Chapter 1: Studies and Design

Question 1

what is epidemiology

Answer 1

study of the occurrence and spread of disease

Question 2

what are the three elements in a confounding variables

Answer 2

1. a causal factor for the outcome, eg.age is a causal factor for death
2. associated to or related to the exposure eg. age is higher in Australia
3. not caused by the exposure eg. the is not caused by living in Australia

Question 3

what is the extreme case of confounding

Answer 3

where all individuals with attribute A also have attribute B

Question 4

how can one stratify

Answer 4

• chop up the population into strata where all individuals have the same level of confound
• this however requires us to know what the confound is

Question 5

how can one undertake randomisation

Answer 5

1. using a systematic method of allocating exposures can fail
2. use a computer

Question 6

what must be bewared when calculating the risk percentage

Answer 6

it is the infection over the n, not the no infections

Question 7

why do we have a control group

Answer 7

• comparison allows us to identify the effects of the response variable
• by forming a baseline for comparison, to detect the effect of any other treatments

Question 8

what is the effect of randomisation

Answer 8

to use randomness to even out the effect of uncontrolled or unknown confounds

Question 9

what is the effect of blocking

Answer 9

1. ensures known factors of confounder evened out by blocking instead of trusting randomisation
2. uses the randomisation of individuals separately within each block to reduce the natural variation by making comparison of similar units
3. removes source of confound for validity and also achieves high precision

Question 10

what is the effect of replication

Answer 10

• have enough individuals in each treatment group so that chase variation can be measured and systematic effect can be seen
• increase replications, increase precision

Question 11

what is the effect of blinding

Answer 11

• for validity
• avoid conscious or unconscious bias from the experimenter or participant
• this would invalidate results

Question 12

what is the effect of balance

Answer 12

• for precision
• simplifies the analysis and gives the most precise comparison
• sometimes defeated by nature eg. dropouts

Question 13

why can causality not be suggested in an observational study

Answer 13

1. exposure not assigned

2. no randomisation

Question 14

what is the point of a cohort study

Answer 14

to compare disease rates in the exposed and unexposed cohort

Question 15

compare cohort studies and case control studies

Answer 15

1. whilst CS uses complete subpopulation, CCS uses sampling from subpopulation
2. whilst CS is usually very expensive, CCS is usually less expensive
3. whilst CS is convenient for studying many diseases, CCS is convenient for studying many exposures
4. whilst CS is usually prospective, CCS is usually retrospective

Question 16

what is a case control study

Answer 16

where the cases and controls are sampled separately, whereas cohort studies obtain information on exposure and other variables from large populations to measure the risk of a disease; in many studies, only a tiny minority who are at risk actually develop the disease

Question 17

in case control studies, why are we matching cases and controls?

Answer 17

clever sampling is used to reduce cost and increase efficiency

the number of controls do not have to be the same as the number of cases. the more controls, the more information we have about the controls

thus, it indicates an association between the response variable

Question 18

what is a cross sectional study

Answer 18

an observational study in which exposure and disease are determined at the same point in time

these are usually cheaper, but cannot determine temporal relationships

they are good for estimating the prevalence of disease

Question 19

what is the hierarchy of evidence

Answer 19

1. clinical trials
2. community trials
3. cohort studies and natural experiments
4. ccs
   cross sect
   anecdotal evidence

Question 20

what are the two types of association

Answer 20

1. a positive relationship between A and B means that if you have A, you are more likely to have B. vice versa
   + sign
2. a negative relationship means between A and C means that if you have A then you are less likely to have C vice verse
   - sign

this is only an association and not a causation

Question 21

what can we tell about an association between A and B

Answer 21

a cause b
b causes a
another factor C cases both a and b
a and b reinforce each other
a and b just happen to be associated

Question 22

what is Bradford hill’s criteria for causation

Answer 22

1. strength: size of association. smoking means a large increase in the chance of cancer
2. consistency: repeat study and get same results. many studies indicate smoking and cancer are related
3. specificity: are there any other causes
4. temporality: exposure precedes outcome. smoking precedes cnacer
5. biological gradient: more causal factor, more outcome. greater smoking level means greater chance ofcancer
6. plausibility: makes sense scientifically. smoking affects lungs
7. coherence: fit with other scientific facts.
8. experiment: can we show it in experiments
9. analogy: similar causal associations?

Question 23

how can confounding be shown in a relationship diagram

Answer 23

in both diagrams, there is a positive relationship between the confound and the outcome variable

whilst there is a positive relationship between the explanatory variable and the response variable in a non confounding diagram, in a confounding diagram there is a negative relationship between the explanatory variable and the response variable

whilst there is no association or causation between the explanatory variable and the confound in a non confounding relationship, there is a negative relationship between the explanatory variable and the confound in a confounding diagram