Upload
others
View
4
Download
0
Embed Size (px)
Citation preview
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: 19 May 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].
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. This is a dynamic report which captures new variables and informationas our understanding of COVID-19 evolves. Please observe the N of each result to note newly addedvariables with fewer data points. 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.
Up to the date of this report, data have been entered for 46929 individuals from 355 sites across 36 countries.
The analysis detailed in this report only includes individuals:
1. for whom data collection commenced on or before 05 May 2020. (We have applied a 14-day rule tofocus analysis on individuals who are more likely to have a recorded outcome. By excluding patientsenrolled during the last 14 days, we aim to reduce the number of incomplete data records and thusimprove the generalisability of the results and the accuracy of the outcomes. However, this limits ourfocus to a restricted cohort despite the much larger volumes of data held in the database.)
AND
2. who have laboratory-confirmed or clinically-diagnosed SARS-COV-2 infection.
The cohort satisfying the above criteria has 25849 cases (97.67% are laboratory-confirmed forSARS-COV-2 infection).
The flow chart in Figure 1 gives an overview of the cohort and outcomes as of 19 May 2020.
1
Demographics and presenting featuresOf these 25849 cases, 15271 are males and 10493 are females – sex is unreported for 85 cases. The minimumand maximum observed ages were 0 and 104 years respectively. The median age is 72 years.
The observed mean number of days from (first) symptom onset to hospital admission was 13, with a standarddeviation (SD) of 7.9 days and a median of 5 days.
The observed mean duration for the number of days from hospital admission to outcome (death or discharge)was 10.5, with SD 10.2 days and a median of 8 days. These estimates are based on all cases which havecomplete records on length of hospital stay (N = 21270).
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 five most commonsymptoms at admission were history of fever, shortness of breath, cough, fatigue/malaise, and confusion.Frequencies of symptom prevalence vary with age.
OutcomesOutcomes have been recorded for 19983 patients, consisting of 12903 recoveries and 7080 deaths. Follow-up isongoing for 4112 patients. Outcome records are unavailable for 1754 patients.
ICU/HDU: A total of 4752 (18%) patients were admitted at some point of their illness into an intensivecare unit (ICU) or high dependency unit (HDU). Of these, 1567 died, 1106 are still in hospital and 1591 haverecovered and been discharged.
The observed mean and median durations (in days) from hospital admission to ICU/HDU admission were2.8 and 1 respectively (SD: 5.9) – estimated from records on cases with complete date records on hospitaladmission and ICU/HDU entry (N = 4454).
The duration of stay in ICU/HDU had a mean of 9.7 days and a median of 7 (SD: 9.3 days) – estimated ononly those cases with complete records for ICU/HDU duration or ICU/HDU start/end dates (N = 3458).Of these 4752 patients who were admitted into ICU/HDU, 1567 died, 1106 are still in hospital and 1591have recovered and been discharged. Outcome records are unavailable for 488 cases. Approximately 42% ofpatients with complete records on ICU admission dates were admitted to ICU within the first day of hospitaladmission. The distribution of the number of days from admission to ICU admission is shown in Figure 11.
TreatmentAntibiotics were received by 16820/20114 (83.6%) patients, and 1771/19505 (9.1%) received antivirals. Thesetreatment categories are not mutually exclusive since some patients received multiple treatments. (Thedenominators differ due to data completeness.) 16760/24861 (67.4%) patients received some degree of oxygensupplementation: of these, 3937/16760 (23.5%) received NIV and 2946/16760 (17.6%) IMV.
Of the patients admitted into ICU/HDU, 3157/3455 (91.4%) received antibiotics and 2582/5164 (50%)antivirals. 4286/4679 (91.6%) received some degree of oxygen supplementation, of which, 2110/4286 (49.2%)received NIV and 2842/4286 (66.3%) IMV.
A total of 3937 patients received non-invasive mechanical ventilation (NIV). The mean and median durationsfrom admission to receiving NIV were 4.2 days and 2 days respectively (SD: 8.3 days) – estimated fromrecords on cases with complete records on dates of hospital admission and NIV onset (N = 3147). The meanand median durations for NIV were 2 days and 0 days respectively (SD: 4 days) – estimated based on onlythose cases which have complete NIV duration records (N = 1837).
A total of 2946 patients received invasive mechanical ventilation (IMV). The mean and median durationsfrom admission to receiving IMV were 3.2 days and 2 days respectively (SD: 6 days) – estimated from recordson cases with complete records on dates of hospital admission and IMV onset (N = 2647). The mean, medianand SD for the duration of IMV – estimated based on all 1751 cases with complete records on IMV stays –were 11.2 days, 10 days and 8.2 days respectively.
