Question 1

In a series of 100 women with breast cancer, 30 reported having a university education.From this we can conclude:&#10;A)Belonging to a high social class increases the probability of developing breast cancer&#10;B)University education is rarely associated with breast cancer&#10;C)In this particular series of women with breast cancer, 30% had a university education&#10;D)A, B and C are all true&#10;E)None of the above statements is true

Accepted Answer

In this particular series of women with breast cancer, 30% were educated to
university level.As we do not know anything about the educational status of
comparable women without breast cancer we cannot draw any conclusions about
education and breast cancer.

Question 2

You are told to design a low&#8208;cost ecological study comparing your country to 20 others, looking at the correlation between meat consumption and coronary heart&#10;Disease (CHD).Which of the following design features would you incorporate:&#10;A)A nutrition survey in each country to measure meat intake accurately&#10;B)Use available morbidity data on CHD for each country&#10;C)Use routine mortality data from each country (generally available with a lag of 2&#8208;3 years) to measure CHD&#10;D)Ask the FAO (United Nations Food and Agriculture Organisation) to provide you with crude food 'disappearance' data from 10 years ago for each country&#10;E)Hire a really good statistician to help with the analyses

Accepted Answer

Cost considerations rule out A despite it being far
more accurate than D.(With a smaller number of countries you might have recent
nutrition surveys that you can use, although they are likely to have used different
methods to capture diet.) You have also seen that good morbidity data for CHD are
mostly not available, so it is safer to go with mortality data which will be of fairly
uniform quality.(A possible drawback of this is if clinical care of CHD differs among the
countries, as this plays an important role in limiting deaths.) And a good statistician is
always an asset!

Question 3

A controversy occurred between the proponents of drug therapy versus remedial reading for patients with dyslexia.To support their position, one group wrote: "Of
111 patients with dyslexia, 91 showed improvement following remedial reading
Courses." Their inference that in patients with dyslexia, remedial reading is the
Therapy of choice is:

A)Correct
B)Incorrect because it is not based on proportions or rates
C)Incorrect because there is no comparison group
D)Incorrect because of both B and C
E)None of the above

Incorrect because there is no comparison group

Accepted Answer

The statement is incorrect because there is no comparison group (we can
calculate the rate, 91/111).We do not know how many children would have shown
improvement if they had been given drug therapy.However, a useful practical
observation is that given the large proportion who have benefited from remedial
reading, there is not a lot of room for further improvement by adding drugs to the
reading program.

Question 4

The sample of women with breast cancer in the previous question was drawn at random from all women attending one surgeon's breast cancer clinic at a private
Hospital in Montreal, Canada.To which target population might the prevalence result
Apply?

A)Attendees of this clinic only
B)All women with breast cancer attending this hospital
C)All Montreal women with breast cancer
D)All Canadian women with breast cancer
E)Any women with breast cancer anywhere in the world

Accepted Answer

This is, as is often the case when trying to identify to whom a set of
findings applies (their generalisability), and/or whom it was intended to apply to, not
wholly clear.We would certainly hope that it would apply to all women at the clinic in
which the study was done, as this is also the sampling frame, and probably to all women
attending that hospital (unless there was something special about the clinic chosen).
But it is likely the authors hoped that their findings were more widely applicable, and at
least to Montreal women overall.However, given the sample comes from only one
clinic in one private hospital in the city, even this may be dubious, as women who
attend a private hospital are likely to have more education than those attending public
hospitals.And generalising more widely is even more hazardous.[Note: if the finding
from the clinic was related to biology, e.g.measuring hormone levels in these women,
this might indeed hold among all women, everywhere.]

Question 5

Many countries now routinely monitor the health of their population through:&#10;A)Cohort studies&#10;B)Prevalence surveys&#10;C)Case&#8208;control studies&#10;D)Population experiments&#10;E)None of the above

Accepted Answer

The various national health surveys described in the text are all cross‐
sectional prevalence surveys of health conditions and related factors.Some may include
or add‐on prospective follow‐up of participants, thus creating a cohort study, but this is
not a principal feature overall.

