivors of whom about 10% only underwent noninvasive ventilation [61]. A similar proportion of noninvasively ventilated patients was included in a HR-QoL investigation in “classic” ARDS survivors [62]. Nevertheless, to account for the presence of both patients who only received noninvasive ventilation, as opposed to patients who were invasively ventilated, we performed two sensitivity analyses, presented in the supplementary materials. We first compared the demographic, socio-economic, and clinical characteristics of the two subgroups of patients, and then we repeated all the analyses only in the subgroup of invasively ventilated patients. Quite interestingly, the two groups had similar demographics and acute physiologic derangements, although noninvasively ventilated patients had lower stays in the ICU and hospital. Albeit our results seem to suggest similar HR-QoL patterns in patients with C-ARDS irrespective of the presence of invasive mechanical ventilation, we caution against such conclusions since the present study is underpowered to address this issue.

Other limitations include the lack of a baseline measurement of HR-QoL before the acute illness and the use of a questionnaire for socio-economic outcomes not previously used in the Italian context. Moreover, we were not able to evaluate potential predictors that may have influenced the reported HR-QoL, such as the dose and the duration of sedatives and neuromuscular blocking agents that have been shown to affect long-term outcomes. COVID-19 pandemic increased unemployment rates and reduced income in general; we lack data for non-COVID-19 controls, and we cannot exclude that the socio-economic effect for both patients and caregivers, employee status and income, is different than for the general population.