Modelling the COVID-19 pandemic in context: an international

AguasR, etal. BMJ Global Health 2020;5:e003126. doi:10.1136/bmjgh-2020-003126

Modelling the COVID-19 pandemic in

context: an international participatory

approach

Ricardo Aguas ,

1,2

Lisa White,

2,3

Nathaniel Hupert,

Rima Shretta ,

Wirichada Pan- Ngum,

Olivier Celhay,

Ainura Moldokmatova,

Fatima Ari,

Ali Mirzazadeh,

Hamid Shari,

Keyrellous Adib,

Mohammad Nadir Sahak,

Caroline Franco,

Renato Coutinho,

CoMo Consortium

Practice

To cite: AguasR, WhiteL,

HupertN, etal. Modelling the

COVID-19 pandemic in context:

an international participatory

approach. BMJ Global Health

2020;5:e003126. doi:10.1136/

bmjgh-2020-003126

Handling editor Seye Abimbola

► Additional material is

published online only. To view,

please visit the journal online

(http:// dx. doi. org/ 10. 1136/

bmjgh- 2020- 003126).

Received 11 June 2020

Revised 24 September 2020

Accepted 28 September 2020

For numbered afliations see

end of article.

Correspondence to

Lisa White;

lisa. white@ ndm. ox. ac. uk

employer(s)) 2020. Re- use

permitted under CC BY.

Published by BMJ.

ABSTRACT

The SARS- CoV-2 pandemic has had an unprecedented

impact on multiple levels of society. Not only has the

pandemic completely overwhelmed some health systems

but it has also changed how scientic evidence is

shared and increased the pace at which such evidence

is published and consumed, by scientists, policymakers

and the wider public. More signicantly, the pandemic

has created tremendous challenges for decision- makers,

who have had to implement highly disruptive containment

measures with very little empirical scientic evidence to

support their decision- making process. Given this lack

of data, predictive mathematical models have played an

increasingly prominent role. In high- income countries,

there is a long- standing history of established research

groups advising policymakers, whereas a general lack

of translational capacity has meant that mathematical

models frequently remain inaccessible to policymakers

in low- income and middle- income countries. Here, we

describe a participatory approach to modelling that aims

to circumvent this gap. Our approach involved the creation

of an international group of infectious disease modellers

and other public health experts, which culminated in

the establishment of the COVID-19 Modelling (CoMo)

Consortium. Here, we describe how the consortium was

formed, the way it functions, the mathematical model used

and, crucially, the high degree of engagement fostered

between CoMo Consortium members and their respective

local policymakers and ministries of health.

INTRODUCTION

The novel coronavirus, SARS- CoV-2, which

causes COVID-19, has affected at least 213

countries/regions, with more than 17 million

confirmed cases and in excess of 660 000

deaths globally.

As a new clinical entity, the

global impact of COVID-19 is characterised

by both uncertainty and rapid discovery,

laying the grounds for mathematical model-

ling to emerge as the prominent field of

research used to provide advice for pandemic

containment strategies.

2 3

High- income Asian

countries were able to call on their system

responsiveness and experience with recent

pandemics, rapidly enforcing efficient testing

and quarantining/isolation strategies. In

contrast, European countries took a much

more measured approach (easily confused

with lack of preparedness) early on, tapping

into their modelling expertise to predict

the outcome of the pandemic and what the

best containment strategies moving forward

might be. As a result, most European coun-

tries converged and introduced suppression

Key questions

What is already known?

► The optimal approaches to tackle the COVID-19 pan-

demic depend on several contextual factors, includ-

ing population age structure, variations in available

resources (including infrastructure, nancial and

human resources) and sociocultural considerations.

► Governments across the world have been advised

by mathematical modelling projections that do

not necessarily take those contextual factors into

consideration.

What are the new ndings?

► We describe the creation of a participatory mod-

elling approach platform, the COVID-19 Modelling

Consortium, and illustrate some of its use cases.

► We demonstrate how the participatory nature of the

consortium has been critical in its success, in terms

of addresing the contextual factors that underpin

health policy interventions and gaining decision

makers' trust.

What do the new ndings imply?

► We advocate a participatory modelling approach,

where in- country experts play an essential iterative

role, being policy- facing in its dealings with policy-

makers and simultaneously delivering or facilitating

reactive modelling that can feed back, in real time,

into the decision- making processes.

on August 3, 2024 by guest. Protected by copyright.http://gh.bmj.com/BMJ Glob Health: first published as 10.1136/bmjgh-2020-003126 on 23 December 2020. Downloaded from on August 3, 2024 by guest. Protected by copyright.http://gh.bmj.com/BMJ Glob Health: first published as 10.1136/bmjgh-2020-003126 on 23 December 2020. Downloaded from on August 3, 2024 by guest. Protected by copyright.http://gh.bmj.com/BMJ Glob Health: first published as 10.1136/bmjgh-2020-003126 on 23 December 2020. Downloaded from

AguasR, etal. BMJ Global Health 2020;5:e003126. doi:10.1136/bmjgh-2020-003126

BMJ Global Health

strategies (including everything from the shutdown

of basic economic activity, in many instances enforced

by hastily promulgated laws or executive orders, to the

wholesale reorganisation of medical and hospital- based

care) centred around non- pharmaceutical interventions

(NPIs),

4–6

fearing that pandemic mitigation would cost

too many lives, with the notable exception of Sweden.

Interestingly, several low- income and middle- income

countries (LMICs) followed suit, adopting health poli-

cies informed by early modelling from developed coun-

tries, without considering how the modelling predic-

tions might be affected by contextual factors. Modelling

devoid of local context has been shown to produce far

from optimal/useful projections in the Ebola, H5N1 and

H1N1 pandemics/outbreaks.

2 8

For such high priority

and complex topics, a participatory modelling approach

seems to be particularly well suited, as it provides policy-

makers with timely and dynamic support, enabling the

modelling process to be built around the coproduction

of knowledge between modellers and policymakers.

