Ethics committee approval statement

This study has been approved by our institutional research ethics committee before started (Comissão de Ética para a Saúde do Centro Hospitalar Universitário de São João; number of approval: 22/21), and it has been conducted in accordance with the principles set forth in the Helsinki Declaration. Written informed consent was obtained from all patients.

Study design

Prospective cohort study with consecutive inclusion of patients admitted to one of four ICU of an Intensive Care Department in a tertiary-care center, between May/2020 and September/2021.

Inclusion criteria were age ≥ 18 years old and ICU admission diagnosis of acute respiratory distress syndrome (ARDS) due to SARS-CoV-2, requiring invasive mechanical ventilation (IMV) for ≥ 48 hours (h). Patients who died during ICU stay were excluded.

After hospital discharge, all patients were reached by telephone and asked to attend a specific post-COVID-19 PRM outpatient appointment. When it was not possible to contact the patients through a single telephonic contact, five more attempts, in different days and at different day hours (from 10am to 5pm, on weekdays) were performed. First PRM appointment was scheduled at six months after ICU discharge. A re-evaluation visit was performed at 12 months after ICU discharge. If any of the appointments were missed, the patient was given two opportunities to reschedule in a one-month period.

The appointments were performed at the outpatient clinic of the PRM department of our center. All participants were interviewed face-to-face by the same PRM physician.

Definitions

ARDS was defined in accordance with the Berlin definition, as an acute syndrome of lung inflammation and increased alveolar-capillary permeability associated with severe hypoxia and bilateral infiltrates on chest radiographs, without evidence of left heart failure [14].

A COVID-19 ARDS case was assumed when a positive result on real-time reverse-transcriptase–polymerase-chain-reaction (RT-PCR) assay of nasal and pharyngeal swab specimens was depicted in the first 24h after hospital admission.

Data collection and outcome measures

Data regarding socio-demographic characteristics, functional status, comorbidities, characteristics of the critical respiratory illness and complications during ICU stay were collected by the main investigator from the electronic clinical records (ECR) using a predefined form.

Comorbidities included hypertension, diabetes mellitus, hyperlipidemia, obesity, smoking habits, atrial fibrillation, ischemic heart disease, heart failure, peripheral vascular disease, chronic pulmonary obstructive disease, asthma, sleep apnea, psychiatric pathology, oncologic pathology and immunosuppression.

To describe the characteristics of the critical respiratory illness, information regarding severity scores at ICU admission, namely Acute Physiology and Chronic Health Evaluation (APACHE) and Simplified Acute Physiology Score II (SAPS II) was retrieved. Additionally, data regarding the number of days at the ICU and total length of stay, number of days under IMV, need and number of days under vasopressors, renal replacement therapy, extracorporeal membrane oxygenation (ECMO) and the need for prone positioning were also collected. Several potential complications during ICU stay were assessed, namely neurological, abdominal, cardiovascular, cutaneous and infectious. Neurological complications (signs, symptoms and syndromes) comprised aphasia, dysarthria, dysphonia, dysphagia, focal weakness, delirium, seizures, cerebrovascular diseases, encephalopathy, encephalitis, myelitis, peripheral neuropathies and signs of corticospinal tract dysfunction (CSTD). Cardiovascular complications included bradyarrhythmia, tachyarrhythmia (atrial fibrillation, flutter, others), tachycardia-bradycardia syndrome, secondary myocardial injury, cardiac arrest, pericarditis, pericardial effusion, endocarditis, acute heart failure and cardiogenic chock. Abdominal complications encompassed hepatitis, gastrointestinal bleeding, pseudo-obstruction and obstruction, diarrhea and constipation. Skin complications included the presence of pressure injuries. Infectious complications were considered in the presence of ICU-acquired infections, superimposed to the primary infectious diagnosis.

Data regarding the aforementioned characteristics was gathered on a database where each patient received a code number to secure their anonymity.

At the PRM appointments (6 and 12 months after discharge), a detailed physical examination was performed, and a predefined set of scales were applied, aiming to comprehensively evaluate the three domains of PICS (physical, mental, and cognitive). Additionally, patients’ functionality was systematically assessed. The chosen outcome measures were preferentially the ones recommended by the core outcome set for survivors of acute respiratory failure [15].

