Module Five Flashcards

1
Q

Three components of internal validity?

A

Chance, bias, confounding

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

Components of chance to identify?

A

95% confidence interval (sample size), Null value, Clinical vs Statistical significance

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

What does increasing the sample size do?

A

Reduces sampling variability, increases likelihood of getting a representative table and increases precision of parameter estimate

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

What do you do if the p-value is below 0.05?

A

Reject the null

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

How do you interpret a confidence interval?

A

x times as likely… we are 95% confident that… and since p-value is less than 0.05 the association is statistically significant, we reject null, unlikely due to chance…

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

What is a type 1 error?

A

Rejects a null hypothesis that is actually true

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

What is a type 2 error?

A

Failing to reject the null hypothesis when it’s actually false

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

If a CI includes the null?

A

p-value > 0.05, fail to reject null, findings not statistically significant

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

If a CI does not include the null?

A

p-value < 0.05, reject null, findings statistically significant

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

What are p-values problemtatic?

A

Arbitrary threshold, only about Ho, nothing about importance

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

What is bias?

A

Any systematic error in an epidemiological study that results in an incorrect estimate of the association between exposure and risk of disease

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

Three types of bias?

A

Selection bias, Information bias, Publication bias

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

What can bias cause?

A

Precision parameter being bias towards or away from the null, causing an over/under estimation

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

When can selection and information bias be controled?

A

Only in design phase

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

When can publication bias be controled?

A

Only in analysis phase

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

How do you minimise loss to follow up?

A

Alternative contact, regular contact and several attempts at contact

17
Q

How can you minimise recall bias

A

Objective measures, validate self-reported measures with other information, memory aids

18
Q

How can you minimise interviewer/observer bias?

A

Clearly define study protocol and measures, structure questionnaire and standard prompts, training of interviewers, blinding

19
Q

What are the requirements for something to be a confounder?

A

Independently associated with exposure, Independently associated with outcome, not on the causal pathway

20
Q

What is an overestimation?

A

RR is further from null

21
Q

What is an underestimation?

A

RR is closer to null

22
Q

How can you control for a confounder in the design stage?

A

Randomisation, Matching, Restriction (to a strata)

23
Q

How can you control for a confounder in the analysis stage?

A

Stratification, Multivariable analysis, Age standardisation

24
Q

What are the pros of stratification?

A

Easy for a small number of potential confounders with limited strata, can evaluate impact of confounding, can identify effect modification

25
What are the cons of stratificaition?
Can leave residual confounding, not feasible when dealing with lots of potential confounders with many strata
26
What are the cons of multivariable analysis?
Can leave residual confounding, not feasible when dealing with lots of potential confounders with many strata, can only control what you've measured
27
When is something a counfounder?
When it adjusts the measure of association by more than 10%
28
What is effect modification?
The effect of exposures and outcomes varies between strata
29
What is a sufficient cause?
Combination of component causes that cause disease
30
What does the B in BESTCDS stand for?
Biological plausibility - Is there a plausible mechanism for the association?
31
What does the E in BESTCDS stand for?
Experimental evidence - Is it an RCT?
32
What does the S in BESTCDS stand for? (1st S)
Specificity - is the exposure specifically associated with a particular outcome and not others?
33
What does the T in BESTCDS stand for?
Temporal sequencing - which came first, exposure or outcome?
34
What does the C in BESTCDS stand for?
Consistency - is study consistent with other studies?
35
What does the D in BESTCDS stand for?
Dose-response relationship - does increasing dose cause worse outcome?
36
What does the S in BESTCDS stand for? (2nd S)
Strength of association - stronger association means less likely caused by chance