EBM

Question 1

Types of Bias

Answer 1

• Systematic distortion of the estimated intervention away from the truth, due to inadequacies in the design, conduct, or analysis of a trial
  
  • Selection bias: selection in a non-random way to either group, leading to some members of the population being less likely to be recruited (ex. only recruiting in a hospital)
  • Measurement bias: different measurements in the intervention and control group
  • Recall bias: patients know where they are allocated and it influences their report of symptoms (prevented by blinding)
  • Response bias: when those who enroll do not represent the population as a whole (prevented by random sampling from a population)
  • Analysis bias: maintain all participants within the initial group for analysis (prevented with intention to treat model)
  • Publication bias: negative studies less likely to be published
  • Hawthorne effect: process of studying influences the outcome

Question 2

Study Designs

Answer 2

Question 3

Prevalence vs. Incidence

Answer 3

Prevalence = proportion of a people with a condition at one point in time (5000 preterm births in last 6 months, out of 60,000, prevalence = 5000/60000)

Incidence = Risk of developing a new condition (120 newborns followed, 45 newborns developed jaundice in that time)

Question 4

Relative Risk

Answer 4

Question 5

Absolute Risk Reduction vs. Relative Risk Reduction

Answer 5

Question 6

Relative Risk Reduction

Answer 6

Question 7

Absolute Risk Reduction

Answer 7

Question 8

NNT

Answer 8

Question 9

Precision vs. Accuracy

Answer 9

Question 10

Sensitivity vs. Specificity

Answer 10

SnNOUT = sensitivity high, negative rules out
SpPIN = specificity high, positive rules in

Question 11

PPV and NPV

Answer 11

*Both affected by prevalence therefore changing populations (ED vs. clinic) can change the PPV/NPV

Question 12

Odds Ratio

Answer 12

Question 13

Likelihood Ratio

Answer 13

Question 14

Null Hypothesis and P value

Answer 14

Question 15

Errors

Answer 15