DL: Statistical Illiteracy Among medical prof....

Probability of getting a result as extreme (far way from the mean) as you got if the null hypothesis is true.

• P value

• standard deviation

Number of patients needed to treat to prevent one additional bad outcome.

• ARR

• RR

• NNT

Answer explanation

Number needed to treat

Events/people

• ARR

• RR

• NNT

Answer explanation

absolute risk reduction:

41-32=9

9deaths/1000 people = 0.009 ARR

How much risk was reduced/original risk with no intervention

• ARR

• RRR

• NNT

Answer explanation

Risk of death was reduced by 9 (41-32=9)

9 less deaths/41 deaths w/out med = 0.22 RRR

The inverse of ARR

• RRR

• NNT

Answer explanation

NNT=1/ARR or inverse of ARR

1000 people/9 deaths = 111.11 --> 112 people needed to treat

The true positive rate

• sensitivity

• specificity

The true negative rate

• sensitivity

• specificity

9/100 people will test positive but will NOT have the disease

• specificity is 91%

• sensitivity is 91%

10/100 people WILL have the disease but test negative.

• sensitivy 90%

• specificity 90%

Which test is best for screening?

• high sensitivity test

• high specificity test

Answer explanation

sensitivity tests for true positives

(sensitivity with 90% means 90/100 that test positive have the disease and 10/100 people that test positive do not have the disease)