International Severe Acute Respiratory and Emerging InfectionsConsortium (ISARIC)
A global federation of clinical research networks, providing a proficient, coordinated, and agile research responseto outbreak-prone infectious diseases
COVID-19 Report: 27 April 2020
SummaryThe results in this report have been produced using data from the ISARIC COVID-19 database. Forinformation, or to contribute to the collaboration, please contact [email protected].
Up to the date of this report, data have been entered for 27424 individuals from 278 sites across 30 countries.
We thank all of the data contributors for collecting standardised data during these extraordinary times. Weplan to issue this report of aggregate data weekly for the duration of the SARS-CoV-2/COVID-19 pandemic.
Please note the following caveats. Information is incomplete for the many patients who are still being treated.Furthermore, it is likely that that we received more cases of severely ill individuals than those with relativelyless severe illness; outcomes from these data, such as the proportion dying, must therefore not be used to inferoutcomes for the entire population of people who might become infected. Some patients may be participantsin clinical trials of experimental interventions. Many of the included cases are from the United Kingdom.Additional caveats are provided in the in the ‘Caveats’ section below.
The analysis detailed in this report only includes individuals for whom data collection commenced on orbefore 13 April 2020. We have applied a 14-day rule to focus analysis on individuals who are more likelyto have a recorded outcome. By excluding patients enrolled during the last 14 days, we aim to reduce thenumber of incomplete data records and thus improve the generalisability of the results and the accuracy ofthe outcomes. However, this limits our analysis to this restricted cohort despite the much larger volumes ofdata held in the database.
The cohort comprises 19463 individuals, including 11688 males and 7684 females – sex is unreported for 91cases. SARS-COV-2 infection has been confirmed by laboratory testing in 15844 of these individuals.3619 individuals are recorded as suspected of SARS-COV-2 infection, without laboratory confirmation at thetime of data analysis.
The median age (calculated based on reported age) is 71 years. The minimum and maximum observed ageswere 0 and 104 years respectively.
Outcomes have been recorded for 11873 patients, consisting of 7595 recoveries and 4278 deaths. Follow-up isongoing for 6464 patients. Outcome records are unavailable for 1126 patients.
The observed mean number of days from (first) symptom onset to hospital admission was 11.7, with a SD of7.4 days and a median of 5 days.
1
The observed mean duration for the number of days from hospital admission to outcome (death or discharge)was 8.7, with a standard deviation (SD) of 8.1 days and a median of 7 days. These estimates are based on allcases which have complete records on length of hospital stay (N = 12829).
The symptoms on admission represent the policy for hospital admission and containment at that time plus,whatever the case definition was. As time passes for most countries these will change. The four most commonsymptoms at admission were fatigue and malaise alongside cough, history of fever and shortness of breath.
2670 patients received non-invasive mechanical ventilation (NIV). The mean and median durations fromadmission to receiving NIV were 4.6 days and 2 days respectively (SD: 12 days) – estimated from records oncases with complete records on dates of hospital admission and NIV onset (N = 2307).
The mean and median durations for NIV were 2 days and 0.5 days respectively (SD: 3.5 days) – estimatedbased on only those cases which have complete NIV duration records (N = 1151).
3752 patients were admitted at some point of their illness into an intensive care unit (ICU) or high dependencyunit (HDU). The observed mean and median durations (in days) from hospital admission to ICU/HDUadmission were 3.3 and 1 respectively (SD: 7) – estimated from records on cases with complete date recordson hospital admission and ICU/HDU entry (N = 3709).
The duration of stay in ICU/HDU had a mean of 7.4 days and a median of 5.5 (SD: 7.5 days) – estimated ononly those cases with complete records for ICU/HDU duration or ICU/HDU start/end dates (N = 1968).Of these 3752 patients who were admitted into ICU/HDU, 989 died, 1514 are still in hospital and 831 haverecovered and been discharged. Outcome records are unavailable for 418 cases.
2249 patients received invasive mechanical ventilation (IMV). The mean and median durations from admissionto receiving IMV were 3.5 days and 2 days respectively (SD: 7.2 days) – estimated from records on cases withcomplete records on dates of hospital admission and IMV onset (N = 2101).
