Exam Two - Observational (prognostic) studies Flashcards
prognosis relates to ….
- evaluation of the timecourse for a disease
- potential influence of exposure to various factors
prognosis in PT is used to…
- identify the risk of developing a future problem
- predict the ultimate outcomes of impairments or functional limitations
prognosis is often evaluated using observational study designs such as…
- prospective cohort
- retrospective cohort
- case control
T or F: statistics used in prognostic studies are also common in experimental studies with categorical outcomes
T
Relative risk definition
- the likelihood that someone who has been exposed to a factor will develop the outcome of interest
- cumulative over an entire study, using a defined endpoint (only use beginning and end, nothing in the middle!)
how to calculate RR?
top box has yes/no condition left side has if they were exposed or not. (label a,b,c,d)
rr = (a/(a+b))/(c/(c+d))
how to interpret RR?
<1 = decreased likelihood
>1 = increased likelihood
=1 = no prognostic value
how much does exposure matter?
rr (increase or reduction) = |(1-rr)| x 100
when is rr statistically significant?
when confidence intervals do not contain one
hazard ratios
- represent instantaneous risk during the study time period
- better for indicating risks that happen at any time during the study interval
- know that this is different than RR!
hazard ratios and survival cues, what is it?
“survival cues” represent the percentage of each group surviving versus time
_________ is concerned with the total events over time while _________ is concerned with likelihood of events at any time.
relative risk
hazard ratio
Odds ratio
odds of experiencing an exposure among people with and without the outcome
- retrospective estimate of relative risk
- commonly used in retrospective cohort studies or case control studies
how to calculate odds ratio
cross multiply the box
- only useful when one factor is assessed, if multiple factors then you should use logistic regression
odds ratio answers the question of:
is it more likely to find an exposure among people with the outcome
T or F: odds ratio uses the same interpretation as relative risk
true
<1 = decreased likelihood
>1 = increased likelihood
=1 = no prognostic value
What factors should you consider when looking at prognosis study quality?
enrollment
follow up
measurement of outcome and blinding
outcome reporting
what are some questions you should be concerned about when analyzing prognosis study quality
- was the cohort assembled at a common point in their course of disease?
- was patient follow up sufficiently long?
- was patient follow up complete?
- how objective is the outcome?
- results reporting?
why should you be concerned if the cohort was assembled at a common point in the course of their disease?
- best if enrolled at a uniform early time in the disease - called inception cohort
- limited survivor bias (good outcomes over-represented because more advanced disease associated with poorer prognosis)
why should you be concerned if the patient follow up was sufficiently long?
- length of time needs to be long enough to detect the outcome of interest and this varies depending on the outcome
why should you care if patient follow up is complete?
- because you don’t want high drop out rates
- if people do drop out, you need to explain why
why should you care about how objective the outcome is?
- how is variable operationally defined
- are results open to interpretation
- were results determined by rates who are masked to prognostic factors?
why should we care about how results are reported?
- survival curves and hazard ratios are preferred/better/more contemporary
- RR or odds ratios only give beginning and ending values