Physical examination encompassed the evaluation of the patient’s swallowing function, muscle strength, sensory response, signs of CSTD, balance and gait. Data regarding muscle strength, sensory response and signs of CSTD was obtained from physical examination and afterwards formally analyzed through statistical methods. Data regarding swallowing function, balance and gait were analyzed using specific instruments, as described below. Muscle strength was evaluated through manual muscle testing and graded in accordance with the Medical Research Council Sum Score (MRC-SS). The maximum score at this metric is 60 points reflecting maximal strength in all evaluated segments bilaterally, so this value was used as reference category [16]. The sensory exam included the evaluation of light touch sensitivity in all dermatomes from C2-S2 bilaterally. The presence of hyper, hypo or anesthesia, and/or the complains of dys or paresthesia during the sensory examination, were considered as sensory impairments. Signs of CSTD were evaluated through the examination of deep tendon reflexes (DTR) and Babinski sign. Bicipital, tricipital, brachioradialis, patellar and achilles DTR were appraised using a predefined T-shaped reflex hammer. The grading of reflex response was in accordance with an adapted form of the National Institute of Neurological Disorders and Stroke (NINDS) Myotatic Reflex Scale in: 0) absent, 1) hyporeflexia, 2) normal, 3) hyperreflexia, 4) hyperreflexia with unsustain clonus (≤ five beats), 5) hyperreflexia with sustain clonus (> five beats) [17]. Babinski sign was evaluated using the reflex hammer dull point by running up, with light pressure, the lateral plantar side of the foot, from heel to toes. The response of hallux and toes was recorded as extensor (Babinski sign), flexor or neutral [18]. The presence of signs of CSTD was defined as 1) Babinski sign in at least one extremity or 2) hyperreflexia in at least two extremities [19]. All data recorded from the physical examination was included in the physical domain.

To further characterize the physical domain, the European-Quality-of-Life-5-Dimensions-3-Level (EQ-5D-3L) questionnaire was used to measure health-related QoL. The first part of this questionnaire is a three-question component that explores five dimensions: mobility, self-care, usual activities, pain/discomfort and anxiety/depression. Each dimension is rated on a scale from 1 to 3. In accordance with Larsson IM et al. study we have considered the result of at least two points per question as cut-off value for impairment (total score ≥ 6) [20]. The second section is a visual analogue scale (EQ-VAS), which is a measure of self-rated overall health status, ranging from 0 to 100%. According to Ferreira PL et al. study in the Portuguese population, we have considered values below 75% as an impairment [21].

Fatigue was evaluated using the Portuguese version of Fatigue Assessment Scale (P-FAS) [22]. The P-FAS is a self-reported, 10-item ordinal questionnaire that varies from 10 to 50. A total P-FAS score ≥ 22 indicates the presence of fatigue [23].

Swallowing function was evaluated using the Functional Oral Intake Scale (FOIS) and the Portuguese Eating Assessment Tool (P-EAT-10) [24, 25]. FOIS is a continuous scale that ranges from 1 to 7. In accordance with Sassi FC et al. analysis, we divided patients in 1) resolved dysphagia, when FOIS levels were of 6 or 7; or 2) non-resolved dysphagia, if the FOIS levels were from 1 to 5 [26]. The P-EAT-10 is a continuous scale, that varies from 0 to 40, with scores over 3 in each question indicating increase risk of dysphagia, as detailed in Zhang PP et al. metanalysis [27]. In accordance, we have considered 30 as the reference value for dysphagia and 0 as the reference for the absence of swallowing complains.

Gait was analyzed and described during physical examination and classified in accordance with the Hauser Ambulation Index (HAI) for analytic purposes [28]. This index ranges from 0 to 9. We have considered 0 as the reference category, thereby classifying patients with scores ≥ 1 as having a gait impairment.

Balance was evaluated objectively in the physical examination and classified accordingly to the Berg Balance Scale (BBS). The BBS assesses the functional balance based on 14 items, with a maximum score of 56 points. In accordance, we have considered that scoring ≤ 55 points was suggestive of having a balance impairment [29]. Additionally, we have considered the cut-off for higher risk of falls (46 points) as reference for the presence of a significative balance impairment [30].

To summarize the obtained results, we have created a composite variable (PICS-physical) that included all subdomains related to the physical function. We have considered that there was an involvement of the physical domain when any of the subdomains was altered.

For the mental domain, we used the Hospital Anxiety and Depression Scale (HADS). We considered 0 points as reference value, thought significant anxiety was defined as HADS-anxiety score ≥ 8 and significant depression as HADS-depression score ≥ 8 [31]. When significant depression or anxiety was reported, an impairment on the mental domain as assumed.

To evaluate the cognitive domain, the Montreal Cognitive Assessment (MoCA) was applied. This score range is 0–30. A cut-off of 26 was used to define cognitive dysfunction in accordance with the literature [32].

Functionality was evaluated through self-reported Functional Independence Measure (FIM). The FIM is a validated and objective assessment of functional status [33]. It is an 18-item ordinal scale, in which the global score varies from 18 to 126 [33]. Since multidisciplinary FIM assessment was not possible in this setting, we have considered self-reported FIM values following previous reports of moderate agreement between self-reported and observed FIM [34]. When the total punctuation at FIM was below 126 points, we considered that the patient was not fully independent on ADL. Even though PICS original definition allocated autonomy on ADL in the physical domain, in our analysis functionality was evaluated separately, since mental, physical and cognitive dysfunction can impact functionality and autonomy on ADL.