The mean, median and SD for the duration of IMV – estimated based on all 1028 cases with complete recordson IMV stays – were 9.6 days, 9 days and 6.5 days respectively.
Of 11407 patients with a recorded outcome and details of treatments received, 72.4% received an antibioticand 8.8% received antivirals. These treatment categories are not mutually exclusive since some patientsreceived multiple treatments. 68.6% of patients received some degree of oxygen supplementation: of these,22.2% received NIV and 12.2% IMV.
Of 1804 patients admitted into ICU/HDU with a recorded outcome and details of treatments, 79.9% receivedantibiotics and 20.5% antivirals; and 93.0% received some degree of oxygen supplementation, of which 43.7%was NIV and 54.1% IMV.
2
Patient CharacteristicsFigure 1: Age and sex distribution of patients. Bar fills are outcome (death/discharge/ongoing care) at thetime of report.
Males Females
0−4
5−9
10−14
15−19
20−24
25−29
30−34
35−39
40−44
45−49
50−54
55−59
60−64
65−69
70−74
75−79
80−84
85−89
90+
1500 1000 500 0 500 1000 1500Count
Age
gro
up
Outcome
Discharge
Ongoing care
Death
3
Figure 2: Top: Frequency of symptoms seen at admission amongst COVID-19 patients. Bars are annotatedwith a fraction representing the number of patients presenting with this symptom over the number of patientsfor whom presence or absence of this symptom was recorded. Middle: The distribution of combinations ofthe four most common symptoms, amongst all patients for whom these data were recorded. Filled and emptycircles below the x-axis indicate the presence or absence of each comorbidity. The “Any other” categorycontains all remaining symptoms in the top plot. Bottom: Heatmap for correlation between symptoms. Fillcolour is the phi correlation coefficient for each pair of symptoms, calculated amongst patients with recordedpresence or absence of both.*
74/17880
90/17791
102/17876
152/17868
217/17878
224/17882
282/17885
440/13781
672/17899
1008/17870
1437/17897
1444/17901
1484/17898
1869/17885
2177/17906
2877/17910
3001/17894
3053/17915
3608/17918
2924/13781
6672/17898
6135/13781
11665/18059
12397/18039
Conjunctivitis
Ear pain
Lymphadenopathy
Bleeding
Seizures
Skin rash
Skin ulcers
Cough (bloody sputum / haemoptysis)
Runny nose
Joint pain
Wheezing
Abdominal pain
Sore throat
Headache
Chest pain
Vomiting / Nausea
Muscle aches
Diarrhoea
Altered consciousness / confusion
Cough (with sputum)
Fatigue / Malaise
Cough (no sputum)
Shortness of breath
History of fever
0.00 0.25 0.50 0.75 1.00Proportion
Sym
ptom
Symptompresent
No
Yes
4
0.000
0.025
0.050
0.075
0.100
0.125
Cough (no sputum)Fatigue / Malaise
Shortness of breathHistory of fever
Any other
Symptoms present at admission
Pro
port
ion
of p
atie
nts
Runny noseSore throat
Ear painDiarrhoea
Vomiting / NauseaAbdominal pain
Joint painMuscle aches
Fatigue / MalaiseHeadache
Shortness of breathHistory of fever
WheezingCough (no sputum)
Cough (with sputum)Cough (bloody sputum / haemoptysis)
Chest painLymphadenopathy
ConjunctivitisBleeding
Skin ulcersSkin rashSeizures
Altered consciousness / confusion
Run
ny n
ose
Sor
e th
roat
Ear
pai
nD
iarr
hoea
Vom
iting
/ N
ause
aA
bdom
inal
pai
nJo
int p
ain
Mus
cle
ache
sFa
tigue
/ M
alai
seH
eada
che
Sho
rtne
ss o
f bre
ath
His
tory
of f
ever
Whe
ezin
gC
ough
(no
spu
tum
)C
ough
(w
ith s
putu
m)
Cou
gh (
bloo
dy s
putu
m /
haem
opty
sis)
Che
st p
ain
Lym
phad
enop
athy
Con
junc
tiviti
sB
leed
ing
Ski
n ul
cers
Ski
n ra
shS
eizu
res
Alte
red
cons
ciou
snes
s / c
onfu
sion
−1.0
−0.5
0.0
0.5
1.0phi coefficient
* We are working to gain a greater understanding of patients reported as having no presenting symptoms.
