Identifying risk factors and predictors of poor health: Aetiology and case-control designs

Question 1

How can 2 variables be described when considering their relationship?

Answer 1

-One = outcome
-Other = predictor/exposure

Question 2

Use smoking status to think of 2 scenarios where this could be an outcome OR a predictor/exposure variable.

Answer 2

*Sex (m/f) vs smoking status:
-Sex (m/f) = predictor
-Smoking status = outcome

*Smoking status vs symptoms of breathlessness:
-Smoking status = predictor
-Breathlessness = outcome

Question 3

What is a key concept of causality?

Answer 3

-Making connections between variables
–> however with the wider knowledge that in reality many predicting factors can influence a single outcome

Question 4

Summarise upstream & downstream events that influence individual & population health in the life course approach.

Answer 4

Question 5

Sufficient-component cause definition

Answer 5

Attempts to explain probabilistic phenomena via unknown component causes
–> on both views - heavy smoking can be said a cause of lung cancer only when the existence of unknown deterministic variables is assumed

Question 6

What is a sufficient cause?

Answer 6

A complete causal mechanism - minimal set of conditions & events that are sufficient for an outcome to occur
–> i.e., complete pie - called a causal mechanism = a sufficient cause
-May be multiple conditions needed e.g., for a disease - each with several component causes
-If none occur = no outcome
-Can be unknown causes involved

Question 7

What is a component cause?

Answer 7

One member (slice) of a set of causes - creates a sufficient cause
–> i.e., individual factors = component causes
-If block the component cause = no outcome

Question 8

What are necessary causes?

Answer 8

A component that occurs in every pie or pathway

Question 9

What is PICO/PECO?

Answer 9

-Population – Women aged between 60 and 70
-Exposure (or intervention) - NSAIDs
-Comparison – No NSAIDs (ps not taking NSAIDs = to compare with)
-Outcome – Atrial fibrillation

Question 10

What are case reports/case series?

Answer 10

-Observational, descriptive
-Detailed description - often of single case/series of cases
-Analysis = narrative description, simple descriptive stats (case series)

Question 11

Case report vs case series?

Answer 11

-Case report = detailed report of diagnosis, treatment, response to treatment, & follow-up after treatment of 1 patient

-Case series = group of case reports involving patients given similar treatment

Case reports & case series usually contain demographic information about patient(s): e.g., age, gender, ethnic origin

Question 12

What are strengths of case reports/case series?

Answer 12

Question 13

Describe how a case report/case series is conducted.

Answer 13

Look at people with a history of the exposure/predictor (NSAID use) - then look to see if outcome of interest is shown (atrial fib)

???

Question 14

Example of case series - cholera.

Answer 14

Rev Whitehead -> looked at name, age, living arrangements, consumed water, hour of onset - of people who died of cholera

Question 15

Weaknesses of case report/case series?

Answer 15

Question 16

What is a cohort study?

Answer 16

-Observational (not experimental)
-Expensive
-Can last v. long time

-> follow people who have & haven’t had the exposure & follow them up to observe the outcome of interest
-So follow people given NSAIDs & not given them - see if develop AF

Question 17

Limitations of a cohort study?

Answer 17

Sufficient no. of outcomes w/ small risks require
-Large no. of subjects &/or
-Long period of follow-up
-Cohort study can become inefficient

Question 18

What is a case-control study?

Answer 18

-Observational, analytical
-Compares cases (diseased ps) & controls (non-diseased ps) with respect to their level of exposure to a suspected risk factor

-Look at people who have developed the outcome/disease of interest (e.g., AF) & then look back to see if they were or weren’t exposed to the exposure/predictor of interest (e.g., NSAID use) - prior history of exposures
-Retrospective study design

Question 19

Examples of case-control studies?

Answer 19

Question 20

Advantages of case-control studies?

Answer 20

Question 21

Limitations of case-control studies?

Answer 21

Question 22

Key elements in case-control design?

Answer 22

-Assembling case series
-Choosing appropriate controls
-Ascertaining prior exposures
-This is NOT a test of causality (this is inferred)
-Confounding is the alternative explanation to causality

Question 23

Compare a cohort study VS a case-control study.

Answer 23

-Cohort = exposure –> outcome (so outcome has not yet happened @ start of study - & exposure may or may not be present)
-Case-control = outcome –> exposure (so outcome has already happened @ start of study - as has exposure if was present)
(when conducting study - the order you go in)

Question 24

What are some critical considerations for when selecting controls?

Answer 24

-Control definition
-Source of controls
-When to sample controls - beginning or later?
-How many controls per case? (can you get away with fewer per case or same no.)
-Match controls with cases or not - e.g., on age, sex (could act as confounding factors otherwise)

Question 25

What are the 4 types of controls?

Answer 25

-Hospital controls
-Relative & friend controls
-Population controls

-Matching

Question 26

What are some critical considerations for when selecting cases?

Answer 26

-Case definition (what counts as a case? - newly diagnosed?) - must make case criteria clear
-Sources of cases
-Identifiable source population
-Incident or prevalent cases
-Representativeness

Question 27

How to ascertain exposure status?

Answer 27

-Source of exposure info/method of data collection –> personal interview, self-administration questionnaire, proxy respondents, records, objective/chemical measurement
-Time & dose
–> how far back in past to go for exposure info? how best to quantify exposure?

Question 28

Problems with how source of exposure info/method of data collection - is done?

Answer 28

-Info biases - especially if is self-report –> recall bias (memory), social desirability
-Ethical considerations - use of medical records

Question 29

Problems with how to measure time & dose of exposure?

Answer 29

-Lack of consideration of induction/start of exposure
–> may cause choosing of incorrect time period for exposure
-Lack of consideration of latent period (disease start to detection)
–> may cause measuring of exposure until after disease process is underway
-Lack of quantification of exposure -> may prevent detection of dose-response effects

Question 30

What are confounding variables?

Answer 30

Variables that influence outcome - other than the variable of interest
–> so outcome/measurement of effect of exposure on outcome = distorted by association of outcome with other factor(s)

Question 31

What can confounding lead to?

Answer 31

Under or over - estimation of association between exposure & outcome/disease - & can sometimes reverse direction of observed effect

Question 32

Hierarchy of research?

Answer 32