This document describes a study that evaluated patient care and outcomes for critically ill COVID-19 patients admitted to intensive care or high-care units in Africa. The study found that mortality for these patients was 54.7%, higher than global rates. Mortality was associated with limited critical care resources and increased organ dysfunction at admission. Factors like need for ventilation, multiple organ support, and vasopressor use predicted higher mortality, while HIV status did not. The results suggest African patients have worse outcomes due to resource constraints and highlight the need for early intervention and triage tools.
Call Girl Service Bidadi - For 7001305949 Cheap & Best with original Photos
Publication on ACCCOS.pdf
1. See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/345336480
An African, Multi-Centre Evaluation of Patient Care and Clinical Outcomes for
Patients with COVID-19 Infection Admitted to High-Care or Intensive Care
Units
Preprint · November 2020
CITATIONS
0
READS
436
35 authors, including:
Some of the authors of this publication are also working on these related projects:
GlobalSurg Randomised Trial (RCT-1) View project
Performance based financing and its impact on service quality and utilization View project
Bruce Biccard
University of Cape Town and Groote Schuur Hospital, Cape Town, South Africa
293 PUBLICATIONS 9,504 CITATIONS
SEE PROFILE
Malcolm G Miller
University of Cape Town
18 PUBLICATIONS 409 CITATIONS
SEE PROFILE
Adesoji O Ademuyiwa
University of Lagos
189 PUBLICATIONS 2,111 CITATIONS
SEE PROFILE
Aniteye Ernest
University of Ghana Medical School National Cardiothoracic Centre
87 PUBLICATIONS 453 CITATIONS
SEE PROFILE
All content following this page was uploaded by Ahmed Y. Azzam on 05 November 2020.
The user has requested enhancement of the downloaded file.
2. 1
Title page
An African, multi-centre evaluation of patient care and clinical outcomes for patients with COVID-19
infection admitted to high-care or intensive care units
Bruce M Biccard,1 Malcolm Miller,2 William L Michell,3 David Thomson,4 Adesoji Ademuyiwa,5 Ernest
Aniteye,6 Greg Calligaro,7 Hailu Tamiru Dhufera,8 Mohamed Elfagieh,9 Mahmoud Elfiky,10 Muhammed
Elhadi,11 Maher Fawzy,12 David Fredericks,13 Meseret Gebre,14 Abebe Genetu Bayih,15 Anneli Hardy,16 Ivan
Joubert,17 Fitsum Kifle,18 Hyla-Louise Kluyts,19 Kieran DM Macleod,20 Zelalem Mekonnen,21 Mervyn Mer,22
Akinyinka Omigbodun,23 Christian Owoo,24 Fathima Paruk,25 Jenna L Piercy,26 Juan Scribante,27 Yakob
Seman,28 Elliott H Taylor,29 Dawid EA van Straaten,30 P. Dean Gopalan,31 on behalf of the ACCCOS
Investigators*
1. Prof Bruce M Biccard PhD, Professor and 2nd Chair, Department of Anaesthesia and Perioperative
Medicine, Groote Schuur Hospital, Faculty of Health Sciences, University of Cape Town, South Africa
2. Dr Malcolm Miller, Intensivist, Division of Critical Care, Department of Anaesthesia and Perioperative
Medicine, Groote Schuur Hospital, Faculty of Health Sciences, University of Cape Town, South Africa
3. Emeritus Associate Professor William L Michell FFA(SA) (Crit Care), Honorary Specialist Division of
Critical Care, Department of Anaesthesia and Perioperative Medicine, Groote Schuur Hospital, and
Senior Scholar, Department of Surgery, Faculty of Health Sciences, University of Cape Town, South
Africa
4. Dr David Thomson, MMed, Consultant, Division of Critical Care, Department of Anaesthesia and
Perioperative Medicine, Groote Schuur Hospital, Faculty of Health Sciences, University of Cape
Town, South Africa
5. Prof Adesoji O Ademuyiwa, FACS, Professor of Surgery (Paediatric and Surgical Epidemiology),
Department of Surgery, College of Medicine, University of Lagos and Lagos University Teaching
Hospital, Lagos, Nigeria
6. Dr Ernest Aniteye, Associate Professor of Anaesthesia, Cardiothoracic Anaesthetist and Intensivist,
Korle-Bu Teaching Hospital and University of Ghana Medical School, Accra, Ghana
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
3. 2
7. A/Prof Greg Calligaro MMed (Centre for Lung Infection and Immunity, Division of Pulmonology,
Department of Medicine and UCT Lung Institute and South African MRC/UCT Centre for the Study of
Antimicrobial Resistance, University of Cape Town, Cape Town, South Africa.
8. Dr Hailu Tamiru Dhufera MD, Medical Service Directorate General, Ministry of Health, Ethiopia
9. Prof Mohamed Ahmed Elfagieh FRCS MD, NCI, Misrata, Libya
10. Dr Mahmoud Elfiky MD, Kasr Al Ainy Faculty of Medicine, Cairo University, Ali Ibrahim St, Manial,
Cairo 11432.
11. Dr Muhammed Elhadi MBBCh, Faculty of Medicine, University of Tripoli, University Road, Furnaj,
13275 Tripoli, Libya
12. Prof Maher Fawzy MD, Emeritus Professor of Anesthesia , ICU and Pain Management, Faculty of
Medicine, Cairo University, Egypt
13. Dr David Fredericks Cert Crit Care(SA), Critical Care Subspecialist, Department of Critical Care
Medicine, Groote Schuur Hospital, Faculty of Health Sciences, University of Cape Town, South
Africa.
