Introduction to Fundamentals of

Epidemiology (FE01)
EPM101 Fundamentals of Epidemiology
Course: PG Diploma/ MSc Epidemiology

This document contains a copy of the study material located within the computer
assisted learning (CAL) session.

If you have any questions regarding this document or your course, please contact
DLsupport via DLsupport@[Link].

Important note: this document does not replace the CAL material found on your
module CDROM. When studying this session, please ensure you work through the
CDROM material first. This document can then be used for revision purposes to
refer back to specific sessions.

These study materials have been prepared by the London School of Hygiene & Tropical Medicine as part of
the PG Diploma/MSc Epidemiology distance learning course. This material is not licensed either for resale
or further copying.

© London School of Hygiene & Tropical Medicine September 2013 v1.3

Section 1: Introduction to Fundamentals of Epidemiology
Aim
To introduce the discipline of epidemiology

Objectives
After completing this session, you should be able to:

explain what epidemiology is and understand the concept of population in

epidemiology

identify an exposure and an outcome in any given study

explain the difference between descriptive and analytic studies, and between
observational and interventional epidemiology

understand the importance of epidemiology to public health practice

appreciate the way in which an epidemiological investigation is carried out

express delight and enthusiasm for this study module and epidemiology in general,
and rush off to start the course immediately

This session should take you about 1 to 2 hours to complete.

Section 2: What is Epidemiology?

Epidemiology is often in the news. Click below to see!

Interaction: button: Show Me

Output: (these topics play in a ‘slideshow’ fashion at about 5 second intervals, with
respective headline graphics appearing on the right)

1. Are mobile phone users exposing their brains to dangerous levels of

radiation?
2. Does eating British beef really cause Creutzfeld-Jacob disease (CJD)?

3. How can we control Legionnaires' disease?

4 Could a daily diet of prunes prevent you from getting cancer?

Interaction: button: © button

Output (pop up):

Williams A. Prunes every day could help keep cancer away.
The Mirror 1999 March 15; p.17
Anon, This is how a mobile phone heats your brain.
Sunday Mirror 1999 March 7; p.1
Anon, Killer that lurks in the showers.
Evening Post 1988 July 26
Anon, Mad cow can kill you.
Daily Mirror 1996 March 20

2.1: What is Epidemiology?

Epidemiology is a key part of public health.

According to Last's Dictionary of Epidemiology, epidemiology is:

"The study of the distribution and determinants of health-related states or events in

specified populations and the application of this study to control of health problems."

The focus of this study module is primarily the contribution of epidemiology to public
health medicine. However, epidemiology also plays a role in clinical medicine and in
pharmacological studies. These applications of epidemiology are not covered in detail
in EP101, although the concepts and methods we discuss apply equally to studies in
these fields.

Epidemiologists often ask "what is the denominator?" - meaning, what is the

population from which the cases arose?

By counting the number of health-related events which occur within a specified time
in different populations we can make estimates of the frequency of these events,
which we can compare in different populations.

For example, we can compare the infant mortality rate in different countries and
at different times.

In England and Wales the infant mortality rate was 6.6 per 1000 live births in 1992.
The corresponding figure for Brazil was 36 per 1000 live births.

2.2: What is Epidemiology?

There is a difference between the concept of 'population' in epidemiology and
demography.

In epidemiology, we define our population as the collection of units from which our
sample is drawn. This could be people, institutions or events.

In demography, we use the term 'population' to mean the inhabitants of a particular

country or area.

2.3: What is Epidemiology?

Comparison is fundamental to epidemiology.

By examining the differences between those people within a particular population

who have a disease and those who do not, we can improve our understanding of the
determinants of a disease.

For example, in Vienna, in the 1840s, the obstetrician Ignaz Semmelweis noticed
that many women were dying of puerperal fever.

He noticed that there were major differences in the maternal mortality rate from
puerperal fever between two clinics within one hospital. He then looked for
differences between the two clinics which could account for this.

The clinic with the high mortality rate was used to train medical students, and the
clinic with the lower mortality rate was used to train midwives.

Semmelweis observed that, as part of their training, medical students performed

genital examinations far more often than midwives. Hand-washing was not routine,
and Semmelweis suggested that the students' unwashed hands transmitted infection
from woman to woman.

He tested his hypothesis by introducing hand-washing with an antiseptic at the first

clinic: the maternal mortality rate subsequently fell to below that in the clinic that
trained midwives.

2.4: What is Epidemiology?

The fundamental things we need to know about a disease in a population are who,
where, and when. In other words, we are interested in describing the distribution of
a disease in terms of person, place and time.

For example, in describing HIV infection in Uganda, we might say:

"The prevalence of HIV infection among pregnant women of all ages in

Kampala, Uganda in 1995 was 20%."

Click the "show" button below to highlight the person, place and time in the
description above.

Interaction: button: Show (bold, bold italic, & bold underline styles indicate correct
words below)

"The prevalence of HIV infection among pregnant women of all ages in Kampala,
Uganda in 1995 was 20%."

Now identify the parts of this description of leukaemia in northern England which
specify person, place and time:
"The incidence of lymphoblastic leukaemia among children aged under 15 years
in Northumberland and Durham (two counties in northern England) between
1951 and 1960 was 3.1 per 100 000 person-years."

Exercise

Person correct response output:

Children under 15 years of age

Place correct response output:

Northumberland and Durham

Time correct response output:

1951 and 1960

For all: incorrect response output:

Try again.

2.5: What is Epidemiology?