5
Figure 3: Top: Frequency of comorbidities or other concomitant conditions seen at admission amongstCOVID-19 patients. Bars are annotated with a fraction representing the number of patients presenting withthis comorbidity over the number of patients for whom presence or absence of this comorbidity was recorded.Bottom: The distribution of combinations of the four most common such conditions, amongst all patients forwhom these data were recorded. Filled and empty circles below the x-axis indicate the presence or absenceof each comorbidity. The “Any other” category contains all remaining conditions in the top plot, and anyothers recorded as free text by clinical staff.
79/19463
86/18107
395/17941
542/17807
660/17677
911/14482
1511/17659
1644/18115
1770/18099
1905/17875
1958/18009
2378/18134
2553/18118
2933/18146
3518/18157
5025/18161
Pregnancy
AIDS/HIV
Malnutrition
Liver disease
Chronic hematologic disease
Smoking
Rheumatologic disorder
Malignant neoplasm
Chronic neurological disorder
Obesity
Dementia
Asthma
Chronic kidney disease
Chronic pulmonary disease
Diabetes
Chronic cardiac disease
0.00 0.25 0.50 0.75 1.00Proportion
Con
ditio
n Conditionpresent
No
Yes
6
0.0
0.1
0.2
0.3
Chronic kidney diseaseChronic pulmonary disease
DiabetesChronic cardiac disease
Any other
Conditions present at admission
Pro
port
ion
of p
atie
nts
7
Variables by ageFigure 4: Comorbidities stratified by age group. Boxes show the proportion of individuals with eachcomorbidity, with error bars showing 95% confidence intervals. The size of each box is proportional to thenumber of individuals represented. N is the number of individuals included in the plot (this varies betweenplots due to data completeness).
0.0
0.1
0.2
0.3
0.4
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
Pro
port
ion
with
asth
ma
N = 17122
0.0
0.1
0.2
0.3
0.4
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
Pro
port
ion
with
mal
igna
ncy
N = 16964
0.0
0.1
0.2
0.3
0.4
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
Pro
port
ion
with
HIV
N = 16862
0.0
0.1
0.2
0.3
0.4
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
Pro
port
ion
with
obes
ity
N = 15855
0.0
0.1
0.2
0.3
0.4
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
Pro
port
ion
with
diab
etes
mel
litus
N = 17213
0.0
0.1
0.2
0.3
0.4
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
Pro
port
ion
with
dem
entia
N = 17024
0.0
0.1
0.2
0.3
0.4
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
Pro
port
ion
who
smok
e
N = 10141
8
Figure 5: Symptoms recorded at hospital presentation stratified by age group. Boxes show the proportionof individuals with each comorbidity, with error bars showing 95% confidence intervals. The size of each boxis proportional to the number of individuals represented. N is the number of individuals included in the plot(this varies between plots due to data completeness).
0.00
0.25
0.50
0.75
1.00
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
Fev
er
N = 16847
0.00
0.25
0.50
0.75
1.00
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
Cou
gh
N = 13604
0.00
0.25
0.50
0.75
1.00
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
Sho
rtne
ss o
f bre
ath
N = 17821
0.00
0.25
0.50
0.75
1.00
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
Con
fusi
onN = 15149
0.00
0.25
0.50
0.75
1.00
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
Gas
troi
ntes
tinal
sym
ptom
s
N = 15483
9
Figure 6: Box and whisker plots for observations at hospital presentation stratified by age group. Outliersare omitted. N is the number of individuals included in the plot (this varies between plots due to datacompleteness).