2
Figure 1: Overview of cohort and outcomes as of 19 May 2020.
All patients in ISARIC database (N=46929)
ANALYSED, 55%EXCLUDED, 45%
>14-days follow-up and
positive for COVID-19(N=25849)
22% 23%
>14-days follow-up and
negative or not confirmed(N=10799)
<14-days follow-up (N=10281)
82%
No ICU/HDU or ICU/HDU status unknown
(N=21097)
Dischargedalive (N=11312)
Current status unknown (N=1266)
In hospital(N=3006)
Deceased(N=5513)
14% 26% 6%54%
ICU/HDU(N=4752)
Dischargedalive (N=1591)
Current status unknown (N=488)
In hospital(N=1106)
Deceased(N=1567)
23% 33% 11%33%
18%
3
Patient CharacteristicsFigure 2: 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+
2000 1500 1000 500 0 500 1000 1500 2000Count
Age
gro
up
Outcome
Discharge
Ongoing care
Death
4
Figure 3: 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.
82/25070
114/24804
117/25064
252/25061
294/25074
313/25063
432/24971
51/1933
730/25077
597/19115
78/1932
1369/25057
1823/25077
1887/25085
2102/25093
2463/25073
2979/25094
4022/25080
4143/25101
4293/25094
3968/19115
5561/25106
9371/25084
8288/19115
16181/25231
16936/25207
Conjunctivitis
Ear pain
Lymphadenopathy
Bleeding
Skin rash
Seizures
Skin ulcers
Disturbance or loss of smell
Runny nose
Cough (bloody sputum / haemoptysis)
Disturbance or loss of taste
Joint pain
Sore throat
Wheezing
Abdominal pain
Headache
Chest pain
Muscle aches
Vomiting / Nausea
Diarrhoea
Cough (with sputum)
Altered consciousness / confusion
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
5
0.00
0.04
0.08
0.12
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
Disturbance or loss of tasteDisturbance or loss of smell
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
Dis
turb
ance
or
loss
of t
aste
Dis
turb
ance
or
loss
of s
mel
lC
onju
nctiv
itis
Ble
edin
gS
kin
ulce
rsS
kin
rash
Sei
zure
sA
ltere
d co
nsci
ousn
ess
/ con
fusi
on
−1.0
−0.5
0.0
0.5
1.0phi coefficient
6
Figure 4: 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 patientsfor whom 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. 16% of individuals had no comorbidities reported on admission(some due to missing data).
105/25130
135/25849
381/24751
565/25011
1002/24568
1274/21785
2348/25230
2295/24571
2670/25226
2678/24992
3168/25069
3277/25238
3814/25243
4168/25262
4796/25165
7312/25269
1012/2212
AIDS/HIV
Pregnancy
Liver disease
Malnutrition
Chronic hematologic disease
Smoking
Malignant neoplasm
Rheumatologic disorder
Chronic neurological disorder
Obesity
Dementia
Asthma
Chronic kidney disease
Chronic pulmonary disease
Diabetes
Chronic cardiac disease
Hypertension*
0.00 0.25 0.50 0.75 1.00Proportion
Con
ditio
n Conditionpresent
No
Yes
*Caution when interpreting this result as the sample size is small due to it being a new variable in the dataset.
7
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
8
Variables by ageFigure 5: 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 = 23685
0.0
0.1
0.2
0.3
0.4
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)P
ropo
rtio
n w
ithm
alig
nanc
y
N = 23486
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 = 21833
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 = 23619
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 = 23591
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
curr
ently
sm
oke
N = 15578
9
Figure 6: Symptoms recorded at hospital presentation stratified by age group. Boxes show the proportionof individuals with each symptom, 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 theplot (this varies between plots due to data completeness). Top: Left-hand column shows symptoms of fever,cough and shortness of breath, and right-hand column shows the proportions experiencing at least one ofthese symptoms. Bottom: The following symptoms are grouped: upper respiratory is any of runny nose,sore throat or ear pain; constitutional is any of myalgia, joint pain, fatigue or headache.
