association and survival analysis Flashcards
positive association
higher value for on factor leads to higher value for the other factor (risk factor)
negative association
higher value for one factor leads to lower value for other factor (protective factor)
if 60/100 smokers get lung cancer, what is risk?
60/100 = 60%
if 60/100 smokers get lung cancer, what are the odds?
60/40 = 1.5
CIR = 1
CIR > 1
CIR
=1 = risk is same in exposed and unexposed >1 = risk is greater in exposed (risk factor)
IDR = 1
IDR > 1
IDR
=1 = rate is same in exposed and unexposed >1 = rate is greater in exposed (risk factor)
OR = 1
OR > 1
OR
=1 = odds are same in exposed and unexposed >1 = odds are greater in exposed (risk factor)
equation to estimate RR from OR
RR = OR/ 1-Rc + (Rc x OR)
attributable risk
rate of disease in exposed that can be attributed to the exposure
attributable risk proportion
proportion by which the incidence rate in the exposed would be reduced if the exposure was eliminated
population attributable risk
rate of disease attributed to an exposure in a population
population attributable risk proportion
proportion by which the incidence rate in a population would be reduced if the exposure was eliminated
current/period life table
summarizes mortality of the population over a brief period
cohort/generational life table
described the actual survival of a group of people born around the same time
clinical life table
describes outcomes in a cohort according to their treatment/exposure history
internal validity
degree to which the index and comparison groups are well selected/compared such that any difference between them can be attributed to the hypothesized effect under investigation
external validity
able to produce unbiased inferences regarding a target population
another term for external validity
generalizability
chance
an event that occurs at random and affects the outcome
how is chance accounted for in statistics?
P-value
define confidence interval 95%
95/100 times, the true value will fall into the range
if CI contains___, fail to reject null hypothesis
null value
null value for RR, OR, any ratio
1
null value for AR, PAR
0
what is selection bias?
bias due to systemic differences between those who take part in the study and those who do not
what is response bias?
bias due to systemic difference between those who choose to take part in the study and those who do not
what is another term for response bias?
self-selection bias
what is information bias?
flaw in measuring exposure/outcome data that results in differing accuracy of information between groups
examples of information bias (8)
interviewer, recall, reporting, instrumental error, measurement, digit preference, selective loss to f/u, misclassification
non-differential misclassification
same misclassification happens in all groups, tendency toward null value
differential misclassification
varied missclassifications in groups, more problematic
3 qualifications for a confoudning variable
1) risk factor for the outcome
2) associated with the exposure
3) not an intermediate in the causal pathway
interaction
interdependent operation of 2+ causes to produce or prevent outcome
effect modifier
factor that modifies the effect of a causal factor under study
how to control for 3rd variables in experimental designs?
randomization
restriction
limiting participation to individuals in a specific category of the confounder
matching
make study groups and comparison groups similar with respect to confounders
stratification
separate sample into many sub-sample classes
pooling
rarely used
multiple linear regression used for___
continuous dependent outcomes
multiple logistic regression used for ___
dichotomous outcomes
cox/poisson used for___
rate from cohort or experimental study are compared