7.1 Key Statistics In Evidence based Medicine

Question 1

What is incidence?

Answer 1

Number of new cases of a disease arising in a population

Incidence Increase

Question 2

What is Prevalence?

Answer 2

The number of cases of disease existing in a given population at a given time

Question 3

Why is incidence important?

Answer 3

COVID example

Important for planning number of COVID tests e.g. number of people likely to be entering the hospital

Question 4

Why is prevalence important?

Answer 4

Number of people that are ill currently, e.g. important for u dears tan ding how the workforce will be affected

Question 5

How risk measured?

Answer 5

Risk ratio
Odds ratio

Question 6

What is a risk ratio?

Answer 6

Ratio of incidence in group A vs group B

Question 7

What is an odds ratio?

Answer 7

Ratio of the outcome in exposed vs unexposed

Odds Outcome

Question 8

Risk ratio example

Answer 8

A: 100 people, 25 people die 0.25
B: 100 people, 10 people die 0.1

Risk ratio of 2.5

Question 9

Odds ratio example

Answer 9

A: 100 people, 25 die Odds 1:3 risk of death
B: 100 people, 10 die Odds 1:9 risk of death

Odds ratio of 3, group A 3x more likely than B

Question 10

What is absolute risk?

Answer 10

Risk of acquiring a given disease over a given period of time

Question 11

What is absolute risk difference?

Answer 11

Risk in group A - group B
Control vs intervention groups

Question 12

Why are relative risks used?

Answer 12

More powerful to say smoking doubles risk of heart disease rather than saying increases it by 15%

Question 13

COVID example of absolute risk vs relative risk

Answer 13

Relative risk of death of strain B was 30% higher than strain A

Absolute risk
If 100,000 people infected 20 would die strain A
26 would die in strain B 0.006% vs 30%

Not as powerful as saying 30% increase in risk of death

Question 14

What is a P-value?

Answer 14

Probability that the effect could have occurred by chance

Small P value implies a small chance of that effect not being a real effect of the drug and vive.

So if something is 0.01, 99% chance of that occurring due to drug or 1% due to chance

Question 15

What are P values usually?

Answer 15

0.05, 5% chance may be due to chance
95% due to drug

Question 16

What are the issues with P-values?

Answer 16

Can never completely rule out chance

Give no indication as to size of the effect, need to know what the size of effect is

p-values give no idea as to the range of uncertainty around the effect you have etsimated

Question 17

What is a confidence interval?

Answer 17

The uncertainty around the estimate e.g.

0.5 rate, uncertainty around this estimate

Imagine the confidence interval is 0.2 to 0.8

95% probability that drug A vs B may be as much as 0.2, 80% reduction in mortality or 0.8 20% reduction in mortality

Finding would be statistically significant as the confidence interval for relative risk does not include 1

If 1.0 is included in the range, it is not significant

Question 18

What is Absolute difference?

Answer 18

Drug 1 leads to 1kg loss
Drug 2 leads to 0.9kg loss

Absolute difference 0.1 kg

Measuring effect between drugs

If the range crosses 0.0 not significant