20
40
60
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
Res
pira
tory
rat
e (m
in.−1
)
N = 15494
85
90
95
100
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
O2
satu
ratio
n in
roo
m a
ir (%
) N = 10193
50
100
150
200
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
Hea
rt r
ate
(min
.−1)
N = 16346
100
150
200
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)Sys
tolic
blo
od p
ress
ure
(mm
Hg) N = 16292
34
36
38
40
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
Tem
pera
ture
(°C
)
N = 16610
10
Figure 7: Box and whisker plots for laboratory results within 24 hours of hospital presentation stratified byage group. Outliers are omitted. N is the number of individuals included in the plot (this varies betweenplots due to data completeness). ALT, Alanine transaminase; APTT, Activated partial thromboplastin time;CRP, C-reactive protein; WCC, white cell count
0
10
20
30
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
WC
C (1
09 /L) N = 5861
0.0
2.5
5.0
7.5
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)Lym
phoc
ytes
(109 /L
)
N = 5562
0
5
10
15
20
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)Neu
trop
hils
(109 /L
)
N = 5604
0
5
10
15
20
25
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)U
rea
(mm
ol/L
) N = 4747
0
100
200
300
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
CR
P (
mg/
L)
N = 5514
8
12
16
20
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)Pro
thro
mbi
n tim
e (s
)
N = 2163
10
20
30
40
50
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
AP
TT
(s)
N = 2030
0
10
20
30
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)Bili
rubi
n (µ
mol
/L) N = 4623
0
30
60
90
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
ALT
(un
its/L
) N = 4383
11
Hospital stays and outcomesFigure 8: Distribution of length of hospital stay, according to sex. This only includes cases with reportedoutcomes. The coloured areas indicate the kernel probability density of the observed data and the box plotsshow the median and interquartile range of the variable of interest.
0
20
40
60
80
Male Female
Sex
Leng
th o
f hos
pita
l sta
y
Sex
Male
Female
Figure 9: Distribution of length of hospital stay, according to patient age group. This only includes caseswith reported outcomes. The coloured areas indicate the kernel probability density of the observed data andthe box plots show the median and interquartile range of the variable of interest.*
0
20
40
60
80
0−9 10−19 20−29 30−39 40−49 50−59 60−69 70+
Age group
Leng
th o
f hos
pita
l sta
y
Age
0−9
10−19
20−29
30−39
40−49
50−59
60−69
70+
* We are working to gain a greater understanding of the patient pathway for individuals recorded as havingextremely long hospital stays.
12
Figure 10: The distribution of patient status by number of days after admission. Patients with “unknown”status have left the site at the time of report but have unknown outcomes due to missing data. Patients stillon site at the time of report appear in the “ongoing care” category for days which are in the future at thattime. (For example, a patient admitted 7 days before the date of report and still on site by the date of thereport would be categorised as “ongoing care” for days 8 and later.) The black line marks the end of 14 days;due to the cut-off, only a small number of patients appear in the “ongoing care” category left of this line.
0.00
0.25
0.50
0.75
1.00
0 10 20 30Days relative to admission
Pro
port
ion
Status
Discharged
Transferred
Unknown
Ongoing care
Ward
ICU
Death
13
Figure 11: Patient numbers and outcomes by epidemiological week (of 2020) of admission (or, for patientsinfected in hospital, of symptom onset). The rightmost bar, marked with an asterisk, represents an incompleteweek (due to the 14-day cutoff).
*0
2000
4000
01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16Epidemiological week of admission/symptom onset (2020)
Cas
es
Outcome
Discharge
Ongoing care
Death
14
TreatmentFigure 12: Top: Treatments used. This only includes patients for whom this information was recorded.Bottom: The distribution of combinations of antimicrobial treatments and steroids administered duringhospital stay, across all patients with completed hospital stay and recorded treatment data. Filled and emptycircles below the x-axis indicate treatments that were and were not administered.
47/11595
53/11555
114/11560
359/11504
390/11588
600/11575
611/11365
962/11593
1238/11677
1304/11609
1348/11598
1746/11524
7719/11704
9337/11781
Extracorporeal support
Inhaled nitric oxide
Tracheostomy
Antifungal agent
Renal replacement therapy
Prone ventilation
Other
Inotropes / vasopressors
Invasive ventilation
Antiviral agent
Non−invasive ventilation
Corticosteroid agent
Oxygen therapy
Antibiotic agent
0.00 0.25 0.50 0.75 1.00Proportion
Trea
tmen
t Treatment
No
Yes
15
0.0
0.1
0.2
0.3
AntifungalAntiviral
CorticosteroidAny oxygen provision
Antibiotic
Treatments used during hospital admission
Pro
port
ion
of p
atie
nts
16
Intensive Care and High Dependency Unit TreatmentsFigure 13: Top: Treatments used amongst patients admitted to the ICU. This only includes patients forwhom this information was recorded. Middle: The distribution of combinations of treatments administeredduring ICU/HDU stay. Filled and empty circles below the x-axis indicate treatments that were and werenot administered respectively.* Bottom: Distribution of lengths of stay for patients who were admitted toICU/HDU: total length of stay for this group and length of stay within intensive care. This only includes caseswith reported completed stays. The coloured areas indicate the kernel probability density of the observeddata and the box plots show the median and interquartile range of the variable of interest.