Question 6

If the age&#8208;adjusted incidence rate of breast cancer among the grandchildren of Japanese immigrants to the USA is much closer to that of American women (high&#10;Rates) than Japanese women (low rates) what does this tell us?&#10;A)The most important causes of breast cancer are genetic&#10;B)The most important causes of breast cancer are environmental (i.e.not genetic)&#10;C)Breast cancer is diagnosed more accurately in the USA&#10;D)Dietary change is the most likely explanation for the increase in risk among migrants&#10;E)Women who think they are at risk of breast cancer are more likely to move to the USA

Accepted Answer

The most important causes of breast cancer are environmental (i.e.not
genetic).If breast cancer rates among the Japanese immigrant population have gone
from low to high in two generations this suggests that the main causes are not genetic
(because a genetic change cannot happen this quickly).It tells us nothing about the
accuracy of diagnosis or about the women who move to the USA.Nor does it give us
any data on what important factors differ in the different environments; diet may be
one of the causes, but it requires further analytic studies to test the theory directly.

Question 7

Vital statistics are important sources of data on which of the following? (Select all that apply)&#10;A)Morbidity&#10;B)Mortality&#10;C)Health&#8208;related behaviours&#10;D)Injury and disability&#10;E)Outpatient health&#8208;care usage.

Accepted Answer

Vital statistics refer to data on birth, death, marriage, and divorce.Vital
statistics are, therefore, the primary source of data on mortality, but not on morbidity
(illness), behaviours, injury (other than fatal injuries), or health‐care usage.

Question 8

Descriptive epidemiology encompasses which of the following?&#10;A)Making a community diagnosis&#10;B)Reporting sex&#8208;specific trends in causes of death in Alaska&#10;C)Producing a table ranking European countries according to their wine consumption&#10;D)Graphing those consumption rates against heart disease mortality rates&#10;E)All of the above

Accepted Answer

Each measure or activity tells us something about the state of health, or
distribution of possible influences on the health of a nation, state or community, and B
specifically includes the descriptive elements of person, time and place.By jointly
graphing country-specific patterns of a possible causal or preventive agent and related
disease as in D, the population-based descriptive data additionally allow hypotheses to
be made about the relation between the two factors.

Question 9

Potential sources of data on morbidity in the population at large include which of the following? (Select all that apply)&#10;A)Environmental monitoring data&#10;B)Hospital discharge data&#10;C)Laboratory results&#10;D)Notifiable disease reports&#10;E)Vital records

Accepted Answer

Each of these is a source of morbidity (illness) data.Notifiable
diseases should be reported no matter where they occur in a community; this may also
be true if laboratory data come from all labs in a region, this is just not specified clearly
here.Hospital discharge data (B) are excellent for recording illnesses that are treated in
hospital, but not for conditions treated in the community, by a GP (primary care doctor)
for example.A includes only exposure data, not illnesses; and vital records (E) do not
include morbidity data.

Question 10

If an ecological study shows a strong positive correlation between per capita alcohol consumption and breast cancer incidence rates in European countries we can draw
The following conclusion:

A)Alcohol consumption is a risk factor for breast cancer
B)Breast cancer rates are higher in countries with higher alcohol consumption
C)We cannot draw any conclusions because the association could be due to selection bias
D)We cannot draw any conclusions because the measurement of alcohol consumption is likely to be unreliable
E)Women who drink more alcohol are more likely to develop breast cancer but this does not mean that the association is causal

Accepted Answer

Breast cancer rates are higher in countries with higher alcohol consumption.
This ecological study compared the prevalence of breast cancer and alcohol
consumption among different European countries; it did not relate alcohol
consumption to breast cancer directly in individual women.It is impossible to
determine whether the two factors are truly associated, and thus also impossible to
determine causality from an ecological study because there may be other explanations
for the observed association.However it does raise the hypothesis as one worth
pursuing.(And in fact alcohol has been recognised by the World Cancer Research Fund
as a cause of breast cancer, albeit a weak one.)

Deck 3: Who, What, Where and When Descriptive Epidemiology