DH301: Basic Epidemiology

Dr Kalyani Addya
&
Dr Sandip Mandal

Indian Institute of Technology - Bombay 8 January 2025

Epidemiology
• Epidemiology is a branch of science that focuses on studying
the distribution, patterns, and determinants of health and
disease conditions within specific populations.
• It explores how diseases spread, why certain groups are more
affected than others, and what factors contribute to the
occurrence and outcomes of diseases.
• Epidemiologists collect and analyze data to identify the causes
and risk factors for diseases, understand how diseases are
transmitted, and assess the impact of interventions or
preventive measures.
• By studying large groups or populations, epidemiologists can
uncover trends, make predictions, and develop strategies to
prevent or control diseases, ultimately improving public health.
Main content

• Basic Epidemiology

• Mathematical Epidemiology
 Course content (Basic
Epidemiology)
• Introduction to Epidemiology – Definition & 0bjectives of epidemiology, principles and methods of
epidemiology to investigate disease distribution.
• Using epidemiology methods to study the cause, extent and prevention of various infectious and non-
infectious diseases.
• Dynamics of disease transmission: modes of transmission, attack rate, disease outbreak investigation

• Disease surveillance and measures of morbidity and mortality: Active and passive surveillance, emergency
warning systems.
• Stages of a disease, incidence and prevalence of disease, mortality rates, case fatality

• Assessing screening tests: Validity of tests, Tests with binary and continuous outcomes, sequential testing,
sensitivity and specificity measures.
• Different types of study design, including randomized trials, case-control and cohort studies, risk estimation
and causal inferences
Course content (Mathematical
Epidemiology)
• Model development
• Rates, proportion, average duration, competing hazards
• Model building
• Force of infection, reproduction number

• Model calibration
• Fitting with data

• Modelling interventions
• Treatment
• Prevention
Examination pattern and marks
division
• Basic Epidemiology – 60 Marks

• Mathematical Epidemiology – 40 Marks

No re-examination
DH301: Basic Epidemiology

Mathematical
Epidemiology
Sandip Mandal
Email: [email protected]
 Model
A simplified abstract view of the complex reality

8
Epidemic modelling
• A way of capturing natural history and transmission of disease

• A way of capturing how interventions shape transmission

• A framework for quantifying costs and epidemiological impact

Caution:
• A model is no substitute for surveillance and primary data!

• Offers a framework for

• making sense of this data, and
• Simulating potential control policies 9
Outline
• Story Ronald Ross and the first useful mathematical model
• McKendric and Kermack
• Story of Covid-19 modelling at ICMR-HQ
• Types of mathematical models
• Content to be covered in this section of the course (DH301)
• Concept of compartmental models
Understanding of data
Model development
Calibration
Intervention
• Reference books
Story of Ronald Ross
Sir Ronald Ross India chapter

• Indian Medical Service (1881)

• Bangalore life (1883)
• First experiment with mosquitoes
• ‘Mosquitoes carry malaria just as
they carry filariae’ - Patrick Manson.
• Second experiment in Bangalore
• Third experiment (on birds)
1857 - 1932
• Nobel Prize (1902)
MALARIA: Life cycle of Plasmodium parasite
Human host, Mosquito vector, Malaria parasite
The latency period of the malaria parasite
within the mosquito [extrinsic incubation
period (EIP)]: 10-21 days

The latency period of the malaria

parasite within humans, also known as
the incubation period, is the time
between the mosquito bite (infection)
and the appearance of symptoms: 9 –
40 days

The average longevity of Anopheles

mosquito: 10- 21 days

Female mosquitoes usually live

longer than males, as males typically
die shortly after mating.
Link between insects and infections
• Ross’s hypothesis - Removing mosquitoes was the key to controlling
malaria
• Practical experience – Sierra Leone (a country in West Africa), 1901
• Experience during Suez Canal Project –Ismailia (Egypt)

• This was against popular wisdom – It was impossible to get rid of

every last mosquito, which meant there would always be some
insects left, and hence potential for malaria to spread.
• Need a stronger argument to persuade Ross’s idea of mosquito
control
Was it really possible to control
malaria without removing every
mosquito?
Assumptions:
Village of 1000
population • 1 in 4 mosquitoes would manage to bite
someone
• 1 of 3 mosquitoes survived to be infectious
12,000 bites
human 12
4 Even if there were
Among all bites 1 48,000 mosquitoes in
48,000 Survives
only 12 bites the area, on average
mosquitoes to be
would be such Bites they would generate
that they can infectious another only one new human
pick up the human infection.
parasite
 Concluding analyses
• More mosquitoes/more infected human  more new infections per
month
• Process to counteracts the number  Estimated around 20% of
humans infected with malaria would recover each month
• Condition for malaria to remain endemic in the population:
Infection and recovery would need to balance each other
• If the recoveries outpaced the rate of new infections, the level of
disease eventually would decline to zero.

