Long-Term Outcomes in COVID-19 ICU Patients: A Prospective Cohort Study

The COVID-19 pandemic causes high rates of intensive care unit (ICU) admissions. After ICU-discharge patients and family members can suffer from persisting impairments known as ‘Post Intensive Care Syndrome’ (PICS) and PICS-family. Since COVID-19 is relatively new, there is barely any knowledge on the long-term outcomes of COVID-19 ICU-survivors and their family members. This study aims to gain insight in the long-term physical, social and psychological functioning of COVID-19 ICU-survivors and their family members at three- and six-months following ICU discharge. A single-center, prospective cohort study was conducted in COVID-19 ICU-survivors and their family members. Enrolled participants received questionnaires at three and six months after ICU discharge. The MOS Short-Form General Health Survey, Clinical Frailty Scale, spirometry tests, McMaster Family Assessment Device (FAD-GF6+), the Hospital Anxiety and Depression Scale and return to work were used to evaluate physical, social and psychological functioning. and social functioning, worse employment status and persisting symptoms in 90%. In addition, family members also report long term effects expressed by a reduction in return to work and impaired well-being. Further research needs to extend the follow up and to study the effects of standardized rehabilitation in COVID-19 patients and their family members. balance problems, hair loss, smell/taste disorder, skin problems and/or hoarseness.


Introduction
Since December 2019, Coronavirus disease 2019 (COVID-19) has spread rapidly around the world, affecting more than 126 million people so far [ 1 ]. Like previous major outbreaks of viral infection in the 21th century, the COVID-19 pandemic is expected to have signi cant long-term clinical consequences for survivors [ 2,3,4 ].

Patients And Methods
The 'COVID-19 Follow-up Intensive Care Studies' (COFICS) aims to give insight in the long-term physical, social and psychological functioning, of COVID-19 ICU-survivors and their family members. The COFICS is a single-center, prospective cohort study, conducted at the ICU of University Medical Center Groningen (Groningen, The Netherlands). Ethical approval for this study has been given by our hospital's Medical Ethical Committee (METc 201800422) according to Dutch and European legislation. All patients gave their informed consent prior to the data collection.

Study participants
All COVID-19 patients admitted to the ICU between March 19 th and September 30 th 2020 and their family members were eligible to participate in this study. COVID-19 was diagnosed according to World Health Organization (WHO) de nition and was con rmed by RNA detection of the SARS-CoV-2 using the polymerase chain reaction (PCR)-based technique. Family members could be partners, children, other family members, or friends who were identi ed by the patient as important.

Procedure
All eligible patients were contacted by telephone by experienced research nurses for participation in this study, three months after discharge from the ICU. In addition, patients were asked if they agreed to have their family members contacted for participation. Multiple family members could participate per patient. Questionnaires were sent by ordinary mail at three and six months after ICU discharge. In case of no response, reminders were sent after three weeks. Results of lung function tests were retrieved from local hospitals after obtaining informed consent from the patients for spirometry results.
Outcomes Page 4/26 The outcome is the response to a series of questionnaires and spirometry tests to measure three domains: physical, social and psychological functioning ( Figure 1).
Physical functioning was scored with the Dutch version of the 9-point Clinical Frailty Scale (CFS) [ 22 ]. The CFS consists of nine pictographs, ranging from 'very t' (1) to 'terminally ill.' CFS-scores from 1 till 4 were classi ed as 'nonfrail' and from 5 till 9 as 'frail.' Additionally, physical functioning was evaluated by requesting the medical data from spirometry tests at six months follow-up. The following respiratory function parameters were measured: forced expiratory volume in 1 second (FEV1); forced vital capacity (FVC); forced expiratory ratio (FEV1/FVC); total lung capacity (TLC) and diffusing lung capacity for carbon monoxide (DLCO %). A measured value of more than 80% of the predicted value was considered as normal.
Symptoms were collected by the question which burden was experienced by the patient at six months follow-up (Supplement 1).
The MOS Short-Form General Health Survey (SF-20) was used to study general health outcomes related to physical and social functioning [ 23,24 ]. The SF-20 measures six QoL domains: physical functioning, role functioning, social functioning, mental health, general health perceptions and pain. The subscale mental health was not part of this study. The SF-20 total score was transformed linearly to a 0-100 scale where 0 represents the lowest and 100 the highest possible score. Return to work was measured as the proportion of previously employed ICU-survivors reporting return to work after critical illness, including work percentage and change of work activities.
Psychological functioning was measured with the Hospital Anxiety and Depression Scale (HADS) to study anxiety and depression [ 28 ]. The HADS contains a seven-item sub-scale for anxiety and a seven-item sub-scale for depression, with a four-point Likert scale for each question. Total scores per subscale range from 0 to 21, with the sums categorized as normal (0-7), mild (8-10), moderate (11)(12)(13)(14) and severe (15)(16)(17)(18)(19)(20)(21). Fear of reinfection was questioned on a scale of 0 ('no fear') to 10 ('high level of fear'). In uence of limited visiting possibilities was derived from a description of family members how they felt about the physical distances from their relative and if it had affected their well-being (Supplement 1). Baseline patient characteristics, including age, gender and clinical data, such as length of hospital stay, comorbidities and delirium, were retrieved from the electronic health record. Patient demographics, such as educational level and marital status, as well as healthcare consumption and family characteristics were addressed in the three-month questionnaire.

