Features of 20 133 UK patients in hospital with covid-19 using the ISARIC WHO Clinical Characterisation Protocol: prospective observational cohort study

Objective To characterise the clinical features of patients admitted to hospital with coronavirus disease 2019 (covid-19) in the United Kingdom during the growth phase of the first wave of this outbreak who were enrolled in the International Severe Acute Respiratory and emerging Infections Consortium (ISARIC) World Health Organization (WHO) Clinical Characterisation Protocol UK (CCP-UK) study, and to explore risk factors associated with mortality in hospital. Design Prospective observational cohort study with rapid data gathering and near real time analysis. setting A case report ISARIC time


Introduction
The outbreak of disease caused by the novel severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) was declared a pandemic by the World Health Organization on 11 March 2020. 1 The WHO situation doi: 10.1136/bmj.m1985 | BMJ 2020;369:m1985 | the bmj report dated 30 April 2020 stated 3 090 445 people had confirmed coronavirus disease 2019 (covid- 19) and 217 769 people had died across the world. 2 In the wake of the influenza A H1N1 pandemic (2009) and the emergence of Middle East respiratory syndrome coronavirus (2012), it was recognised that the effectiveness of a response to a future pandemic threat would critically depend on the speed and focus of that response. The United Kingdom set up and maintained a "sleeping" prepandemic suite of protocols, documents, and agreements in preparation for future outbreaks. The International Severe Acute Respiratory and emerging Infections Consortium (ISARIC) WHO Clinical Characterisation Protocol UK (CCP-UK) study was a core component of this portfolio. 3 Further details about ISARIC WHO CCP-UK can be found at https://isaric4c.net and in the online supplement.
In response to the emergence of SARS-CoV-2 and its pandemic potential, the ISARIC WHO CCP-UK study was activated on 17 January 2020, in time to enrol the first wave of patients with covid-19 admitted to hospitals in England and Wales. The first confirmed patient with covid-19 in the UK was reported on 31 January 2020.
Hospital admission rates for patients with covid-19 have been difficult to estimate because rates depend on the prevalence of community testing and admission criteria, which vary between countries. However, an estimated one in 10 to one in five adults have illnesses of sufficient severity to warrant hospital admission. 4 Patients have mostly been admitted with severe acute respiratory infection or severe acute respiratory syndrome according to the previous WHO case definitions. 5 6 The provision of intensive care also varies between countries. Studies first from China, and more recently from Europe and the United States, have found rates of admission to intensive care range from 5% to 32%. 7 8 Old age, chronic major comorbidity, and male sex have consistently been associated with increased mortality. [9][10][11][12] In this first report of the ISARIC WHO CCP-UK study, we characterise the clinical features of patients admitted to hospital with covid-19 in England, Scotland, and Wales during the growth phase of the first wave of this outbreak, up to 19 April 2020. Future reports will include Northern Ireland. We describe all patient outcomes as known on 3 May 2020 and explore risk factors associated with mortality in hospital.

Participants
Inclusion criteria were people of all ages who were admitted to one of 208 acute care hospitals in England, Scotland, and Wales with proven or high likelihood of infection with a pathogen of public health interest, defined as SARS-CoV-2 for this event by Public Health England. Reverse transcriptase polymerase chain reaction was the only mode of testing available during the period of study. The decision to test was at the discretion of the clinician attending the patient, and not defined by protocol. The enrolment criterion "high likelihood of infection" reflects that a preparedness protocol cannot assume that a diagnostic test will be available for an emergent pathogen. Site training emphasises that only patients who tested positive for covid-19 were eligible for enrolment.
National guidance was provided by Public Health England and other UK public health agencies that advised who to test based on clinical case definitions for possible covid-19 (online supplement). We also included patients who had been admitted for a separate condition but had tested positive for covid-19 during their hospital stay. We collected additional biological samples for research purposes when consent was given (please see online supplement for details of consent procedures and biological samples). These samples are currently undergoing analysis and we will present the results when they become available. Patients were only enrolled during their index admission. We used three tiers in the ISARIC WHO CCP-UK protocol. Patients in tier 0 had clinical information from their routine health records uploaded into the case report form. Consent was not required for collection of depersonalised routine healthcare data for research in England and Wales. A waiver for consent was given by the Public Benefit and Privacy Panel in Scotland. Tier 1 and 2 of the protocol involve additional biological sampling for research purposes for which consent by, or assent for, participants was obtained.

