Post-Intensive Care Syndrome Associated With Coronavirus Disease: A Single Institutional Study In Tokyo

Aya Banno St Luke's International Hospital Toru Hifumi (  hifumitoru@gmail.com ) Kagawa Daigaku https://orcid.org/0000-0002-9627-3702 Yuta Takahashi St. Luke's International Hospital Mitsuhito Soh St. Luke's International Hospital Ayako Sakaguchi St. Luke’s International Hospital Shodai Shimano St. Luke's International Hospital Yoshie Miyahara St. Luke’s International Hospital Shutaro Isokawa St. Luke's International Hospital Kenji Ishii St. Luke's International Hospital Kazuhiro Aoki St. Luke's International Hospital Norio Otani St. Luke's International Hospital Shinichi Ishimatsu St. Luke's International Hospital


Background
A year has passed since the World Health Organization declared the coronavirus disease (COVID-19) pandemic on March 11, 2020 [1].Although there has been extensive research on the acute phase of the disease, limited studies have investigated its long-term health effects [2][3][4][5][6].Regarding intensive care unit (ICU)-treated patients, there is a need to evaluate the occurrence of post-intensive care syndrome (PICS) [7] because it has become recognized and is the next care target in the eld of critical care medicine [8][9][10].
PICS is de ned as a new or worsening impairment in physical, mental, or cognitive health status that arises and persists after hospitalization for critical illness [7].Although there have been recent reports regarding long-term consequences, including fatigue, insomnia, and anxiety or depression, among COVID-19 survivors at or 6 months after acute infection, information regarding critically ill patients remains missing [2,5,6].Small-scale studies investigating ICU treated patients with COVID-19 have suggested that respiratory, mobility, and cognitive consequences are common at approximately a month after hospital discharge [3,6].However, PICS occurrence in COVID-19 survivors remains largely unclear.There is a need to evaluate longterm PICS occurrence in this population with respect to all three PICS components.
This study aimed to investigate the full spectrum of PICS, including its physical, mental, and cognitive aspects, in COVID-19 survivors hospitalized at our single-center ICU.

Study design and subjects
This prospective study enrolled critically ill patients with COVID-19 who were consecutively hospitalized in the ICU of a single institution in central Tokyo, Japan between March 19 and April 30, 2020.The hospital provided permission for collecting data from electronic health records; moreover, this study was approved by the institutional review board (approval number 20-R102).
The inclusion criteria included ICU admission and laboratory or radiological con rmation of COVID-19 [11].We excluded patients who died during hospitalization.A laboratory diagnosis was based on positive reverse-transcriptase polymerase chain reaction test results for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) using samples such as nasal swabs, pharyngeal swabs, or sputum [12].A radiological COVID-19 diagnosis was based on the presence of bilateral groundglass opacities on chest computed tomography [13].

Questionnaire details
A telephone call was made to each patient by a team of doctors, nurses, and a physical therapist who were all ICU staff members.First, the patient's ability to complete the mailed questionnaire was assessed.In the case of severe mental distress or patient refusal, the questionnaire was not mailed.The survey comprised the following items: the post-COVID-19 functional status (PCFS) scale [14], modi ed medical research council dyspnea scale (mMRC) [15], impact of event scale-revised (IES-R) [16], and hospital anxiety and depression scale (HADS) [17].Additionally, the participants were asked about changes in living or working status; moreover, they conducted self-assessments of their concentration, memory, and forgetfulness (Additional le 1).The survey booklet and an informed consent form were sent to the patients' homes in mid-August 2020.Responses from patients with valid consent documents were used for assessment.A second survey with similar questions was sent to consenting patients in mid-October 2020.