The word derives from the Greek word epidemos - epi meaning "upon" and demos
meaning "people".

The ancient Greek physician Hippocrates (born around 460 BC) used the word
epidemeion to refer to diseases that visit the community, meaning they occurred
from time to time, in contrast to endemeion, diseases which resided within the
community. The expression "endemic disease" comes from this word.

The earliest use of the word "epidemiology" was probably in Spain: in 1598, a book
about plague called Epidemiología was already in its second edition.

The first documented appearance of the English form of the word was in 1850 when
the London Epidemiological Society was formed.

2.6: What is Epidemiology?

Hippocrates described the distribution of diseases by season, age, climate and
personal behaviour - which is much the same perspective as modern epidemiologists
have.

"Whoever wishes to pursue properly the science of medicine should proceed thus.
First he ought to consider what effect each season of the year can produce … He
must consider … how the natives are off for water, whether they use marshy, soft
waters, or such as are hard and come from rocky heights, or brackish and harsh. The
mode of life also of the inhabitants… whether they are heavy drinkers, taking lunch
and inactive, or athletic, industrious, eating much and drinking little."

[Hippocrates: On Airs, Waters and Places]

2.7: What is Epidemiology?

In 1839, a physician called William Farr was made responsible for medical statistics
in the Registrar General's office for England and Wales. He set up a system for the
routine collection of data on the number and causes of deaths and other vital
statistics.

Farr used his data to compare the mortality rates among different populations. For
example, he compared mortality rates among people of different occupations. The
graph below shows his figures comparing the mortality rates between clergymen and
publicans (men who ran public bars), subdivided by age-group.

2.8: What is Epidemiology?

This is an example of descriptive epidemiology.

It is also an early example of the use of routine data sources - data about health-
related events which are collected systematically. Routine data sources available in
many countries today include census and population registers, birth certificates,
death certificates, cancer registries, notifications of infectious diseases and registers
of congenital abnormalities.
Farr recognised that these data could be used to help understand diseases. He
wrote:

"Different classes of the population experience very different rates of mortality … the
principal causes of these differences, besides the sex, age and hereditary
organization, must be sought in three sources: ordinary occupations of life, supply of
warmth and of food, and differential exposure to poisonous effluvia and to
destructive agents" (Farr, 1885)

Section 3: Historical Evolution of Epidemiology

Another important figure in the history of epidemiology was John Snow, an
anaesthetist in London in the 19th century who is famous for having administered
chloroform to Queen Victoria at the births of two of her children. However he is
better known to epidemiologists for his studies of cholera in London.

During the 19th century, cholera periodically swept across Europe. London was
rather different then - there was no electricity, there were cows and horses in the
streets, and standards of hygiene were poor.

In 1848-9 there was a severe cholera epidemic in London, with 15,000 recorded
deaths. There was considerable debate about the cause of cholera - at the time, the
existence of microbes was hotly contested.

3.1: Historical Evolution of Epidemiology

William Farr, a different epidemiologist, had his own ideas on this. He noticed that
some areas of London had far more cholera deaths than others (see first graph
below ), and he wondered what factors might explain this variation. He found that
there were more deaths from cholera in the lower areas near the river (see second
graph below).

Farr thought that cholera might be caused by breathing unclean air at lower
altitudes.

3.2: Historical Evolution of Epidemiology

During the 19th century, drinking water was supplied to houses by private
companies via a network of pipes. There was competition between the water
companies, resulting in overlap between the areas supplied by the different
companies.

Toilets were widely introduced in London between 1830 and 1850: main sewers were
introduced in the 1840s. The sewers emptied into the River Thames.

During the 1848-9 cholera epidemic, mortality from cholera was particularly high in
the districts supplied by two particular water companies, the Southwark & Vauxhall
and the Lambeth companies, both of which took water direct from the River Thames
where it flowed through London.

Click below to view part of the original map of this area, including an area where
both companies supplied water. The green area was served by the Southwark and
Vauxhall company, the pink area by the Lambeth company, and the grey area in
between is where the two companies' pipes were intermingled.

Interaction: hyperlink: Map

Output (pop up):

3.3: Historical Evolution of Epidemiology
Based on these observations, John Snow formulated an hypothesis on the nature and
mode of communication of cholera:

that cholera can be communicated

from the sick to the healthy
that disease is communicated by
"morbid matter" which has the property
of multiplying in the body of the person
it attacks
that the morbid matter producing
cholera must be introduced into the
alimentary canal
water supplies appeared to be able to
disseminate the morbid matter from the
sick to the healthy

There was no further cholera in London until 1853. During this period, the Lambeth
company moved its source so that it now drew water from the Thames upstream of
London (and thus uncontaminated by London sewage). The Southwark and Vauxhall
company continued to draw water from the Thames in London. This provided John
Snow with an ideal opportunity to test his hypothesis.

3.4: Historical Evolution of Epidemiology

Cholera reappeared in London in June 1853. John Snow asked permission to obtain
(from William Farr) the addresses of people who died of cholera in the districts which
received water supplies from both the Southwark and Vauxhall and the Lambeth
companies.

He then visited the homes of all recorded cholera deaths in these districts, to get
information about which company supplied water to the household. The table shows
his results for the first 334 deaths.

On the basis of these figures, which company is more likely to be transmitting

"morbid matter" causing cholera? Think about this for a moment, and write down the
reason for your answer before you continue.

