Study Design Introduction and Case Control Studies.

Question 1

When carrying out epidemiological research what do you need to describe in the first instance?

Answer 1

The time, place and persons that the disease effects.

Question 2

After aetiological factors of a disease have been determined and intervention procedures have been developed what is the next step in epidemiological research?

Answer 2

Carry out clinical trials to evaluate the intervention.

Question 3

What is it vital that you do before stating a clinical trial?

Answer 3

Define an endpoint.

Question 4

Is epidemiology primarily quantitative or qualitative?

Answer 4

Quantitative.

Question 5

What two types of questions are often answered with qualitative studies?

Answer 5

Those starting with ‘why’ and ‘how’.

Question 6

What is more important in a qualitative study: the quality of the information or the size of the study?

Answer 6

The quality of the information.

In quantitative the amount of information is more important.

Question 7

Do descriptive or analytical studies generate hypothesis?

Answer 7

Descriptive.

Question 8

What type of study are case control, case series and cross sectional studies all examples of?

Answer 8

Descriptive.

Question 9

What would a medical professional report in a case report?

Answer 9

An unusual disease case.

Question 10

If a doctor notices multiple cases of an usual disease what would be produced?

Answer 10

A case series.

Question 11

Are patterns of disease in case series studies at the population or the individual level?

Answer 11

Individual.

Question 12

Case reports are brief explanations of consecutive unusual disease cases noticed by a medical professional. True or false?

Answer 12

False. They are detailed.

Question 13

Once a case report has been made how can it be used to produce an hypothesis on a wider population?

Answer 13

It can lead to the study of medical histories of patients that can trigger the production of new hypothesis.

Question 14

What is another name for a cross sectional study?

Answer 14

A prevalence study.

Question 15

What can make uncommon diseases look common in a prevalence study?

Answer 15

A greater survival rate of the disease.

Question 16

What two things are measured simultaneously in prevalence studies?

Answer 16

Disease and exposures.

Question 17

Name three advantages of a cross sectional study.

Answer 17

1. Good for estimating the population burden of a disease.
2. General population studied.
3. Good for constant exposures.

Question 18

Name three disadvantages of a cross sectional study.

Answer 18

1. Description limited to a specific time point.
2. Cannot determine is exposure proceeded prevalence.
3. Influenced by the survival factor of a disease.

Question 19

Data collected in a ecological study can not confer something about an individual. Why?

Answer 19

Data is at the aggregate/ population level.

Question 20

What are ecological studies also known as?

Answer 20

Correlation studies.

Question 21

What has to be limited for correlation studies to be good?

Answer 21

Variation within a population.

Question 22

What are compared in ecological studies?

Answer 22

Groups, not individuals.

Question 23

Ecological studies do not include time intervals. True or false?

Answer 23

False, they can look at the same population at different time intervals.

Question 24

Ecological studies tend not to be misleading. true or false?

Answer 24

False, they can be.

Question 25

What is the biggest benefit of ecological studies?

Answer 25

Exposure is often easier measured at the population level compared to the individual level, eg with air pollution.

Question 26

What is ecological fallacy?

Answer 26

The failure of ecological data to reflect the true effect of a exposure at the individual level.

Question 27

Collinearity is an issue with correlation studies. Why?

Answer 27

Some social and demographic factors are correlated more at the population/group level that at the individual level.

Question 28

What is the overall aim of analytical studies?

Answer 28

To investigate factors leading to the observed distribution of a disease by testing specific hypotheses relating to the aetiology of that disease.

Question 29

Can ecological and cross sectional studies address analytical hypothesis?

Answer 29

Yes, even though they are descriptive studies. This is the case when data is available on an exposure and an endpoint.

Question 30

Name three examples of an observational study.

Answer 30

1. Case-control. 2. Cohort. 3. Nested case control.

Question 31

T/F: can interfere with a population in an observational study?

Answer 31

No

Question 32

Can cohort or case control studies be retrospective?

Answer 32

Cohort.

Question 33

Can incidence rate be determined in case control or cohort studies.

Answer 33

Cohort.

Question 34

What is the other name for a cohort study?

Answer 34

A longitude study.

Question 35

Cohort studies provide large amounts of useful data that can be used collaboratively. Despite this it would be hard to get funding to start one. Why?

Answer 35

A handful of large cohort studies are available and they are exceptionally expensive to do.

Question 36

For a retrospective study to be produced what is it vital to have?

Answer 36

An exposure that is able to be measured in a clear-cut and non biased way.

Question 37

What type of study is best to study rare diseases?

Answer 37

Case control.

Question 38

Can you study multiple outcomes in a case control or cohort study?

Answer 38

Cohort.

Question 39

Do subjects of field trials already have the disease?

Answer 39

No.

Question 40

Name the hierachy of evidence (assuming all have been conducted correctly.

Answer 40

1. Systematic review and meta analysis. 2. Randomised control trials. 3. Cohort. 4. Case control. 5. Cross sectional. 6. Ecological. 7. Case reports/ case series. 8. Expert opinion.

Question 41

Bias primarily happens at the data collecting stage of a study. True or false?

Answer 41

False. It can also happen at the analysis and interpretation stage.

Question 42

To be a confounding factor what three things must a confounder be?

Answer 42

1. It must be associated with the disease. 2. It must be associated with the exposure in different amounts across the exposure groups. 3. Not on the causal pathway.

Question 43

What sort of relationships do confounding factors mix up?

Answer 43

Causal and non causal.

Question 44

Factor 1 is known to be associated with a disease. A positive association is then found with another factor (factor 2) and the disease. In order to test wether this is another associated factor what must you do?