Data collection
We collected information regarding baseline characteristics, comorbidities, illness severity, mortality prediction scores (Sequential Organ Failure Assessment [SOFA] score, Acute Physiology and Chronic Health Evaluation [APACHE] II and III scores, Simpli ed Acute Physiology Score [SAPS II]), ICU therapies (systemic steroids, invasive mechanical ventilation, extracorporeal membrane oxygenation, renal replacement therapy, and continuous benzodiazepine administration), and clinical outcomes.Information regarding delirium was collected from electronic ICU charts.ICU nurses performed routine assessments using the confusion assessment method for the ICU to determine delirium occurrence.Additionally, delirium was de ned based on a positive Richmond agitation and sedation scale score [8,18] or bedside nurse's judgment of the presence of agitation, hallucination, or dangerous behavior.Moreover, outcomes such as the duration of ICU stay, hospital stay, and place of discharge were investigated.

Study endpoints
The following PICS components were assessed: physical PICS, mental PICS, and cognitive PICS.Physical PICS was de ned by a PCFS or mMRC score ≥ 1 [2,19].Mental PICS was assessed using IES-R and HADS scores.An IES-R score ≥ 25 indicated post-traumatic stress disorder (PTSD) [16], while scores ≥ 8 for the anxiety and depression components of the HADS survey indicated anxiety and depression, respectively [17].A composite endpoint of these score results was considered as mental PICS.Cognitive PICS was de ned if the patient had complaints of deteriorated concentration, memory, or forgetfulness.Care was offered to patients with positive results for PICS.

Statistical analysis
Each subgroup of patients with PICS was compared with the corresponding non-PICS groups.Speci cally, we compared the following variables: baseline demographics, comorbidities, living and marital status, illness severity and mortality prediction scores, ICU treatments, delirium occurrence, and outcomes.Continuous variables are presented as median and interquartile ranges.Categorical variables are presented as the number and percentages.Between-group comparisons of continuous variables were performed using the Kruskal-Wallis test or Mann-Whitney U-test; moreover, categorical variables were compared using Fisher's exact test or chi-square test, as appropriate.Statistical analyses were performed using the JMP version 12 statistical software (SAS Institute, Cary, NC, USA).Statistical signi cance was set at a two-sided p-value < 0.05.Missing data were not replaced or estimated.

Results
Among 27 eligible patients, 20 patients responded with a valid consent form and were included in the initial analysis.The second survey, which was sent 6 months after ICU discharge, was completed by 15 patients who were included in the subsequent assessment (Fig. 1).

Baseline characteristics
Table 1 presents the baseline and demographic characteristics.All patients were of Japanese ethnicity.The median age was 57.5 (interquartile range: 48.8, 71.3) years; moreover, 80% and 85% of participants were male and had comorbidities, respectively.At the time of hospital admission, 80% and 20% of patients were employed/self-employed and unemployed/retired, respectively.Additionally, 55% of patients were married, while 2 (10%) patients were widowed.None of the patients lived in nursing homes or long-term care facilities before hospitalization.All patients were physically independent in daily life activities and none were receiving social services.No patients were previously diagnosed with mental diseases or cognitive impairments.Changes in social status at 4 months Among 16 (80%) patients who were employed or self-employed, 12 fully recovered their pre-hospitalization working status.
The remaining 4 patients partially recovered; moreover, all patients recovered or none were unemployed.All patients, except one patient who was discharged to a long-term care hospital, returned to their previous living status at 4 months after ICU discharge.

Occurrence of each PICS component
Figure 2 shows the distribution of patients based on each PICS subgroup.Physical, mental, and cognitive PICS occurred in 14 (78%), 9 (45%), and 11 (55%) patients, respectively.Nine patients presented with anxiety, PTSD, and depression; speci cally, patients with anxiety had PTSD, depression, or both.Regarding the cognitive PICS component, 7 patients claimed deterioration in both concentration and memory and additionally claimed deterioration in forgetfulness.Speci cally, all patients who claimed deterioration in forgetfulness also claimed worsening of concentration and memory.
Overall PICS occurrence at 4 months Figure 3 shows the proportion of patients with physical, mental, and cognitive PICS.Sixteen (80%) patients had at least one PICS type.All 9 (45%) patients with mental PICS had physical PICS.Eight (40%) patients had all PICS components.