Interaction: hotspot: Southwark and Vauxhall

Output (pop up):

It looks as if this may be so, but we don’t have enough information to be

sure. Although there were more deaths in houses supplied by the
Southwark and Vauxhall company, this could just be because this company
supplied water to a greater number of people. In order to know whether
one company is more likely to transmit contaminated water, we need to
know, for each company, the number of deaths as a proportion of the
number of people supplied with water.

Interaction: hotspot: Lambeth

Output (pop up):

In fact we don’t have enough information to answer this question. Although

there were more deaths in houses supplied by the Southwark and Vauxhall
company, this could just be because this company supplied water to a
greater number of people. In order to know whether one company is more
likely to transmit contaminated water, we need to know, for each company,
the number of deaths as a proportion of the number of people supplied with
water.
Interaction: hotspot: Can’t tell

Output (pop up):

That’s right. Although there were more deaths in houses supplied by the
Southwark and Vauxhall company, this could just be because this company
supplied water to a greater number of people. In order to know whether
one company is more likely to transmit contaminated water, we need to
know, for each company, the number of deaths as a proportion of the
number of people supplied with water.

3.5: Historical Evolution of Epidemiology

We might just note that John Snow did a very thorough job - he managed to get
information from 330 out of the 334 households. He did this by going from house to
house to make enquiries. Today, this type of investigation is often called "shoe-
leather epidemiology", meaning that it involves so much walking that your shoes
may wear out!

In order to estimate the denominator, Snow obtained information on the number of

houses in London whose water was supplied by each of the two water companies.

Snow noted the source of water in the houses of all those who died of cholera from
8th July to 26th August 1854, as shown in this table:

3.6: Historical Evolution of Epidemiology

Take a few minutes to think about whether there are problems with using the
number of houses supplied by each company as a denominator, then click on the
button below for our interpretation.

Interaction: button: ‘thought bubble’ button

Output:

Ideally the denominator should be all people supplied by water from each company.
Different houses may contain different numbers of people. If, for example, the
Lambeth company supplied an area where there were consistently more people per
house than the area supplied by the Southwark and Vauxhall company, then we
might be misled by using the number of houses as a denominator.

However for this example we will assume there was no difference in the number of
people in each house in the areas supplied by the two water companies.

End interaction.

How much more dangerous was it to drink Southwark and Vauxhall water than
Lambeth water (to the nearest whole number)? Click below if you need a hint.

It was

Interaction: calculation:

Incorrect output:

The number of deaths per house supplied was:

for Southwark and Vauxhall, 1263 / 40046
= 31.5 per 1000 houses
for Lambeth, 98 / 26107
= 3.8 per 1000 houses
So it was 31.5 / 3.8 = 8 times more dangerous to drink Southwark and Vauxhall
water than Lambeth water.

Correct output:

There were 31.5 cholera deaths per 1000 houses supplied by Southwark and
Vauxhall, and 3.8 cholera deaths per 1000 houses supplied by Lambeth, so it was
31.5 / 3.8 = 8 times more dangerous to drink Southwark and Vauxhall water than
Lambeth water.

End interaction.

times more dangerous to drink Southwark and Vauxhall water than Lambeth
water.

Interaction: button: Hint

Output:

The best way to estimate this is to calculate, for each company, the number of
cholera deaths per house supplied, and then compare the two figures. This could be
expressed as the number of times more deaths per household in the Southwark and
Vauxhall area than the Lambeth area.
3.7: Historical Evolution of Epidemiology
So John Snow started with descriptive epidemiology, obtaining information on the
numerator (the number of cholera deaths) and the denominator (the number of
people supplied with water) so that he could describe the number of cases of cholera
in different areas in relation to the size of the population at risk.

He then went on to compare the death rates from cholera in the two areas. He used
this information to calculate how much more risky Southwark and Vauxhall water
was compared to Lambeth water. In other words, he looked for an association
between water source and the risk of death from cholera.

We call this analytical epidemiology meaning studies in which we attempt to

explain differences in patterns of disease by examining associations, and identifying
possible causes of the disease.

We shall return to John Snow's investigations later.

Section 4: Exposures and Outcomes

The two key elements that we measure in most epidemiological studies are the
exposure and the outcome.

• The exposure is the process by which an agent comes into contact with a person or
animal in such a way that the person or animal may develop the relevant outcome,
such as a disease

• The outcome is the disease, or event, or

health-related state, that we are
interested in.

An exposure can be any factor that may influence the outcome.

Take a moment to think of some exposures that may be relevant to whether or not
an individual develops lung cancer, then press the button to compare with our list.

Interaction: button: ‘thought bubble’ button

Output:

• Smoking habits
• Exposure to asbestos
• Doctor's advice on smoking
• Cultural background
• Region of residence
• Socio-economic class
• Government legislation on tobacco advertising
• Price of cigarettes
- and these are just a few examples - you may have thought of others which are also
correct.

4.1: Exposures and Outcomes

The outcome can be any health-related event or state - or it can be a risk factor
for, or a precursor to, a disease.

Example

In a study of the effect of cigarette smoking on lung cancer, cigarette smoking is the
exposure, or risk factor, that we are interested in, and the outcome is lung cancer.

However, in a study of the effect of cigarette advertising on smoking, the exposure

of interest is cigarette advertising, and cigarette smoking is the outcome.

4.2: Exposures and Outcomes

In a study of the influence of advice from a midwife on whether a mother breast-
feeds or not, is breast-feeding the exposure or the outcome?

Interaction: hotspot: Exposure

Output:

No, in fact the exposure that we are interested in is the advice from the midwife. We
would then examine whether or not the mother breast-fed her baby. So here
breast-feeding is the outcome.