Data collection
We collected baseline demographic data on a paper case report form (version 9.2; supplementary material 2) that was developed by ISARIC and WHO for use in outbreak investigations. Data were uploaded from admission, and usually before hospital episodes were complete, to a REDCap database (Research Electronic Data Capture, Vanderbilt University, US, hosted by University of Oxford, UK). We aimed to record measures of illness severity and routine blood test results at a minimum of four time points: day of hospital admission (day 1), day 3, day 6, day 9, and day of any admission to critical care. We recorded relevant treatments that patients received in hospital, level of care (ward based, high dependency unit, or intensive care unit), complications, and details of discharge or death while in hospital. Further information about these variables can be found in the online supplement.

Outcomes
The main outcomes were critical care admission (high dependency unit or intensive care unit) and mortality in hospital or palliative discharge. We chose a priori to restrict analysis of outcomes to patients who were admitted more than two weeks before data extraction (3 May 2020) to enable most patients to finish their hospital admission.
bias Research nurses relied on local covid-19 test reports to enrol patients. Capacity to enrol was limited by staff resources at times of high covid-19 activity. Otherwise we are unable to comment on the potential selection bias of our cohort. We are in the process of linking to routine administrative healthcare data and will be able to make comparisons at that point.

Missing data
The nature of the study means that a large amount of data were missing, particularly during the later parts of the growth curve of the UK outbreak. Because this paper is mainly descriptive, we have not performed any imputation for missing data, and describe the data as they stand. To reduce the impact of missing data on outcome analyses, we restricted these analyses to patients who had been admitted for at least two weeks before data extraction.

statistical analyses
Continuous data are summarised as median (interquartile range) and categorical data as frequency (percentage). For univariate comparisons, the Mann-Whitney U test or Kruskal-Wallis test were used. We compared categorical data by using the χ 2 test.
We used several approaches to model survival. Discharge from hospital was considered an absorbing state, meaning that once discharged, patients were considered no longer at risk of death. Patients who were discharged were not censored and held within the risk set, therefore accounting for the competing risk of discharge on death. We checked this approach by using a formal Fine and Gray competing risks approach. Hierarchical Cox proportional hazards approaches included geographical region (clinical commissioning group or health board) as a random intercept. We used a parsimonious criterion based model building approach based on several principles: clinically relevant explanatory variables were identified a priori for exploration; population stratification was incorporated; interactions were checked at first order level; final model selection was informed by log likelihood tests and the concordance statistic, with appropriate assumptions checked including the distribution of residuals and requirement for proportional hazards. We set statistical significance at 5%. All tests were two sided. We analysed data by using R (R Core Team version 3.6.3, Vienna, Austria), with packages including tidyverse, finalfit, survival, cmprsk, and coxme.

Patient and public involvement
This was an urgent public health research study in response to a Public Health Emergency of International Concern. Patients or the public were not involved in the design, conduct, or reporting of this rapid response research.

results
On behalf of ISARIC WHO CCP-UK, 2468 research nurses, administrators, and medical students enrolled 20 133 patients who were admitted with covid-19 to 208 hospitals in England, Scotland, and Wales between 6 February and 14:00 on 19 April 2020 (table 1 and fig E1). This figure represents 34% of the 59 215 covid-19 admissions in these countries. The median time from onset of symptoms of covid-19 in the community to presentation at hospital was 4 days (interquartile range 1-8; n=16 221).

age and sex
The median age of patients was 73 years (interquartile range 58-82, range 0-104; fig 1); 310 patients (1.5%) were less than 18 years old and 194 (1.0%) were less than 5 years old. More men (59.9%, n=12 068) than women (40.1%, n=8065) were admitted to hospital with covid-19. One hundred women (10%) of reproductive age (n=1033) were recorded as being pregnant. Clusters of symptoms on admission were apparent ( fig E2). The most common symptom cluster encompassed the respiratory system: cough, sputum, shortness of breath, and fever. We also observed three other clusters: one encompassing musculoskeletal symptoms (myalgia, joint pain, headache, and fatigue); a cluster of enteric symptoms (abdominal pain, vomiting, and diarrhoea); and less commonly, a mucocutaneous cluster. Twenty nine per cent (5384/18 605) of all patients complained of enteric symptoms on admission, mostly in association with respiratory symptoms; however, 4% of all patients described enteric symptoms alone.    E4). For patients who died, we found no association between age and time to death, with around 80% dying before day 14 of hospital admission.