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 = 23491
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 = 18832
0.00
0.25
0.50
0.75
1.00
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
Cou
gh o
rfe
ver
N = 23810
0.00
0.25
0.50
0.75
1.00
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
Sho
rt o
f bre
ath
N = 24828
0.00
0.25
0.50
0.75
1.00
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
Cou
gh, f
ever
or
shor
t of b
reat
h
N = 24918
10
0.00
0.25
0.50
0.75
1.00
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
Upp
er r
espi
rato
rysy
mpt
oms
N = 18700
0.00
0.25
0.50
0.75
1.00
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
Con
fusi
on
N = 21137
0.00
0.25
0.50
0.75
1.00
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
Con
stitu
tiona
lsy
mpt
oms
N = 21029
0.00
0.25
0.50
0.75
1.00
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
Nau
sea
orvo
miti
ng
N = 20646
0.00
0.25
0.50
0.75
1.00
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
Dia
rrho
ea
N = 20612
0.00
0.25
0.50
0.75
1.00
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
Abd
omin
alpa
in
N = 19939
11
Figure 7: 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 = 22186
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 = 14378
50
100
150
200
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
Hea
rt r
ate
(min
.−1)
N = 23172
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 = 23220
34
36
38
40
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
Tem
pera
ture
(°C
)
N = 23495
12
Figure 8: 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
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
WC
C (1
09 /L) N = 9704
0
3
6
9
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)Lym
phoc
ytes
(109 /L
)
N = 9326
0
5
10
15
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)Neu
trop
hils
(109 /L
)
N = 9424
0
10
20
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)U
rea
(mm
ol/L
) N = 8123
0
100
200
300
400
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
CR
P (
mg/
L)
N = 9178
8
12
16
20
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)Pro
thro
mbi
n tim
e (s
)
N = 4078
0
10
20
30
40
50
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
AP
TT
(s)
N = 3570
0
10
20
30
40
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)Bili
rubi
n (µ
mol
/L) N = 7862
0
30
60
90
<10 10− 20− 30− 40− 50− 60− 70− ≥ 80
Age group (years)
ALT
(un
its/L
) N = 7232
13
Hospital stays and outcomesFigure 9: 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. White dots are outliers.
0
30
60
90
120
Male Female
Sex
Leng
th o
f hos
pita
l sta
y
SexMaleFemale
Figure 10: 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. White dots are outliers.
0
30
60
90
120
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 Age0−910−1920−2930−3940−4950−5960−6970+
14
Figure 11: Distribution of time (in days) from hospital admission to ICU admission, excluding outliers.
0.0
0.1
0.2
0.3
0.4
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Time (in days) from admission to ICU
Den
sity
15
Figure 12: 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 30 40Days relative to admission
Pro
port
ion
Status
Discharged
Transferred
Unknown
Ongoing care
Ward
ICU
Death
16
Figure 13: 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
6000
01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19Epidemiological week of admission/symptom onset (2020)
Cas
es
Outcome
Discharge
Ongoing care
Death
17
TreatmentFigure 14: 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.
77/20367
109/20437
18/1015
383/20383
685/20405
812/20311
1141/19089
1261/20404
1560/20402
1771/20427
2091/20522
2668/20468
3122/20311
245/1029
13996/20574
16820/20636
Inhaled nitric oxide
Extracorporeal
Off−label / compassionate use medications*
Tracheostomy
Renal replacement therapy
Antifungal agent
Other
Prone ventilation
Inotropes / vasopressors
Antiviral agent
Invasive ventilation
Non−invasive ventilation
Corticosteroid agent
High flow oxygen therapy*
Nasal / mask oxygen therapy
Antibiotic agent
0.00 0.25 0.50 0.75 1.00Proportion
Trea
tmen
t Treatment
No
Yes
18
0.0
0.1
0.2
0.3
0.4
AntifungalAntiviral
CorticosteroidAny oxygen provision
Antibiotic
Treatments used during hospital admission
Pro
port
ion
of p
atie
nts
*Caution when interpreting this result as the sample size is small due to it being a new variable in the dataset.
19
Intensive Care and High Dependency Unit TreatmentsFigure 15: 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.
2/111
73/3439
109/3455
367/3452
372/3451
423/3259
513/3447
724/3485
860/3463
1196/3468
1541/3507
1543/3453
55/106
2055/3542
3157/3542
3234/3573
Off−label / compassionate use medications*
Inhaled nitric oxide
Extracorporeal
Antifungal agent
Tracheostomy
Other
Renal replacement therapy
Antiviral agent
Corticosteroid agent
Prone ventilation
Non−invasive ventilation
Inotropes / vasopressors
High flow oxygen therapy*
Invasive ventilation
Antibiotic agent
Nasal / mask oxygen therapy
0.00 0.25 0.50 0.75 1.00Proportion
Trea
tmen
t Treatment
No
Yes
20
0.00
0.05
0.10
0.15
0.20
0.25
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
0
50
100
Total hospital stay ICU
Location
Leng
th o
f sta
y (d
ays)
*Caution when inter-preting this result as the sample size is small due to it being a new variable in the dataset.