Answer 44

See if factor 2 is still associated with the disease in the absence of factor 1. If this is not the case then factor 2 and the disease are statistically linked, not causally linked. In this case factor 1 is causal, factor 2 is confounding.

Question 45

At the study design stage what can you do to minimise confounding (4 things)?

Answer 45

1. Randomise individuals to avoid selection bias. 2. Restrict entry into study so you only compare comparable groups. 3. Match individuals/ whole populations. 4. Analyse subgroups separately and adjust the data accordingly.

Question 46

Matching is done at the individual level. True of false?

Answer 46

False. It is done at both the population and the individual level.

Question 47

Name four things you can match on.

Answer 47

1. Sex. 2. Race. 3. Study centre. 4. Age.

Question 48

It is harder to match for continuous factors. How can you overcome this problem?

Answer 48

Spilt the continuous variables into categories/ brackets.

Question 49

What study design answers the following question? How common is the finding in a disease?

Answer 49

Case series.

Question 50

What study design answers the following questions? How common is a disease in a population?

Answer 50

Cross sectional.

Question 51

What study design answers the following questions? What explains the differences between populations?

Answer 51

Ecological studies.

Question 52

What study design answers the following questions? What factors are associated with a disease?

Answer 52

Case control studies.

Question 53

What study design answers the following questions? 'How many people will get the disease', 'What factors caused the disease'.

Answer 53

Cohort studies.

Question 54

What setup is used in the analysis of case control studies?

Answer 54

2 by 2 tables.

Question 55

What are the four main stages in a case control study?

Answer 55

1. Select cases with the disease. 2. Select controls without the disease. 3. Obtain information on past exposures and other relevant factors. 4. Compare proportion of exposures in cases and controls.

Question 56

Name 4 design issues of a case control study.

Answer 56

1. Definition of cases and controls. 2. Selection of cases and controls. 3. How do you measure an exposure? 4. How do you account for bias?

Question 57

Name 3 analytical issues of a case control study.

Answer 57

1. What is the strength of association? 2. Confounding issues. 3. Logistic regression.

Question 58

What is the main assumption made in a case control study?

Answer 58

That the observations seen in the cases and the controls are representative of the whole population.

Question 59

Why are newly diagnosed cases preferable in a case control study?

Answer 59

Less recall bias.

Question 60

Name 3 situations in which cases are not representative.

Answer 60

1. Cases may be regularly undiagnosed in countries that lack screening. Only seeing certain cases. 2. Endpoints may be misdiagnosed. May not be what you think it is. 3. When diseases have more aggressive processes. These aggressive cases won't be represented in the study.

Question 61

Self reporting of a case makes bias highly likely. This method does however come with an advantage. What is it?

Answer 61

Presence of a disease is pretty much guaranteed.

Question 62

What can minimise bias in self reporting?

Answer 62

Getting surrogates to report on the case too.

Question 63

What issue comes with using blood samples when looking at cases in case cohort studies?

Answer 63

Blood profiles can change with disease state and may not be representative of the blood profile prior to the disease state.

Question 64

How do you calculate the odds ratio of exposure (2 equations)?

Answer 64

odds exposure amongst cases/ odds exposure amongst controls. (Exposed cases x non exposed controls)/ (Exposed controls x non exposed cases).

Question 65

How do you calculate odds ratio of disease?

Answer 65

Odds of disease amongst cases/ Odds of disease amongst controls.

Question 66

What does Woolf's method state?

Answer 66

Ih the 95% CI contains 1 than the odds ratio is not statistically significant.

Question 67

State the equation for working out the 95% confidence interval for the odds ratio.

Answer 67

exp(ln(OR)) +/- 1.96 ( 1/a +1/b +1/c +1/d)^1/2

Question 68

OR of (non ordinal category 1) vs (non ordinal category 2) = ?

Answer 68

OR non ordinal category 1/ OR non ordinal category 2.

Question 69

What does the Mantel- Haenszel test examine?

Answer 69

The trend in disease risk with increasing levels of exposure.

Question 70

What sort of trends does the Mantel- Haenszel test?

Answer 70

Linear trends.

Question 71

How do you calculate the expected number of cases in the Mantel- Haenszel chi squared test for ordinal data.

Answer 71

(Total number of people in each category x total number of cases)/ Total number of participants.

Question 72

What is are 3 advantages of analysing a continuous variable as an ordinal variable?

Answer 72

1. Cut offs may have biological significance. 2. Association less effected by skews in the biological data. 3. Dose dependant relationships can be made clear.

Question 73

If analysis of categorical data shows a linear relationship what is it necessary for you to do?

Answer 73

You should analyse the variables as continuous variables in regression models.

Question 74

Splitting linear data into ordinal categories increases statistical significance. True or false?

Answer 74

False. Less data is used so statistical significance is reduced.

Question 75

Why does volunteer bias exist?

Answer 75

Willingness to volunteer maybe associated with confounding variables or systematic differences compared to the general population.

Question 76

Why is recall bias different in cases and controls?

Answer 76

Cases are more likely to recall more factors/ be more inclined to recall more as the results directly effect them.

Question 77

If a patient does not look ill it is favourable to blind the interviewer. Why?

Answer 77

They may ask leading questions/ if you're looking for something you are more likely to find it (outcome bias).

Question 78

What is survival bias?

Answer 78

Risk can be underestimated because most severe cases die before being identified and recruited.

Question 79

Name three advantages of case control studies.

Answer 79

1. No loss to follow up. 2. Results are acquired quickly. 3. Relatively inexpensive.

Question 80

Name 5 disadvantages of a case control studies.

Answer 80

1. Bias in exposure assessment. 2. Limited interpretation of causality. 3. Temporal relationship unclear. 4. No estimates of incidence of disease. 5. Only one outcome tested.