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 ncov@isaric.org.

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 ncov@isaric.org.

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