21
Statistical AnalysisFigure 16: 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.
0.000
0.025
0.050
0.075
0.100
0.125
0 10 20 30
Time (in days) from symptom onset to admission
Den
sity
22
Figure 17: 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 30 60 90 120
Time (in days) from admission to death or recovery
Den
sity
23
Figure 18: Probabilities of death (red curve) and recovery (green curve) over time. The black line indicatesthe case fatality ratio (CFR). 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). The point estimate ofthe CFR is 0.35 (95% CI: 0.34-0.36).
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
24
Country ComparisonsFigure 19: Number of sites per country.
1 1 2 4 4
27
1 1 3 2 2 6 2 4 313
210
4 2
19
5 3 1 18
1 2 1 28
1
191
17
10
50
100
150
200
Argen
tina
Austra
lia
Austri
a
Belgium
Brazil
Canad
aChil
e
Colom
bia
Ecuad
or
Estonia
Franc
e
Germ
any
Greec
e
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
Afri
ca
South
Kor
eaSpa
in
Turk
ey UKUSA
Viet N
am
Country
Site
s
25
Figure 20: Distribution of patients by country and outcome.
6
282
15
188
22
101
52
13 14 12
302
26
151
7 9
396
7
182
317
44
260
92
103
266
2
0
100
200
300
400
Austri
a
BelgiumBra
zil
Canad
a
Estonia
Franc
e
Germ
any
Greec
e
Hong
KongIn
dia
Irelan
dIsr
aelIta
ly
Japa
n
Kuwait
Nethe
rland
s
New Z
ealan
d
Norway
Peru
Poland
Portu
gal
Roman
ia
South
Afri
ca
South
Kor
eaSpa
inUSA
Viet N
am
Cas
es
21566
0
5000
10000
15000
20000
UK
Outcome
Discharge
Ongoing care
Death
Country
26
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).
27
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 parentheses and have been rounded to two decimal places.
Table 1: Patient Characteristics
Description ValueSize of cohort 25849
By sexMale 15271 (0.59)Female 10493 (0.41)Unknown 85 (0)
By outcome statusDead 7080 (0.27)Recovered (discharged alive) 12903 (0.5)Still in hospital 4112 (0.16)Transferred to another facility 1303 (0.05)Unknown 451 (0.02)
By age group0-9 195 (0.01)10-19 137 (0.01)20-29 462 (0.02)30-39 1018 (0.04)40-49 1919 (0.07)50-59 3546 (0.14)60-69 4280 (0.17)70+ 13886 (0.54)Unknown 406 (0.02)
Admitted to ICU/HDU?Yes 4752 (18)No/Unknown 21097 (82)
28
Table 2: Outcome by age and sex.
Variable Still in hospital Death Discharge Transferred UnknownAge0-9 25 (0.01) 2 (0) 162 (0.01) 4 (0) 2 (0)10-19 15 (0) 4 (0) 110 (0.01) 7 (0.01) 1 (0)20-29 57 (0.01) 14 (0) 379 (0.03) 9 (0.01) 3 (0.01)30-39 153 (0.04) 38 (0.01) 786 (0.06) 25 (0.02) 16 (0.04)40-49 325 (0.08) 120 (0.02) 1384 (0.11) 64 (0.05) 26 (0.06)50-59 607 (0.15) 424 (0.06) 2314 (0.18) 132 (0.1) 69 (0.15)60-69 792 (0.19) 915 (0.13) 2300 (0.18) 203 (0.16) 70 (0.16)70+ 2070 (0.5) 5475 (0.77) 5247 (0.41) 848 (0.65) 246 (0.55)
SexMale 2424 (0.59) 4503 (0.64) 7316 (0.57) 767 (0.59) 261 (0.58)Female 1673 (0.41) 2553 (0.36) 5547 (0.43) 533 (0.41) 187 (0.41)Unknown 1 (0) 12 (0) 22 (0) 1 (0) 2 (0)
Table 3: Prevalence of Symptoms