Statistical analysis
The Age-adjusted Charlson Comorbidity Index (ACCI) was calculated based on age and comorbidities. ACCI is a simple scoring system in which the factor age is included in the Charlson Comorbidity Index (CCI) [ 29, 30 ]. All outcomes were assessed for the total group of ICU-survivors and divided into three subgroups according to the ACCI (ACCI 0-1, ACCI 2-3 or ACCI ≥4). Outcomes of the family members were assessed for the total group. Quantitative data is reported as median with interquartile range (IQR), mean with corresponding standard deviation, or number with percentage. Descriptive analyses were performed with SPSS Statistics version 23.0 for Windows. The qualitative data was analyzed by two researchers (NV, IM) using ATLAS TI version 9 for Windows. Firstly, the data was coded inductively where the researcher used the words of the participant as label (in vivo coding). Second, codes were categorized to a list of symptoms.

Results
Baseline characteristics A total of 94 patients diagnosed with COVID-19 were admitted to the ICU during the study enrolment period. Seventythree (78%) patients were alive and eligible for inclusion three months post ICU discharge. Sixty (82%) COVID-19 ICUsurvivors returned the 3-months questionnaire and 50 (68%) returned the 6-months questionnaire. A total of 102 family members of COVID-19 ICU-survivors were asked to participate in this study, of which 78 (76%) and 67 (66%) completed the 3-months and 6-months questionnaire respectively ( Figure 2).
Participant characteristics are given in Table 1.A. Participants are subdivided to ACCI scores 0-1, 2-3 and ≥ 4. Table  1.B shows the characteristics of all family members. The vast majority of ICU-survivors had a BMI above 25 (n=56; 93%) and 55 (92%) ICU-survivors had a ACCI higher than 2. The median length of ICU stay was 19.4 days (IQR 12.3-31.7), of which 16.3 days (10.6-26.5) on mechanical ventilation. Almost 50 percent of the patients suffered a delirium during ICU stay. Most of the family members were partner of the patient, female and the median age was 56 years (IQR 41.0-63.0).

