Patients infected with SARS-CoV-2 (COVID-19 disease) present with respiratory tract infection and symptoms such as fever, cough, fatigue, sputum production and/or breathlessness.1 The spectrum of COVID-19 varies from asymptomatic infection through mild upper respiratory tract illness, to severe and potentially fatal viral pneumonia with respiratory failure. The Chinese Centre for Disease Control and Prevention reported that 70,420 of the 80,928 confirmed cases in China are “cured and discharged from hospital,” while 3,245 have died.2 However, in the early stage of this pandemic, and due mainly to the lack of understanding of the properties of the virus, inadequate medical protection, high infectivity and absence of effective treatment there was a dramatic increase in the number of patients exceeding medical resources. As a result, the initial patient discharge rate was reported to be relatively low.3 Scientists and health care workers around the world are working to improve treatments in order to reduce mortality, and improved recovery rates are anticipated. Chronic morbidity due to permanent organ damage in some patients is likely, and many patients who have recovered from the acute illness will need ongoing health care and allied health support.
Twenty-three percent of patients with severe acute respiratory syndrome caused by SARS-CoV-1 (SARS) had impaired lung function and a reduced exercise capacity after 1 year compared to normal predicted values.4,5 Similarly, there is a strong suggestion of prolonged lung function deficits in some patients with COVID-19. A recently published study6 on 50 middle-aged (median [interquartile range] 54 y [46 to 62 y]; 44% female) patients reported that 27 (54%) had impaired lung function 30 days after the onset of symptoms. In another study, 14 of 55 “recovered” COVID-19 patients had lung function abnormalities 3 months after resolution of the acute disease, and radiological abnormalities were present in 39 patients.7 Given the intensive medical management for people with severe loss of respiratory function including prolonged mechanical ventilation, sedation and use of neuromuscular blockade, these patients are likely to be at high risk of intensive care unit acquired weakness (ICU-AW).8 This disease has independent long-term effects on symptoms and physical function.9 It is therefore desirable to define the clinical trajectory in these patients so as to provide them with appropriate rehabilitative interventions from a symptomatic, physical and psychological perspective.
The primary aim of the proposed study is to therefore provide a multidimensional analysis of recovery over 12 months following diagnosis and hospital admission in adults with COVID-19, with possible extension if recovery trajectories are more prolonged. The specific objectives are to document the duration of symptoms and the time course of physical and psychological recovery during and following the acute (in-hospital or at home) phase of the disease; detect risk factors for patients who deteriorate in spite of treatment; and identify factors and specific elements of pragmatic rehabilitation programs that are associated with recovery of symptoms (fatigue and shortness of breath), physical function (functional capacity and peripheral muscle force), HRQoL, feelings of anxiety and depression, and post-traumatic stress over 12 months.
This will be a prospective, longitudinal, multi-centre cohort study led by Allied Health staff at Fiona Stanley Hospital, in collaboration with Royal Perth Hospital and Sir Charles Gairdner Hospital in Perth, Western Australia (WA). It is designed to complement the Western Australian Health Translation Network (WAHTN) COVID Research Response (CRR). The WAHTN is a state-wide collaborative network of WA’s universities, medical research institutes, public and private hospitals, PathWest and the WA Department of Health. International collaboration is also a possibility.
Adults > 18 years of age and who have tested positive for COVID-19 admitted to the participating hospitals will be eligible to participate. Informed consent will be sought. Exclusion criteria will be pre-existing neuromuscular disorders thought to affect the measures of physical function or known mental illness, or significant communication or cognitive impairment thought to affect responses to self-report measures. Participants will be given a unique study ID, and may withdraw consent at any time, but data accumulated to that point will be retained with assent from the patient.
Outcome assessment Protocol and Timeline
This study assessment protocol has been designed in two parts. The participant flow through this study’s assessment protocol is outlined in Figure 1. The outcome assessment protocol is outlined in Table 1.
Outcomes and Measures
A detailed description of the tests and tools used to assess each outcome measure in Parts A and B is included in the Supplementary Information (after references).
Part A: Acute Hospital Period
Part A will be undertaken by participants during their hospital stay. They will receive standard treatment but will have additional clinical and symptom assessments every 48 hours. The assessments will be performed via phone communication with the participant (who will be in isolation); with the assistance of the attending nurse and/or from distant observation of participant’s performance (in the tests related to physical function) via a glass window. This will be on a case-by-case basis. In the case of patient admission to the intensive care unit (ICU) and the need for mechanical ventilation, the assessments will not be attempted while intubated but will resume following extubation from mechanical ventilation in ICU. Routine collection of symptoms and an abridged physical assessment will provide valuable insight into the acute course of the disease to the development of adaptive algorithms to predict those who deteriorate. The first assessment will occur on recruitment and will be completed every 48 hours during the hospital stay.
Symptoms of fatigue and shortness of breath will be assessed using The Edmonton Symptom Assessment Scale (ESAS).10,11 The physical assessments will comprise measures of functional capacity, using the 1-minute sit-to-stand test (1STS),12 and peripheral muscle power (handgrip power) measured using a hydraulic hand dynamometer (Jamar; JA Preston Corporation; MI, USA).13 In view of the safety challenges and physical distancing requirements for COVID-19 positive patients, the physical testing has been abridged to clinical tests which can be completed by researchers and treating clinicians (physiotherapists or nurses).
Part B: Discharge to 12 months
Participants will undergo assessments (Table 1) at four time points T1 – T4: T1, at hospital discharge; T2, T3 and T4, at 3, 6 12 months dated from the date of admission. Part B assessments T2 to T4 will be performed either in person, or via the use of electronic data capture including telehealth systems, REDCap electronic data capture tool and/or the South Metropolitan Health Service Virtual Clinic System (VCS) (data capture via smartphone).
