key: cord-0814311-nk6w423x
authors: Alhilani, Michel; Cohn, Martin; Nakhoul, Maria; Than, Jonathan; Sim, Sing Yue; Choi, Byung; Jegatheeswaran, Lavandan; Minocha, Amal; Mutengesa, Ernest; Zala, Ashik; Karagiannis, Georgios
title: Predictors of mortality and ITU admission for COVID‐19 patients admitted to a London district general hospital: A retrospective cohort study
date: 2021-10-01
journal: Health Sci Rep
DOI: 10.1002/hsr2.404
sha: d2af4ab1cbf79c958db3f84d5e7ff593debabc17
doc_id: 814311
cord_uid: nk6w423x

nan

The coronavirus disease 2019 outbreak that originated in Wuhan, Hubei Province in China, was classified as a global pandemic by the World Health Organisation (WHO) on March 11, 2020. 1 As of April 8, 2021, the causative virus SARS-CoV-2 has spread to 219 countries and territories around the world, with 133 897 605 individuals infected and 2 904 686 deceased as a result. 2 COVID-19 manifests as a wide spectrum of disease ranging from asymptomatic infection to multiple organ failure requiring ITU admission and potentially leading to death. [3] [4] [5] In the United Kingdom (UK), the first confirmed cases were reported on January 27, 2020 6 and London has comprised the epicenter of its outbreak, bearing the highest rate of admission and mortality from COVID-19 in the country. 7 A number of cohort studies from Italy, China, the United States (USA), and early UK studies have demonstrated the clinical characteristics and outcomes of patients with COVID-19 within their individual populations. 5, [8] [9] [10] [11] Risk factors for ITU admission and mortality identified by these studies vary, with factors such as increasing age and various comorbidities showing consistent association, whereas other factors such as gender, ethnicity, antihypertensive medication use, and hematological laboratory results are inconsistently reported to predict adverse outcomes.

We therefore present the demographic, laboratory, and clinical features of patients with COVID-19 admitted to a district general hospital in London and assess predictors of mortality and ITU admission outcomes, in order to identify parameters aiding risk stratification of such patients in a secondary care setting. 14, 2020. We have included all patients aged >18 years with reverse transcription-polymerase chain reaction (RT-PCR) confirmed SARS- 

The demographics, clinical characteristics, imaging data, and laboratory parameters on presentation, and the study outcomes were extracted retrospectively from electronic medical health records. Demographic characteristics collected included age, sex, and ethnicity. Ethnicity was divided into four main groups: Asian, black, white, and other (comprising mixed/multiple ethnic groups and other ethnic groups). 12 Clinical characteristics collected included pre-existing medical conditions and presenting symptoms. The following chronic comorbidities were recorded: hypertension (HTN), ischemic heart disease (IHD), diabetes mellitus (DM), chronic kidney disease (CKD), asthma, chronic obstructive pulmonary disease (COPD), active cancer (Ca), cerebrovascular accident (CVA), dementia, and congestive cardiac failure (CCF). Information regarding current smoking status and use of angiotensin-converting enzyme inhibitors (ACEis) or angiotensin II receptor blockers (ARBs) were also collected. All information related to comorbidities was checked against previous hospital admission letters, clinic letters, and GP health records where accessible.

Presenting symptoms collected on admission included shortness of breath (SOB), fever (sublingual temperature of >37.9 C), cough, myalgia, production of sputum, fatigue, diarrhoea, nausea and vomiting, and chest pain.

Blood test results collected included routine hematology and biochemistry panels. Arterial blood gas results were collected where performed. In cases where venous instead of arterial blood gas was performed, only pH, lactate, and bicarbonate measurements were reviewed. 13 Chest X-rays (CXRs) of our patient cohort were reported by the Hospital Radiology Department according to the British Society of Thoracic Imaging CXR reporting guidelines, 14 which deemed a CXR (CVCX3). In our study, we categorzsed CXR reports into two groups: 

Data collected from our patient cohort were subject to analysis using R version 3.6.1. Characteristics of the cohort were described with categorical variables represented as n (%), and continuous variables were represented as median with interquartile range (IQR). Wilcoxon ranked tests, Student's t-tests, chi-squared tests, and fisher-exact tests were used to compare differences between features for surviving and deceased patient groups when appropriate. The normality of variables was assessed using the Shapiro test, and all P-values were corrected using Bonferroni correction. Ethnicity was analyzed using two approaches: The first approach was to compare all ethnic groups against each other, whereas the second was to compare white ethnicity vs all other ethnicities.