Furthermore, information regarding PRM interventions was also collected. Through the ECR, data regarding in-hospital (ICU and wards) rehabilitation programs was retrieved. In each follow-up visit, information concerning PRM programs performed after hospital discharge was obtained, namely, setting (in or outpatient), modalities (uni or multimodal, including rehabilitation nursing, physical therapy, occupational therapy, speech therapy and/or neuropsychology), length (number of months under PRM intervention) and intensity (number of sessions per week and each session duration).

Statistical analysis

Statistical analysis was performed using Statistical Package for the Social Sciences (SPSS) statistics software (version 27).

Categorical variables are summarized as frequencies and percentages. Continuous variables are summarized as means and standard deviations (variables with normal distribution) or medians and interquartile ranges (variables with skewed distributions). Normal distribution was checked using histogram visual inspection and the Shapiro–Wilk test.

The chi-square test or Fisher’s exact test was used, as appropriate, to compare categorical variables. Continuous variables comparison between “included” and “lost to follow-up” patients was performed with independent sample t-test or the Mann-Whitney U test, according to the variable distribution.

To compare patient’s values with normative data (reference category), one sample t-test or one sample Wilcoxon Signed Rank Test were used, in accordance with the variable distribution. Due to the absence of specific cut-offs for the applied scales in this post-critical COVID-19 sample, and the diversity of results (regarding the used metric and the timing of examination) available in the literature, we compared our results with the classic normative values of each scale, as previously referred as the reference category.

Continuous variables were compared regarding its evolution throughout the 6 month follow-up period using paired-samples t-test or related samples Wilcoxon Signed Rank Test, in accordance with the variable distribution.

All reported p values are two-tailed, with a p-value <0.05 indicating statistical significance.

Sample characteristics

A total of 92 critical COVID-19 patients were admitted and discharged alive from our ICU, between May/2020 and September/2021.

Within this eligible sample, 5 patients died after ICU discharge (5.4%), 35 refused to be observed in the appointment (38%) and 10 patients were not reachable (10.9%). As so, 42 patients were included (45.7%). No differences were found between patients included and those lost to follow-up regarding socio-demographic characteristics (gender and age), the severity of disease at admission (APACHE and SAPS II scores), number of days at the ICU and total length of stay (S1 Table).

In the included sample, 66.7% were males with a mean age of 62 years old (standard deviation (SD) = 13.5). All patients were previously independent on ADL. Obesity (57.1%), hypertension (54.8%), and hyperlipidemia (45.2%) were the comorbidities more frequently identified.

The mean SAPS II score was 40.7 (SD = 15.3), with 73.8% (n = 31) of patients scoring over 30 points at ICU admission. Moreover, the median of ICU stay was 31.5 days (interquartile range (IQR) = 15.5–51.3) and the median number of days under IMV was 25 (IQR = 10.0–43.0). During ICU stay, several complications were registered, namely neurological (54.8%), cardiovascular (14.3%), abdominal (42.9%), cutaneous (47.6%) and infectious (71.4%).

Data regarding socio-demographic characteristics, comorbidities, characterization of critical respiratory illness and complications at ICU are detailed in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Socio-demographic and clinical characteristics of the sample.

https://doi.org/10.1371/journal.pone.0284597.t001

PRM care characterization

Throughout ICU and hospital wards stay, all included patients were evaluated and included in PRM inpatient programs, which consist of one (physical therapy, occupational therapy, speech therapy, and rehabilitation nursing) or more treatment modalities, pending on each patient sequelae. After discharge, 88% (n = 37) maintained PRM treatments: 7 patients (16.7%) were admitted at PRM facilities for inpatient rehabilitation (followed by outpatient rehabilitation), and the remaining 30 patients were included in outpatient rehabilitation programs. A total of 5 patients were not included in PRM programs after hospital discharge since a full recovery in neuromotor domains and autonomy on ADL was achieved. In most cases, PRM outpatient intervention was unimodal (physical therapy; 70.2%; n = 26). The frequency of PRM interventions ranged between 2 to 5 sessions per week, in which each session length of 30 to 60 minutes. At 6 months post-discharge, 62.2% remained under PRM intervention, while at 12 months 32.4% maintained PRM outpatient programs.

PICS analysis

Table 2 details the clinical status at 6 and 12 months, univariate analysis between the achieved scores and the reference values and a comparative analysis (“evolution” between scores at 6 and 12 months.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Scores at 6 and 12 months and comparison with normative values.

https://doi.org/10.1371/journal.pone.0284597.t002

PICS—Physical domain

PICS physical domain was evaluated through multiple metrics in our analysis. Indeed, 76.2% (n = 32) of our sample had at least one physical impairment at 6 months decreasing to 71.4% (n = 30) at 12 months (p-value = 0.475).