From T1 to T4 measures to be collected are: (i) functional capacity by the 1STS (face-to-face if the participant remains in contact with clinical services, or remotely by video link); (ii) fatigue related to activities by the Fatigue Severity Scale (FSS);14 (iii) shortness of breath during activities by the Modified Medical Research Council Dyspnoea Scale (MMRC);15 (iv) health-related quality of life by the EuroQol 5 dimensions – five response level (EQ-5D-5L);16 (v) feelings of anxiety and depression: Hospital Anxiety and Depression Scale (HADS);17 and (vi) posttraumatic stress: Impact of Events Scale–6 (IES-6).18,19
Clinical data and other information
Clinical data will be available in the form of Tier 0 clinical data as per the ISARIC / WHO RAPID minimum dataset (https://isaric.tghn.org/COVID-19-CRF/) captured during the WAHTN COVID-19 Research Response program. The data will consist of patient demographics (age, gender, body-mass index at admission), pre-existing comorbidities, short term outcomes such as number of days in intensive care unit, duration of mechanical ventilation, length of hospital stay, recorded mobility status (average distance walked per day), and average daily distance during hospital stay. These data will assist predictive modelling and to identify factors affecting clinical variability after hospital discharge. At T2, patients will be asked to provide a total number of days in isolation as well as information on use of so-called “wellbeing apps” (exercise, mindfulness, meditation, etc.).
Study data will be recorded and held electronically in the REDCap electronic data capture tool hosted at the Department of Health Western Australia.20,21 REDCap is a secure, web-based software platform designed to support data capture for research projects. This data will only be accessible to the recruiters and senior co-investigators who have established and, or been granted permission to access the database.
Sample size and statistical analysis plan
In this observational trial all suspected and proven COVID-19 positive inpatients in each facility will be given the opportunity to join the study. A 50% recruitment rate is envisaged, owing to the complexity of engaging patients in isolation and availability of recruitment staff resources.
Descriptive summaries will include mean ± SD or median [interquartile range] for continuous data, and frequency distributions for categorical data [counts or percentage]. Comparisons of patient baseline characteristics (gender, age, comorbidities) and clinical characteristic data by disease severity (mild or severe/critical) will be undertaken using parametric independent t-tests or oneway ANOVA for normally distributed continuous data and non-parametric Mann-Whitney U or Kruskall-Wallis H tests for continuous data not confirmed to display a normal distribution. Chi-squared or Fisher’s Exact tests, as appropriate, will be applied for categorical data.
Associations of patient and clinical characteristic predictors with longitudinal in-hospital symptomatic and physical functioning outcomes (including ESAS, 1STS, handgrip force) will be examined using generalised linear mixed models (GLMM) with random subject effects, adjusting for disease severity, length of stay in ICU and in hospital, and selected baseline (admission) characteristics. Associations of patient characteristic, clinical pathway and rehabilitation pathway predictors with post-discharge recovery trajectories (including 1STS, FSS, MMRC, EQ-5D-5L, HADS, IES-6 and clinical sequalae) will be examined using GLMM with random subject effects, adjusting for disease severity and selected baseline (discharge) characteristics. All models will be implemented as univariable and predictive multivariable models, with all covariates univariately significant at alpha=0.15 considered as candidate predictors in the multivariable models. Potential confounding variables informed by the literature will also be included in the multivariable models. Results will be summarised using estimated marginal means and 95% CI for continuous outcomes or odds ratios (OR) and 95% CI for categorical outcomes. Model fit will be assessed using residuals checking and cross-validation of models will be undertaken by splitting the dataset to form training and test sets if the sample is large. Missing data points in longitudinal data will be accounted for by the use of maximum likelihood estimation (MLE) methods in the models. Temporal outcomes including time to death and time to discharge in the in-hospital analysis and time to reach recovery milestones, defined according to EQ5D age and gender appropriate norms, in post-discharge analysis will be assessed using Kaplan-Meier survival probabilities and compared between relevant patient characteristic, clinical pathway and rehabilitation pathway predictors using Log rank tests. Effects of these predictors on temporal outcomes will be examined using Cox proportional hazards regression models, censoring at 12 months post-admission and adjusting for relevant baseline patient and clinical characteristics. Results will be summarised using hazard ratios (HR) and 95% CI. Sensitivity analyses will be performed on patients who are engaged in any other forms of exercise/wellbeing apps/activities and those who are not. Stata MP version 16.0 (StataCorp LP, College Station, TX) will be used for data analysis. All hypotheses will be two-sided and significance will be set at alpha = 0.05. All analyses will be conducted by senior biostatisticians (coinvestigators AJ and HJC).
Provenance: Externally reviewed
Ethics & Dissemination: The protocol has been granted Ethical approval by the South Metropolitan Health Service Human Research Ethics Committee (HREC) (REG 0000004040), with reciprocal approval at the other sites. Results will be published after peer review and disseminated in the community by the consumer group representatives in the LATER-19 advisory group.
Conflicts or declarations of interest: None declared
Funding interests: Supported by the WA Department of Health and the WA Health Translation Network (COVID-19 Research Grants Program). EuroQOL has licensed use of the EQ-5D-5L.
Author Contribution: DWE, LN, AM, LVDL and VC contributed to study conception and design. PG obtained Ethical approval and wrote the initial manuscript. AJ prepared the statistical analysis. The LATER-19 Coinvestigator Group has advised regarding study design and development.
Corresponding author: Dale Edgar, Departments of Physiotherapy and Exercise Physiology, Fiona Stanley Hospital, Murdoch, Western Australia. firstname.lastname@example.org