10. ECM 2 - Understanding Causation

Question 1

what does PICO stand for

Answer 1

P - POPULATION, PATIENTS
I - INTERVENTION
C - CONTROL INTERVENTION, non-exposed or no risk factor
O - OUTCOME

Question 2

how would you change this bad question ‘Does smoking cause heart attacks?’ into a GOOD QUESTION using PICO

Answer 2

In adults aged 35+ (POPULATION), Smoking 1+ Cigarettes per day (INTERVENTION/EXPOSURE) compared to no smoking (CONTROL) increase the incidence of fatal and non-fatal acute myocardial infarction? (OUTCOME)

Question 3

CONFOUNDING occurs when the association between an EXPOSURE and OUTCOME becomes DISTORTED by a THIRD VARIABLE

what are the 3 CRITERIA for a CONFOUNDER

Answer 3

1. ASSOCIATED WITH EXPOSURE
2. ASSOCIATED WITH DISEASE/OUTCOME (risk factor)
3. NOT PART OF THE CAUSAL CHAIN FROM EXPOSURE TO OUTCOME (not caused by the exposure and thus causing the outcome)

eg.
exposure - consumption of ice-cream
outcome - timing of riots

possible confounder: good weather

eg2.
exposure - Baldness
outcome - Coronary Heart Disease

Confounder: Age

Question 4

if RISK RATIO (RR) GREATER THAN 1 what does it mean

Answer 4

there is an ASSOCIATION between the variables

Higher above 1 means stronger association

risk ratio = 1 means no difference / no association

Question 5

what are 2 ways researchers can use to adjust for CONFOUNDING in DATA ANALYSIS

Answer 5

1. REGRESSION ANALYSIS
2. STRATIFICATION (separating the data i.e data for association between baldness and CHD separately by age group, young and old men)

Question 6

can HYPERTENSION be a CONFOUNDER for the ASSOCIATION between OBESITY and CVD

Answer 6

NO, as it is on the CAUSAL CHAIN

Obesity -> Hypertension -> CVD

Question 7

to understand if the EXPOSURE is ASSOCIATED with the OUTCOME what do we need

Answer 7

STUDY DESIGNS

Question 8

2 Main aspects to understands different STUDY DESIGNS

Answer 8

• is the OUTCOME MEASURED BEFORE or AFTER measuring the EXPOSURE?
• How are you dealing with CONFOUNDING?

Question 9

STUDY DESIGN type that OBSERVES both EXPOSURE and OUTCOME AT THE SAME TIME

Answer 9

CROSS-SECTIONAL

Question 10

STUDY DESIGN type that OBSERVES OUTCOME FIRST (selects people with the outcome ie heart disease) and then looks back at if they have EXPOSURE (ie then look at their shoe size)

Answer 10

CASE-CONTROL DESIGN

Question 11

STUDY DESIGN type that 1ST OBSERVES EXPOSURE and then FOLLOWS them over time until there is an OUTCOME (future)

Answer 11

COHORT DESIGN

Question 12

STUDY DESIGN TYPE that SETS THE EXPOSURE (not observes) ie selects people who don’t already have the outcome (ie heart disease) and put them into RANDOM groups, some are GIVEN EXPOSURE (ie drug) and some are given PLACEBO. then FOLLOW them over time

Answer 12

RANDOMISED TRIAL Design

Question 13

what is the BEST STUDY DESIGN for ensuring CAUSAL EFFECT and NOT CONFOUNDED

Answer 13

RANDOMISED TRIAL

Question 14

for which STUDY DESIGNS use
OBSERVATIONAL DESIGN (observe the exposure not cause)
and the METHOD used to account for CONFOUNDERS is STRATIFICATION, MATCHING OR ADJUSTMENT

Answer 14

CROSS-SECTIONAL

CASE-CONTROL

COHORT

Question 15

for RANDOMISED TRIAL DESIGN

1. do you observe or cause the exposure?
2. what method is used to account for confounders?

Answer 15

1. EXPERIMENTAL DESIGN - cause the exposure
2. RANDOMIZATION

Question 16

for CUASI-EXPERIMENTAL DESIGN

1. do you observe or cause the exposure?
2. what method is used to account for confounders?

Answer 16

1. EXPERIMENTAL DESIGN - cause the exposure
2. STRATIFICATION, MATCHING OR ADJUSTMENT
   (not randomising the exposure so not as good at defining if confounders or not)

Question 17

for CROSS-SECTIONAL DESIGN

1. do you observe or cause the exposure?
2. what method is used to account for confounders?

Answer 17

1. OBSERVATIONAL DESIGN
2. STRATIFICATION, MATCHING OR ADJUSTMENT

and same for CASE-CONTROL and COHORT DESIGN

Question 18

what is the BRADFORD-HILL CRITERIA for ‘ASSOCIATION or CAUSATION?’

Answer 18

• STRENGTH of effect?
• Dose-RESPONSE?
• OVERTIME and Overseas ie. consistency
• FOLLOWS exposure?
• DESIGN of Experiment
• ANIMAL model established?
• REVERSIBILITY?
• BIOLOGICAL PLAUSIBILITY?

(BRADORDS acronym)