Regarding QoL, at 6 months after ICU discharge, COVID-19 survivors had significantly higher EQ-5D-3L median (7.2 (IQR = 5.0–8.5), p-value = 0.034), and a lower mean EQ-VAS (67.4 ± 21.6, p-value = 0.060), in comparison with reference values. Nevertheless, one year after ICU discharge, we found no difference either in EQ-5D-3L nor in EQ-VAS compared to normative values (p-value = 0.180 and p-value = 0.573, respectively). In both QoL metrics, there was a tendency for improvement through the follow-up period, which was statistically significant only for the EQ-VAS (p-value = 0.038).

Concerning fatigue, COVID-19 survivors had a mean P-FAS of 34.5 ± 32.0 at 6 months, a significantly superior value in comparison with reference values (p-value = 0.009). At 12 months, the median FAS decreased to 18.5 points (SD = 50.0; p-value = 0.043).

Swallowing function was also analyzed in our sample, as part of the physical domain. Both at 6 and 12 months, the median of FOIS and P-EAT-10 were similar to the reference category. At 6 months after ICU discharge, 5% (n = 2) presented swallowing impairments whereas at 12 months none had this complication.

A total of 31% of COVID-19 survivors had muscle weakness at 6 months, and 28.6% maintained this impairment at 12 months (p-value = 0.002). In brief, both at 6 and 12 months after ICU discharge, median values at MRC-SS were significantly below the normative value (6 months: median = 60.0, IQR = 48.0–60.0, p-value = 0.002; 12 months: median = 60.0, IQR = 56.8–60.0, p-value = 0.001). Nonetheless, during the follow-up period, there was a significant improvement on this parameter (p-value = 0.003).

Gait, assessed objectively through the HAI, was also significantly impaired, in comparison with reference values, both at 6 and 12 months after ICU discharge (6 months: median = 1.0, IQR = 0.0–2.3, p-value<0.001; 12 months: median = 0.0, IQR = 0.0–1.0, p-value<0.001). Similarly, there was a significative improvement in this subdomain through the follow-up period (p-value = 0.026).

Regarding functional balance, median BBS was 51.0 (IQR = 42.3–56.0) at 6 months and 56.0 (IQR = 50.0–56.0) at 12 months, values that were significantly reduced in comparison with the reference category (p-values<0.001). Nevertheless, this parameter also evolved positively through time (p-value = 0.001). Moreover, when analyzing the BBS through a cut-off for higher risk of falls, at 6 months 28.6% had a higher risk of this adverse event, a value that decreased to 18.9% at 12 months (p-value = 0.004).

Sensory impairments were present in 16.7% at 6 months and in 7.1% at 12-months. No significant improvement on this subdomain was depicted through the study period (p-value = 0.331).

Signs of CSTD were present in 52.3% (n = 22) at 6 months, and in 26.2% (n = 11) at one year follow-up (p<0.001).

PICS–Mental domain

Overall, at 6 months after ICU discharge, 31.0% (n = 13) had impairments on the mental domain of PICS, value that was reduced to 26.2% (n = 11) at 12 months (p-value = 0.159).

Both at 6 and 12 months after ICU discharge, the mean score at the HADS–Anxiety (5.2±4.5 and 4.0±4.3, respectively) and at the HADS–Depression scores (6.3±3.7 and 5.2±4.9, respectively) was significantly superior to the reference value (p-value<0.001). Using the 8 points cut-offs for significant anxiety and depression, a significant difference from our values was also reported (p-value = 0.003 and 0.022), pointing to the absence of significant impairments, despite the presence of alterations. In both anxiety and depression scores, there was a significative positive evolution through the follow-up period (anxiety: p-value = 0.012; depression: p-value<0.001).

PICS—Cognitive domain

The prevalence of cognitive dysfunction at 6 months after ICU discharge was of 79.2%, value that shifted to 64.9% at 12-months (p-value = 0.003). In fact, significant disablement was noted in comparison with reference values both at 6 months (median = 20.0, IQR = 11.3–24.8, p-value<0.001) and 12 months (median = 2.03; IQR = 16.0–26.5; p-value<0.001). Yet, a global improvement in this 1-year follow-up was noted (p-value = 0.0031).

Functionality

In our sample, 54.8% of the patients were fully active on ADL 6 months after ICU discharge, with a median FIM of 126.0 (IQR = 125.0–126.0), a value significantly inferior to the reference category (p-value = 0.018). Moreover, at 12-months after ICU discharge, functionality significantly improved, with 74% of patients being totally independent (p-value = 0.002). Nevertheless, median FIM was still significantly diminished (median = 126.0, IQR = 126.0, p-value = 0.002).