Symptoms Present Absent UnknownHistory of fever 16936 (0.66) 6934 (0.27) 1979 (0.08)Shortness of breath 16181 (0.63) 9040 (0.35) 628 (0.02)Cough 12853 (0.5) 6262 (0.24) 6734 (0.26)Fatigue / Malaise 9371 (0.36) 10544 (0.41) 5934 (0.23)Altered consciousness / confusion 5561 (0.22) 15918 (0.62) 4370 (0.17)Diarrhoea 4293 (0.17) 16659 (0.64) 4897 (0.19)Vomiting / Nausea 4143 (0.16) 16834 (0.65) 4872 (0.19)Muscle aches 4022 (0.16) 14550 (0.56) 7277 (0.28)Chest pain 2979 (0.12) 17317 (0.67) 5553 (0.21)Headache 2463 (0.1) 15988 (0.62) 7398 (0.29)Abdominal pain 2102 (0.08) 18154 (0.7) 5593 (0.22)Wheezing 1887 (0.07) 17315 (0.67) 6647 (0.26)Sore throat 1823 (0.07) 16345 (0.63) 7681 (0.3)Joint pain 1369 (0.05) 16400 (0.63) 8080 (0.31)Runny nose 730 (0.03) 17076 (0.66) 8043 (0.31)Skin ulcers 432 (0.02) 18707 (0.72) 6710 (0.26)Seizures 313 (0.01) 20017 (0.77) 5519 (0.21)Skin rash 294 (0.01) 18933 (0.73) 6622 (0.26)Bleeding 252 (0.01) 19776 (0.77) 5821 (0.23)Lymphadenopathy 117 (0) 18695 (0.72) 7037 (0.27)Ear pain 114 (0) 17439 (0.67) 8296 (0.32)Conjunctivitis 82 (0) 18750 (0.73) 7017 (0.27)
29
Table 4: Prevalence of Comorbidities
Comorbidities Present Absent UnknownChronic cardiac disease 7312 (0.28) 16914 (0.65) 1623 (0.06)Diabetes 4796 (0.19) 19185 (0.74) 1868 (0.07)Chronic pulmonary disease 4168 (0.16) 19982 (0.77) 1699 (0.07)Chronic kidney disease 3814 (0.15) 20225 (0.78) 1810 (0.07)Asthma 3277 (0.13) 20765 (0.8) 1807 (0.07)Dementia 3168 (0.12) 20633 (0.8) 2048 (0.08)Obesity 2678 (0.1) 19346 (0.75) 3825 (0.15)Chronic neurological disorder 2670 (0.1) 21210 (0.82) 1969 (0.08)Malignant neoplasm 2348 (0.09) 21491 (0.83) 2010 (0.08)Rheumatologic disorder 2295 (0.09) 20866 (0.81) 2688 (0.1)Smoking 1274 (0.05) 12058 (0.47) 12517 (0.48)Hypertension 1012 (0.04) 1069 (0.04) 23768 (0.92)Chronic hematologic disease 1002 (0.04) 22188 (0.86) 2659 (0.1)Malnutrition 565 (0.02) 22220 (0.86) 3064 (0.12)Liver disease 381 (0.01) 22955 (0.89) 2513 (0.1)Pregnancy 135 (0.01) 25146 (0.97) 568 (0.02)
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 UnknownAntibiotic agent 16820 (0.65) 3294 (0.13) 5735 (0.22)Oxygen therapy 16760 (0.65) 8101 (0.31) 988 (0.04)Nasal / mask oxygentherapy
13996 (0.54) 5858 (0.23) 5995 (0.23)
Non-invasiveventilation
3937 (0.15) 20791 (0.8) 1121 (0.04)
Corticosteroid agent 3122 (0.12) 16250 (0.63) 6477 (0.25)Invasive ventilation 2946 (0.11) 21831 (0.84) 1072 (0.04)Antiviral agent 1771 (0.07) 17734 (0.69) 6344 (0.25)Inotropes /vasopressors
1560 (0.06) 17652 (0.68) 6637 (0.26)
Prone ventilation 1261 (0.05) 17853 (0.69) 6735 (0.26)Other 1141 (0.04) 16276 (0.63) 8432 (0.33)Antifungal agent 812 (0.03) 18563 (0.72) 6474 (0.25)Renal replacementtherapy
685 (0.03) 18562 (0.72) 6602 (0.26)
Tracheostomy 383 (0.01) 18762 (0.73) 6704 (0.26)Extracorporealmembraneoxygenation (ECMO)
310 (0.01) 24428 (0.95) 1111 (0.04)
30
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
10.5 10.2 8 9 19 (17.9, 20.4)
Symptomonset toadmission
13 7.9 5 9 7.9 (7.5, 8.7)
Admissionto ICUentry
2.8 5.9 1 3 3.7 (3.5, 4.1)
Durationof ICU
9.7 9.3 7 11 NA
Admissionto IMV
3.2 6 2 4 3.9 (3.7, 4.3)
Durationof IMV
11.2 8.2 10 10 NA
Admissionto NIV
4.2 8.3 2 5 4.9 (4.6, 5.3)
Durationof NIV
2 4 0 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.
31