14. Dr Meseret Gebre, MD, Pediatrician and Researcher, Armauer Hansen Research Institute, Addis
Ababa, Ethiopia
15. Dr Abebe Genetu Bayih, Director General, Armauer Hansen Research Institute, Addis Ababa, Ethiopia
16. Ms Anneli Hardy MS Statistics, Statistical Consultant, Department of Statistical Sciences, Faculty of
Science, University of Cape Town, South Africa
17. A/ Prof Ivan Joubert, FCA(SA), Head of Critical Care, Division of Critical Care, Department of
Anaesthesia and Perioperative Medicine, Groote Schuur Hospital, Faculty of Health Sciences,
University of Cape Town, South Africa
18. Dr Fitsum Kifle Belachew, Lecturer, BSc, Debre Berhan University College of Medicine Department
of Anaesthesia, Debre Berhan Comprehensive Specialized Hospital, Debre Berhan, Ethiopia
19. Prof H Kluyts MMed (Anaes), Department of Anaesthesiology, Sefako Makgatho Health Sciences
University, South Africa
20. Dr Kieran DM Macleod MbChB, Glasgow Royal Infirmary, Glasgow, United Kingdom
21. Dr Zelalem Mekonnen (MD,MSc), Armauer Hansen Research Institute, Addis Ababa, Ethiopia
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
4. 3
22. Prof Mervyn Mer PhD, Department of Medicine, Divisions of Critical Care and Pulmonology,
Charlotte Maxeke Johannesburg Academic Hospital and Faculty of Health Sciences, University of the
Witwatersrand, Johannesburg, South Africa
23. Prof Akinyinka O Omigbodun FWACS, Professor of Obstetrics and Gynaecology, College of
Medicine, University of Ibadan, Nigeria
24. Dr Christian Owoo, FGCS, Consultant Anaesthetist and Intensivist, University of Ghana Medical
School, Korle Bu Teaching Hospital and University of Ghana Medical Centre, Accra, Ghana
25. Prof Fathima Paruk PhD, Associate Professor, Clinical and Academic Head of Department of Critical
Care, Steve Biko Academic Hospital, Faculty of Health Sciences, University of Pretoria, South Africa
26. Dr Jenna L Piercy, Cert Crit Care, Head Clinical Unit, Division of Critical Care, Department of
Anaesthesia and Perioperative Medicine, Groote Schuur Hospital, Faculty of Health Sciences,
University of Cape Town, South Africa
27. A/Prof Juan Scribante, PhD Associate Professor, Department of Anaesthesiology, School of Clinical
Medicine. Faculty of Health Sciences, University of the Witwatersrand, South Africa
28. Dr Yakob Seman A MPH, Director General, Medical Services, Ministry of Health, Ethiopia
29. Elliott H Taylor BSc (Hons), Oxford University Global Surgery Group, Nuffield Department of
Surgical Sciences, University of Oxford, United Kingdom and Global Surgery Division, Department of
Surgery, University of Cape Town, Cape Town, South Africa
30. Mr Dawid EA van Straaten BA, Safe Surgery South Africa NPC, 5 14th Street, Parkhurst, Gauteng,
2193, South Africa
31. Dr P. Dean Gopalan, PhD, Chief Specialist and Head of Discipline: Anaesthesiology & Critical Care,
School of Clinical Medicine, University of KwaZulu Natal, Durban, South Africa
*ACCCOS Investigators listed in Supplementary Material
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
5. 4
Corresponding author
Professor Bruce Biccard
Department of Anaesthesia and Perioperative Medicine,
Groote Schuur Hospital and University of Cape Town,
South Africa.
E-mail: bruce.biccard@uct.ac.za
Telephone: +27 (0) 76 160 6387
Facsimile: +27 (0) 21 4066589
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
6. 5
Research in context
Evidence before this study
There is little data to guide the management of critically ill COVID-19 patients in under-resourced
environments. Published systematic reviews confirmed that there were no published outcomes data from Africa,
and little data on factors associated with mortality or survival in under-resourced environments. We expected
the outcomes to be potentially worse in Africa as the ability to provide care may be compromised by a limited
workforce, limited intensive care facilities and critical care resources. In order to understand the management of
critically ill COVID-19 infected patients in under-resourced environments, we designed an African continental
study to determine which resources, patient comorbidities and critical care interventions may be associated with
either mortality or survival in these patients. It is hoped that these data may provide some evidence for the
development of realistic management strategies for these critically ill patients in under-resourced settings. We
also updated a global critical care COVID-19 mortality meta-analysis to provide context to our findings.
Added value of this study
The mortality following critical care admission for suspected and confirmed COVID-19 infection in this African
cohort is 54·7% (95% CI 51·9-57·6). The meta-analysis reports a global mortality of 31·4% (95% CI 24·6-
38·2), with the African data reporting an excess mortality of between 18 and 29 deaths per 100 patients when
compared to other regions. The excess mortality may be explained by the lack of critical care resources. Only
one in two patients referred for critical care were admitted. Patients were admitted to units with limited access to
dialysis, proning, extracorporeal membrane oxygenation (ECMO), arterial blood gases, and pulse oximetry.
Furthermore, at the patient level, access to interventions such as dialysis, proning, and ECMO were estimated to
be between seven- and 14-fold lower than what is needed to manage critically ill COVID-19 patients. Adjusted
analyses suggest that critical care mortality does not appear to be associated with patient comorbidities (with the
exception of increasing age) but rather with the severity of organ dysfunction on presentation to critical care,
and the initial need for maximal respiratory and cardiovascular support. The quick sequential organ failure
assessment (SOFA) score at admission was strongly associated with patient mortality, and may be a simple and
feasible triage tool to use in an under-resourced environment.
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
7. 6
Implications of all the available evidence
As mortality is strongly associated with organ dysfunction and the organ support needed at critical admission,
the role of early warning systems to identify the need for early interventions prior to critical care referral are
needed. The use of the quick SOFA score may provide some guidance for appropriate triage at the time of
referral to critical care in an under-resourced setting when managing critically ill COVID-19 patients. It is likely
that patient outcomes will continue to be severely compromised until the limited critical care resources are
addressed.
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
8. 7
Abstract
Background
There is little data on critically ill COVID-19 patients in under-resourced environments, and none from Africa.
The objectives of this study were to determine resources, patient comorbidities and critical care interventions
associated with mortality in critically ill COVID-19 African patients.
Methods
African multi-centre, prospective observational cohort study of adult patients referred to intensive care or high-
care units with suspected or known COVID-19 infection. Patient follow up was until hospital discharge,
censored at 30 days. The study recruited from March to September 2020.
Findings
1243 patients from 38 hospitals in six countries participated. The hospitals had a median of 2 (interquartile
range (IQR) 1-4) intensivists, with a nurse to patient ratio of 1:2 (IQR 1:3 to 1:1). Pulse oximetry was available
to all patients in 29/35 (82·9%) sites, and 21/35 (60%) of sites could provide dialysis or proning. The 30-day
mortality following critical care admission was 54·7% (95% confidence interval (CI) 51·9-57·6). Factors
independently associated with mortality were an increasing age (odds ratio (OR) 1·04, 95% CI 1·02-1·05,
p<0·001), a quick SOFA score of 3 (OR 3·61, 95% CI 1·41-9·24, p=0·01), increasing respiratory support
defined as the need for continuous positive airway pressure (OR 5·86, 95% CI 1·47-23·35, p=0·01), invasive
mechanical ventilation (OR 16·42, 95% CI 4·52-59·65, p<0·001), three organ systems requiring support at
admission (OR 5·52, 95% CI 1·13-27·01, p=0·04), cardiorespiratory arrest within 24 hours prior to admission
(OR 4·43, 95% CI 1·01-19·54, p=0·05) and vasopressor requirements (OR 2·73, 95% CI 1·71-4·36, p<0·001).
Human immunodeficiency virus was not associated with mortality (OR 1·84, 95% CI 0·99-3·40, p=0·05).
Interpretation
Mortality in critically ill COVID-19 African patients is higher than any other region, with an excess mortality of
18 and 29 deaths per 100 patients compared to other regions. Mortality is associated with limited critical care
resources and severity of organ dysfunction at admission.
Funding:
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
9. 8
African Covid-19 Critical Care Outcomes Study (ACCCOS) was partially supported by a grant from the Critical
Care Society of Southern Africa.
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
10. 9
Introduction
COVID-19 caused by coronavirus SARS-CoV-2 (severe acute respiratory syndrome coronavirus 2) was
declared a pandemic and an international healthcare emergency by the World Health Organization. It has spread
across the globe, overwhelming healthcare systems by causing high rates of critical illness. The global case
fatality rate is approximately 3%,1 with older people with comorbidities being extremely vulnerable.2 There is a
concern that there will be further mortality with a second wave across countries.