association of pre-existing patient characteristics and survival
The online supplement (table E4) describes univariable and multivariable associations with mortality. Figure  5 shows variables that remained significant in the multivariable model. Increasing age was a strong predictor of mortality in hospital after adjusting for major comorbidity (reference age <50 years): 50-59 years, hazard ratio 2.63 (95% confidence interval 2.06 to 3.35, P<0.001); 60-69 years, 4.99 (3.99 to 6.25, P<0.001); 70-79 years, 8.51 (6.85 to 10.57, P<0.001); ≥80 years, 11.09 (8.93 to 13.77, P<0.001). Female sex was associated with lower mortality (0.81, 0.75 to 0.86, P<0.001). Chronic cardiac disease, chronic non-asthmatic pulmonary disease, chronic kidney disease, obesity, chronic neurological disorder (such as stroke), dementia, malignancy, and liver disease were also associated with increased hospital mortality. An interactive infographic is available at https:// isaric4c.net/info. This information must not be used as a predictive tool in practice or to inform individual treatment decisions.

Principal findings
Patients with covid-19 usually presented with fever, cough, and shortness of breath, and met the WHO case definitions for severe acute respiratory infection or severe acute respiratory syndrome. The most common previous major comorbidities were chronic cardiac disease, diabetes, and chronic non-asthmatic pulmonary disease. Seventeen per cent of patients were admitted to critical care (high dependency unit or intensive care unit). Mortality in hospital was at least 26%, with 34% of the cohort still in hospital at the time of analysis; these proportions increased with escalating level of care. Factors associated with mortality in hospital were increasing age, male sex, and major comorbidities (cardiac disease, non- The data presented in this study describe patients admitted to hospital during the growth phase of the SARS-CoV-2 pandemic in the UK. The first 101 patients were enrolled in the early phase of the outbreak as part of a high consequence infectious disease containment strategy that ended on 10 March 2020. These patients and others who were identified through screening in hospital, or who contracted covid-19 after admission (hospital acquired infection), are included in the 855 patients who were admitted without covid-19 symptoms. The impact these patients have had on the overall cohort characteristics has diminished as numbers have increased, and we believe it is important to keep these patients in the study. Other patients in our cohort without covid-19 symptoms are those who were diagnosed with the disease at the discretion of the clinician looking after them while staying in hospital for other reasons.

ICU/HDU admission Count
The pattern of disease we describe broadly reflects the pattern reported globally. 7 Patients in our study had a higher median age and higher rates of chronic obstructive pulmonary disease and asthma than patients in China 8 and the US. 11 12 The prevalence of obesity in our study (11%) was considerably lower than the overall UK prevalence (29%). 13 This proportion could reflect the relatively elderly male population admitted to hospital and misclassification or under reporting by admitting physicians. Our patients presented with a relatively short time interval between onset of symptoms and admission to hospital, which might also be a function of the older and vulnerable patient population.
The current case definition of cough and fever, if strictly applied, would miss 7% of our inpatients. A smaller proportion, 4% of patients, presented with enteric symptoms only. This figure could be an underestimate because these patients fall outside standard criteria for testing. This enteric presentation risks misclassification of patients, and assignment to non-covid-19 care areas, which could pose a nosocomial transmission risk. Severe SARS-CoV-2 infections are rare in people younger than 18 years, comprising only 1.5% of those admitted to hospital. Only 1.0% of those in our study were younger than 5 years. The J shaped age distribution is starkly different to the U shaped age distribution seen in seasonal influenza and the W shaped distribution observed in the 2009 influenza pandemic. 14 The reason why SARS-CoV-19 has mostly spared children is not clear, but we speculate this could be because angiotensin converting enzyme 2 receptors are expressed differently in younger lungs.