* We are working to gain a greater understanding of patients reported as having been admitted to ICU/HDUbut having no intensive treatments recorded.
46/2108
49/2099
106/2101
161/2112
227/2021
307/2108
503/2132
568/2117
608/2135
768/2138
954/2119
1219/2186
1933/2209
1953/2223
Extracorporeal support
Inhaled nitric oxide
Tracheostomy
Antifungal agent
Other
Renal replacement therapy
Corticosteroid agent
Prone ventilation
Antiviral agent
Non−invasive ventilation
Inotropes / vasopressors
Invasive ventilation
Antibiotic agent
Oxygen therapy
0.00 0.25 0.50 0.75 1.00Proportion
Trea
tmen
t Treatment
No
Yes
17
0.0
0.1
0.2
Renal replacement therapyCorticosteroid
InotropesInvasive ventilationAny antimicrobials
Any oxygen provision
Treatments used
Pro
port
ion
of p
atie
nts
adm
itted
to in
tens
ive
care
18
0
25
50
75
Total hospital stay ICU
Location
Leng
th o
f sta
y (d
ays)
19
Statistical AnalysisFigure 14: Distribution of time from symptom onset to admission. The blue curve is the Gamma distributionfit to the data. The black dashed line indicates the position of the expected mean. The expected mean estimatehere differs from the observed mean indicated in the summary text due to the differences in estimation: themean shown in the figure below is the mean of the fitted Gamma distribution whereas the observed mean (inthe summary text) is the arithmetic mean.
0.00
0.04
0.08
0.12
0 10 20 30
Time (in days) from symptom onset to admission
Den
sity
20
Figure 15: Distribution of time from admission to an outcome - either death or recovery (discharge). Theblue curve is the Gamma distribution fit to the data. The black dashed line indicates the position of theexpected mean. The expected mean differs from the observed mean in that it accounts for unobservedoutcomes.
0.00
0.02
0.04
0.06
0 25 50 75 100
Time (in days) from admission to death or recovery
Den
sity
21
Figure 16: Probabilities of death (red curve) and recovery (green curve) over time. The black line indicatesthe case fatality ratio (black). The method used here considers all cases, irrespective of whether an outcomehas been observed. For a completed epidemic, the curves for death and recovery meet. Estimates were derivedusing a nonparametric Kaplan-Meier–based method proposed by Ghani et al. (2005).
0.00
0.25
0.50
0.75
1.00
0 10 20 30 40 50Days after admission
Cum
ulat
ive
prob
abili
ty
Legend
Deaths
DischargesCase fatalityratio
22
Country ComparisonsFigure 17: Number of sites per country.
2 1 3 2
15
1 2 2 4 4 111
29
2 2
164 3 1 1 4 1 1 2
71
169
50
50
100
150
Austra
lia
Austri
a
Belgium
Brazil
Canad
a
Colom
bia
Ecuad
or
Franc
e
Germ
any
Hong
KongIn
dia
Irelan
dIsr
aelIta
ly
Japa
n
Kuwait
Nethe
rland
s
New Z
ealan
d
Norway
Peru
Poland
Portu
gal
Roman
ia
Saudi
Arabia
South
Kor
eaSpa
in
Turk
ey UKUSA
Country
Site
s
23
Figure 18: Distribution of patients by country and outcome.