This multipronged approach with collaboration among

experts of different disciplines working together to

incorporate all relevant contextual factors into pandemic

modelling has been discussed at length in Rhodes et al.

Here, we describe a participatory approach to model-

ling that coalesced around these contextual consider-

ations and resulted in the creation of an international

consortium of infectious disease modellers and other

public health experts, the COVID-19 Modelling (CoMo)

Consortium. Putting in- country experts at the forefront

of model development, we underscore the need to incor-

porate contextual factors (including population age

structure, resource availability—including infrastruc-

ture, financial and human resources—and sociocultural

considerations) into an iterative policy informing tool.

Our approach demands a social–ecological component

comprising a combination of the contextual factors listed

below as an integral part of the epidemiological model,

thus addressing some of the limitations observed in other

modelling exercises.

2 3 8 10 11

We appreciate a compromise

between accuracy, transparency, flexibility and timeliness

remains, but are confident a participatory approach is

the best avenue to minimise those trade- offs.

CONTEXTUAL FACTORS

Contextual factor 1: population age structure

It became clear during the early stages of the COVID-19

pandemic that a disproportionate number of older

individuals are at higher risk of severe disease and

mortality,

12 13

with 80% of deaths associated with the

disease occurring in those aged more than 65 years in

outbreaks in China and the USA.

13 14

In Italy, the propor-

tion of deaths occurring in people over the age of 70

reached 88%, presumably due to an aged population

in the Lombardy region.

Data from the Chinese and

Italian outbreaks suggest that the case fatality rate (CFR)

was <1% in the under- 50s, rising to almost 15% and 20%

in the over- 80s, respectively.

15 16

Children appear to suffer

from less severe symptoms

but are at a similar risk of

being infected as the general population; therefore, the

role played by children in the transmission of COVID-19

should be considered when developing control strate-

gies.

This variation in the age- dependent severity of

COVID-19 has important implications for the impact of

the disease in any given country.

Contextual factor 2: uncertainties around the characteristics

of the disease

As a newly emerged disease, considerable uncer-

tainty remains around some of the basic parameters of

COVID-19 infection. Evidence suggests that the median

incubation period is approximately 5 (95% CI: 4.5 to 5.8)

days.

The duration of the infectious period is extremely

uncertain, with some studies suggesting that people

become infectious before developing symptoms

20–22

and others finding that viral shedding in clinical cases

can persist for more than 20 days.

12 23

In fact, some

evidence suggests infectiousness begins before symptoms

develop and is likely to peak around the time of symptom

onset.

23–25

Several estimates for the serial interval (time

between transmission chains),

26 27

taken together with

estimates for the incubation period, strongly suggest

that asymptomatically infected people can transmit the

virus. A study from China found the median duration

from first symptoms to dyspnoea, hospital admission

and acute respiratory distress syndrome was 5, 7 and 8

days, respectively,

although data from New York City

suggest a more rapid progression.

The infection fatality

rate (IFR) measures the percentage of infected individ-

uals who later succumb to the disease; a meta- analysis

suggests an IFR for COVID-19 of around 0.20% (mean

across all ages), given all available data as of 22 March

2020.

However, these estimates are full of uncertain-

ties particularly to what concerns the denominator due

to challenges in reliably ascertaining how many people

are/have been infected with the virus. Two streams of

scientific research are trying to resolve this underlying

burden of infection: one relies on the use of inference

and predictive models,

30–32

making the best use of avail-

able data to disentangle the unobserved number of infec-

tions driving the force of infection; the other focuses on

diagnostic tool development to enable reliable mass sero-

logical studies to be carried out.

33–35

Reported symptomatic CFR vary widely by country, as

criteria and capacity for testing can vary considerably.

Burdens of comorbidity differ, as do demographics, and

cause of death attribution is not uniform. Both IFR and

CFR values may also be dynamic in a single setting, as

delays in deaths tend to result in the underestimation

of CFR early on in an epidemic, with surges in lethality

during healthcare system stress.

Those at higher risk of developing severe disease

include individuals with comorbidities such as hyper-

tension, cardiovascular disease and diabetes.

Obesity,

especially in younger patients, is emerging as a risk factor

on August 3, 2024 by guest. Protected by copyright.http://gh.bmj.com/BMJ Glob Health: first published as 10.1136/bmjgh-2020-003126 on 23 December 2020. Downloaded from

AguasR, etal. BMJ Global Health 2020;5:e003126. doi:10.1136/bmjgh-2020-003126

BMJ Global Health

for more severe clinical manifestations in cohorts in

the USA.

There is conflicting evidence regarding the

contribution of pre- existing respiratory disease to clin-

ical severity.

Given the relative paucity of detailed and

accurate parameter data, it is vital that local modellers

who have intimate knowledge of their own context(s) are

engaged to ensure the optimal application and commu-

nication of any model and its outputs.

Contextual factor 3: differences in health system capacity

The COVID-19 pandemic presents myriad challenges

for healthcare systems around the world,

including the

need to repurpose and train healthcare staff, increase the

number of both general hospital beds and intensive care

unit beds, purchase equipment (particularly ventilators,

high- flow oxygen systems and oxygen concentrators), and

hire carers needed for specialist care and/or treatment

for comorbidities.

Anticipating health system demand

in comparison to capacity under various intervention

scenarios is a key aspect of national and regional stra-

tegic decision- making.

This underscores one of the key

roles of the disease- modelling approach in the context of

COVID-19 pandemic preparedness and response.

Contextual factor 4: socioeconomic and cultural differences

The connection between cultural values (uncertainty

avoidance, power distance, individualism vs collectivism)

and infectious diseases is well established, primarily

with literature on antibiotic prescribing and treatment

seeking behaviours.

Different countries have adopted

a variety of strategies and combinations of NPIs to meet

the challenges posed by COVID-19. While these differ-

ences in approach are highly dependent on local cultural

contexts and values,

their differential effect in the

setting of COVID-19 is debated among social scientists.

Early empirical work is emerging suggesting disease

spread was slower in countries with strong institutional

systems and hierarchical cultures.

