Biostats Flashcards
What parameter helps you interpret the significance of a single result in a single patient, regardless of prevalence?
Likelihood ratio
The LR tells you how likely a single result is to be truly positive or negative.
Positive LR = sensitivity / (1 - specificity)
Negative LR = (1 - sensitivity) / specificity
Describe verification bias.
Verification bias occurs when the gold standard workup is applied selectively to a given population. Most often this happens when the gold standard test is invasive or costly. Applying it selectively will miss positive results in patients not worked up and may make the selected population seem sicker.
This is also called workup bias.
What is a standardized incidence ratio?
It is a way of knowing if the incidence of something in a given population is higher or lower than expected.
The formula:
SIR = (observed cases) / (expected cases)
What is an effect modifier?
An effect modifier is a factor that modifies the effect of something else. For instance, age in Reye syndrome is an effect modifier for aspirin.
Explain in words what an odds ratio greater than 1 means.
Exposure to something increases your risk of developing a defined outcome
In a negatively skewed distribution, what are the relationships (i.e., greater than or less than)?
Mean ____ median ____ mode
less than; less than
“Skewing” refers to where the mean sits in relation to the bulk of the population. In a negatively skewed plot, the mean is less than the median and mode because there are low outliers.
In a positively skewed distribution, what are the relationships (i.e., greater than or less than)?
Mean ____ median ____ mode
greater than; greater than
“Skewing” refers to where the mean sits in relation to the bulk of the population. In a positively skewed plot, the mean is greater than the median and mode because there are high outliers.
If you are finding the average incidence of something occurring across two unequal groups, this is called a ________ average.
weighted
_____________ = (incidence of disease in exposed group) / (incidence of disease in unexposed group)
Relative risk
A type I error is a false __________.
positive
(“Pos1tive”)
A type II error is a false ____________.
negative
(“Ne2ative”)
Smaller studies have a higher rate of type ____ errors compared to larger studies.
II
The smaller population means lower power.
SMR stands for what?
Standardized mortality ratio
This is a value that looks at the observed number of deaths from a particular cause divided by the expected number of deaths. It helps to evaluate if there is a higher or lower than expected incidence of death from a specific cause.
One problem with studies looking at mortality rates among working people is that _______________.
working people are generally healthier than non-working people, something called the healthy worker effect (HWE)
Review the formulas for positive and negative likelihood ratios.
PLR = sensitivity / (1 - specificity)
NLR = (1 - sensitivity) / specificity
Just remember you “like sensitive specifics” and then the formulas can tell you which is negative and positive. Having a high specificity makes it more likely to be a true positive.
