Term
Total# of CV disease:
Control Group (N=8102): 546
Experimental (N=8506): 694
What is the absolute risk (or event rate) of each group? |
|
Definition
Control Event Rate (CER) = 546/8102 = 1.5%
Experiment Event Rate (EER) = 166/8506 = 2.0%
Note how the EER is bigger. This implies that experiment is doing harm (as opp. to being beneficial). |
|
|
Term
Total# of CV disease:
Control Group (N=8102): 546
Experimental (N=8506): 694
What is the relative risk (RR)? |
|
Definition
[image]
Experiment Control
RR = (694/8506) / (546/8102) = 1.21
Note:
if RR > 1 => causes harm
if RR < 1 => causes treatment
if RR = 0 => no difference |
|
|
Term
Total# of CV disease:
Control Group (N=8102): 546
Experimental (N=8506): 694
What is the Relative Risk Increase (RRI)? |
|
Definition
RRI = Experiment - Control
Control
RRI = (694/8506 - 546/8102)
546/8102
= 21%
Note: use RRI when RR > 1; otherwise use, RRR: Control-Experiment
Control
|
|
|
Term
Total# of CV disease:
Control Group (N=8102): 546
Experimental (N=8506): 694
What is the Absolute Risk Increase (ARI)? |
|
Definition
ARI = Experiment - Control
ARI = 694 - 546
8506 8102
= 1.5%
Note: when RR < 1, use ARR: Control - Experiment |
|
|
Term
Total# of CV disease:
Control Group (N=8102): 546
Experimental (N=8506): 694
What is the Number Needed to Harm (NNH)? |
|
Definition
NNH = 1/ARI
= 1/(EER-CER)
=1/(694/8506 - 546/8102)
= 70
=> You need to treat 70 people before one adverse side effect from treatment will occur
=> Bigger numbers are better for NNH (as opp. to NNT)
|
|
|
Term
Total# of fractures:
Control Group (N=8102): 788
Experimental (N=8506): 650
What is the absolute risk (event rate)? |
|
Definition
Control Event Rate (CER) = 788/8102 = 9.7%
Experiment Event Rate (EER) = 650/8506 = 7.6%
Notice how the control is bigger! This implies experiment is beneficial (as opp. to harm).
|
|
|
Term
Total# of fractures:
Control Group (N=8102): 788
Experimental (N=8506): 650
What is the relative risk (RR)? |
|
Definition
[image]
RR = (650/8506) / (788/8102)
= 0.79
Note:
if RR > 1 => causes harm
if RR < 1 => causes treatment
if RR = 0 => no difference
|
|
|
Term
Total# of fractures:
Control Group (N=8102): 788
Experimental (N=8506): 650
What is the Relative Risk Reduction? |
|
Definition
RRR = Control - Experiment
Control
RRR = (788/8102) - (650/8506)
788/8102
= 21%
Note: use RRR when RR < 1; otherwise use, RRI: Experiment-Control
Control
|
|
|
Term
Total# of fractures:
Control Group (N=8102): 788
Experimental (N=8506): 650
What is the Absolute Risk Reduction? |
|
Definition
ARR = Control - Experiment
ARR = 788 - 650
8102 8506
= 2.1%
Note: when RR > 1, use ARI: Experiment- Control |
|
|
Term
Total# of fractures:
Control Group (N=8102): 788
Experimental (N=8506): 650
What is the Number Needed to Treat (NNT)? |
|
Definition
NNT = 1/ARR
= 1/ (CER - EER)
= 1/(788/8102 - 650/8506)
= 48
=> You need to treat 48 people to prevent one bad outcome.
=> Smaller numbers are better (as opp. to NNH)
|
|
|
Term
For UGI bleed, what risk factors show harm? benefit?
[image] |
|
Definition
Harm: Smoking, Dyspepsia, Uncomplicated ucler, Complicated Ucler
Benefit: Female |
|
|
Term
For UGI bleed, which factor has the largest number of people? fewest?
[image] |
|
Definition
Most: female
Fewest: Complicated Ulcer |
|
|
Term
Which risk factor is the most convincing?
[image] |
|
Definition
|
|
Term
Which risk factor requires further investigation?
[image] |
|
Definition
|
|