In China, where this

novel coronavirus first emerged, a strict policy including

measures such as quarantine, self- isolation and contain-

ment immediately implemented.

This ultimately

involved strict physical distancing measures in social

settings, or a ‘lockdown’, as Chinese local authorities

imposed travel restrictions and severely restricted the

movement of people. Physical distancing and isolation

measures, aided by extensive testing and contact tracing,

were also used effectively in South Korea to bring their

COVID-19 epidemic under control.

Singapore under-

took intensive surveillance and contact tracing, followed

by isolation of suspected and confirmed cases to halt

transmission chains.

Some Muslim countries in the

Middle East have made a historical decision to cancel

Friday and congregational prayers and to close their

holy shrines.

Other countries, such as the UK, started

pursuing strategies aimed at allowing herd immunity to

gradually develop while reducing the demand on the

health service, also known as ‘flattening the curve’, but

did eventually enforce a strict lockdown policy. Notably,

in Northern Europe, the onus of practicing efficient

containment measures was transferred to the individual,

while more relaxed lockdown versions were enforced,

characterised by a minimal disruption to society. Several

recent studies have tried to estimate the impact of these

different containment/suppression strategies in different

countries,

4 47–52

but little insight has been gained into the

optimal long term strategy, assuming that a vaccine may

not be available before the end of the year.

Socioeconomic differences can critically underpin

the potential adherence of any infection containment

measure. A significant proportion of the population in

LMICs are daily- wage earners who cannot work from

home. They often rely on street vendors or local markets

for their meals, which are usually overcrowded places

where hygiene and physical distancing measures are

difficult to enforce. Large and intergenerational families,

migrants and refugees who live in densely populated areas

mean physical distancing is virtually impossible, ineffec-

tive and may cause more harm than good. Religious and

cultural festivals can seriously disrupt physical distancing

measures usually people would gather in the thousands

to celebrate Easter, Chinese New Year, Ramadan.

Contextual factor 5: a rapidly developing situation

Published models frequently remain inaccessible to poli-

cymakers in LMICs due to the lack of translational capacity

in many of these countries. This can present difficulties in

converting a prepublication or peer- reviewed model into

a practical, real- time decision- making tool. Critically, any

model requires continuous updating, usually on a daily

basis, if it is to keep pace with the situation unfolding in

a given country and the science relating to the infectious

agent it is modelling, and therefore meet the needs of

policymakers in that country. Publishing scientific papers

and online tools alone is insufficient to engage with and

inform health policymakers, because in a rapidly devel-

oping situation, the model and the modeller cannot be

separated, and policy responses must be couched within

the specific country or subnational context. This can only

be achieved via a combination of technology, training

and communication. User- friendly platforms facilitating

real- time data analysis and scenario prediction by local

epidemiologists is key for effective planning and policy

decision making to mitigate the COVID-19 pandemic.

THE COVID-19 MODELING (COMO) CONSORTIUM

CoMo Consortium—policy rationale

During the acute phase of the pandemic, the focus of

most CoMo Consortium modelling work has been to

address immediate questions driven by policymakers

(figure 1). These have focused on exploring optimal strat-

egies to achieve specific aims, such as minimising cases,

mortality or demand on the health system. As the situa-

tion evolves, more thought is now being directed towards

economic impacts and the direct and indirect costs of

the disease and the measures taken to control its spread.

on August 3, 2024 by guest. Protected by copyright.http://gh.bmj.com/BMJ Glob Health: first published as 10.1136/bmjgh-2020-003126 on 23 December 2020. Downloaded from

AguasR, etal. BMJ Global Health 2020;5:e003126. doi:10.1136/bmjgh-2020-003126

BMJ Global Health

The challenges of increasing and repurposing health

system capacity, in addition to worker absenteeism due

to COVID-19 morbidity and mortality, represent some

of the direct costs associated with the pandemic.

The

economic and behavioural responses adopted to reduce

SARS- CoV-2 transmission form a substantial proportion

of the indirect costs associated with the pandemic. These

include suspension of manufacturing and production;

employment losses and reductions in consumer spending

in many sectors, notably travel and tourism;

interrup-

tions in services and the supply chain of products for the

prevention and control of other major diseases, such as

vaccine- preventable diseases, HIV, malaria and active

cancers. A fully comprehensive set of cost efficiency/

effectiveness analyses will be of great value to inform

countries seeking scientific evidence evaluating current

and future containment strategies.

CoMo Consortium—a participatory approach

Historically, modelling expertise has been concentrated

in high- income countries (HICs), where multiple model-

ling groups, often with large teams, have tended to form

consortia. While these consortia mainly inform policy

in HICs, they often act as external service providers to

LMICs. Modelling inputs to LMICs have, therefore, gener-

ally involved two groups of professionals: modellers from

HICs and policymakers from LMICs; a situation which

is far from ideal. We sought to avoid this by adopting a

participatory approach when establishing the interna-

tional CoMo Consortium. A participatory approach is

key for policymakers to fully appreciate the uncertainties

subjacent to assumed parameter values, implemented

mechanisms of action and general model structure. The

immediate consequence of that understanding is clarity

on the relevance of critical data in circumventing uncer-

tainties, and the understanding that continual validation

frameworks are key to guarantee the best possible policy

is implemented at all times. The use of a participatory

approach can even be viewed as part of the intervention

package, as argued in other studies,

and is highly desir-

able,

as evidenced by the number of policymakers from

multiple countries around the world who have requested

to actively participate in the CoMo Consortium. Being a

Consortium member also facilitates information sharing

among countries with comparable contexts that might be

addressing similar questions.

The CoMo Consortium mathematical model was devel-

oped by three groups of professionals, with each group

forming one of three nodes: a development node, an

in- country expert node and a policymaker node. Each

node comprises a variety of relevant professionals (table 1,

figure 2). Where there were existing in- country experts,

the CoMo Consortium sought to build on existing close

working relationships (or establish such relationships)

with these experts.

