# biostats

## Terms

undefined, object

copy deck

- define sensitivity
- # of true positives divided by all people with disease
- define specificity
- # of true negatives divided by all people without disease
- positive predictive value
- # of true positives divided by all people who tested positive
- negative predictive value
- # of true negatives divided by all people who tested negative
- what characteristic is desirable for a confirmatory test?
- high specificity
- what characteristic is desirable for a screening test
- high sensitivity
- define odds ratio
- estimate of relative risk because don't have incidence data (case-control)
- define relative risk
- incidence of disease in person exposed to risk factor / incidence of disease in person not exposed to risk factor
- what type of studies use odds ratio?
- case-control studies
- what type of studies use relative risk?
- cohort studies
- how are samples chosen for a case-control study
- chosen based on presence (cases) or absence (controls) of disease
- what type of information is being collected in a case-control study?
- risk factors
- how are samples chosen in a cohort study?
- presence or absence of risk factors
- what effect does a meta-analysis have on limitations or bias in studies?
- can't overcome limitations
- what does a meta-analysis try to achieve?
- greater statistical power
- what type of information is being collected in a cohort study?
- development of disease
- define clinical study
- compares therapeutic benefits of 2 or more treatments or treatment/placebo
- bias: subjects permitted to choose whether to go into drug group or placebo group rather than being assigned randomly
- selection bias
- knowledge or presence of disorder alters remembering by subjects (i.e. parents of kids w/ birth defects)
- recall bias
- subjects who volunteer to be in a study are not representative of population being studied
- sampling bias
- information gathered at an inappropriate time
- late-look bias
- name 4 ways to reduce bias
- blind studies, placebo responses (having control group), crossover studies, randomization
- define crossover study
- each subjects acts as own control (patient receives drug then later on receives placebo, and vice versa)
- define prevalence
- total cases in population at given time/total population
- define incidence
- new cases in population over a given time / total population at risk during time
- is prevalence or incidence greater for chronic disease?
- prevalence
- is incidence or prevalence greater for acute disease?
- incidence
- what is the relationship between prevalence and incidence (mathematically)?
- prevalence = incidence x disease duration
- what factor do you need to account for in calculating incidence?
- people previously positive for disease are no longer considered at risk
- how do you mathematically calculate the false negative rate?
- 1-sensitivity
- how do you mathematically calculate the false positive rate?
- 1-specificity
- when does the odds ratio approximate the relative risk?
- when the prevalence of the disease is not too high
- define precision
- how reliable is the estimate (are all values closer together?)
- define accuracy
- how close is the estimate to the true mean
- what is the difference between random error and systematic error?
- random error: reduced precision in test; systematic error: reduced accuracy in test
- what is the relationship between mean, median, and mode in a gaussian distribution?
- all are equal
- what is a bimodal statistical distribution?
- 2 modal peaks - two distinct populations
- what is the relationship between mean, median and mode in a positive skew
- mean>median>mode
- where is the tail of a positive skew?
- to the right
- what is the relationship between mean, median, and mode in a negative skew?
- mean<median<mode
- where is the tail in a negative skew?
- on the left
- define null hypothesis
- hypothesis of no difference
- define alternative hypothesis
- hypothesis that there is a difference
- define type 1 error
- stating there is a difference where no exists
- what greek letter defines type I error?
- alpha
- define p value
- probability of making type 1 error (say there's a difference when there isn't)
- define type II error
- stating there is no difference where one exists
- what greek letter defines type II error?
- beta
- what is the mathematical equation for power?
- 1-beta
- define power
- probaiblity of rejecting null hypothesis when it is indeed false
- how do you increase the power?
- increase sample size
- what 3 factors does power depend on?
- total number of end points experienced by population, difference in compliance between treatment groups, size of expected effect
- how many people are within 1 standard deviation
- 68%
- how many people are within 2 standard deviations
- 95%
- how many people are within 3 standard deviation?
- 99.7%
- what does SEM stand for?
- standard error of the mean
- what is the mathematical equation for standard error of the mean?
- standard deviation/square root (sample size)
- what is greater, standard error of the mean or standard deviation?
- standard deviation
- what is the relationship between sample size and standard error of the mean
- SEM decreases as n increases
- definition: range of values in which a specified probability of the mean of repeated samples would be expected to fall
- confidence interval
- if the 95% CI for a mean difference includes 0, what does this signify?
- no significant difference, null hypothesis not rejected
- if the 95% CI for an odds ratio or RR includes 1, what does this signify?
- null hypothesis is not rejected
- evaluate presence of statistically significant differences between 2 groups
- T-test
- test that evaluates the presence of statistically significant differences between 3 or more groups
- ANOVA (analysis of variance)
- analyze categorial data or compare proportions
- chi squared
- what is the mathematical expression for coefficient of determination
- r(squared)
- between what 2 integers is r always located?
- -1 and 1
- what does the absolute value of "r" indicate?
- strength of correlatoin between 2 variables
- what is already known when calculating sensitivity?
- person has disease
- what is already known when calculating specificity?
- person doesn't have disease
- what is already known when calculating PPV
- preson already tested positive
- what is already known when calculating NPV
- person already tested negative
- what factors does prevalence have a relationship with?
- positive predictive value, negative predictive value
- what is the relationship with prevalence and PPV
- higher the prevalence, higher the PPV
- what is the relationship between prevalence and NPV
- lower the prevalence, higher the NPV
- how do you calculate confidence interval
- mean of sample plus/minus (Z score * SEM). 95% CI (z=2). 99.7% CI (z=3)
- what is the relationship between accuracy, precision, and the width of the CI
- wider the CI, the less precise, the more accurate (wider CI more likely to contain mean)
- how to calculate an odds ratio
- (person with risk factor that has "disease" x person without risk factor that does not have "disease) / multiple two opposites