Statistics Flashcards
Cohort study
Population followed over time- longitudinal study
- provides estimates of risk associated with a suspected causative factor
- rare exposure vs people without exposure and see in future what outcomes emerge
EX: Group split into people who have risk factor and who do not for certain disease and then followed up later to see who gets that disease and who doesn’t
- can calculate odds ratio and risk ratio
Case-control
- Retrospective study
- Have disease (rare) and don’t have disease then look back in time and try to identify exposure that may have led to condition
- can only calculate odds ratio
Relative risk
probability of an event occurring to all possible events (risk of developing lung cancer in people exposed or not exposed to smoke)
1= no difference between the two groups, no increased risk no association
>1= positive association, increased risk
<1= negative association, decreased risk
further from 1 the stronger the association
—If P value is not less than 0.05 or if confidence interval includes 0 the RR is not significant
RR= probability that an exposed person gets disease/probability that an unexposed person gets disease
Odds ratio
OR=Odds that a case was exposed/odds that a control was exposed
OR=1, exposure not associated with disease
OR >1, exposure is positive associated with disease
OR<1, expose is negatively associated with disease
further OR from 1 the stronger the association
Odds ratio
When is odds ratio similar to relative risk?
When the disease does not occur frequently among the exposed (disease is rare)
When does odds ratio overestimate risk?
When the outcome is more common such as in hyperlipidemia
Type 1 error
Occurs when the null hypothesis is rejected when it should have been maintained. Which means difference only due to chance
Power
the probability of finding the difference between two samples
- probability of rejecting the null hypothesis when it should be rejected
A probability of 1 means it will occur, a probability of 0 means what?
it will not
Regression analysis
method of predicting the value of one variable in relation to anther variable based on observed data
Incidence
the number of new cases/total number of people at risk
central tendency
central value in a distribution around which other values are arranged (mean, median, mode)
ANOVA
- set of statistical procedures that compares two groups and determines if the differences are due to experimental influence or chance
Regression analysis
using data to predict how the value of one variable in relation to another
null hypothesis
assumption that there are no differences between two samples of polulation
correlation coefficient
measurement of the direction and strength of the relationship between two variables
- -1 to +1
- closer to +1 or -1 stronger relationship
- shows nothing about cause and effect
attributable risk
absolute incidence of the illness in patients exposed to the condition that can be attributed to the expose
predictive validity
diagnosis allows the doctor to predict clinical course and treatment response
construct validity
the diagnosis is based on underlying pathophysiology and use of biomarkers to confirm disease
analysis of variance
set of statistical procedures that compares two groups and determines if the differences are due to experimental influence or chance
x^2
- binary predictor variable and one binary outcome variable
binary variable
two possible values, yes or no
continuous varialbe
will fall on range- height or weight
independent variables
manipulated by the experimenter
dependent variable
variables not manipulated by experimenter
T test
one binary predictor variable and one continuous outcome variable
ANOVA
- two or more binary predictor variable and one continuous outcome variable
correlation
one continous predictor variable and one continous outcome variable
validity
the degree to which an instrument measures what it is intended to measure
face validity
diagnosis based on a general consensus among experts
descriptive validity
based on characteristic features that distinguish it from other disorders
predictive validity
a diagnosis will allow clinicians to accurately predict treatment response and clinical course
construct validity
diagnosis is based on an understanding of the underlying pathophysiology
positive predictive power
ability of a positive test to predict a positive disease
true positives/(true positive+false positives)
kappa
number used for binary data and tells whether a given procedure or test produces reliable or reproducible results
correlation coefficient
reliability for nonbinary data such as continuous measurement
period prevalence
looks at the number of cases both existing and new during a specific time period
lifetime prevalence
proportion of people who have ever had a specific condition during their lifetime