Results on health domains
Physical functioning Physical functioning of COVID-19 ICU-survivors was low three months post discharge with a median score of 33.3 out of a maximum of 100 (IQR 16.7-66.7) on the physical functioning subscale and 35.0 (IQR 25.0-50.0) on experienced health. Scores slightly improved -but remained low -at six months with a median score of 50 (IQR 16.7-83.3) and 50.0 (IQR 35.0-71.3), respectively. Patients had a median pain score of 50 out of a maximum of 100 at three-and sixmonths follow-up, but differed between ACCI categories. One third of the ICU-survivors considered themselves 'mildly frail' to 'frail.' Table 2.A summarizes the 6 months data including FEV1, FVC and DLCO% which were impaired in 18%, 20% and 69% respectively (   (Table 2A). Three (10%) of the 30 pre-ICU employed survivors fully returned to work whereas 10 (43%) were still too ill to work at six months post ICU-discharge (Table 1 and Table 2.A). Employment rate was decreased for the vast majority of patients at six months post ICU-discharge ( Figure 3A). Social functioning in family members scored high on role activities (median 100; IQR 50.0-100), social functioning (median 100; IQR 70.0-100) and family functioning (median 3.8; IQR 3.1-4.0). Of the 40 pre-ICU employed family members 26 (65%) fully returned to work at three post ICU discharge whereas nine (23%) were re-integrating or did not returned to work yet. At six months 23 of the 36 (64%) family members fully returned to work and 4 (11%) were reintegrating ( Table 2.B). Median employment rate in family members decreased with 18.3% and 7.7% on three and six months respectively compared to pre-ICU admission (Table 1.B and Table 2.B) but differed between family members ( Figure 3B).
Psychological functioning Psychological functioning in ICU-survivors was good with median scores ≤5.0 on the anxiety and depression subscales at three and six months ( Table 2.A).
Family members showed good psychological functioning as well, with median scores ≤4.0 on the anxiety and depression subscales at three and six months ( Table 2.B). The fear of reinfection scored 5.0 and 6.0 for ICU-survivors and family members respectively at six months.
Sixty-three percent of the family members reported impaired well-being due to the mandatory physical distance to their relative at the time of ICU admission ( Table 2.B). Fifty-four (68%) family members were distressed by the physical distances, mostly described as 'very di cult,' 'helpless,' 'terrible,' 'heavy' and/or 'terrifying.' Twelve family members were not distressed and described that 'the patient was in good hands' or 'reasonably well through contact.' The majority of family members referred to the telephone and video contact as positive and supporting. In contrast, some family members experienced the telephone and video contact negatively, describing this as 'tense' or 'di cult to get in contact'.
Quote family member: "Saying goodbye to my husband via an iPad before he was put into a coma, was horrible. I had EMDR therapy afterwards. It was a trauma." Quote family member: "The distance was di cult, but we understood the necessity. The situation was acceptable. In addition, the (digital / telephone) contact with the nurses made up for a lot." Health care consumption All ICU-survivors received care during the follow-up period. The physiotherapist, general practitioner and pulmonologist were mostly visited by ICU-survivors (Table 2.A). The general practitioner and social worker were mostly visited by the family (Table 2B).

Discussion
Our results showed that COVID-19 ICU-survivors experienced limitations in physical functioning, reduced diffusion lung capacity and 90% endorsed at least one symptom after six months. Impaired social functioning was present as 90% of ICU-survivors did not reach their pre-ICU employment level at six months. In addition, family members experienced worse employment status 34%. Psychological functioning was normal in ICU-survivors and their family members however 63% of the family members reported ongoing impaired well-being due to the COVID-19 related mandatory physical distances to their relatives.
To our knowledge, this is the rst study to assess the health consequences of COVID-19 ICU-survivors at two follow-up times. It is known from previous Corona outbreaks, such as severe acute respiratory syndrome (SARS) and Middle East Respiratory Syndrome (MERS), that survivors suffer from pulmonary dysfunction, psychological impairment and reduced exercise capacity [ 2 ]. Survivors of acute respiratory distress syndrome (ARDS) are also known to experience a high prevalence of functional disability, cognitive impairment, posttraumatic stress disorder, and impaired QoL [ 31, 32, 33, 34 ]. Hence, high levels of physical, cognitive and psychosocial impairments among COVID-19 survivors can be expected and require anticipation.
Although a majority of the COVID-19 ICU-survivors in our study were impaired in the physical domain at 3-and 6months follow-up, only a third of the patients considered themselves as 'mildly frail' to 'frail.' A possible explanation might be the multidimensional concept of frailty, which includes the social and psychological domain as well [ 35 ].
DLCO was lower than 80% from normal values in 69% of ICU-survivors. This percentage is higher than in ward-based Delayed return to work is common after critical illness and is likely a consequence of post-ICU impairments. After ICUadmission, 20%-36% of survivors experiences job loss, 17%-66% changed their occupation and 5%-84% had a worsening of employment status [ 38 ]. Pre-existing comorbidities are a potential risk factor for delayed return to work. Although 46% of the pre-ICU employed patients of our study cohort had returned to work at six months, a majority of these patients is not back to their pre-COVID-19 level. Our ndings correspond with nding in other studies where 33-47% of ICU-survivors had returned back to work at 3 months follow-up [ 16,17 ]. Burden in the social domain for the family is re ected by one third of the family members that has not fully returned to work yet, at 3 and 6 months. Despite limited evidence on this topic, it is known that 85% of caregivers had returned to their previous work level after one year [ 39, 40 ].
Although psychological symptoms are likely to occur in two thirds of survivors of acute respiratory distress syndrome at 12 months follow-up [ 41 ], our results showed no impairments in the psychological domain. A possible explanation for this can be a mixture of effective family support, re ected by the high mean family functioning and the relatively high health care consumption. The amount of family care compared to professional care was not studied but this might be interesting for follow-up research.
In our study cohort, 100% and 96% of the patients consumed professional health care at 3 and 6 months follow-up respectively. This is in line with 57% of patients needing healthcare assistance after prolonged mechanical ventilation at 1-year follow-up [ 42 ]. It is known that frail patients consume more health care services compared to patients who are not frail [ 43 ], but in our study cohort nearly all COVID-19 ICU-survivors consumed health care services post-discharge at 6 months follow-up.
The responses of the family members in this study show that the profound nature of the situation was overwhelming. However, the telephone and video contact were positively evaluated. The majority of the family members (68%) declared that the physical distance was 'very di cult' to handle. However, it is known that experiences of family members are not well represented in existing standardized questionnaires [ 44 ].
An easy tool to evaluate the functional outcomes in COVID-19 survivors has been proposed [ 45 ]. Several other instruments are also used in COVID-19 studies and studies are therefore di cult to compare. This fact argues in favor of using a standardized set of validated instruments in an international research context [ 46 ]. A standardization of follow-up will bene t an organized way of determining functional recovery over time and can improve health care worldwide.