3 4297
30 245 113 42 14 231 26 140 2 4454
7 154 3 68 34 187 1 3 89
15936
249
0
5000
10000
15000
Austra
lia
Austri
a
Belgium
Brazil
Canad
a
Franc
e
Germ
any
Hong
Kong
Irelan
dIsr
aelIta
ly
Japa
n
Kuwait
Nethe
rland
s
New Z
ealan
d
Norway
Peru
Poland
Portu
gal
Roman
ia
Saudi
Arabia
South
Kor
eaSpa
in UKUSA
Country
Cas
es
Outcome
Discharge
Ongoing care
Death
24
BackgroundIn response to the emergence of novel coronavirus (COVID-19), ISARIC launched a portfolio of resources toaccelerate outbreak research and response. These include data collection, analysis and presentation toolswhich are freely available to all sites which have requested access to these resources. All data collectiontools are designed to address the most critical public health questions, have undergone extensive review byinternational clinical experts, and are free for all to use. Resources are available on the ISARIC website.
The ISARIC-WHO COVID-19 Case Record Form (CRF) enables the collection of standardised clinical datato inform patient management and public health response. These forms should be used to collect data onsuspected or confirmed cases of COVID-19. The CRF is available in multiple languages and is now in useacross dozens of countries and research consortia, who are contributing data to these reports.
To support researchers to retain control of the data and samples they collect, ISARIC also hosts a dataplatform, where data can be entered to a web-based REDCap data management system, securely stored, andused to produce regular reports on their sites as above. Data contributors are invited to input on the methodsand contents of the reports, and can also contribute to the aggregated data platform which aggregatessite-specific data from all other sites across the world who are using this system. For more information, visitthe ISARIC website.
All decisions regarding data use are made by the institutions that enter the data. ISARIC keeps contributorsinformed of any plans and welcomes their input to promote the best science and the interests of patients,institutions and public health authorities. Feedback and suggestions are welcome at [email protected].
MethodsPatient details were submitted electronically by participating sites to the ISARIC database. Relevantbackground and presenting symptoms were recorded on the day of study recruitment. Daily follow-up wasthen completed until recovery or death. A final form was completed with details of treatments receivedand outcomes. All categories that represent fewer than five individuals have been suppressed to avoid thepotential for identification of participants.
Graphs have been used to represent the age distribution of patients by sex and status (dead, recovered & stillin hospital), the prevalence of individual symptoms on admission, comorbidities on admission, the lengthof hospital stay by sex and age group and the distribution of patient statuses by time since admission. Inaddition, the number of cases recruited by country and site, as well as the case count by status, has beenrepresented.
Using a non-parametric Kaplan-Meier-based method (Ghani et al., 2005), the case- fatality ratio (CFR)was estimated, as well as probabilities for death and recovery. This method estimates the CFR with theformula a/(a + b), where a and b are the values of the cumulative incidence function for deaths and recoveriesrespectively, estimated at the last observed time point. In a competing risk context (i.e. where there aremultiple endpoints), the cumulative incidence function for an endpoint is equal to the product of the hazardfunction for that endpoint and the survival function assuming a composite endpoint. It is worth noting thatthis method assumes that future deaths and recoveries will occur with the same relative probabilities as havebeen observed so far. Binomial confidence intervals for the CFR were obtained by a normal approximation(See Ghani et al., (2005)).
To obtain estimates for the distributions of time from symptom onset to hospital admission and the time fromadmission to outcome (death or recovery), Gamma distributions were fitted to the observed data, accountingfor unobserved outcomes. Parameters were estimated by a maximum likelihood procedure and confidenceintervals for the means and variances were obtained by bootstrap.
All analysis were performed using the R statistical software (R Core Team, 2019).
25
CaveatsPatient data are collected and uploaded from start of admission, however a complete patient data set is notavailable until the episode of care is complete. This causes a predictable lag in available data influenced bythe duration of admission which is greatest for the sickest patients, and accentuated during the up-phase ofthe outbreak.
These reports provide regular outputs from the ISARIC COVID-19 database. We urge caution in interpretingunexpected results. We have noted some unexpected results in the report, and are working with sites thatsubmitted data to gain a greater understanding of these.
Summary TablesProportions are presented in parantheses. Proportions have been rounded to two decimal places.