End interaction.

Interaction: hotspot: Outcome

Output:

Yes, that’s right. In this example, the exposure is the midwife’s advice, and the
outcome is breast-feeding.

End interaction.

In a study of the effect of breast-feeding on diarrhoeal disease in childhood, is

breast-feeding the exposure or the outcome?

Interaction: hotspot: Exposure

Output:

That’s right. Here breast-feeding is the exposure, and diarrhoeal disease is the
outcome.

End interaction.

Interaction: hotspot: Outcome

Output:

This time we are interested in the effect that breast-feeding has on whether a child
gets diarrhoea. So here breast-feeding is the exposure and diarrhoeal disease is the
outcome.

End interaction.

4.3: Exposures and Outcomes

As we mentioned, the outcome need not be a disease. Some further examples of
outcomes which are not diseases include:

growth

intelligence

multiple pregnancy

fertility

In John Snow's investigation of cholera in 1852 (see section 3, what was the
exposure of interest and what was the outcome?

Interaction: drag & drop:

Options:

Cholera deaths
Cases of cholera
Area of residence
Elevation
Water source

to

Exposure
Outcome
None of the above

Correct responses:

Water Source  Exposure

Cholera deaths  Outcome
Area of residence  None of the above
Cases of cholera  None of the above
Elevation  None of the above

Section 5: Observational and Interventional Epidemiology

So far we have discussed one major type of epidemiological investigation:
observational epidemiology in which we describe patterns of health and disease
of a population, without doing anything to change the factors which influence them.

Observational epidemiology includes both descriptive and analytical studies.

5.1: Observational and Interventional Epidemiology

Observational epidemiology has its complications. In the real world, there may be
more than one possible risk factor for an outcome. Inconveniently, these factors may
be related to each other, and this can cause confusion.

Unless we take this problem into account, we cannot be sure that an association
observed between an exposure and an outcome is genuine, and not the consequence
of a third factor.

We will spend several sessions in this study module (FE09 - FE12) discussing how
to identify and avoid these and other potential pitfalls in epidemiological studies.

Interaction: tabs: Example 1

For example, William Farr found an association between elevation above sea level
and the risk of death from cholera. This association was because it so happened that
people living in the lower areas near the river were more likely to be supplied with
contaminated water.

So the apparent association between elevation above sea level and cholera was in
fact caused by differences in water supply, which happened to be related to the
elevation above sea level.

Interaction: tabs: Example 2

As another example, if we did a study of the effect of drinking alcohol on lung

cancer, we would find that there was indeed an association. This is because cigarette
smokers are often also drinkers of alcohol. We know that cigarette smoking causes
lung cancer.

So our finding of an association between drinking alcohol and lung cancer is due to
the cigarettes smoked by our subjects who drink alcohol, rather than due to the
alcohol itself.

5.2: Observational and Interventional Epidemiology

Another person whose name appears in the hall of epidemiological fame is James
Lind.
He was a physician in the early 18th century, when scurvy was a major problem
among sailors on long sea voyages. At that time, the cause of scurvy was not known.
Bad air, congenital laziness and indigestible food were all suggested as possible
causes. Lind observed that the sailors' diet was very poor, consisting of biscuits and
salted fish or meat.

In 1747 he conducted an experiment at sea with 12 patients suffering from scurvy.

Lind divided the patients into six groups of two. All received the same basic diet, in
addition to which he assigned different supplements to each group over a six day
period. Click on each of the supplements listed below to find out what he discovered.

Interaction: hyperlink: A quart of cider daily

Output (pop up):

The sailors that received the cider showed no improvement in their condition. Try the
other supplements.

Interaction: hyperlink: Elixir vitriol

Output (pop up):

"Elixir vitriol" means sulphuric acid mixed with spices, and was the standard
treatment for scurvy at the time. However, this group of sailors showed no change in
their condition.

Interaction: hyperlink: Vinegar

Output (pop up):

Patients who were given the vinegar supplement showed no improvement in their
condition.

Interaction: hyperlink: An “electuary”

Output (pop up):

An "electuary" is a medicine composed of various ingredients. In this case it was

recommended by a hospital surgeon as a cure for scurvy, and was made from garlic,
mustard seed, charlock, balsam of Peru and gum myrrh.

However there was no change in the condition of the patients who took this
supplement.

Interaction: hyperlink: Sea water

Output (pop up):

Sea water was another standard treatment for scurvy in 1747, but Lind's group
showed no improvement as a result of taking it.

Interaction: hyperlink: Two oranges and a lemon daily

Output (pop up):

Lind found that the two sailors receiving the oranges and the lemon improved
rapidly, whereas there was no change in the condition of the others.

5.3: Observational and Interventional Epidemiology

Lind's experiment is an early example of interventional epidemiology (also known
as experimental epidemiology). Lind divided his population into groups and allocated
different treatments to each. Effectively Lind allocated a specific exposure (type of
food supplement) to each group and then observed the outcome (whether or not
scurvy improved).

What was particularly important was that he had comparison or "control" groups of
patients who did not receive the intervention of interest, which meant that he could
compare the outcome in those who received the intervention to those who did not.

Does Lind's experiment prove that oranges and lemons cure scurvy? Write down
some reasons for your answer before you click on the button below.

Interaction: button: Yes

Output:

Not quite. The allocation to different treatments was not random, so we cannot be
sure that all the patients were all equally unwell at the start of the experiment. Also
the numbers in each group were very small, and in these circumstances, pure
chance may play a major role. Lind was correct, but we would not design a clinical
trial in quite the same way today!