CoMo Consortium—development phase

The CoMo Consortium was established as a response to

the analytical demands of LMICs trying to prepare for

the COVID-19 epidemic. Several in- country experts were

approached for support by their own policymakers and

later reached out to the Oxford Modelling for Global

Health group for additional technical advice and support.

This sparked an initiative to provide technical support

and mentoring to these individuals which evolved into a

precursor for the CoMo consortium. The consortium was

officially formed when the mathematical model and its

accompanying online application were introduced to the

first in- country teams. The model is an age- dependent

susceptible- exposed- infected- recovered model, adapted

from

to reflect SARS- CoV-2 transmission/virological

traits and the interventions being deployed in different

countries—a detailed model description can be found

in the online supplemental materials. A key focus was to

Figure 1 CoMo Consortium participants. Individual country participants, colour coded by the stages of engagement with

policymakers (table2). *Refers to the 22 countries/territories using the CoMo model through the WHO Regional Ofce for the

Eastern Mediterranean (EMRO). The designations employed and the presentation of the material on this map do not imply the

expression of any opinion concerning the legal status of any country, territory, city or area or of its authorities or concerning the

delimination of its frontiers or boundaries. CoMo, COVID-19 Modelling.

on August 3, 2024 by guest. Protected by copyright.http://gh.bmj.com/BMJ Glob Health: first published as 10.1136/bmjgh-2020-003126 on 23 December 2020. Downloaded from

AguasR, etal. BMJ Global Health 2020;5:e003126. doi:10.1136/bmjgh-2020-003126

BMJ Global Health

build a SARS- CoV-2 specific model that could be seam-

lessly updated as new information became available and

different sets of interventions were considered. A stand-

alone Excel- based tool, the Cornell COVID-19 Caseload

Calculator with Capacity and Ventilators model, formed

the basis for the healthcare components of the CoMo

Consortium model.

From inception, the goal was to allow the user to define

critical aspects of the model structure, model inputs, both

in terms of parameter values and interventions consid-

ered, interface options and output reporting. This was

key to the model being well accepted and easily adopted

by so many different countries.

The consortium has grown organically since 16 March

2020 and now includes approximately 100 members,

representing more than 30 countries (figure 1). Regular

communication channels were established, and several

developmental working groups were formed. These

working groups are either country- specific, question-

specific or technique- specific, with the latter including

working groups for the mechanistic model, developing

the web- based interface, exploring spatial formulations

of the model and a hospital capacity simulation tool.

Question- specific working groups include those centred

around lockdown- release strategies, or screening and

diagnostic strategies. The in- country experts are engaged

in continuous communication with their respective poli-

cymakers, enabling the rapid adaptation of the CoMo

Consortium model and other in- country models (where

available) to address the fast- paced changes occurring

during the course of the outbreak in each particular

setting.

CoMo Consortium—dissemination phase

Following its initial development phase, the CoMo

Consortium moved into the dissemination phase, once

the model code had been sufficiently scrutinised and the

interface—figure 3—had been redesigned to facilitate

model calibration to data. A health economics working

group was formed to identify the most appropriate use of

available funds and inform future cost efficiency analyses.

In- country working groups continued to work on their

local COVID-19 situations while also contributing to and

benefitting from the question- specific and technique-

specific working groups. The full model code was made

available to all members and a special online ‘code

reading’ session was held. Members were then able to

take the code and adapt it for their own local context. For

Table 1 The three nodes of the CoMo Consortium Development phase

Development node In- country expert nodes Policymaker nodes

► Lead modeller

► Experienced modellers

► Clinician modeller (public health

specialist)

► App developer

► Coordinator (junior modeller)

► Economist

► Research clinician/preparedness modellers

► Epidemiologists/surveillance specialists

(eld/public health)

► Modellers (early stage/senior)

► Public health specialists

► Health economics modellers

► Medical statisticians

► Representatives from WHO/other NGOs

► State government

► State ministries of public health

► National ministries of public

health

► Local governing bodies and their

health departments

CoMo, COVID-19 Modelling.

Figure 2 CoMo Consortium outlook and interaction ows. This diagram illustrates how the different partners interact in order

to digest policy questions into model simulations through the in- country expert node and the development node and ultimately

result in actionable predictions informing policy decisions. CoMo, COVID-19 Modelling.

on August 3, 2024 by guest. Protected by copyright.http://gh.bmj.com/BMJ Glob Health: first published as 10.1136/bmjgh-2020-003126 on 23 December 2020. Downloaded from

AguasR, etal. BMJ Global Health 2020;5:e003126. doi:10.1136/bmjgh-2020-003126

BMJ Global Health

example, the Brazil group has incorporated a modelling

fitting algorithm while simplifying the hospital simulator

submodel, while the Nigeria group has changed the age

structure to reflect the age classes used by the Nigerian

disease surveillance system.

An ongoing series of training sessions on the use of the

model and the communication of results to policymaking

partners was initiated by the consortium leader and then

taken up by the members. Code version release notes are

continually made available to all consortium members,

highlighting any changes made and the reasons for those

changes. A concerted effort was made to bridge the

model code with the app code and test the performance

of the model across code versions to ensure consistency.

At least one consultation was held between the devel-

opment node and each country’s expert node to discuss

parameter value assumptions and the contextual nature

of some parameters, given how interventions are imple-

mented locally. Some of these meetings instigated

updates to the model structure and app interface custo-

misation options.

CoMo Consortium—use-cases

Three use- cases have emerged for the CoMo Consor-

tium model and its accompanying web- based interface.

First, in settings where there is already considerable

translational capacity, the CoMo Consortium model is

used by in- country modellers to crosscheck the models

they have written, thus helping to refine their outputs

and to maintain high accuracy and validity. Second, in

settings where there is some coding expertise but less

capacity to develop bespoke models, the CoMo Consor-

tium’s primary code can be modified by in- country

modellers to create country- specific models that are

being used to assist policymakers’ decision- making.