We expect the patient outcomes to be potentially worse in Africa as the ability to provide sufficient care will be
compromised by a limited workforce,3 and limited intensive care facilities and critical care resources. It is
estimated that there are 0·8 (95% confidence interval 0·3 to 1·45) critical care beds per 100,000 population in
Africa.4 It is likely that the volume of unplanned admissions associated with COVID-19 will further adversely
affect critical care outcomes in Africa,5 especially as the ability of healthcare systems in Africa to adapt and
expand during the pandemic to meet the clinical workload is unknown. Finally, the patient outcomes following
COVID-19 critical care are not documented in this under-resourced environment,6 with a call for prevention and
response measures on the individuals in low- and middle-income countries without impacting on response
activities.7
As there were little data on how to manage these critically ill patients in early April 2020,8 we designed an
African continental study to determine which resources, patient comorbidities and critical care interventions are
potentially associated with either mortality or survival in Africa. Rapid dissemination of these findings may help
inform appropriate resource prioritisation and utilisation necessary to manage critically ill COVID-19 patients in
Africa during this pandemic. These points provided the rationale for the African Covid-19 Critical Care
Outcomes Study (ACCCOS). This objective remains relevant as a recent meta-analysis of critically ill COVID-
19 patients reported an in-hospital mortality of 41·6 (95% confidence interval (CI) 34·0-49·7) deaths per 100
patients with no data on outcomes from Africa,6 and no data on how to manage these patients in a resource-
limited setting.
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
11. 10
The objectives of this research were to determine which; i) critical care resources, ii) patient comorbidities, and
iii) hospital interventions are associated with in-hospital mortality in patients with suspected or known COVID-
19 infection referred for critical care in Africa.
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
12. 11
Methods
Study design, participants and setting
This was an African multi-centre prospective observational cohort study of adult (≥18 years) patients referred to
intensive care or high-care units in Africa with suspected or known COVID-19 infection. Patient follow up was
until hospital discharged, censored at 30 days if still in-hospital. The study was to recruit from April to
December 2020, with an interim analysis after 250 to 300 deaths registered in the study. The rationale for the
interim analysis was to provide data which may potentially be associated with improved critical care outcomes
in Africa, in a timeous manner for possible implementation during this COVID-19 pandemic. This study was
registered on ClinicalTrials.gov (NCT04367207). The primary ethics approval was from the Human Research
Ethics Committee of the University of Cape Town (HREC 213/2020).
Eligible patients included all consecutive adult patients at participating centres referred for high-care unit or
intensive care unit admission with suspected or known COVID-19 infection. We planned to recruit as many
centres as possible in Africa. All centres were requested to include all eligible patients in the study if possible,
and requested to recruit as long as possible, with the understanding that sites may stop recruiting at any point if
they were overwhelmed by clinical service commitments.
The primary ethics committee approved a ‘delayed consent’ process. As we expected that most patients would
be unable to consent at the time of admission to critical care due to their condition and the critical care
management needed, the ‘delayed consent’ provided for consent by the patient (following stabilisation or
recovery in critical care) or a legal representative or proxy (should the patient be unable to provide consent). If
there was no opportunity to acquire ‘delayed consent’ (i.e. the patient did not recover sufficiently to provide
consent and there was no legal representative or proxy to provide consent for participation) before the study
outcome was reached (i.e. 30 days in hospital, or the patient died in hospital), then the ethics committee
approved the inclusion of the patient’s data in the study. The justification for this consent process and data
inclusion were to minimise the risk of a non-consecutive patient enrolment which would result in a biased
sample with artificially low estimates of adverse outcomes in African COVID-19 critical care patients. All sites
had to fulfil local ethics and regulatory requirements to participate.
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
13. 12
Variables
Centre specific data were collected once for each hospital including: secondary/tertiary centre, number of
hospital beds, number and level of critical care beds, details about the reimbursement status of the hospital and
other local factors affecting patient care during the study period e.g. nurse to patient ratio. All patient data were
collected on a one-page case record form (CRF) (Supplementary Material S1). The definitions document is
shown in Supplementary Material S2. The management of the patients was left to the discretion of the
healthcare providers at participating sites.
Bias
To ensure a representative sample, we planned to include as many sites as possible with the requirement for
inclusion of all consecutive patients, and a ‘delayed consent’ with inclusion of data where a patient or proxy was
unable to provide consent.
Outcomes
The primary outcome was in-hospital mortality in adult patients referred to intensive care or high-care units
following suspected or known COVID-19 infection in Africa. The secondary outcome was to determine the
factors (human and facility resources, patient comorbidities and critical care interventions) associated with
mortality in adult patients with suspected or known COVID-19 infection in Africa.
Statistical analysis
The sample size was dependent upon the number of centres recruited and their respective caseloads. Each centre
was requested to complete a screening tool of all eligible patients during study participation. The duration of
enrolment at a site was determined by the local lead investigator, and the circumstances allowing participation
during the COVID-19 pandemic. With approximately 25 to 30 variables in the dataset (Supplementary Material
S1) which may be associated with mortality in COVID-19 patients requiring critical care admission, we planned
to do this interim analysis once 250 to 300 deaths were registered in the study to allow assessment of all these
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
14. 13
risk factors in a multivariable regression while ensuring that we did not violate the principle of approximately
10 events per variable.9 A decision was taken to lock the database on 7 September 2020 for this interim analysis,
with the hope that these data could inform further management in Africa during the ongoing COVID-19
pandemic.
A statistical analysis plan was written and published on ClinicalTrials.gov prior to data evaluation
(Supplementary Material S3). Data are presented at a continental African level with all institutions anonymised.
Categorical variables are described as proportions and compared using chi-square tests. Continuous variables
are described as mean and standard deviation (SD) if normally distributed or median and inter-quartile range
(IQR) if not normally distributed. Comparisons of continuous variables between groups are performed using t-
tests, one-way ANOVA or equivalent non-parametric tests as appropriate. The main model included patients
with complete outcome data (i.e. excluded patients who were still in hospital receiving therapy, and had not
reached the outcome definition of death, discharge, or alive in-hospital at 30 days). A three-level generalized
linear mixed model (GLMM) using a logit link was fitted to identify independent risk factors for the binary
outcome of mortality, with patients being at the first level, hospital at the second and country at the third level,
to account for the expected correlation in outcomes within hospitals and countries. A fully conditional
specification (FCS) method was used to impute missing values for variables. The FCS method uses an iterative
Markov chain Monte Carlo (MCMC) method. In each iteration, the FCS method sequentially imputes missing
values in the order specified in the variable list. We used a predictive mean matching (PMM) method for scale
variables. Five imputed datasets were constructed. All risk factors were considered for entry into the model as
the number of reported deaths exceeded 10 events (deaths) per variable,9 provided there was no evidence of
collinearity. The variables considered for inclusion in the model included; i) subject variables, ii) resource
variables and iii) therapy variables. Subject related variables included age, sex, body mass index (BMI),
coronary artery disease, congestive heart failure, hypertension, stroke or transient ischaemic attack, diabetes
mellitus, cancer, current smoker, chronic lung disease, active tuberculosis, chronic liver disease, human
immunodeficiency virus/ acquired immunodeficiency syndrome (HIV/AIDS), chronic or previous malaria,
chronic kidney disease, cardiorespiratory arrest in 24 hours prior to critical care referral, quick sequential organ
failure assessment score (qSOFA score)10 on referral, and sequential organ failure assessment score (SOFA
score)11 on referral/ admission. Resource related variables included whether admission was delayed due to lack
of resources (e.g. bed, staffing etc.), nurse to patient ratio in critical care, ability to provide invasive ventilation
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
15. 14
for the patient if required, and whether a physician was available on site 24/7 for patient care. Therapy related
variables included organ support at admission (defined by the number of organ systems requiring support i.e.