Interaction: button: No

Output:

That’s right. The allocation to different treatments was not random, so we cannot be
sure that all the patients were equally unwell at the start of the experiment. Also the
numbers in each group were very small, and in these circumstances, pure chance
may play a major role. Lind was correct, but we would not design a clinical trial in
quite the same way today!

5.4: Observational and Interventional Epidemiology

More recently, the method of randomised trials was introduced, in which
participants are randomly assigned to receive the intervention or an alternative,
either a placebo or the standard treatment for the condition in question, which acts
as a control.

Randomised controlled trials were first used in the 1950s – an early example was in
the investigation the effect of streptomycin in the treatment of tuberculosis. This
method has been increasingly used to evaluate new treatments and interventions
since then.

5.5: Observational and Interventional Epidemiology

Another category of intervention study is the quasi-experimental study. This is
where the allocation of the intervention does not happen randomly.

For example, a comparison of the incidence of measles before and after the
introduction of measles vaccine would be a quasi-experimental study.

We will discuss the role of chance in epidemiological studies in the statistics with
computing module, intervention studies in session FE17, and how to calculate
sample sizes in the practical epidemiology module.

(Centre remains static throughout 6)

Section 6: What is the Role of Epidemiology?

Epidemiology has three major functions:
to describe patterns of health and disease within populations
to interpret these differences
to apply our results to public health practice, and
to evaluate the effect of
health-related interventions
The first of these functions is to describe differences in the distribution of health and
disease within and between populations.

With descriptive epidemiology, we can measure the burden of illness within a

population.

For example, we might use descriptive epidemiology to examine how the birth-
weight of babies has changed in a particular country over the last 50 years, or to
describe differences in the prevalence of hypertension between men and women, or
to compare the incidence of tuberculosis in different countries.

6.1: What is the Role of Epidemiology?

The second function is to interpret the differences we have described.

With analytical epidemiology, we can investigate risk factors for a disease or an

outcome. Here we ask the question "does the pattern of exposure to certain risk
factors among individuals with or without a specific disease help us to work out the
cause of the disease?"

For example, click below to compare maps of the global prevalence of hepatitis B
infection and of the incidence of primary liver cancer. Notice the similarity in the
distribution of the high risk areas, suggesting that the two diseases are associated.

Interaction: button: View map

Output (pop up):

Liver cancer incidence, 1990 estimates

Interaction: button: Swap Maps (in pop up)

Output (pop up):

Hepatitis B prevalence, 1997 estimates

Interaction: button: Source

Adapted from: World Health Organisation. World Health Report 1996. Geneva:
WHO; 1996. p. 61.

6.2: What is the Role of Epidemiology?

We can do similar analyses using data taken from individuals rather than whole
populations. For example, we might look at the effect of different levels of dust
exposure on the risk of developing industrial lung disease.

However, we must be careful in how we interpret our findings: in analytical

epidemiology, we measure associations between exposures and outcomes. If we
demonstrate an association, that does not necessarily mean that the exposure
caused the outcome.

Click below for an example.

Interaction: button: Example

Output (pop up):

The rate of heart attack varies considerably by geographical region, even within a
single country. The rate of heart attack is higher in Scotland than in England. This
does not necessarily mean that living in Scotland causes heart attacks: it is more
likely that people living in Scotland are exposed to other factors which are more
direct causes of heart disease.

6.3: What is the Role of Epidemiology?

Later in the course (FE13) we will discuss how to judge whether an association is
likely to be causal.

In some cases we can suggest appropriate public health action even if we do not
know the precise biological cause of a disease.

Interaction: tabs: Example 1

We can reduce the risk of lung cancer by not smoking cigarettes, even if we do not
know exactly what it is in cigarettes that causes cancer.

Tabs: Example 2

We can reduce the risk of getting diarrhoea from contaminated water by boiling it,
even if we do not know exactly which pathogen in the water causes the disease.

6.4: What is the Role of Epidemiology?

The third major function of epidemiology is to apply our results, and then to
evaluate the effectiveness of interventions and strategies of health-care delivery.

With observational epidemiology, we can assess the impact of health services on the
health of a community. Click below for an example.

Interaction: button: Example

Output (pop up):

We could look at the differences in maternal mortality rates in communities with and
without access to antenatal care.

With interventional epidemiology, we can assess the effect of a specific health

intervention.

We can use individually-randomised controlled trials (where each participant is

randomised separately, as an individual) to compare a new treatment with an
established one.

Click below for an example.

Interaction: button: Example

Output (pop up):

The outcome of an operation using laparascopic surgery (the use of fibreoptic

technology to enable operations to be performed through a very small incision) could
be compared to conventional surgery by means of an individually-randomised
controlled trial. Each individual patient would be randomised to have either
conventional or laparascopic surgery.

6.5: What is the Role of Epidemiology?

Sometimes it may be more suitable to randomise at the level of the community
rather than the individual.

For example, in a study in Tanzania to investigate the effect of improved services for
treatment of STIs on the incidence of HIV infection, communities rather than
individuals were randomised.

So in this case, a number of suitable villages were selected, and then each village
(with all its inhabitants) was randomised to receive either improved STD services or
the usual services.

We will discuss intervention studies in session FE17.

6.6: What is the Role of Epidemiology?

The challenge of epidemiology is that we do not make these comparisons in a highly-
controlled environment, such as a laboratory. In a real life situation we work with
real populations and have to compare exposures and outcomes in situations where
there may be many sources of confusion and error. This is a particular problem of
observational studies.