Third, in contexts where there is a desire among poli-

cymakers to use modelling to inform development of

their strategies based on local parameter values, but

limited capacity in terms of modelling or coding, the

primary code is used via the CoMo Consortium model’s

web- based interface.

Regardless of the use- case, CoMo operations are ruled

by the flows of interaction depicted in figure 2. Clearly, the

main catalysts are policy questions that shape the shared

model that is adapted by each in- country expert team to

generate country specific models. The in- country expert

teams liaise with the technical experts to validate and

debug their models and later proceed to communicate the

model predictions to the policymakers in a comprehen-

sible way. We cannot emphasise enough how critical the

role played by the in- country experts is. It falls on them to

engage with policymakers and explain to them the features

and capabilities of the model, guide them in addressing

appropriate modelling questions, and later package those

questions alongside all relevant contextual factors to the

technical team to ensure the model can represent the

desired context appropriately. A few examples of the appre-

ciation of the contextual nature of questions addressed by

countries follow:

► In Syria, the main concern in the early stages of the

epidemic was how the virus would spread in displaced

populations and refugee camps. This could not be

explored with the standard CoMo model, so a task

team was put together to develop a bespoke refugee

camp model.

Figure 3 The CoMo model online application. Users can either upload a lled- in template or input all parameter values in the

app directly. User can specify up to 30 intervention periods, dening the start and end dates, as well as the assumed coverage

for each.

on August 3, 2024 by guest. Protected by copyright.http://gh.bmj.com/BMJ Glob Health: first published as 10.1136/bmjgh-2020-003126 on 23 December 2020. Downloaded from

AguasR, etal. BMJ Global Health 2020;5:e003126. doi:10.1136/bmjgh-2020-003126

BMJ Global Health

► In New York, special attention was paid to the very

fast upsurge in patients with respiratory distress in

most city hospitals. This raised serious concerns on

how countries with lower health capacity would fare,

specifically to what concerns the number of available

intensive care unit beds and ventilators. As a conse-

quence, we invested quite some time developing the

hospital submodel and have made it a central point of

discussion with in- country modellers, as it is their role

to assess the current and likely health capacity in their

country and convey the message of how those limi-

tations impact on the predicted epidemic mortality

burden to their respective policymakers.

► In Afghanistan and Timor- Leste there was a lot of

concern regarding the role of migrants in causing

local outbreaks and seed local transmission. This led

to the addition of a user defined parameter setting

the number of daily imported infections.

► Interventions are not implemented in the same way

everywhere, as country contextual idiosyncrasies crit-

ically affect the potential efficiency and practicability

of any intervention.

– Shielding of the elderly should be a major com-

ponent of mortality burden reduction strategies

in HICs, but are not feasible to employ in most

LMICs, where people live in large familial house-

holds and it would not be possible for the elder-

ly to isolate (and in some countries not culturally

acceptable).

– Self- isolation was a core component of initial con-

tainment strategies in Europe where it was quite

straightforward and easily achievable for the indi-

vidual person. In LMICs that is not that case, for

the reasons highlighted above. Due to these issues

with household structures, in Thailand, the min-

istry of health decided to isolate people fitting a

clinical algorithm in governmental facilities, where

they were kept for up to 14 days, or until they test

negative for the virus. This type of testing centre

was implemented early on in South Korea and

Singapore to great effect.

The stage of engagement with local policymakers is

quite heterogeneous across CoMo Consortium members

(table 2). Members with existing channels of communi-

cation with Ministries of Health have gone through the

process of providing feedback to their relevant authori-

ties at a much faster rate. Some members of the academic

community have needed to foster links with appropriate

points of contact at Ministries of Health, specially appointed

government entities, or COVID-19 taskforces, and establish

appropriate channels of communication and gain the trust

of policymakers. In some instances, it has been necessary

for members to navigate a complex landscape, sometimes

involving finding consensus among several different model-

ling groups offering advice to their government.

At the time of writing, 11 of the 22 countries from the

WHO Regional Office for the Eastern Mediterranean

(EMRO) have been using the CoMo Consortium model.

Table 2 CoMo Consortium member countries’ stages of engagement with policymakers

Stage 1

Preliminary analyses and model calibration to explore optimal containment strategies

► Argentina

► Bangladesh

► Brazil

► Cambodia

► Ecuador

► Ethiopia

► Nigeria

► Taiwan

► Sierra Leone

► Malaysia

Stage 2

Have engaged with local Ministries of Health (MoH) or relevant policymakers and are in

the process of analysing the different strategies under consideration

► Philippines

► Portugal

► Myanmar

► Northwest Syria

Stage 3

Have on at least one occasion presented CoMo Consortium model results to the local

MoH or relevant policymakers

► Afghanistan

► Cameroon

► Haiti

► Iran

► Lao PDR

► Tabasco Province, Mexico

► Queretaro State, Mexico

► Thailand

► Mozambique

► New York State, USA

► Orange County, California, USA

► Timor- Leste

► Nepal

Stage 4

The local MoH has made policy decisions based on CoMo Consortium model predictions

► Kyrgyzstan

► North Sumatra Province, Indonesia

CoMo, COVID-19 Modelling.

on August 3, 2024 by guest. Protected by copyright.http://gh.bmj.com/BMJ Glob Health: first published as 10.1136/bmjgh-2020-003126 on 23 December 2020. Downloaded from

AguasR, etal. BMJ Global Health 2020;5:e003126. doi:10.1136/bmjgh-2020-003126

BMJ Global Health

Some are engaged in independent research using either

the CoMo Consortium model (table 2), their own models,

or both, while others are receiving active support from

the WHO EMRO COVID-19 Modelling Support Group.

The countries are distributed across all stages of policy

engagement documented in table 2.

CONCLUSION

Faced with the most significant pandemic in more than a

century, we chose a participatory approach to create the

international CoMo Consortium and develop a dynamic

infectious disease model that addressed a global need.