“Malaria cannot persist in a community unless the Anopheles are so numerous that the number of new
infections compensates for the number of recoveries.” – The Prevention of Malaria (1910)
 Ross model (1911)
Human

Susceptible
Infected (Ih)
(Sh)

Mosquito

Susceptible
Infected (Im)
(Sm)

16
 Two ways to approach disease
analysis
Descriptive methods
• This involved starting with real life data and working
backwards to identify predictable patterns.
• Example: William Farr’s analysis of London smallpox
outbreak, in 1830 and in 1840.
• Farr’s method focused what shape epidemics take,
not why they take the shape.
Mechanistic methods
• This started outlining the main processes that influenced transmission.
• Example: Ross’s analysis by applying the knowledge of malaria transmission.
• Ross’s conceptual model of transmission using mathematical equations, to make
conclusions about likely outbreak patterns. It can answer ‘what if’ without doing
real experiments.
McKendrick and Kermack: Mathematics
of disease
Anderson Gray McKendrick William Kermack

1876 - 1943 1898 - 1970

Extended Ross’s ideas to look at epidemics in general
Mathematical model of disease
transmission
What causes epidemics to end?

Two popular explanations:

• Transmission ceased because there were no susceptible people left to infect
• Pathogen itself became less infectious as the epidemic progressed

Neither explanation was correct

Day 1 After one week After one month

20
Day 1 After one week After one month
21
Mathematical model of disease
transmission
Compartments
according to Simulated outbreak using SIR model
disease status

Susceptible

Infectious

Recovered
“An epidemic, in general comes to an end before
the susceptible population has been exhausted” –
SIR Model Kermack and McKendrick
COVID-19 modelling
Some practical application
Novel Coronavirus (COVID-19)

• The first known cases were identified in Wuhan, China, in December

2019.

• This was spreading rapidly in various European countries and USA

• Many unknowns - transmissibility (R0), infectiousness

24
 TIME LINE
1st National
sero survey 2nd (Aug-Sep 3rd (Dec 2020- 4th (Jun-
(May-Jun 2020) 2020) Jan 2021) July 2021)

(7 May 2021)
(25 Jan 2022)

(16 Sep 2020)

Unlock start Vaccination began

First report (1 June 2020) (16 Jan 2021)
(30 Jan 2020)
Lockdown start
(25 Mar 2020)
With limited data
(Early February, • How much reduction is possible
2020) by isolating infected persons?
• Would the thermal screening
be effective?

Reduction of Optimistic Pessimistic

scenario scenario
Peak 89% 8%
prevalence
Cumulative 62% 2%
incidence

Intervention: 50% of symptomatic cases

Ref: Mandal el al. Indian Journal of Medical Research, 2020
are isolated within 3 days
27
28
 TIME LINE
1st National
sero survey 2nd (Aug-Sep 3rd (Dec 2020- 4th (Jun-
(May-Jun 2020) 2020) Jan 2021) July 2021)

(7 May 2021)
(25 Jan 2022)

(16 Sep 2020)

Unlock start Vaccination began

First report (1 June 2020) (16 Jan 2021)
(30 Jan 2020)
Lockdown start
(25 Mar 2020)
With seroprevalence
data • Model calibration with
• Strategy for removing lockdown seroprevalence: estimation
of number of cases
Aligning the data with reality:

8% seroprevalence in
mid-September 2020
Ref: Mandal el al. Scientific Reports, 2021
31
 November- December
2020

Vaccination strategies: priority groups

 TIME LINE
1st National
sero survey 2nd (Aug-Sep 3rd (Dec 2020- 4th (Jun-
(May-Jun 2020) 2020) Jan 2021) July 2021)

(7 May 2021)
(25 Jan 2022)

(16 Sep 2020)

Unlock start Vaccination began

First report (1 June 2020) (16 Jan 2021)
(30 Jan 2020)
Lockdown start
(25 Mar 2020)
Vaccination strategy against COVID-19
amongst different priority groups