An important part of the process of learning about epidemiology is to understand

how such mistakes may arise and to know how to minimise them. We will return to
these issues later in the course.

Section 7: Types of Epidemiological Investigation

We can summarise the main types of epidemiological investigation in this diagram.
Click on any of the white boxes in the "family tree" for a definition of that type of
study, and an example.
Interactive: buttons: Observational

Output:

A study in which the distribution of an exposure and/or an outcome are examined

without any attempt by the investigator to influence them.

Interactive: buttons: Interventional

Output:

A study designed to test a hypothesis by modifying an exposure within the study

population.

Interactive: buttons: Aggregated Data

Output:

Data concerning exposures and outcomes in which the unit of analysis is the
population rather than individuals.
Interactive: buttons: Individual-based Data

Output:

Data concerning exposures and outcomes derived from individuals within the
population.

Interactive: buttons: Randomised Controlled Trial

Output:

A study in which individuals are randomly assigned to receive the exposure under
investigation or to be a control who does not receive the exposure. See FE17.

Example: in a randomised controlled trial to investigate the effect of pneumococcal

polysaccharide vaccine in HIV-infected people, participants were randomly assigned
to receive the active vaccine or to be a control (no vaccine).

Interactive: buttons: Quasi-experimental

Output:

A study where the allocation of the intervention does not happen randomly

For example, a comparison of the incidence of measles before and after the
introduction of measles vaccine would be a quasi-experimental study.

Interactive: buttons: For example, disease mapping

Output:

For example, we might examine data on the overall prevalence of hepatitis B

infection and the incidence of primary liver cancer in different countries rather than
in individual people.

Interactive: buttons: Ecological Study

Output:

A study in which data on average exposure and outcome for a population are used to
compare with similar data for other populations in order to look for associations
between the exposure and the outcome.

For example, in an ecological study of the effect of circumcision on susceptibility to

HIV infection, the prevalence of circumcision and the prevalence of HIV infection in
different populations would be compared in order to look for an association. See
FE18.

Interactive: buttons: Cross-sectional Study

Output:(for cross-sectional study under analytic)

A study in which the prevalence of an exposure and/or an outcome are measured in
a given population at a specified point in time. The data may be analysed to look for
an association between the exposure and the outcome. See FE14.

Example: the prevalence of onchocerciasis in a particular community could be

determined by a descriptive cross-sectional survey. If data on possible risk factors
for onchocerciasis are collected from the same people at the same time then they
could be used in an analytic study to look for associations between the risk factors
and the disease.

Output:(for cross-sectional study under descriptive)

A study in which the prevalence of an exposure and/or an outcome are measured in

a given population at a specified point in time. See FE14.

Example: the prevalence of onchocerciasis in a particular community could be

determined by a descriptive cross-sectional survey.

Interactive: buttons: Cohort Study

Output:(for cohort study under analytic)

A study in which one or more groups of individuals are followed up over a period of
time to determine the frequency of a particular outcome in the group(s).

For example, in a cohort study, the effect of smoking on fatal diseases was studied
by determining the smoking habits of a large group of doctors, and then following
them up to determine the causes of death. The frequency of specific causes of death
(such as lung cancer and heart disease) could then be compared between smokers
and non-smokers.

Output:(for cohort study under descriptive)

A study in which one or more groups of individuals are followed up over a period of
time to determine the frequency of a particular outcome in the group(s).

For example, the incidence of age-related eye disease in a particular population

could be determined by a cohort study.
Interactive: buttons: Case-control Study

Output:

A study in which individuals with and without the outcome of interest are identified.
Their status with respect to exposures of interest is then determined in order to look
for associations between these exposures and the outcome of interest.

For example: we could use a case-control study to determine risk factors for
diarrhoeal disease among children. We would identify children with and without
diarrhoea and obtain information concerning exposures of interest (breast-feeding,
water supply etc). We would then analyse the data to see if breast-feeding or water
supply was associated with diarrhoea.
Section 8: Using Epidemiology to Investigate a Clinical Problem
Sometimes the investigation of a clinical problem involves a sequence of
epidemiological studies of different types. We can illustrate this with the example of
the iodine deficiency disorders.

Iodine deficiency affects over half a billion people world-wide. It arises in areas
where people depend upon food supplies grown on iodine-deficient soil. Iodine is an
essential micronutrient required by the thyroid gland to make thyroid hormone,
which controls metabolic activity in adults, and is important in foetal growth and
development.

Iodine-deficient soils are most likely to occur at high altitude (e.g. the Andes in
South America and the highlands in Papua New Guinea) where the soil is leached by
water running off the mountain slopes, and in areas far from the sea (e.g. Western
China).

The best-known effects of iodine deficiency are goitre and cretinism.

Goitre is recorded in historical documents and art from many regions.

8.1: Using Epidemiology to Investigate a Clinical Problem

The ancient Greeks used to treat goitre with seaweed, which is rich in iodine. Iodine
itself was not isolated until 1811, and in 1821 a Swiss physician called Coindet used
iodine to treat goitre.

The diagram on the right shows the global distribution of iodine deficiency disorders
(IDDs). Shaded regions are areas where IDDs are prevalent.

World-wide distribution of iodine deficiency disorders (IDD) in developing countries.