The key to the success of this participatory approach lies

with the in- country expert node. The in- country experts

include professionals with a wide range of expertise that

play an essential iterative role, being both policy- facing in

its dealings with policymakers and simultaneously deliv-

ering or facilitating reactive modelling that can feed back,

in real time and based on the latest data, into the decision-

making processes. Importantly, this continuous cooper-

ation and feedback loop has been a valuable part of the

process to facilitate collaboration and develop trust. A static

online tool alone would not be sufficient to achieve this.

The biggest strength of participatory approaches can ulti-

mately be its largest limitation, as everything is reliant of the

in- country expert being able to reach policymakers and/

or gain access to the relevant data, perform data quality

control, and liaise with the model development team to

ensure the interventions being implemented in the field

are well captured in the model.

Critically, our approach allows for tailoring of the

model and online app to meet each country’s needs

and facilitates translation of the analytical requirements

into easily digestible outputs that can inform policy. The

CoMo Consortium approach brings modelling to a broad

range of people who will benefit from its participatory

nature, through a combination of technology, training

and effective communication.

Author afliations

Nufeld Department of Medicine, University of Oxford Centre for Tropical Medicine

and Global Health, Oxford, UK

MAEMOD, Mahidol Oxford Tropical Medicine Research Unit, Bangkok, Thailand

Center for Tropical Medicine and Global Health, University of Oxford Centre for

Tropical Medicine, Oxford, UK

Weill Cornell Medicine, Cornell Institute for Disease and Disaster Preparedness,

New York, New York, USA

Nufeld Department of Medicine, University of Oxford, Oxford, UK

Department of Epidemiology, Florida International University, Miami, Florida, USA

School of Medicine, University of California San Francisco, San Francisco,

California, USA

WHO Collaborating Center for HIV Surveillance, Institute for Futures Studies in

Health, Kerman University of Medical Sciences, Kerman, Iran, Kerman, Iran (the

Islamic Republic of)

Independent Researcher (No afliation), no afliation, UK

Regional Ofce for the Eastern Mediterranean, World Health Organization, Kabul,

Afghanistan

Waves and Non- Linear Patterns Research Group, São Paulo State University

(UNESP), Institute of Theoretical Physics, Sâo Paulo, Sao Paulo, Brazil

Centre for Mathematics, Computation and Cognition, Federal University of ABC

Center of Mathematics Computing and Cognition, Santo Andre, São Paulo, Brazil

Twitter Rima Shretta @rimashretta and Caroline Franco @_francocarol

Collaborators CoMo Consortium: Proochista Ariana; Penny Hancock; Roberto

A Kraenkel; Sompob Saralamba; Nantasit Luangasanatip; Sheetal Prakash Silal;

Jared Norman; Rachel Hounsell; Sai Thein Than Tun; Yu Nandar Aung; Bakare

Emmanuel A; Biniam Getachew; Sandra Adele; Semeeh A Omoleke; Rashid U

Zaman; Nicholas Letchford; Daniel M Parker; Sunil Pokharel; Dipti Lata; Siyu Chen;

Shwe Sin Kyaw; Inke N D Lubis; Ivana Alona; John Robert C Medina; Chris Erwin

G Mercado; Sana Eybpoosh; Ibrahim Mamadu; Manar Marzouk; Nicole Feune de

Colombi; Lorena Suárez- Idueta; Francisco Obando; Luzia Freitas; Michael G Klein;

David Scales; Dooronbekova Aizhan; Chynar Zhumalieva; Aida Estebesova; Aibek

Mukambetov; Shamil Ibragimov; Aisuluu Kubatova; Phetsavanh Chanthavialy;

Amel H Salim; Sudhir Venkatesan; Sarin K C; Priyanka Shrestha; Sayed Ataullah

Saeedzai; Jenny Hsieh; Mick Soukavong; Yuki Yunanda; Handoyo Harsono; Mahnaz

Hossain Fariba; Viviana Mabombo; Nicole Advani; Nusrat Jabin; Reshania Naidoo;

Parinda Wattanasri; Amen- Patrick Nwosu; Sopuruchukwu Obiesie.

Contributors RA, LW and WP- N developed the CoMo model. OC developed the

online model application. CF and RC consulted on the model structure. NH, RS,

AMo, FA, AMi, HS, KA and MNS are early model users that helped revise the model

structure and the online application. RA and LW wrote the initial manuscript draft.

All authors revised the draft manuscript.

Funding RA is funded by the Bill and Melinda Gates Foundation (OPP1193472).

LW is funded by the Li Ka Shing Foundation. CF is funded by grant #2017/26770-8,

São Paulo Research Foundation (FAPESP). The CoMo Consortium has support from

the Oxford University COVID-19 Research Response Fund (ref: 0009280). Scientic

writing assistance and editorial support was provided by Adam Bodley, according

to Good Publication Practice guidelines.

Map disclaimer The depiction of boundaries on this map does not imply the

expression of any opinion whatsoever on the part of BMJ (or any member of its

group) concerning the legal status of any country, territory, jurisdiction or area or of

its authorities. This map is provided without any warranty of any kind, either express

or implied.

Competing interests None declared.

Patient consent for publication Not required.

Provenance and peer review Not commissioned; externally peer reviewed.

Data availability statement There are no data in this work.

Supplemental material This content has been supplied by the author(s). It has

not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been

peer- reviewed. Any opinions or recommendations discussed are solely those

of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and

responsibility arising from any reliance placed on the content. Where the content

includes any translated material, BMJ does not warrant the accuracy and reliability

of the translations (including but not limited to local regulations, clinical guidelines,

terminology, drug names and drug dosages), and is not responsible for any error

and/or omissions arising from translation and adaptation or otherwise.

Open access This is an open access article distributed in accordance with the

Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits

others to copy, redistribute, remix, transform and build upon this work for any

purpose, provided the original work is properly cited, a link to the licence is given,

and indication of whether changes were made. See:https:// creativecommons. org/

licenses/ by/ 4. 0/.

