Clinical Epi Flashcards
“Risk” in research
Equivalent to incidence. Risk = #events or cases or outcomes / # in study arm or group
Risk measures involving SUBTRACTION:
- Risk difference
- Absolute risk reduction
- Attributable risk
Risk difference:
Incidence of disease in one group minus the incidence in another
Absolute risk reduction:
Amount of patients spared in an adverse outcome related to treatment (same as risk difference, but specific for treatment studies)
Attributable Risk:
(Similar to risk difference). Equals the Incidence of exposed minus the incidence of unexposed
Number needed to treat:
= how many patients need to be treated to prevent one outcome event.
NNT = 1/ARR
If baseline risk goes up, NNT goes down
Risk measures involving DIVISION
- Relative risk
- Relative risk reduction
- Attributable risk percent
Relative risk:
(also called risk ratio)
relative risk = risk of outcome given an exposure, as compared to risk of outcome without exposure (incidence exposed / incidence unexposed)
Relative risk reduction:
(percentage of baseline risk that was removed or prevented with therapy)
Relative risk reduction = Incidence control - incidence treatment) / incidence control. Also = 1-RR
Attributable risk percent:
(the percent of risk in the exposed group that is attributable to the exposure or among the exposed, the percentage of risk that would be eliminated if the exposure had not occurred)
AR% = (incidence exposed - incidence unexposed) / incidence exposed
Type I error (alpha):
Incorrectly conclude that there is a difference, when there is not (false positive)
Type II error (beta):
Fail to find a difference when a true difference exists (false negative)
**determines statistical power
Standard deviation:
a measure of degree of variability in individuals in the sample (difference between an individual’s mean and the sample mean)
**not related to sample size
Standard error:
an indication of degree of uncertainty in the mean values of each group ( = SD / sq rt of sample size)
A higher sample size means a smaller SE
Test statistic:
= (effect size x sqrt sample size) / SD
The greater the test statistic, the lower the p value