Interaction: button: © button

Output:

Hetzel BS. Overview of the Prevention and Control of Iodine Deficiency Diseases. In:
Hetzel BS, Dunn JT, Stanbury JB, editors. The Prevention and Control of Iodine
Deficiency Diseases. Amsterdam: Elsevier; 1987. p. 16.

8.2: Using Epidemiology to Investigate a Clinical Problem

Sequence of epidemiological studies

Descriptive Studies

Interaction: tabs: Part 1

One area that was severely affected by iodine-deficiency in the past was Switzerland.
At the end of the 18th century Napoleon ordered a survey of cretinism to be carried
out in part of Switzerland because many young men from the area were found to be
unfit for military service.

This survey revealed 4 000 people with cretinism in a population of 70 000. This
could be considered as an early descriptive study, quantifying the problem of
cretinism in a particular population.

Interaction: tabs: Part 2

Descriptive studies are usually the first step in the investigation of a clinical problem.
Another area with a high frequency of iodine-deficiency disorders is Papua New
Guinea (PNG). Surveys in the 1950s revealed a high frequency of goitre and
cretinism.

8.3: Using Epidemiology to Investigate a Clinical Problem

Sequence of Epidemiological Studies

Descriptive Studies

Population-Based Analysis

The next step was to compare the frequency of goitre and cretinism in different PNG
communities, and to relate this to the iodine level in the food eaten in each
community.

In other words, population-based analytic (ecological) studies were carried out, to

examine the association between iodine-deficient food in the diet and the prevalence
of goitre and endemic cretinism in each community.

At this stage, communities, rather than individuals, were compared so this is an

example of a population-based (or ecological study).

8.4: Using Epidemiology to Investigate a Clinical Problem

Sequence of Epidemiological Studies

Descriptive Studies

Population-Based Analysis

Individual-Based Analysis
(Case-Control Studies)

As we will see later in the course, ecological studies have drawbacks and are better
used to generate, rather than test, hypotheses. Further evidence is needed in the
form of individual-based analytic studies to confirm the suspected association.

Case-control studies were therefore carried out to establish the association between
cretinism in children and the iodine / thyroid status of their mothers.

8.5: Using Epidemiology to Investigate a Clinical Problem

Sequence of Epidemiological Studies
Descriptive Studies

Population-Based Analysis

Individual-Based Analysis

Intervention Studies

Once the association between iodine deficiency and cretinism was established, the
next step was to apply this knowledge, and introduce an intervention. Experimental
studies were carried out, in the form of randomised controlled trials, to investigate
the effect of iodine supplementation on the incidence of cretinism.

In the late 1960s, randomised controlled trials were carried out in western PNG to
investigate the effect of giving iodine injections to pregnant women on the incidence
of cretinism among their children. The intervention was highly successful.

8.6: Using Epidemiology to Investigate a Clinical Problem

Sequence of Epidemiological Studies

Descriptive Studies

Population-Based Analysis

Individual-Based Analysis

Intervention Studies

Change in Public Health Policy

Following these studies, individual supplementation programmes were launched in

some countries, followed by the introduction of iodized salt.

8.7: Using Epidemiology to Investigate a Clinical Problem

Sequence of Epidemiological Studies

Descriptive Studies

Population-Based Analysis

Individual-Based Analysis

Intervention Studies

Change in Public Health Policy

Evaluation of Public Health Intervention

Once the benefits of supplementation have been established, the next stage is the
evaluation of iodine supplementation programmes.

This could be, for example, by monitoring the prevalence of goitre and cretinism in
areas where supplementation has been implemented.

Section 9: The concept of population in epidemiology

As epidemiologists, we deal with populations. Our basic questions with respect to an
outcome in a population are "who, where and when"?

Epidemiologists are sometimes criticised for having too narrow a focus, in that we
identify and measure risk factors for a particular outcome, but tend to ignore the
social context of those risk factors. Thus perhaps as well as "who, where and when?"
we should be asking "and in what social context?"

Epidemiological findings about risk factors such as smoking, high blood cholesterol or
unsafe sexual practices are not always translated into effective disease prevention.
This may be because we do not pay enough attention to the underlying social and
political processes that account for the distribution of these risk factors within a
population.

These social contexts may vary from one population to another, and this may need
to be taken into account in developing strategies to control disease.

9.1: The concept of population in epidemiology

The epidemiologist Geoffrey Rose pointed out that we need to distinguish two types
of determinants of disease in a population:

• factors that determine which individuals

within a population get a disease
 • factors that determine which populations
get a disease

For example, if we examine the effect of the hardness of the water supply on death
rates due to cardiovascular disease in Scotland, we will not detect any effect,
because the water is soft throughout Scotland.

The effect can only be detected if we study a larger region with a wider range of
exposure. Click below to see a graph of the correlation between water hardness and
mortality, from Rose's paper.

The graph shows a negative association between water hardness and cardiovascular
mortality - in other words, the harder the water, the lower the risk of death from
cardiovascular disease.

Interaction: hyperlink: View graph

Interaction: button: © button

Output:

Pocock SJ. et al. British Regional Heart Study: geographic variations in

cardiovascular mortality, and the role of water quality. British Medical Journal 1980;
280: 1243.

9.2: The concept of population in epidemiology

It follows that if we want to control disease on a population level, we need to
understand the determinants of that disease at a population level, not just at an
individual level.

For example, we might be asked to develop a public health strategy to reduce heart
disease attributable to high cholesterol levels. There are two approaches that we
could use:
● we could decide to target the people with very high cholesterol levels, since they
are the ones who are most at risk (top diagram)
● we could decide to try to reduce the average cholesterol level throughout the
population (bottom diagram)

9.3: The concept of population in epidemiology

In the previous example, a population strategy would be more effective than a high-
risk strategy in reducing cardiovascular mortality attributable to high cholesterol
levels.