No of patients
Other studies have not widely reported that obesity as recognised by clinical staff is associated with mortality in hospital after adjustment for other comorbidities, age, and sex. Obesity was recognised as a risk factor in the 2009 influenza A H1N1 pandemic, but not for the 2012 Middle East respiratory syndrome coronavirus. 15 16 The proportion of pregnant women in our cohort was small (10%), similar to the estimated proportion of pregnant women in the community. 15 Pregnancy was not associated with mortality, in apparent contrast to influenza. 17 comparison with other studies The proportion of patients admitted to critical care in our study was similar to that reported in Italy (17%), 18 19 and New York (14.2%), 11 12 but higher than China. 8 At the time of enrolment, the Intensive Care Society had issued guidance to its members that there would be no rationing of critical care admission until all capacity in the country had been exhausted. As far as we are aware, critical care capacity was not exceeded in the UK during the period of the study. We do not believe that any equipment shortages existed during this period that might have prompted more aggressive futility discussions.
Mortality in our cohort was high in patients admitted to general wards who were not admitted to critical care, which suggests that advanced care planning occurred. We were unable to capture treatment limiting decisions about level of care. The high median age of patients who died in the cohort (80 years) could partly explain the high mortality rate. Mortality rates were extremely high for patients who received invasive mechanical ventilation in the intensive care unit compared with the 2009 influenza A H1N1 pandemic, for which mortality in intensive care was 31%. 15 Our data were in line with the initial ICNARC (intensive care national audit and research centre) audit reports, which represent intensive care units in England, Wales, Scotland, and Northern Ireland. 20 Outcome analyses only included patients who were admitted before 19 April to allow most patients to complete their hospital admission. However, an inherent reporting bias exists because the sickest of patients, particularly those admitted to intensive care, have the longest hospital stays; mortality rates in hospital could therefore increase. These mortality rates were considerably higher than the 24% mortality rate in hospital seen in patients in intensive care units in Italy 19 and the US. 11 12 The lower rate in the US could in part be explained by differences in healthcare systems and the proportion of intensive care unit beds to hospital beds between the two countries. In Italy,  19 The finding of independent associations of advancing age, male sex, chronic respiratory (nonasthmatic) disease, chronic cardiac disease, and chronic neurological disease with mortality in hospital is in line with early international reports. 9 10 However, although age adjusted mortality rates are high in elderly patients, most of these patients were admitted to hospital with symptoms of covid-19 and would not have been in hospital otherwise. Enhanced severity in male patients was seen across all ages.
strengths and limitations of study ISARIC WHO CCP-UK stood ready to conduct large scale studies of pandemic outbreaks for eight years, enabling us to enrol 34% of all patients with covid-19 admitted to 208 acute care hospitals across England, Wales and Scotland in the early phase of the pandemic.
Our study has some limitations. We do not currently have data on the inpatients that were not enrolled, or people managed in community settings, such as usual domestic residences and older people's care homes. We are unable to comment on community risk factors that drive hospital admission except by inference from expected representation at admission. We will be linking to routine administrative healthcare datasets which will enable us to assess the presence of any selection bias.
A large amount of data were missing and we suggest there are two main reasons for this. Firstly, enrolment occurred in the nonlinear growth phase of the outbreak, and outcomes for recent admissions have not been reported yet; these admissions account for 18% of the total number of patients enrolled. Secondly, the research network was dealing with unprecedented numbers of patients at a time when many were seconded to clinical practice or themselves off sick. This study is ongoing, and further data are being added to case report forms.
We suggest it is possible that the sickest patients were enrolled in our study, and this could partly explain our high mortality rates in hospital. Some of the sickest patients in the study had the longest lengths of hospital stay and we do not have outcome data for all of these patients yet.
conclusions and policy implications This large and rapidly conducted study of patients admitted to hospital in England, Wales, and Scotland with covid-19 shows the importance of putting plans in place for the study of epidemic and pandemic threats, and the need to maintain these plans. Our study identifies sectors of the population that are at greatest risk of a poor outcome, and reports the use of healthcare resources. Most patients with covid-19 experience mild disease. However, in our cohort, of those who were admitted to hospital two weeks before data extraction, less than half have been discharged alive and a quarter have died. The remainder continued to receive care at the date of reporting. Seventeen percent of patients admitted to hospital required critical care. Factors associated with mortality in hospital were increasing age, male sex, obesity, and major comorbidities.
ISARIC Coronavirus Clinical Characterisation Consortium 21 investigators have submitted regular reports to the UK Government's New and Emerging Respiratory Virus Threats Advisory Group (NERVTAG) 22 and the Scientific Advisory Group for Emergencies (SAGE). 23 Patient level data have been shared and independently analysed by the Scientific Pandemic Influenza Group on Modelling (SPI-M) 24 and other investigators. Aggregated data have been shared with WHO in the ISARIC covid-19 report.
Studies such as this cannot be developed, approved, and opened from the start of a pandemic in time to inform case management and public health policy. Our study has shown the importance of forward planning and investment in preparedness studies. Over the next few months we will issue reports in The BMJ on specific topics and analyses that are key to understanding the impact of covid-19 and focus on improving patient outcomes.
autHOr aFFiliatiOns