ORCID iDs

RicardoAguas http:// orcid. org/ 0000- 0002- 6507- 6597

RimaShretta http:// orcid. org/ 0000- 0001- 5011- 5998

REFERENCES

1 COVID-19 Dashboard: Johns Hopkins University. Available: https://

coronavirus. jhu. edu/ map. html

2 Van Damme W, Dahake R, Delamou A, etal. The COVID-19

pandemic: diverse contexts; different epidemics- how and why? BMJ

Glob Health 2020;5:e003098.

3 Rhodes T, Lancaster K. Mathematical models as public troubles in

COVID-19 infection control: following the numbers. Health Sociology

Review 2020;29:177–94.

4 Flaxman S. Report 13: estimating the number of infections and the

impact of non- pharmaceutical interventions on COVID-19 in 11

European countries, 2020.

on August 3, 2024 by guest. Protected by copyright.http://gh.bmj.com/BMJ Glob Health: first published as 10.1136/bmjgh-2020-003126 on 23 December 2020. Downloaded from

AguasR, etal. BMJ Global Health 2020;5:e003126. doi:10.1136/bmjgh-2020-003126

BMJ Global Health

5 Davies NG, Kucharski AJ, Eggo RM, etal. Effects of non-

pharmaceutical interventions on COVID-19 cases, deaths, and

demand for hospital services in the UK: a modelling study. Lancet

Public Health 2020;5:e375–85.

6 Cowling BJ, Ali ST, Ng TWY, etal. Impact assessment of non-

pharmaceutical interventions against coronavirus disease 2019 and

inuenza in Hong Kong: an observational study. Lancet Public Health

2020;5:e279–88.

7 Habib H. Has Sweden's controversial covid-19 strategy been

successful? BMJ 2020;369:m2376.

8 Rhodes T, Lancaster K, Lees S, etal. Modelling the pandemic:

attuning models to their contexts. BMJ Glob Health 2020;5:e002914.

9 Freebairn L, Rychetnik L, Atkinson J- A, etal. Knowledge

mobilisation for policy development: implementing systems

approaches through participatory dynamic simulation modelling.

Health Res Policy Syst 2017;15:83.

10 Mansnerus E. Using model- based evidence in the governance of

pandemics. Sociol Health Illn 2013;35:280–91.

11 Leach M, Scoones I. The social and political lives of zoonotic

disease models: narratives, science and policy. Soc Sci Med

2013;88:10–17.

12 Zhou F, Yu T, Du R, etal. Clinical course and risk factors for mortality

of adult inpatients with COVID-19 in Wuhan, China: a retrospective

cohort study. Lancet 2020;395:1054–62.

13 CDC COVID-19 Response Team. Severe Outcomes Among Patients

with Coronavirus Disease 2019 (COVID-19) - United States, February

12- March 16, 2020. MMWR Morb Mortal Wkly Rep 2020;69:343–6.

14 nychealth / coronavirus- data, 2020. Available: https:// github. com/

nychealth/ coronavirus- data/ blob/ master/ by- age. csv

15 Onder G, Rezza G, Brusaferro S. Case- Fatality rate and

characteristics of patients dying in relation to COVID-19 in Italy.

JAMA 2020;323:1775-1776.

16 Oke J, Heneghan C. Global Covid-19 case fatality rates, 2020.

Available: https://www. cebm. net/ covid- 19/ global- covid- 19- case-

fatality- rates/

17 Wu Z, McGoogan JM. Characteristics of and Important Lessons

From the Coronavirus Disease 2019 (COVID-19) Outbreak in

China: Summary of a Report of 72 314 Cases From the Chinese

Center for Disease Control and Prevention. JAMA 2020;323:1239-

1242.

18 Bi Q, Wu Y, Mei S, etal. Epidemiology and transmission of

COVID-19 in Shenzhen China: analysis of 391 cases and 1286 of

their close contacts. medRxiv 2020.

19 Lauer SA, Grantz KH, Bi Q, etal. The incubation period of

coronavirus disease 2019 (COVID-19) from publicly reported

conrmed cases: estimation and application. Ann Intern Med

2020;172:577–82.

20 Wei WE, Li Z, Chiew CJ, etal. Presymptomatic Transmission of

SARS- CoV-2 - Singapore, January 23- March 16, 2020. MMWR Morb

Mortal Wkly Rep 2020;69:411–5.

21 Arons MM, Hateld KM, Reddy SC, etal. Presymptomatic SARS-

CoV-2 infections and transmission in a skilled nursing facility. N Engl

J Med 2020;382:2081–90.

22 Tong Z- D, Tang A, Li K- F, etal. Potential presymptomatic

transmission of SARS- CoV-2, Zhejiang Province, China, 2020.

Emerg Infect Dis 2020;26:1052–4.

23 To KK- W, Tsang OT- Y, Leung W- S, etal. Temporal proles of viral

load in posterior oropharyngeal saliva samples and serum antibody

responses during infection by SARS- CoV-2: an observational cohort

study. Lancet Infect Dis 2020;20:565–74.

24 He X, Lau EHY, Wu P, etal. Temporal dynamics in viral shedding and

transmissibility of COVID-19. Nat Med 2020;26:672–5.

25 Li P, Fu J- B, Li K- F, etal. Transmission of COVID-19 in the terminal

stages of the incubation period: a familial cluster. Int J Infect Dis

2020;96:452–3.

26 Nishiura H, Linton NM, Akhmetzhanov AR. Serial interval of novel

coronavirus (COVID-19) infections. Int J Infect Dis 2020;93:284–6.

27 Du Z, Xu X, Wu Y, etal. Serial interval of COVID-19 among publicly

reported conrmed cases. Emerg Infect Dis 2020;26:1341–3.

28 Wang D, Hu B, Hu C, etal. Clinical characteristics of 138

hospitalized patients with 2019 novel coronavirus- infected

pneumonia in Wuhan, China. JAMA 2020;323:1061.

29 Goyal P, Choi JJ, Pinheiro LC, etal. Clinical characteristics of

Covid-19 in New York City. N Engl J Med 2020;382:2372–4.