However this is not always true. For example, in the prevention of STD transmission
by promoting condom use, in some situations it may be more effective to target
high-risk groups than to aim the intervention at the whole population. Statistical
modelling may help to predict what strategy will be most effective.

9.4: The concept of population in epidemiology

Epidemiology is a young science that has developed particularly rapidly over the last
50 years, as new techniques of analysis have been developed. Much progress has
been made, particularly in understanding the causes of "non-communicable"
diseases (for example, the links between [Link] and gastric ulcer, and between EBV
(Epstein-Barr virus) and Burkitts lymphoma).
However new challenges continue to emerge. In recent years, HIV, Legionnaires’
disease, cholera O139, and new variant CJD (Creutzfeld-Jacob disease) have all
emerged as challenges to the public health. As an epidemiologist, you are unlikely to
become bored!

Section 10: Exercise

Let us now return to John Snow's investigation of cholera.

John Snow's work investigating the Lambeth and the Southwark & Vauxhall water
companies was interrupted at the end of August 1854 by news of a severe outbreak
of cholera in Soho, nearer the centre of London. There were 616 fatal cases which
had their onset between 19 August and 30 September 1854.

John Snow made this map of the area, with a line showing where each person who
had had a fatal case of cholera had lived, and the position of the public water pumps.

(remains through cards 1-3 of page 10)

Snow’s Map of Cholera Cases in Soho

Interaction: button: Source

Snow, J. Snow on Cholera. New York: The Commonwealth Fund; 1936.

10.1: Exercise
Looking at the geographical distribution of cases (the black dots), what was the most
likely source of the outbreak?

Interaction: hotspots: Pump A

Output:

Yes, Pump A on Broad Street is the one on which the fatal cases appear to be
centred.

Interaction: hotspots: Pump B

Output:

From the map, it seems more likely that Pump A in Broad Street is the one on which
fatal cases are centred.

Interaction: hotspots: Pump C

Output:

From the map, it seems more likely that Pump A in Broad Street is the one on which
fatal cases are centred.

10.2: Exercise
John Snow also observed that:
• none of the workers at the Broad Street
brewery had cholera: they were very
close to pump A, but drank beer rather
than water
• an elderly lady and her niece in West
Hampstead (an area some distance
away, which was free of cholera) liked
the taste of Broad Street water, so they
had a bottle brought to them every day
from the pump. They both died of cholera.

Snow became convinced that the Broad Street pump was the source of the outbreak,
and he persuaded the local authorities to remove the pump handle on the 8th
September 1854.

10.3: Exercise
The time distribution of the date of onset of the fatal cases of cholera is shown in the
graph opposite. John Snow had the pump handle removed on 8th September 1854.
Why do you think the epidemic stopped? Think about this for a moment, then press
the button to continue.
Interaction: button: ‘thought bubble’ button

Output:

The graph shows that the epidemic was nearly over by 8th September when John
Snow removed the pump handle. This was probably because the contaminating
sewage had been diluted by this stage. Alternatively, perhaps people had left the
area because there was so much cholera. John Snow's action was important, but
probably not the most important reason for the end of the outbreak!

10.4: Exercise
Did John Snow prove that contaminated drinking water causes cholera?

Interaction: hotspot: Yes

Output:

It is better to say that John Snow provided strong evidence in support of this idea:
epidemiologists cannot prove causation, rather we provide evidence for or against a
cause. Causality is a complex subject which we will return to later in this study
module.

Interaction: hotspot: No

Output:

That’s right: epidemiologists cannot prove causation, rather we provide evidence for
or against a cause. Causality is a complex subject which we will return to later in
this study module.
10.5: Exercise
Although Snow worked before the era of bacteriology, his observations and
deductions led him to clear descriptions and valid theories about the nature and the
mode of communication of cholera.

He applied his findings with a control intervention (removal of the pump handle), and
went on to make recommendations for prevention (personal hygiene, boiling of
soiled bedclothes of patients, isolation and quarantine, improved waste disposal,
drainage, provision of clean water etc.).

Section 11: Summary

The main points of this session will appear below as you click through the step card
opposite. Click on any of the list entries below to go back to that card.

What is epidemiology?

Epidemiology is the study of the distribution and determinants of health-related

states or events in specified populations and the application of this study to the
control of health problems.

11.1: Summary
The main points of this session will appear below as you click through the step card
opposite. Click on any of the list entries below to go back to that card.

How do we describe a health-related event in a population?

Exposures and outcomes

Descriptive and analytical epidemiology

Observational and interventional epidemiology

How do we describe a health-related event in a population?

The key elements are to describe the health-related event in terms of its distribution
in person, place and time.

Exposures and outcomes

An exposure is a risk factor for the outcome that we are investigating, which may or
may not be the cause of the outcome.

The outcome is the disease, or event, or health-related state, that we are interested
in.

Descriptive and analytical epidemiology

In descriptive epidemiology, we describe the distribution of an exposure or outcome,
without seeking to explain the distribution by looking for associations.

In analytical epidemiology, we examine associations, often with the aim of

identifying possible causes for an outcome.

Observational and interventional epidemiology

In observational epidemiology we examine the distribution or determinants of an

outcome without any attempt to influence them.

In interventional epidemiology we test a hypothesis by modifying an exposure within

the study population and examining the effect on the outcome.