Model findings
• Keyworkers  Over- 60s  with comorbidities (relatively high transmission setting)

• Keyworkers  with comorbidities  Over- 60s (In settings with weakest transmission, e.g.
sparsely-populated rural areas)
Ref: Mandal el al. BMJ Open, 2021
 March- April 2021

Vaccination strategies: area prioritisation

How can vaccination efforts be best
prioritised in order to help those areas of
the country that need it most?
(A) Time course of test positivity rate
between the first and second
epidemic waves in Delhi

(B) Model projections for mortality

during the second wave in a
hypothetical district in India, where
25% of the population has prior
immunity, with = 2, and under three
different scenarios

Model findings
Flexible, agile vaccination strategies could play an important role in protecting lives and
livelihoods as the COVID-19 pandemic continues to unfold.

Ref: Mandal el al. The Lancet Global Health, 2021

 April - June 2021

Plausibility of third wave

Third wave
concerns
 TIME LINE
1st National
sero survey 2nd (Aug-Sep 3rd (Dec 2020- 4th (Jun-
(May-Jun 2020) 2020) Jan 2021) July 2021)

(7 May 2021)
(25 Jan 2022)

(16 Sep 2020)

Unlock start Vaccination began

First report (1 June 2020) (16 Jan 2021)
(30 Jan 2020)
Lockdown start
(25 Mar 2020)
Some hypotheses for a severe third
wave
1. Restrictive
3. Viral evolution for increased
measures/release
transmission
If restrictions have been successful at
reducing transmission, releasing them Emergence of a new virus with a higher
may afford new opportunities for than previously
transmission

2. Full waning of immunity 4. Viral evolution for immune

Independent of the virus, any acquired escape
immunity in the population so far
decays over time Emergence of a new virus that can fully
escape prior immunity

40
One example scenario of third wave of COVID-19 in India
R0 = 3 and 25% immune escape

Assumptions:

1. Date of emergence of
variant (mid September
2021)
2. R0 = 3 and 25% immune
escape
3. Proportion symptomatic
among all cases (33%-
66%)
4. Vaccine efficacy on
reducing severity (60%)

Ref: Mandal el al. Indian Journal of Medical Research, 2021

Conclusions
 June-July 2021

Local projection with local data

44
45
Lesson learned

• Modelling is an useful tool if used properly

• Model outcomes depend on assumptions

• Always important to know why a model suggests certain conclusions

• Better to think of a model as bringing together our best

understanding of an epidemic and Projecting their implications under
given future scenarios

• ‘Project’ rather than ‘predict’

46
Spatial and temporal scales of
disease dynamics

ENSO
(El Niño
Southern
Oscillation)
Modelling disease prevalence
Mechanistic way of modelling Data-based modelling
Mathematical & Computational Mathematical & Statistical
Modelling Modelling
Discrete & continuous Analyse the prevalence data
generation ecological and of a parasitic disease to
epidemiological models to search for patterns of
understand general features of occurrence under the
eco-epidemiological and influence of environmental
demographic influences variables.

The final aim is to unify model studies with real

observation
 Example of data-based Statistical
Modelling Mangalore
Chennai

Malaria incidence and climate data.

(a) SPR (%) values in Chennai, during January 2002 - December 2004, and
(c) TMC values in Mangalore, during January 2003 - December 2007. The climate variables, Rainfall (cm), Temperature (°C),
Humidity (%), and Population (in millions) are appropriately scaled to fit in the same plot
Mechanistic approach of modelling
Different types of theoretical approaches are used in
epidemiological modelling
1) Standard differential equation based SIR models
2) Discrete models using generic host parasite population
growth models
3) Agent-based models
4) Network-based models – SIR or individuals
5) Lattice based models – continuous or discrete
 Reference Books

Mathematical Epidemiology of Infectious Diseases: Model Building,

Analysis and Interpretation. By Diekman and Heesterbeek. Wiley

Modeling Infectious Diseases in Humans and Animals. By Matt J.

Keeling and Pejman Rohani. Princeton University Press

Mathemtical Models in Biology. By Leah Edelstein-Keshet. SIAM

Classics in Applied Mathematics 46

Mathematical Biology I and II. J D Murray. Springer

Mathematical Models in Molecular and Cellular Biology. Lee Segel.

Cambridge University Press
Thank you for your
patience