none, one, two, three or more), respiratory support (none, oxygen mask, high flow nasal oxygenation,
continuous positive airway pressure), proning (none, when not ventilated, or invasive mechanical ventilation),
ventilatory support (none, non-invasive ventilation, invasive ventilation), intubation (no, yes elective, yes
emergency), inotropes/ vasoconstrictors, dialysis, therapeutic anticoagulation, steroid therapy, repurposed/
experimental COVID-19 drug therapy, extracorporeal membrane oxygenation (ECMO). Collinearity was
assessed using the variance inflation factor (VIF). The VIF showed collinearity associated with intubation,
respiratory and ventilatory interventions, and anticoagulation. Therefore, for the GLMM we created a single
categorical variable for respiratory support (none, oxygen, high flow nasal oxygen, continuous positive airway
pressure, invasive mechanical ventilation), and we removed the variables ‘dialysis’ and ‘ECMO’ which had
collinearity with ‘anticoagulation’. A three-level random-intercept mixed effects logistic regression was
performed on each of the five imputed datasets using the ‘glmer’ function in the ‘lme4’ package12 in R.13
Estimates were combined from the five repeated complete data analyses using the ‘pool’ function from the
‘mice’ package.14 The ‘pool’ function implements the rules developed by Rubin for combining the separate
estimates and standard errors from each of the imputed datasets to provide an overall estimate with standard
error, confidence intervals (CI) and p values.15 A p-value of <0.05 was considered significant. To allow for
comparison with the imputed datasets, the complete case analyses for the regression models are also presented.
The results of the GLMM are reported as adjusted odds ratios (OR) with 95% CI. The following sensitivity
analyses were defined a priori; i) confirmed SARS-CoV-2 positive patients only, and ii) confirmed SARS-CoV-
2 patients only with exclusion of all patients who had life support withdrawn or a decision to limit therapy. A
further sensitivity analysis was also conducted which only considered patients who died or were discharged
alive at 30 days (excluding patients alive, but still in hospital at 30 days). All analyses were conducted by AH
and BB.
A post hoc decision was taken to update the critical care meta-analyses of COVID-19 mortality by region6 and
risk factors for mortality,8 in order to provide further context to the ACCCOS findings. This update was
conducted by ET, KM, ME and JS, and the protocol was registered on PROSPERO (CRD42020212347).
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
16. 15
Univariate analyses and the imputations were performed using the Statistical Package for the Social Sciences
(SPSS) version 24 (SPSS Inc., Chicago, IL, USA). The mixed effects logistic regressions of the original dataset
and imputed datasets were conducted using R.16 The meta-analysis was conducted using StataCorp. 2019. Stata
Statistical Software: Release 16. College Station, TX: StataCorp LLC.
Role of the funding source
The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing
of the report. The corresponding author had full access to all the data in the study and had final responsibility for
the decision to submit for publication.
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
17. 16
Results
40 African countries were invited to participate via the African Perioperative Research Group (APORG), with
26 country leaders accepting the invitation. Only six countries managed to obtain the necessary ethical and
regulatory requirements necessary to participate in this study in time to be included in this interim analysis,
which included a total of 38 hospitals from the six countries (Egypt (9), Ethiopia (7), Ghana (2), Libya (7),
Nigeria (2) and South Africa (11)). Thirty-five (92·1%) of these hospitals provided hospital level data. 29/35
(80·6%) were university affiliated hospitals. 28/35 (80%) were tertiary hospitals, 5/35 (14·3%) were secondary
level hospitals and 2/35 (5·7%) were primary level hospitals. 32/35 (91·4%) were government funded hospitals,
1/35 (2·9%) was privately funded, and 2/35 (5·7%) had dual funding. The resource characteristics of the
participating hospitals are shown in Table 1. The hospitals had a median of 2 (IQR 1-4) intensivists, with a
nurse-to-patient ratio of 1:2 (IQR 1:3 to 1:1). Only 89% of sites could provide pulse oximetry to all patients in
the critical care or high care environment enrolled in the study, and only 60% could provide renal replacement
therapy.
Insert Table 1 near here
Between April and September 2020, 5391 patients were referred to critical care with suspected or known
COVID-19 infection, of which 2687 (49·8%) were admitted to critical care. 1243/2687 (46·3%) of the patients
were entered into the database with a median of 20 (IQR 10 to 36) patients per hospital. The patient recruitment
is shown in Figure 1.
Insert figure 1 near here
The median duration of recruitment was 72 (IQR 46 to 95) days per hospital. 1126/1235 (91.2%) of the cohort
were confirmed SARS-CoV-2 positive either before or during the critical care admission. These patients were
referred from the emergency department (293/1228, 23·9%), from within the hospital (463/1228, 37·7%) or
from another hospital (472/1228, 38·4%). The site of referral was not associated with mortality (p=0·058).
Admission to critical care was delayed in 138/1080 (12·8%) due to a lack of available resources e.g. bed space
at the time of admission.
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
18. 17
The patient characteristics are shown in Table 2. The patients were young with few comorbidities. The most
common comorbidities were hypertension 635/1206 (52·7%), diabetes 474/1197 (39·6%) and HIV/AIDS
120/1189 (10·1%). On univariate analyses, mortality was more common in patients who were older, had
diabetes or HIV/AIDS. Current smokers and a history of malarial infection were associated with a lower
mortality. The severity of disease at critical care presentation was associated with increased mortality; including
cardiorespiratory arrest prior to admission, the need for increasing organ support, and an increasing qSOFA
score.10 A qSOFA score of 3 was associated with 83/100 (83%) mortality in this cohort. As most patients were
tachypnoeic on admission, it was the hypotension and decreased cognition components of the qSOFA that
contributed most to the increasing qSOFA score, and associated mortality. The full SOFA score was also
associated with mortality, although the majority of patients did not have a full SOFA score conducted at
admission (798/1243, 64·2%). Critical care interventions including ventilatory and respiratory support,
inotropes and dialysis were all associated with mortality.
The length of critical care stay was 8 days (IQR 4 to 14) days. The decision to limit therapy was made in
148/1144 (12·9%) patients and therapy was withdrawn in 35/1132 (3·1%) patients. In 32 of the 35 patients in
whom therapy was withdrawn, therapy had already been limited. There was no difference in the length of
critical care stay between patients who had therapy limited or withdrawn and patients who did not (p=0·56).
Primary outcome measure
Incidence of in-hospital mortality in adult patients referred to intensive care or high-care units following
suspected or known COVID-19 infection in Africa was 631/1153 (54·7%, 95% CI 51·9-57·6). Of the alive
patients, 122/1153 (10·6%) were in hospital at 30 days, and 400 (34·7%) discharged alive at 30 days. The
outcome is unknown for 90 patients.