Comparison of PICS versus non-PICS
Table 3 presents the comparison of each PICS subgroup and the corresponding non-PICS group.There were no signi cant differences in the compared variables.However, the use rate of systemic steroids (physical PICS vs. non-physical PICS: 86% vs. 50%, p = 0.2; mental PICS vs. non-mental PICS: 89% vs. 64%, p = 0.32; cognitive PICS vs. non-cognitive PICS, 82% vs. 67%, p = 0.62) and continuous benzodiazepine (physical PICS vs. non-physical PICS: 29% vs. 0%, p = 0.52; mental PICS vs. nonmental PICS: 33% vs. 18%, p = 0.62, cognitive PICS vs. non-cognitive PICS, 36% vs. 11%, p = 0.32) tended to be higher in all the PICS subgroups.The comparison according to each questionnaire is displayed in the Additional File 2. *Includes chronic respiratory disease and asthma.†Includes hypertension, diabetes mellitus, dyslipidemia, and obesity.‡Patients who had repeated positive reverse-transcriptase polymerase chain reaction tests required isolation at an observatory hotel before discharge.

Summary of ndings
This single-center study investigated PICS occurrence in 20 patients who were hospitalized in the ICU for severe COVID-19 during the rst wave of the pandemic in Tokyo.At 4 months after ICU discharge, the occurrence rates of physical, mental, and cognitive PICS were 78%, 45%, and 55%, respectively.Moreover, 80% and 40% of patients had at least one and all PICS components, respectively.At 6 months after ICU discharge, there was an improvement tendency in most questionnaire scores.

Comparison with other studies
To our knowledge, this is the rst study to examine the long-term outcomes (PICS occurrence) among COVID-19 survivors.Martillo et al. investigated COVID-19 survivors of the ICU at 1 month from hospital discharge to discover a 91% prevalence of overall PICS, which was predominated by physical impairments seen in 87% of the patients[6].Mental and cognitive impairments were seen in 22% and 8% respectively.The overall prevalence of PICS and physical impairments was higher than our study, which may be because patients were evaluated brie y after hospital discharge.In contrast, three previous studies have investigated all three PICS components in a non-COVID population after at least 3 months of discharge.All three studies used different methods to assess each component.In these previous studies, physical impairments were investigated using the physical components of the 36-item Form (SF-36) [8], Katz Index of Independence in Activities of Daily Living [20], and Euro-QOL [21], with reported prevalence rates of 31%, 23%, and 42%, respectively.In the present study, we used the mMRC and PCFS since the mMRC is appropriate for assessing post-COVID breathing problems [2]; moreover, PCFS is valid for assessing mobility and the capability of usual activities [19].Regarding the mental component, the SF-36 [8], Beck Depression Inventory-II [20], and combination of HADS and Post-Traumatic Stress Symptom-10 [21] were used, with prevalence rates of 15%, 33%, and 58%, respectively.We used the HADS and IES-R, which conform to the consensus of the Society of Critical Care Medicine [22] The use of HADS was consistent with the study by Maley et al. [21]; moreover, it may have attributed to the comparable occurrence rates of mental PICS.Finally, cognitive impairment was assessed using the Short Memory Questionnaire [8], Repeatable Battery for the Assessment of Neuropsychological Status [20], and selfassessment questions [21], with prevalence rates of 37%, 37%, and 56%, respectively.
Among the three studies, the occurrence of at least one PICS domain ranged from 64-84%; moreover, the rates of having all three PICS types were 2%, 6%, and 33% in each study [7,19,20].Our occurrence rates were most comparable with those reported by Maley et al. who studied the youngest patients (mean age: 59 years) and used the HADS and self-assessment questions for assessing mental and cognitive impairments [21].Kawakami et al. reported a lower prevalence, which could be attributed to the authors comparing SF-36 scores with baseline scores that were reported by patients' proxy recalling the status at 4 weeks before the acute illness.Additionally, Marra et al. excluded patients with preexisting physical, mental, and cognitive impairments, which could explain the lower PICS frequency.Contrastingly, our ndings were based on one-point evaluation.However, none of our included patients had preexisting physical, mental, or cognitive impairments; moreover, the evaluations employed were suitable for assessing long-term consequences in ICU-treated COVID-19 survivors.
The observed high PICS prevalence in COVID-19 survivors could be attributed to several reasons.First, we only included COVID-19 survivors.The included patients were overall younger than the general ICU-treated population [8, 23,24]; moreover, patients with PICS impairments tended to use systemic steroids or benzodiazepines, which are known factors for mental and physical PICS [25].Additionally, long-term health effects across all severities have been recognized in numerous COVID-19 survivors [2,3].Second, our patients were victims of a new and obscure disease and also a phenomenal pandemic.Media in uence and COVID-19-related celebrity deaths may have triggered mental impairments in our patients [26].Finally, our results were based on one-point evaluations and were not compared to the baseline status, which could have led to a higher prevalence.

