W10_02 evidence based medicine 2

Question 1

note that logic, personal experience, and advice from others are not as good as clinical trials

Answer 1

ok

Question 2

recall the case of using the external carotid to bypass the internal carotid, and how that didn’t work

Answer 2

stroke rate didn’t change, but acute deaths increased with this experience

Question 3

recall the case of suppressing arrhythmias post-MI

Answer 3

anti-arrhythmics killed people because antiarrhythimcs paradoxically cause other arrhythmias!

Question 4

recall the case of the fluoride, which increased the bone mineral density, but resulted in more fractures

Answer 4

because fluoride increases cancellous bone, but decreases cortical bone

Question 5

what are the key aspects of RCTs?

Answer 5

1: equal chances of being in treatment or control
2: groups treated equally
3: all subjects accounted for at the end of the study
4: intention to treat
5: double-blind
6: all clinically relevant results reported
7: consider both statistical AND clinical significance

Question 6

what’s the difference in efficacy and effectiveness?

Answer 6

efficacy: does the Rx work under ideal circumstances?
Effectiveness: does the Rx work in usual circumstances?

Question 7

name 4 kinds of outcomes of clinical trials

Answer 7

mortality;
morbidity;
symptomatic status;
surrogate markers

Question 8

what are surrogate outcomes?

Answer 8

parameters that are risk factors for an illness

Question 9

note: be careful with surrogates. Just because someone’s cholesterol goes down doesn’t mean they will avoid the infarct

Answer 9

ok

Question 10

what’s a meta-analysis

Answer 10

quantitative summary of multiple trials for the same therapy, even if their results differ;
convert results into relative risk ratio or odds ratio, then make weighted average

Question 11

what’s type I error (stats)?

Answer 11

probability of a real difference even if there isn’t one. P-value. Alpha < 0.05 “proves” this

Question 12

what’s type II error (stats)?

Answer 12

probability of no difference when there really is one. Power. Power is 1-beta. Threshold is > 0.8.

Question 13

what’s relative risk reduction?

Answer 13

[(risk in untreated)-(risk in treated)] / (risk in untreated)

Question 14

what’s absolute risk reduction?

Answer 14

risk in untreated - risk in treated

Question 15

define “number needed to treat”

Answer 15

number of people to have to be treated before 1 person benefits.
NNN = 1/ARR

Question 16

define “number needed to harm”

Answer 16

NNH = 1/ARI(ncrease)

Question 17

note: good to consider patient values and how they view outcomes to weight the numbers and aide in decision making

Answer 17

ok