Table 1: Patient Characteristics
Description ValueSize of cohort 19463
By sexMale 11688 (0.6)Female 7684 (0.39)Unknown 91 (0.01)
By outcome statusDead 4278 (0.22)Recovered (discharged alive) 7595 (0.39)Still in hospital 6464 (0.33)Tranferred to another facility 694 (0.04)Unknown 432 (0.02)
By COVID-19 statusPositive (laboratory-confirmed) 15844 (0.81)Suspected 3619 (0.19)
By age group0-9 275 (0.01)10-19 173 (0.01)20-29 364 (0.02)30-39 789 (0.04)40-49 1485 (0.08)50-59 2742 (0.14)60-69 3293 (0.17)70+ 10083 (0.52)Unknown 259 (0.01)
26
Table 2: Outcome by age and sex
Variable Still in hospital Death Discharge Transferred UnknownAge0-9 37 (0.01) 0 (0) 189 (0.02) 17 (0.02) 32 (0.07)10-19 23 (0) 2 (0) 118 (0.02) 7 (0.01) 23 (0.05)20-29 69 (0.01) 10 (0) 271 (0.04) 7 (0.01) 7 (0.02)30-39 218 (0.03) 27 (0.01) 507 (0.07) 19 (0.03) 18 (0.04)40-49 454 (0.07) 66 (0.02) 901 (0.12) 35 (0.05) 29 (0.07)50-59 985 (0.15) 239 (0.06) 1357 (0.18) 85 (0.12) 76 (0.18)60-69 1187 (0.18) 572 (0.13) 1336 (0.18) 113 (0.16) 85 (0.2)70+ 3398 (0.53) 3316 (0.78) 2807 (0.37) 407 (0.59) 155 (0.36)
SexMale 3954 (0.61) 2762 (0.65) 4310 (0.57) 401 (0.58) 261 (0.6)Female 2467 (0.38) 1498 (0.35) 3256 (0.43) 293 (0.42) 170 (0.39)Unknown 10 (0) 8 (0) 17 (0) 0 (0) 0 (0)
Table 3: Prevalence of Symptoms
Symptoms Present Absent UnknownHistory of fever 12397 (0.64) 4676 (0.24) 2390 (0.12)Shortness of breath 11665 (0.6) 6386 (0.33) 1412 (0.07)Cough 9499 (0.49) 4282 (0.22) 5682 (0.29)Fatigue / Malaise 6672 (0.34) 7386 (0.38) 5405 (0.28)Altered consciousness / confusion 3608 (0.19) 11746 (0.6) 4109 (0.21)Diarrhoea 3053 (0.16) 11948 (0.61) 4462 (0.23)Muscle aches 3001 (0.15) 10171 (0.52) 6291 (0.32)Vomiting / Nausea 2877 (0.15) 12117 (0.62) 4469 (0.23)Chest pain 2177 (0.11) 12298 (0.63) 4988 (0.26)Headache 1869 (0.1) 11250 (0.58) 6344 (0.33)Sore throat 1484 (0.08) 11350 (0.58) 6629 (0.34)Abdominal pain 1444 (0.07) 12995 (0.67) 5024 (0.26)Wheezing 1437 (0.07) 12137 (0.62) 5889 (0.3)Joint pain 1008 (0.05) 11553 (0.59) 6902 (0.35)Runny nose 672 (0.03) 11883 (0.61) 6908 (0.35)Skin ulcers 282 (0.01) 13455 (0.69) 5726 (0.29)Skin rash 224 (0.01) 13514 (0.69) 5725 (0.29)Seizures 217 (0.01) 14374 (0.74) 4872 (0.25)Bleeding 152 (0.01) 14274 (0.73) 5037 (0.26)Lymphadenopathy 102 (0.01) 13264 (0.68) 6097 (0.31)Ear pain 90 (0) 12298 (0.63) 7075 (0.36)Conjunctivitis 74 (0) 13336 (0.69) 6053 (0.31)
27
Table 4: Prevalence of Comorbidities
Comorbidities Present Absent UnknownOther 7525 (0.39) 9203 (0.47) 2735 (0.14)Chronic cardiac disease 5025 (0.26) 12449 (0.64) 1989 (0.1)Diabetes 3518 (0.18) 13911 (0.71) 2034 (0.1)Chronic pulmonary disease 2933 (0.15) 14480 (0.74) 2050 (0.11)Chronic kidney disease 2553 (0.13) 14761 (0.76) 2149 (0.11)Asthma 2378 (0.12) 14960 (0.77) 2125 (0.11)Dementia 1958 (0.1) 15190 (0.78) 2315 (0.12)Obesity 1905 (0.1) 14059 (0.72) 3499 (0.18)Chronic neurological disorder 1770 (0.09) 15414 (0.79) 2279 (0.12)Malignant neoplasm 1644 (0.08) 15535 (0.8) 2284 (0.12)Rheumatologic disorder 1511 (0.08) 15189 (0.78) 2763 (0.14)Smoking 911 (0.05) 9401 (0.48) 9151 (0.47)Chronic hematologic disease 660 (0.03) 16083 (0.83) 2720 (0.14)Liver disease 542 (0.03) 16326 (0.84) 2595 (0.13)Malnutrition 395 (0.02) 16025 (0.82) 3043 (0.16)AIDS/HIV 86 (0) 16982 (0.87) 2395 (0.12)Pregnancy 79 (0) 18870 (0.97) 514 (0.03)
Table 5: Prevalence of Treatments
The counts presented for treatments include all cases, not only cases with complete details of treatments (asexpressed in the summary).