Secondary outcome measures
The data missingness for the GLMM is shown in Supplementary Material S4. Five variables had more than 5%
missingness, and four variables with >10%. The GLMM for the risk factors (resources, comorbidities and
interventions) associated with mortality in adult patients with suspected or known COVID-19 infection in Africa
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
19. 18
are shown in Table 3. The sensitivity analyses which include the original full dataset, and sensitivity analyses
for confirmed SARS-CoV-2 positive patients only, and confirmed SARS-CoV-2 positive patients in whom
therapy was neither withdrawn nor limited are shown in Supplementary Material S5 and S6. Supplementary
Material S5 presents data for outcome of mortality versus alive (either in hospital or discharged) and
Supplementary Material S6 presents the data for outcome of mortality compared to patients discharge alive at 30
days. Factors independently associated with mortality were increasing age, a qSOFA score of 3, increasing
respiratory support defined as the need for continuous positive airway pressure (CPAP) or invasive mechanical
ventilation, at least three organ systems requiring support at admission, cardiorespiratory arrest within 24 hours
prior to critical care referral and the need for inotropes or vasopressors. The likelihood ratio for mortality with a
qSOFA of 3 was 4·15, and for organ support of ≥3 organ systems requiring support was 2·93. Neither diabetes,
hypertension nor HIV/AIDS were associated with mortality. The sensitivity analyses all supported these
findings. No human resource factors were associated with mortality.
A post hoc analysis to explore relationships between the frequency of interventions in patients invasively
mechanically ventilated showed that 70/594 (11·8%) also received dialysis.
The updated meta-analyses are shown in Supplementary Material S7 to S9. The updated search strategy is
presented in Supplementary Material S7, the PRISMA flow diagram in Supplementary Material S8, and the
updated meta-analysis of mortality for critical ill COVID-19 patients is shown in Supplementary Material S9.
The overall reported global mortality is 31·4% (95% CI 24·6-38·2), and the African data is higher than any
other region (the lower confidence interval of the African region does not overlap with the upper confidence
interval of any other region). The mortality rate per 100 admissions is between 18 and 29 deaths per 100 higher
in Africa. The meta-analysis summary table and references are shown in Supplementary Materials S10 and S11.
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
20. 19
Discussion
The principal finding of this study is that the 30-day mortality following critical care admission for suspected
and confirmed COVID-19 infection in this African cohort was 54·7% (95% CI 51·9-57·6) and this represents an
excess mortality of between 18 and 29 deaths per 100 patients when compared to other regions. This high
mortality does not appear to be driven by limited human resources nor patient comorbidities, with the exception
of increasing age. Comorbidities, such as hypertension, diabetes, HIV and increasing BMI were not associated
with mortality. Mortality was associated with the severity of organ dysfunction on presentation, and the need for
maximal respiratory and cardiovascular support. The qSOFA score of 3 appears to be a simple and effective tool
for triaging COVID-19 patients at risk of mortality when referred for admission to critical care.
Limited critical care resources are potentially important contributors to the high mortality reported in this
African cohort. Firstly, there are inadequate critical care beds, with only 1:2 referrals been admitted to critical
care. Previous estimates have suggested that the number of critical care beds in Africa may be more than 10-fold
lower than Europe and North America, at an estimated 0·8 (0·3–1·45) beds per 100 000 population.4 It is
possible that the limited number of critical care beds may contribute to the admission of a relatively ‘sicker’
patient cohort than that of more resourced environments. Secondly, the data suggest a very under-resourced
critical care setting, when considering the proportion of sites that could provide dialysis, proning, ECMO,
arterial blood gases, and pulse oximetry. These hospital and critical care resources are suggestive of a low-
volume critical care environment, which has been associated with worse outcomes.17 However, thirdly, and
most importantly merely ‘counting’ the available critical care resources does not accurately reflect the
proportion of patients who actually receive the interventions associated with these resources. We estimate that
the patient access to these interventions is between seven (dialysis and proning) and 14-fold lower (ECMO) than
the documented availability of the resource. Dialysis was only available at 60% of sites, but it was only offered
to 102/1175 (8·7%) patients. Acute kidney injury (AKI) has been reported in over 90% of COVID-19 patients
admitted to intensive care units, and of the patients that require invasive ventilation nearly one in four patients
require renal replacement therapy.18 If the need for renal replacement therapy was the same as that reported in
Hirsch et al (23·2%), then the proportion of patients in the African cohort that should receive dialysis should be
double that reported. Similarly, proning was also only available in 60% of patients, and only delivered to
119/1243 (9·6%) patients on invasive mechanical ventilation. Proning is included in acute respiratory distress
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
21. 20
syndrome (ARDS) management strategies, and it has been provided in excess of 90% of patients (e.g. in the
control arm of the EOLIA Trial).19 A global literature survey showed that 75% of COVID-19 patients referred
to critical care developed ARDS.20 If we assume that 75% of the African cohort developed ARDS, then at least
four times more patients should have received proning during invasive mechanical ventilation in the African
cohort. Similarly ECMO was only available in 6/35 (17·1%), yet large registry data supports its use in COVID-
19 patients with refractory respiratory failure, with a reported mortality in these critically ill patients of 38·0%
(95% CI 34·6–41·5),21 a mortality which is significantly lower than the mortality of the African cohort. ECMO
was only offered to 5/416 (1·2%) patients in whom it was potentially available. These limited resources may
partly explain why only half of the referred patients are admitted in Africa, that one in eight of admitted patients
have therapy withdrawn or limited while in critical care, and the mortality is so high in Africa.
In contrast the human resources available to these critical care units were relatively good with respect to
availability of a physician 24/7, and nurse-to-patient ratios which did not differ between day and night.
However, the inability to admit approximately half of the referred patients to the critical care unit may reflect a
unit working at capacity as determined by the available human resources, and therefore the impact of human
resources may be reflected on adverse outcomes outside the critical care unit which we could not assess in this
analysis.
Limitations
There are limitations to this study. Firstly, the study presents data from predominantly tertiary hospitals.
However, even with the predominance of tertiary hospitals, pulse oximetry was not universally available, and
renal replacement therapy was available in less than 2/3 of the critical care units. It is likely that lower level
hospitals would have less resourced critical care units, which would possibly be associated with worse outcomes
than reported in this cohort. It is also possible that there were limited cases managed at lower levels due to
insufficient critical care resources. Although, we did not show a difference in mortality between patients
referred from within and without the hospital providing the critical care, it is possible that the need for critical
care referral to higher level centres would increase the time taken to access critical care which may result in
excess mortality before reaching a critical care unit. Unfortunately, we cannot assess this contribution to
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
22. 21
mortality. It is therefore possible that the mortality following critical care admission for COVID-19 infection in
Africa is actually higher than what we have reported.
