Collecting Data About People

Question 1

In case control studies vs cohort what is key difference?

Answer 1

In case control you select participants hence cannot calculate risk but in cohort study you can

Question 2

What is the odds ratio?

Answer 2

Below 1 exposure is protective
Above 1 exposure causes outcome
Equal to 1 no association

Question 3

Strengths and weaknesses of control studies?

Answer 3

Strengths:
See multiple associations
Very good when disease outcome is rare (because you select in those people)
Minimise selection/information bias (you can decide how similar case and control group are/ you can decide to measure outcomes consistently)
Retrospective-cheaper/shorter in duration

Weaknesses:
Cant calculate prevalence/ incidence (you decide participants)
Not suitable for rare exposure
Data availability poor due to being retrospective
Vulnerable to confounding (different variable affecting both whether they finish high school education and likelihool of cancer)
Suitable ocntrol group hard to find

Question 4

RCT?

Answer 4

Study in which participants are allocated randomly between an intervention and a control group

Question 5

Difference between efficacy and effectiveness?

Answer 5

Efficacy- Does this intervention work in ideal conditions

Effectiveness- does it work in non-ideal normal conditions/ real life

Question 6

To find a causal relationship you would use which study?

Answer 6

RCT, reduce confounding

Question 7

Strengths and weakness of RCT?

Answer 7

Minimise selcetion and information bias
Establish safety efficacy/ effectiveness
Best single study evidence

Weakness:
Time consuming
Expensive
Not immune to bias
Participants drop out 
Can lack generalisability ( selecting people)

Question 8

Data properties?

Answer 8

Categorical- disease being present or not, sexual orientation, ethnicity
Continuous- age, weight, height, no of symptoms

Question 9

Type 1 error?

Answer 9

No true difference, but observed difference

Question 10

Type 2 error?

Answer 10

There is a true difference but no observed difference

Question 11

Protection against type 1?

Answer 11

Setting threshold, i am confident this difference is significant
Allow a 5% chance of error

Question 12

Protection against type 2 error?

Answer 12

Statistical power set at 80-90%, so 10-20% chance of making an error
Larger the sample, larger the statistical power

Question 13

Difference in error perecentage in type1 and type 2?

Answer 13

Because better to be more safe when giving treatment than withholding treatment

Question 14

What is p value?

Answer 14

Probability the differenc eif observe could have occured by chance if the groups compares were really alike,

Large sample- smaller p value

Question 15

For a continuous scale comparison will require tests?

Answer 15

T-test or anova

Question 16

To compare between categorical comparator you will use the test…?

Answer 16

Chi squared

Question 17

T - test?

Answer 17

Is there a Mean difference between two groups?

Question 18

Chi squared test?

Answer 18

Is the association between exposure and outcome statistically significant?

Question 19

Confidence interval?

Answer 19

95% of confidence intervals will be true population 95% of the time