To assess the association between outcomes and variables, univariate and multivariate logistic regressions were used with results presented as odds ratios with a 95% confidence interval. When choosing our predictors for the multivariate regression, the regsubsets R function was used to obtain the different model subsets-chooses variables based on the branch and bound algorithm. 15 Models having either the highest adjusted R 2 , or lowest Akaike information criterion (AIC), or lowest Bayesian information criterion (BIC) were extracted, and the subset of the variables were chosen based on being selected by at least two of the models. Variables with greater than 20% missing data, or for which an odds ratio for the univariate analysis could not be retrieved, were excluded from the multivariate regression model. 10 Since our primary outcomes were of competing risks, this as well as right-censored data were accounted for. Cumulative incidence plots were used to evaluate the proportion of patients who were either deceased or discharged from the hospital over time.

Amongst our cohort of 383 patients were included in the study analysis ( Figure 1 ). The median age was 71 years (54, 82), and 210 (54.8%) were male (Table 1) . White ethnicity was the most prevalent in our study cohort. White patients had a higher median age compared to other ethnicities, and the median age amongst the four groups was statistically different (P < .001), with white having the highest median age of 78.5, followed by Asian 63, other 60.5, and black 54.

Of these 383 patients, 311 (81.2%) had definite outcomes at the end of the study period; 72 (18.8%) were considered censored, or with incomplete outcomes, due to being transferred or still hospitalized at the date of the last recorded follow-up on May 12, 2020 ( Figure 1 ). Of those with definite outcomes, 184 patients (59.2%) were discharged from hospital, and 127 (40.8%) died during their stay. Furthermore, 295 patients had complete outcomes with regard to ITU admission. The median age, sex, ethnicity along with certain comorbidities, symptoms, and laboratory findings were significantly different between those who were admitted to ITU and those who were not ( Table 2) .

Median length of stay for all patients was 7 (IQR 3-12) days.

Median length of stay for deceased patients was 6 (IQR 2-11) days.

White ethnicity was associated with increased length of hospital stay compared to the other ethnicities (P = .0021).

In univariate logistic regression, age was the only significant demographic variable that was associated with an increased risk of mortality (Table 3) .

Within our cohort, black, Asian, and other minority ethnicity (BAME) groups were overrepresented in terms of both hospital and ITU admission; 51.2% of admitted patients, and 76.4% of ITU patients were from BAME groups despite comprising only 39.9% of the London Borough of Hillingdon population. 12 Despite this overrepresentation, there was no statistically significant association between ethnicity and odds of mortality in our study, in both univariable and multivariable analysis controlling for potential confounders including sex, age, and comorbidities.

Cough and myalgia were associated with lower odds of mortality, whereas SOB was linked to increased odds (Table 3) . Classic finding of COVID-19 on CXR was associated with increased odds of mortality. Comorbidities that had a significant association with outcome were HTN, COPD, cancer, CCF, and dementia (Table 3 ). Laboratory results that had a significant association on outcome were WCC >11.0 10 9 /L, neutrophils >7.7 10 9 /L, CRP >100 mg/L, albumin ≤35 g/ L, creatinine >125 μmol/L, and eGFR ≤60 mL/min (Table 3 ).

Significant predictors of outcome from the multivariable regression model were age, CXR reported as classic appearance of COVID-19, SOB, WCC > 11.0 10 9 /L, and eGFR ≤60 mL/min (Table 3 ).

In univariable logistic regression, age, white ethnicity, and other ethnicities were associated with decreased odds of ITU admission F I G U R E 1 Patient cohort pathway (Table 4 ). Male sex was strongly associated with increased likelihood of ITU admission ( Table 4 ). The presence of symptoms on admission such as SOB, fever, cough, and production of sputum were all significant predictors of admission to ITU ( Table 4 ).

The admission blood tests significantly associated with increased odds of ITU admission in univariable logistic regression included ALT >45 U/L, LDH >250 U/L, d-dimer >500 μg/L, and pH >7.45 on blood gas. On the other hand, eGFR ≤60 mL/min, PCO2 > 6.4 kPa, and HCO3 > 28 mmol/L on blood gas were significant predictors of decreased odds of ITU admission. Although raised LDH, d-dimer, PCO2, and HCO3 (increase and decrease) appear to be significantly associated with odds of ITU admission, missingness of data of more than 20% meant that these markers could not be conclusively convincingly interpreted in logistic regression.

In the multivariable logistic regression model, we found that male sex, SOB, production of sputum, and ALT >45 U/L were strongly associated with increased likelihood of ITU admission (Table 4 ).

We present a comprehensive analysis of the baseline characteristics and factors associated with short-term mortality and ITU admission in 383 patients over 18 years old presenting to a London district general T A B L E 1 Demographic, clinical, laboratory, and radiographic findings of patients on admission Note: Only significant results are shown above. Detailed results are found in Table S3 . Variables with greater than 20% missing data were excluded from the regression analysis and this table.