Secondly, only half of the patients admitted to the critical care units in this cohort were entered into this
database. The inability to achieve data submission of all critical care patients admitted into this cohort, may
reflect the demands of conducting research while providing a critical care service in an under resourced
environment during a pandemic. Other contributing factors may include a reluctance to commit to the reasons
for non-admission (as this was part of the screening log but was almost unanimously incomplete), or a lack of
electronic databases to track referrals. This study is also limited by a lack of biomarker data, which would be
impractical in an African setting. We also have little data to understand how one in six patients died without
receiving oxygen, and one in two died without receiving inotropes. It is unclear if these resources were
unavailable, or this was the result of early therapy limitation or withdrawal. Finally, this cohort only represents
six African countries, despite 26 country leaders agreeing to participate. The pandemic resulted in only these six
countries fulfilling ethics and regulatory requirements necessary to allow recruitment into this cohort, and other
barriers associated with conducting research in an under-resourced environment.22 It is therefore difficult to
determine the generalisability of these results, although these data provide the largest cohort of critically ill
COVID-19 patients from under-resourced environments (Supplementary Material S9).
Strengths
The main strengths of this study are the prospective, multi-centre design from a previously unreported African
setting, the large sample size collected over a relatively short period of time, and the only study from a
population with high HIV burden. A recent systematic review could not describe the clinical course of patients
with HIV and COVID-19 infection.23 Our data suggests that HIV/AIDS is not an important contributor to
mortality in patients with confirmed or suspected COVID-19 infection. We would provide some caution with
respect to this comment, as the lower confidence interval just crosses unity with an OR for mortality of 1·84.
This study was adequately powered to adjust for subject, human resource and therapy covariates associated with
mortality, and the statistical analysis plan was published prior to data inspection. All prespecified sensitivity
analyses confirm the main findings. The ability to compare the association between human resources, patient
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
23. 22
comorbidities and critical care interventions on mortality allows us to develop strategies to potentially decrease
mortality in these patients.
Interpretation
Admission to critical care for suspected or known COVID-19 infection is limited in Africa, with only one in two
referred patients admitted. The hospital and critical care resources are limited, and access to core components of
COVID-19 critical care interventions are estimated to be between seven and 14-fold less than what is needed.
Neither human resources nor patient comorbidities appear to be predictive of mortality in Africa, with the
exception of increasing age. The prognosis of these patients is strongly associated with the degree of organ
dysfunction at admission, and the need for invasive mechanical ventilation or inotropic support. The qSOFA is a
simple triage tool which is feasible (with over 95% compliance) and a robust, independent predictor of outcome
in an under resourced environment. A score of 3 had a strong, independent association with mortality, with a
reported mortality of 83% in this cohort. Previously the (full) SOFA score has been shown to have superior
performance to the qSOFA,24 but in this African cohort, the ability to perform a full SOFA was severely limited
due to critical care resource constraints. The performance of the qSOFA in this cohort suggests that it is an
acceptable risk stratification tool for a low resource environment.
Future considerations
As mortality is strongly associated with organ dysfunction and organ support needed at critical admission, the
role of early warning systems and early intervention needs to be urgently evaluated in these patients to avoid
delays in instituting necessary organ support. As critical care resources are severely limited, appropriate triage is
needed at the time of referral, and the qSOFA score may contribute to appropriate triage.
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
24. 23
Other information
Acknowledgements
Rema Ramakrishnan for assistance with the meta-analysis forest plot, and Dilshaad Brey for assistance with the
database search for the meta-analysis.
Registration
ClinicalTrials.gov (NCT04367207): African Covid-19 Critical Care Outcomes Study
Protocol
The protocol and statistical analysis plan are posted at ClinicalTrials.gov (NCT04367207)
Funding
The African Covid-19 Critical Care Outcomes Study was supported by a grant from the Critical Care Society of
South Africa.
Data Sharing Statement: Data will be disclosed only upon request and approval of the proposed use of the data
by the Steering Committee. Data are available to the journal for evaluation of reported analyses. Data requests
from other non-ACCCOS investigators will not be considered until two years after the close out of the trial.
Data will be de-identified for participant, hospital and country, and will be available with a signed data access
agreement.
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
25. 24
References
1. Coronavirus Resource Center. https://coronavirus.jhu.edu/map.html (accessed 30
September 2020).
2. Zhou F, Yu T, Du R, et al. Clinical course and risk factors for mortality of adult inpatients with
COVID-19 in Wuhan, China: a retrospective cohort study. Lancet 2020; 395(10229): 1054-62.
3. Biccard BM, Madiba TE, Kluyts HL, et al. Perioperative patient outcomes in the African
Surgical Outcomes Study: a 7-day prospective observational cohort study. Lancet 2018; 391(10130):
1589-98.
4. Ayebale AET, Kassebaum NJ, Roche AM, Biccard BM. Africa’s Critical Care Capacity before
COVID-19. South Afr J Anaesth Analg 2020; 26(3): 162-4.
5. Skinner DL, De Vasconcellos K, Wise R, et al. Critical care admission of South African (SA)
surgical patients: Results of the SA Surgical Outcomes Study. S Afr Med J 2017; 107(5): 411-9.
6. Armstrong RA, Kane AD, Cook TM. Outcomes from intensive care in patients with COVID-19:
a systematic review and meta-analysis of observational studies. Anaesthesia 2020; 75(10): 1340-9.
7. Gupta M, Wahl B, Adhikari B, et al. The need for COVID-19 research in low- and middle-
income countries. Global Health Research and Policy 2020; 5(1): 33.
8. Taylor EH, Hofmeyr R, Torborg A, et al. Risk factors and interventions associated with
mortality or survival in adult COVID-19 patients admitted to critical care: a systematic review and
meta-analysis. South Afr J Anaesth Analg 2020; 26(3): 116-27.
9. Peduzzi P, Concato J, Kemper E, Holford TR, Feinstein AR. A simulation study of the number
of events per variable in logistic regression analysis. J Clin Epidemiol 1996; 49(12): 1373-9.
10. Seymour CW, Liu VX, Iwashyna TJ, et al. Assessment of Clinical Criteria for Sepsis: For the
Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). Jama 2016; 315(8):
762-74.
11. Vincent JL, Moreno R, Takala J, et al. The SOFA (Sepsis-related Organ Failure Assessment)
score to describe organ dysfunction/failure. On behalf of the Working Group on Sepsis-Related
Problems of the European Society of Intensive Care Medicine. Intensive Care Med 1996; 22(7): 707-
10.
12. Bates D, Maechler M, Bolker B, Walker SA. Fitting Linear Mixed-Effects Models Using lme4.
Journal of Statistical Software 2015; 67(1): 1-48.
13. RStudio. 1.3.959 ed; 2020.
14. van Buuren S, Groothuis-Oudshoorn K. mice: Multivariate Imputation by Chained Equations
in R. Journal of Statistical Software 2011; 45(3): 1-67.
15. Rubin DB. Multiple Imputation for Nonresponse in Surveys. New York.: John Wiley & Sons.;
1987.
16. Team RC. R: A language and environment for statistical computing. VIenna, Austria: R
Foundation for Statistical Computing; 2019.
17. Nguyen YL, Wallace DJ, Yordanov Y, et al. The Volume-Outcome Relationship in Critical Care:
A Systematic Review and Meta-analysis. Chest 2015; 148(1): 79-92.
18. Hirsch JS, Ng JH, Ross DW, et al. Acute kidney injury in patients hospitalized with COVID-19.
Kidney Int 2020; 98(1): 209-18.
19. Combes A, Hajage D, Capellier G, et al. Extracorporeal Membrane Oxygenation for Severe
Acute Respiratory Distress Syndrome. N Engl J Med 2018; 378(21): 1965-75.