Strengths and limitations
A strength of this study is the high response rate at follow-up, from both patients and family members, and the use of validated questionnaires and measurements. In addition, this study includes two follow-up time points after discharge, allowing comparisons over time.
Some limitations are present as well. First, the sample size of this prospective cohort study was limited, increasing the vulnerability for confounding factors. On the other hand, our sample size is larger than the ICU subgroups in other studies [14,17,18]. Increasing the cohort size will allow stronger conclusions. A second limitation of this study was that we did not assess the baseline status of patients prior to development of COVID-19. Many ICU-survivors already The datasets generated during and/or analyzed during the current study are available in the UMCG repository.

Competing interests
The authors declare no potential con icts of interest with respect to the research, authorship and/or publication of the article.
Funding None to declare.

Authors' contributions
All authors were involved in the design of the study. MO was responsible for data collection and coordinating the research nurses. NV and IM prepared the data for analyses, undertook data analysis and wrote the manuscript. Critical revision of the manuscript for important intellectual content was done by all authors. The manuscript has been seen and approved by all authors.       Abbreviations: ACCI = Age-adjusted Charlson Comorbidity Index (including age); ICU = Intensive Care Unit 1 A higher score re ects a better functioning 2 A higher score represents pain in a greater extent. 3       Abbreviations: ACCI = Age-adjusted Charlson Comorbidity Index (including age); ICU = Intensive Care Unit 1 A higher score re ects a better functioning 2 A higher score represents pain in a greater extent. 3 Data is given for 39 COVID-19 ICU survivors (42 participants underwent a spirometry test of which 1 participant did not consent to collect the data from the hospital. For 2 participants data is missing.) All numbers given are the median and interquartile range (IQR), unless otherwise stated.
Abbreviations: ACCI = Age-adjusted Charlson Comorbidity Index (including age); ICU = Intensive Care Unit 1 A higher score re ects a better functioning 2 A higher score represents pain in a greater extent. 3 Data is given for 39 COVID-19 ICU survivors (42 participants underwent a spirometry test of which 1 participant did not consent to collect the data from the hospital. For 2 participants data is missing.) 4 > 5% reported health care consumption are presented. Other health care consumption were mentioned for example; social work, occupational therapist, cardiologist, (vascular) surgeon or otorhinolaryngologists.  All numbers given are the median and interquartile range (IQR), unless otherwise stated. 1 A higher score re ects a better functioning. 2 A higher score represents pain in a greater extent. 3 Employed family members at 3 months n = 40 and at 6 months n = 36. 4 > 5% reported health care consumption at 3 months are presented. Other health care consumption were mentioned for example; home care, physiotherapist, company physician, coach.