Abbreviations: 95% CI, 95% confidence interval; CCF, congestive heart failure; COPD, chronic obstructive pulmonary disease; CRP, C-reactive protein; eGFR, estimated glomerular filtration rate; HTN, hypertension; WCC, white cell count. The main limitation of our study is its single-center nature. We recognize missing laboratory results as a potential limitation of our multivariate regression model (eg, d-dimer, ferritin, troponin, and blood gases), although this was accounted for through use of a maximum cut-off of 20% missing data for each variable. We do recognize that body mass index data on admission was unavailable to the authors, and thus, this study was unable to assess its effect on primary outcomes or exclude its potential confounding nature. We acknowledge that due to adherence to the NICE guidelines on admission to critical care 24 favoring younger healthier patients, this potential confounder should be taken into consideration when assessing ITU regression analysis results. We also recognize the lack of inclusion of initial treatments given to our cohort of patients as a limitation, preventing our study from assessing the impact of treatment on mortality and disease severity.

Our study provides comprehensive characterization of presenting clinical features and predictors of mortality and ITU admission for hospitalized COVID-19 patients within our ethnically diverse population.

Our study shows that features, such as older age, classic COVID-19 chest X-ray, SOB, WCC > 11 10 9 /L, and eGFR ≤60 mL/min, found on admission should prompt the clinician to be aware of an increased odds of mortality for those patients. The risk factors for poor clinical outcomes identified within our study broadly correlate with those previously identified in previous UK and international studies and should facilitate the development of risk stratification algorithms for COVID-19 secondary care management in future. Dr Michel Alhilani had full access to all of the data in this study and takes complete responsibility for the integrity of the data and the accuracy of the data analysis.

Dr Michel Alhilani affirms that this manuscript is an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned (and, if relevant, registered) have been explained.

The authors confirm that the data supporting the findings of this study are available within the article and its supplementary materials.

https://orcid.org/0000-0001-5391-6459

World Health Organization. WHO Director-General's Opening Remarks at the Media Briefing on COVID-19-11

European Centre for Disease Prevention and Control. COVID-19 situation update worldwide

Clinical, laboratory and imaging features of COVID-19: a systematic review and meta-analysis

A familial cluster of pneumonia associated with the 2019 novel coronavirus indicating person-to-person transmission: a study of a family cluster

Epidemiological and clinical characteristics of 99 cases of 2019 novel coronavirus pneumonia in Wuhan, China: a descriptive study

Lessons for managing high-consequence infections from first COVID-19 cases in the UK

Disparities in the Risk and Outcomes of COVID-19. London: Public Health England

Clinical features of patients infected with 2019 novel coronavirus in Wuhan

Presenting characteristics, comorbidities, and outcomes among 5700 patients hospitalized with COVID-19 in the New York City area

Report 17 -Clinical Characteristics and Predictors of Outcomes of Hospitalised Patients with COVID-19 in a London NHS Trust: a Retrospective Cohort Study

Baseline characteristics and outcomes of 1591 patients infected with SARS-CoV-2 admitted to ICUs of the Lombardy region

National Identity and Religion. London: Office for National Statistics

Comparison of arterial and venous pH, bicarbonate, PCO2 and PO2 in initial emergency department assessment

Thoracic imaging in COVID-19

Regressions by leaps and bounds

Features of 16,749 hospitalised UK patients with COVID-19 using the ISARIC WHO clinical characterisation protocol. medRxiv

Clinical features of 95 sequential hospitalised patients with novel coronavirus 2019 disease (COVID-19), the first UKcohort

OpenSAFELY: factors associated with COVID-19-related hospital death in the linked electronic health records of 17 million adult NHS patients

Clinical characteristics of fatal and recovered cases of coronavirus disease 2019 in Wuhan, China: a retrospective study

Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study

Predictors of mortality for patients with COVID-19 pneumonia caused by SARS-CoV-2: a prospective cohort study

Hypertension and COVID-19

National Institute for Health and Care Excellence. COVID-19 Rapid Guideline: Critical Care in Adults. London: National Institute for Health and Care Excellence

B lymphopenia in uremia is related to an accelerated in vitro apoptosis and dysregulation of Bcl-2

IL-10, IL-6, and TNFalpha: central factors in the altered cytokine network of uremia-the good, the bad, and the ugly

Key Workers: Key Facts and Questions. London: The Institute for Fiscal Studies

Ethnicity and COVID-19: an urgent public health research priority

Predictors of mortality and ITU admission for COVID-19 patients admitted to a London district general hospital: A retrospective cohort study