30 Verity R, Okell LC, Dorigatti I, etal. Estimates of the severity of

coronavirus disease 2019: a model- based analysis. Lancet Infect Dis

2020;20:669–77.

31 Russell TW, Hellewell J, Jarvis CI, etal. Estimating the infection and

case fatality ratio for coronavirus disease (COVID-19) using age-

adjusted data from the outbreak on the diamond Princess cruise

SHIP, February 2020. Euro Surveill 2020;25.

32 Mizumoto K, Chowell G. Estimating risk for death from coronavirus

disease, China, January- February 2020. Emerg Infect Dis

2020;26:1251–6.

33 Winter AK, Hegde ST. The important role of serology for COVID-19

control. Lancet Infect Dis 2020;20:758–9.

34 Xiang F, Wang X, He X, etal. Antibody detection and dynamic

characteristics in patients with COVID-19. Clin Infect Dis 2020.

35 Yong SEF, Anderson DE, Wei WE, etal. Connecting clusters of

COVID-19: an epidemiological and serological investigation. Lancet

Infect Dis 2020;20:809–15.

36 Yang J, Zheng Y, Gou X, etal. Prevalence of comorbidities and its

effects in patients infected with SARS- CoV-2: a systematic review

and meta- analysis. Int J Infect Dis 2020;94:91–5.

37 Halpin DMG, Faner R, Sibila O, etal. Do chronic respiratory diseases

or their treatment affect the risk of SARS- CoV-2 infection? Lancet

Respir Med 2020;8:436–8.

38 Emanuel EJ, Persad G, Upshur R, etal. Fair allocation of scarce

medical resources in the time of Covid-19. N Engl J Med Overseas

Ed 2020;382:2049–55.

39 Wang B, Li R, Lu Z, etal. Does comorbidity increase the risk of

patients with COVID-19: evidence from meta- analysis. Aging

2020;12:6049–57.

40 MacIntyre CR, Heslop DJ. Public health, health systems and

palliation planning for COVID-19 on an exponential timeline. Med J

Aust 2020.

41 Touboul- Lundgren P, Jensen S, Drai J, etal. Identication of cultural

determinants of antibiotic use cited in primary care in Europe: a

mixed research synthesis study of integrated design "Culture is all

around us". BMC Public Health 2015;15:908.

42 Hale T, Angrist N, Kira B. “Variation in Government Responses to

COVID-19” Version 5.0. Blavatnik School of Government Working

Paper, 2020.

43 Guan Y, Deng H, Zhou X. Understanding the impact of the COVID-19

pandemic on career development: insights from cultural psychology.

J Vocat Behav 2020;119:103438.

44 Messner W. The institutional and cultural context of cross- national

variation in COVID-19 outbreaks. medRxiv 2020.

45 Shim E, Tariq A, Choi W, etal. Transmission potential and severity of

COVID-19 in South Korea. Int J Infect Dis 2020;93:339–44.

46 Karamouzian M, Madani N. COVID-19 response in the middle East

and North Africa: challenges and paths forward. Lancet Glob Health

2020;8:e886–7.

47 Jarvis CI, Van Zandvoort K, Gimma A, etal. Quantifying the impact

of physical distance measures on the transmission of COVID-19 in

the UK. BMC Med 2020;18:124.

48 Prem K, Liu Y, Russell TW, etal. The effect of control strategies

to reduce social mixing on outcomes of the COVID-19 epidemic

in Wuhan, China: a modelling study. Lancet Public Health

2020;5:e261–70.

49 Tian H, Liu Y, Li Y, etal. An investigation of transmission control

measures during the rst 50 days of the COVID-19 epidemic in

China. Science 2020;368:638–42.

50 Kucharski AJ, Klepac P, Conlan AJK, etal. Effectiveness of isolation,

testing, contact tracing, and physical distancing on reducing

transmission of SARS- CoV-2 in different settings: a mathematical

modelling study. Lancet Infect Dis 2020;20:1151–60.

51 Salje H, Tran Kiem C, Lefrancq N, etal. Estimating the burden of

SARS- CoV-2 in France. Science 2020;369:208–11.

52 Di Domenico L, Pullano G, Sabbatini CE, etal. Expected impact of

lockdown in Île- de- France and possible exit strategies. medRxiv 2020.

53 Shretta R. The economic impact of COVID-19, 2020. Available:

https://www. tropicalmedicine. ox. ac. uk/ news/ the- economic- impact-

of- covid- 19

54 Freebairn L, Atkinson J- A, Kelly PM, etal. Decision makers'

experience of participatory dynamic simulation modelling: methods

for public health policy. BMC Med Inform Decis Mak 2018;18:131.

55 Pan- Ngum W, Kinyanjui T, Kiti M, etal. Predicting the relative

impacts of maternal and neonatal respiratory syncytial virus (RSV)

vaccine target product proles: a consensus modelling approach.

Vaccine 2017;35:403–9.

56 Cornell COVID caseload calculator with capacity and ventilators

(C5V Online*), 2020. Available: https:// covid19. sjsu. edu/ C5V/ default/

user/ login?_ next=/ C5V/ default/ index

on August 3, 2024 by guest. Protected by copyright.http://gh.bmj.com/BMJ Glob Health: first published as 10.1136/bmjgh-2020-003126 on 23 December 2020. Downloaded from

BMJ Global Health 2021;6:e003126corr1. doi:10.1136/bmjgh-2020-003126corr1

Correction

Correction for modelling the COVID-19 pandemic in context:

an international participatory approach

Aguas R, White L, Hupert N, et al. Modelling the COVID-19 pandemic in context: an

international participatory approach. BMJ Glob Health 2020;5:e003126. doi: 10.1136/

bmjgh-2020-003126.

This article has a correction. The group in South Africa has been using their own

models and not the COVID-19 Modelling Consortium model. Their inclusion in the

manuscript was an error.

Open access This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported

(CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose,

provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See:

https:// creativecommons. org/ licenses/ by/ 4. 0/.

BMJ Global Health 2021;6:e003126corr1. doi:10.1136/bmjgh-2020-003126corr1