Treatments Present Absent UnknownOxygen therapy 10747 (0.55) 6559 (0.34) 2157 (0.11)Antibiotic agent 9337 (0.48) 2146 (0.11) 7980 (0.41)Non-invasiveventilation
2670 (0.14) 14484 (0.74) 2309 (0.12)
Invasive ventilation 2249 (0.12) 14964 (0.77) 2250 (0.12)Corticosteroid agent 1746 (0.09) 9236 (0.47) 8481 (0.44)Antiviral agent 1304 (0.07) 9765 (0.5) 8394 (0.43)Inotropes /vasopressors
962 (0.05) 9944 (0.51) 8557 (0.44)
Other 611 (0.03) 9779 (0.5) 9073 (0.47)Prone ventilation 600 (0.03) 10215 (0.52) 8648 (0.44)Renal replacementtherapy
390 (0.02) 10529 (0.54) 8544 (0.44)
Antifungal agent 359 (0.02) 10602 (0.54) 8502 (0.44)Extracorporealmembraneoxygenation (ECMO)
209 (0.01) 16945 (0.87) 2309 (0.12)
Tracheostomy 114 (0.01) 10724 (0.55) 8625 (0.44)Inhaled nitric oxide 53 (0) 10723 (0.55) 8687 (0.45)
28
Table 6: Key time variables.
Unlike the observed mean, the estimation process of the expected mean accounts for all cases, irrespectiveof whether an outcome has been observed. The expected mean is ‘NA’ for those variables for which parameterestimation could not be performed, due to the high proportion of unobserved end dates. The interquartilerange is abbreviated ‘IQR’.
Time (indays)
Mean(observed)
SD(observed)
Median(observed)
IQR(observed )
Expected mean (95%CI)
Length ofhospitalstay
8.7 8.1 7 8 24 (22.8, 26.1)
Symptomonset toadmission
11.7 7.4 5 7 7.4 (7.1, 8.1)
Admissionto ICUentry
3.3 7 1 2.5 3.3 (3.1, 3.5)
Durationof ICU
7.4 7.5 5.5 8.5 NA
Admissionto IMV
3.5 7.2 2 3.5 3.5 (3.3, 3.8)
Durationof IMV
9.6 6.5 9 8 NA
Admissionto NIV
4.6 12 2 4.5 4.6 (4.3, 5)
Durationof NIV
2 3.5 0.5 4.5 NA
AcknowledgementsThis report is made possible through the efforts and expertise of the staff collecting data at our partnerinstitutions across the globe, and the ISARIC Team. For a list of partners and team members, please visithttps://isaric.tghn.org/covid-19-data-management-hosting-contributors/.
References1. A. C. Ghani, C. A. Donnelly, D. R. Cox, J. T. Griffin, C. Fraser, T. H. Lam, L. M. Ho, W. S.
Chan, R. M. Anderson, A. J. Hedley, G. M. Leung (2005). Methods for Estimating the Case FatalityRatio for a Novel, Emerging Infectious Disease, American Journal of Epidemiology, 162(5), 479 - 486.doi:10.1093/aje/kwi230.
2. R Core Team (2019). R: A language and environment for statistical computing. R Foundation forStatistical Computing, Vienna, Austria.
29