20. Tzotzos SJ, Fischer B, Fischer H, Zeitlinger M. Incidence of ARDS and outcomes in hospitalized
patients with COVID-19: a global literature survey. Crit Care 2020; 24(1): 516.
21. Barbaro RP, MacLaren G, Boonstra PS, et al. Extracorporeal membrane oxygenation support
in COVID-19: an international cohort study of the Extracorporeal Life Support Organization registry.
Lancet 2020.
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
26. 25
22. Conradie A, Duys R, Forget P, Biccard BM. Barriers to clinical research in Africa: a
quantitative and qualitative survey of clinical researchers in 27 African countries. British Journal of
Anaesthesia 2018; 121(4): 813-21.
23. Costenaro P, Minotti C, Barbieri E, Giaquinto C, Donà D. SARS-CoV-2 infection in people living
with HIV: a systematic review. Rev Med Virol 2020: e2155.
24. Liu S, Yao N, Qiu Y, He C. Predictive performance of SOFA and qSOFA for in-hospital mortality
in severe novel coronavirus disease. Am J Emerg Med 2020.
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
27. Figure 1. The African Covid-19 Critical Care Outcomes Study (ACCCOS) recruitment
5367 patients referred to critical care
2415 patients admitted to critical care unit
1243 patients in the ACCCOS database
1172 patients not entered into the database
2952 patients not admitted to critical care unit
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
28. Tables
Table 1. The characteristics of the hospitals included in the study
Hospital characteristics and resources N (proportion) or median (IQR)
Population served 2·5 million (IQR 0·3-6·0)
Hospital beds 500 (IQR 330-832)
Critical care beds providing invasive ventilation 16 (IQR 10-40)
Critical care beds unable to provide invasive ventilation 12 (IQR 8-40)
Specialist intensivists 2 (IQR 1-4)
Specialist doctors (not intensivists) 4 (IQR 2-10)
Non-specialist doctors 6 (IQR 4-12)
Nurse to patient ratio (day) 1:2 (IQR 1:2 to 1:1)
Nurse to patient ratio (night) 1:2 (IQR 1:3 to 1:1)
Doctor to patient ratio 1:4 (IQR 1:6 to 1:3)
Doctor on site in critical care after hours 33/35 (94·3%)
Surge capacity
Number of extra intensive care ventilators 5 (IQR 3 to 12)
Number of operating room ventilators available for critical care
upgrade
4 (IQR 0-9)
Haematology laboratory on hospital site 34/35 (97·1%)
Ability to perform arterial blood gases on hospital site 29/35 (82·9%)
Critical care oxygen supply
Vacuum insulated evaporator (VIE) 17/35 (48·6%)
Cylinder oxygen 13/35 (37·1%)
Oxygen concentrator 5/35 (14·3%)
Pulse oximetry
All patients 29/35 (82·9%)
Selected patients 5/35 (14·3%)
No patients 1/35 (2·8%)
Ability to provide prone ventilation 21/35 (60%)
Ability to provide renal replacement therapy 21/35 (60%)
Ability to provide extra-corporeal oxygenation 6/35 (17·1%)
IQR=interquartile range
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
29. Table 2· Description of African Covid-19 critical care cohort
All patients (n=1243) Patients who died (n=631) Patients who survived (n=522) P value
Age (years) 54·72 (15·16) 56·07 (14·19) 53·19 (16·01) 0·001
Male 635/1150 (55·2) 349/629 (55·5%) 286/521 (54·9%) 0·858
Female 515/1150 (44·8%) 280/629 (44·5%) 235/521 (45·1%) -
Body mass index categories
<25 144/789 (18·3%) 60/373 (16·1%) 69/328 (21·0%) 0·100
25-29·9 257/789 (32·6%) 114/373 (30·6%) 115/328 (35·1%)
30-34·9 192/789 (24·3%) 93/373 (24·9%) 77/328 (45·3%)
35-39·9 99/789 (12·5%) 54/373 (14·5%) 34/328 (10·4%)
>40 97/789 (12·3%) 52/373 (13·9%) 33/328 (10·1%)
Missing 454
Comorbidities
Coronary artery disease 96/1199 (8·0%) 48/623 (7·7%) 47/519 (9·1%) 0·452
Congestive heart failure 58/1200 (4·8%) 31/625 (5·0%) 25/518 (4·8%) 1·000
Hypertension 635/1206 (52·7%) 334/627 (53·3%) 266/519 (51·3%) 0·514
Stroke or transient ischaemic attack 50/1195 (4·2%) 25/621 (4·0%) 23/517 (4·4%) 0·768
Diabetes 474/1197 (39·6%) 272/624 (43·6%) 178/515 (34·6%) 0·002
Cancer 27/1194 (2·3%) 13/622 (2·1%) 13/519 (2·5%) 0·693
Current smoker 76/1188 (6·4%) 30/615 (4·9%) 41/517 (7·9%) 0·037
Chronic lung disease 92/1195 (7·7%) 43/620 (6·9%) 42/518 (8·1%) 0·497
Active tuberculosis 14/1193 (1·2%) 8/620 (1·3%) 3/517 (0·6%) 0·362
Chronic liver disease 26/1195 (2·2%) 15/622 (2·4%) 11/518 (2·1%) 0·843
HIV/AIDS 120/1189 (10·1%) 82/615 (13·3%) 29/519 (5·6%) <0·001
Antiretroviral therapy 93/114 (81·6%) 66/81 (81·5%) 22/25 (88·0%) 0·448
Chronic malaria/ malaria within 3 months 56/1195 (4·7%) 15/621 (2·4%) 41/518 (7·9%) <0·001
Chronic kidney disease 78/1192 (6·5%) 49/622 (7·9%) 26/516 (5·0%) 0·056
Condition at admission
Cardiorespiratory arrest in 24 hours prior to critical care referral 30/1138 (2·6%) 25/620 (4·0%) 5/518 (1·0%) 0·001
Organ support required at admission
Respiratory support 139/1233 (11·3%) 588/629 (93·5%) 425/521 (81·6%) <0·001
Cardiovascular support 257/1243 (20·7%) 175/631 (27·7%) 74/522 (14·2%) <0·001
Renal support 87/1243 (7·0%) 62/631 (9·8%) 20/522 (3·8%) <0·001
Other support 139/1243 (11·2%) 53/631 (8·4%) 80/522 (15·3%) <0·001
Number of organ systems requiring support
No organ system 100/1233 (8·1%) 36/629 (5·7%) 63/521 (12·1%) <0·001
One organ system 772/1233 (62·6%) 368/629 (58·5%) 336/521 (64·5%)
Two organ systems 291/1233 (23·6%) 172/629 (27·3%) 106/521 (20·3%)
Three organ systems 59/1233 (4·8%) 46/629 (7·3%) 13/521 (2·5%)
Four organ systems 11/1233 (0·9%) 7/629 (1·1%) 3/521 (0·6%)
Quick SOFA score on presentation
SBP≤100mmHg 192/1211 (15·9%) 135/621 (21·7%) 47/510 (9·2%) <0·001