Clinical implementations
We observed a high prevalence rate of PICS in COVID-19 survivors, which indicates the importance of acute follow-up for ICU patients.Our patients showed various long-term consequences; therefore, there is a need to evaluate all three domains to allow comprehensive care.Furthermore, although we did not observe signi cant differences, there were tendencies of more frequent use of systemic steroids and continuous benzodiazepine in each PICS subgroup compared with the corresponding non-PICS groups.
Given that the aforementioned drugs are known risk factors for PICS [25], physicians should acknowledge their long-term disadvantages in patients.Moreover, substituting either drug for another may be considered [27].For instance, while steroid administration for COVID-19 is controversial, most studies have favored using interleukin-6 (IL-6) inhibitors [28].Additionally, there is accumulating evidence regarding the effectiveness of anti-tumor necrosis factor (TNF) therapy for COVID-19 [29].
There is a need for more studies on both IL-6 and TNF inhibitors and their role in COVID-19; however, they may be an alternative for steroids and also protect against PICS development.

Limitations
This study had several limitations.First, this study was conducted by the ICU staff; therefore, 1-or 3-month post-ICU evaluation could not be performed owing to the heavy workload persistent during the pandemic.Second, we could not perform comparisons with the baseline status.Third, selection bias may have occurred since this was a mailed survey requiring responses from participants.Patients with severe impairments or illness were less likely to provide consent or responses to the survey.Fourth, this study had a small sample size and the 6-month survey response rate was low.Fifth, we could not perform statistical comparisons between the 4-and 6-month questionnaire results.Finally, the PICS de nition remains unclear; moreover, our evaluation methods were limited to questionnaires answerable via postal mail.However, the mMRC was used in accordance with the study by Huang et al. [2]; furthermore, the PCFS is appropriate for physical PICS examination since it is mainly associated with the capability of mobility and usual activities [19].Additionally, the usage of HADS and IES-R conformed with the consensus of the Society of Critical Care Medicine for evaluating mental PICS [22].Our most signi cant limitation was that the evaluation of the cognitive component relied on self-assessment.

Conclusions
The rate of PICS among COVID-19 survivors was as high as 80%; moreover, the co-occurrence of all 3 domains was observed in 40% of patients.Long-term and comprehensive evaluation of all three PICS components is crucial for these patients to provide appropriate care.

Abbreviations
Acute physiology and chronic health evaluation

Table 1
Baseline and demographic characteristics on hospital admission Data are expressed as numbers (percentage) or median (interquartile range).*Includes untreated or simultaneously treated neoplasms.IQR, interquartile range; BMI, body mass index Results of the questionnaire

Table 3
Comparison between PICS subgroups and the corresponding non-PICS subgroup Data are expressed as numbers (percentage), or median (interquartile range).*Includes chronic respiratory disease and asthma.†Includes hypertension, diabetes mellitus, dyslipidemia, and obesity.‡Patients who had repeated positive reverse-transcriptase polymerase chain reaction tests required isolation at an observatory hotel before discharge.