risk and odds ratios and survival analyses - guest video Flashcards
what is risk
the number of participants with outcomes divided by the number of participants
what is odds
the number of participants with the outcomes divided by the number of non-outcomes
what is the same and whats different : risk vs odds
the numerator is the same : number of ppl with the outcome but the denominator is the different
how to calculate risk ratio ?
find the risk of the exposed and the risk of the unexposed
then to find the Relative risk divided the exposed by the unexposed
relative risk is measured by - eg: the relative risk of lunger cancer is 5 times higher in smokers RR= 5
when do we use odds instead of risk
when we don’t have the denominator - number at risk only people who got the disease
- we compare the odds of smokers vs odds of control then divide those
in a cohort study what ratio do we compute?
risk or an odds
in a case-control we can only compute what ratio
odds
pros and cons of risk ratio
- easier to interpret but less desirable statistical properties
risk vs odds ratio
- when the outcome is rare then the odds and risk ratios are similar
- when the outcome is common the odds ratio will be much higher than the risk ratio
- usually an outcome of 15% in exposed and unexposed is “low” - odds and risk are similar
what is a differential follow up?
all participants are followed up for the full 10 years - generally does not occur ppl drop out
what is censoring?
drop out before the end of a cohort study
- called right censoring
what does censoring require?
us to adjust our odds or risk ratios to account for the differences in follow-up times
- if we don’t adjust it creates a bias in our results (especially true if there is a difference in the censoring of exposed and nonexposed)
what is left censoring?
when people experience the outcome before entering the cohort study - not usually a problem
how to adjust for censoring
instead of using participants, we use the person-years to calculate - find the incidence rate
make an incidence rate ratio
- properly accounts for differential followup
the easiest way to account for differential follow-ups
finding incidence rates and finding the incidence rate ratios
- relative risk is not applicable
what are Kaplan Meir curves
when we are studying a binary event with variable follow-up time we have two primary pieces of information
1. whether the event occurred - yes/no
2. the time to when the event or the last follow-up occurred
Why are Kaplan Meir curve graphs important
- summarize incidence rate ratios
- visualize the distribution of survival times with a KM curve
- show the probability of having no event at each time point
What does the Kaplon Meir estimator do?
takes into account the patients who drop out of the study
how do you find the probability of survival - Kaplan meir estimate
1-probability of death
eg: 1/95 ppl died = 0.01 therefore 1-0.01 = 0.98 = the survival rate of that year
calculating survival probability
read off the survival curve
find the year and then find the corresponding probability
curve gives us the info
What and when is the media survival time calculated/
- used as a summary statitsic
this is the point at which 50% of patients have died on KM curve
half of the participants died before this point and half will die after
what happens when follow up times differ in a cohort
any summary of the data must account for this