Respiratory rate≥22 breathes/min 967/1205 (80·2%) 495/615 (80·5%) 404/511 (79·1%) 0·552
Glasgow Coma Score≤14 383/1187 (32·3%) 260/601 (43·3%) 98/509 (19·3%) <0·001
Quick SOFA score
0 risk factor 135/1105 (12·2%) 53/597 (8·9%) 82/508 (16·1%) <0·001
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
30. 1 risk factor 619/1105 (56·0%) 296/597 (49·6%) 323/508 (63·4%)
2 risk factors 251/1105 (22·7%) 165/597 (27·6%) 86/508 (16·9%)
3 risk factors 100/1105 (9·0%) 83/597 (13·9%) 17/508 (3·3%)
Full SOFA score on admission 4·9 (3·00) 6·3 (3·25) 3·5 (2·12) <0·001
Full SOFA missing data 798
ICU resources
Was admission delayed due to lack of resources (e.g. bed, staffing etc) 138/1080 (12·8%) 70/537 (13·0%) 62/469 (13·2%) 1·000
Nurse to patient ratio 0·5 (0·33-1·00) 0·5 (0·33-1·00) 0·5 (0·25-1·00) 0·006
Ability to provide invasive ventilation for patient if required 1120/1215 (92·2%) 592/625 (94·7%) 478/518 (92·3%) 0·114
Physician available on site 24/7 for patient 1144/1208 (94·7%) 598/619 (96·6%) 491/516 (95·2%) 0·229
Respiratory support (highest level of support)
None 135/1064 (12·7%) 91/576 (15·8%) 44/488 (9·0%) <0·001
Oxygen mask 354/1064 (33·3%) 160/576 (27·8%) 194/488 (39·8%)
High flow nasal oxygenation 361/1064 (33·9%) 202/576 (35·1%) 159/488 (32·6%)
CPAP 214/1064 (20·1%) 123/576 (21·4%) 91/488 (18·6%)
Proned
None 752/1243 (60·5%) 374/631 (59·3%) 312/522 (59·8%) 0·001
Not ventilated 372/1243 (29·9%) 174/631 (27·6%) 174/522 (33·3%)
Invasive mechanical ventilation 119/1243 (9·6%) 83/631 (13·2%) 36/522 (6·9%)
Ventilatory support
None 355/1065 (33·3%) 80/575 (13·9%) 275/490 (56·1%) <0·001
Non-invasive ventilation 116/1065 (10·9%) 53/575 (9·2%) 63/490 (12·9%)
Invasive mechanical ventilation 594/1065 (55·8%) 442/575 (74·4%) 152/490 (31·0%)
Intubation
No 517/1114 (46·4%) 155/608 (25·5%) 362/506 (71·5%) <0·001
Yes, elective 171/1114 (15·4%) 121/608 (19·9%) 50/506 (9·9%)
Yes, emergency 426/1114 (38·2%) 332/608 (54·6%) 94/506 (18·6%)
Inotropes/ vasoconstrictors 447/1189 (37·6%) 333/621 (53·6%) 107/519 (20·6%) <0·001
Dialysis 102/1175 (8·7%) 75/615 (12·2%) 23/512 (4·5%) <0·001
Therapeutic anticoagulation 1006/1187 (84·8) 538/624 (86·2%) 419/510 (82·2%) 0·070
Steroid therapy 957/1112 (86·1%) 495/575 (86·1%) 413/484 (85·3%) 0·725
Repurposed/ experimental Covid-19 drug therapy 167/1060 (15·8%) 57/530 (10·8%) 110/489 (22·5%) <0·001
Extracorporeal membrane oxygenation (ECMO) (if available) 5/416 (1·2%) 3/216 (1·4%) 2/187 (1·1%) 0·773
Data are mean (SD) or n (proportion)· Odds ratios were constructed for in-hospital mortality with univariate binary logistic regression analysis· HIV=human immunodeficiency virus· AIDS=acquired
immunodeficiency syndrome· SOFA= sequential organ failure assessment score
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
31. Table 3. Generalised linear mixed model (pooled results of the imputed datasets) for patients referred to critical care with suspected or known Covid-19 infection
with full dataset for mortality. Control are patients alive in hospital and alive and discharged at 30 days.
Odds ratio
(n=1153)
2·5% CI 97·5% CI p-value
Intercept 0·01 0·00 0·12 <0·001
Age (years) 1·04 1·02 1·05 <0·001
Male 0·87 0·59 1·29 0·50
Body mass index
<25 Reference
25-29·9 1·23 0·66 2·30 0·51
30-34·9 1·07 0·52 2·23 0·84
35-39·9 1·07 0·50 2·26 0·87
>40 0·84 0·39 1·82 0·66
Coronary artery disease 0·70 0·33 1·48 0·35
Congestive heart failure 1·68 0·58 4·80 0·33
Hypertension 1·07 0·71 1·61 0·75
Stroke or transient ischaemic attack 2·61 0·83 8·21 0·10
Diabetes 1·42 0·94 2·14 0·09
Cancer 3·09 0·51 18·71 0·21
Current smoker 0·87 0·37 2·04 0·74
Chronic lung disease 0·85 0·42 1·75 0·67
Active tuberculosis 1·74 0·19 15·58 0·60
Chronic liver disease 3·02 0·37 24·92 0·29
HIV/AIDS 1·84 0·99 3·40 0·05
Chronic malaria/ malaria within 3 months 0·84 0·15 4·62 0·84
Chronic kidney disease 1·35 0·61 3·01 0·46
Cardiorespiratory arrest in 24 hours prior to critical care referral 4·43 1·01 19·54 0·05
Quick SOFA score
0 risk factor Reference
1 risk factor 0·80 0·44 1·45 0·45
2 risk factors 1·40 0·71 2·76 0·33
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
32. 3 risk factors 3·61 1·41 9·24 0·01
Was admission delayed due to lack of resources (e.g. bed· staffing etc) 1·12 0·51 2·45 0·78
Physician available on site 24/7 for patient 0·81 0·30 2·24 0·69
Nurse to patient ratio 1·18 0·77 1·82 0·44
Respiratory support
None Reference
Oxygen mask 1·62 0·46 5·73 0·45
High flow nasal oxygenation 3·64 0·96 13·76 0·06
CPAP 5·86 1·47 23·35 0·01
Invasive mechanical ventilation 16·42 4·52 59·65 <0·001
Organ systems requiring support at admission
No organ system Reference
One organ system 1·66 0·56 4·94 0·36
Two organ systems 1·60 0·48 5·28 0·44
Three organ systems 5·52 1·13 27·01 0·04
Four organ systems 0·72 0·07 7·76 0·78
Prone ventilation
No proning Reference
On spontaneous ventilation 0·74 0·45 1·20 0·22
On invasive mechanical ventilation 3·32 1·47 7·49 <0·001
Inotropes/ vasoconstrictors 2·73 1·71 4·36 <0·001
Therapeutic anticoagulation 1·08 0·54 2·16 0·82
Steroid therapy 0·47 0·21 1·04 0·06
Repurposed/ experimental Covid-19 drug therapy 1·07 0·38 2·96 0·90
CI=confidence interval CPAP=continuous positive airway pressure HIV=human immunodeficiency virus AIDS=acquired immunodeficiency syndrome SOFA=sequential organ failure
assessment score
This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3707415
View publication stats
View publication stats