Biostats Flashcards
a test is negative, what is the chance that you really dont have the disease?
NPV
what is “matching” and what bias does it decrease
“matching’ is pairing subects of similar age/SS.. ex: CRC patients and their neighbors of similar race w/o CRC
the goal is to decrease confounding bias
what is the Hawthorne Effect?
Subject changes behavior once they know they are being watched
How do you predict the risk in a person with X behavior? Attributable Risk in Exposed?
ARP= 100* ( RR-1/ RR)
what study method SIMULTANEOUSLY measures exposures and outcomes
Cross Sectional Study- “a snapshot”
the probability of concluding there is no difference when one exists, is what?
Beta error
what is the power of a study
1-B
the probability of concluding there is a difference when there isn’t one (just chance)
alpha error
the patient tests positive, what are the chances that the patient has disease X?
PPV= TP/ TP+FP
what is resistant to outliers?
MODE
incidence v prevelance
new cases/pop@risk vs total cases/total pop
what is relative risk?
(the risk w/ drug X of getting disease Y)/ (the risk w/o X of getting Y)
T/F assistive walking devices show reduced risk of falls
FALSE- they do increase mobility. but need medication review in elderly to decrease falls
stratification decreases risk of what bias?
confounding
case control studies use WHAT stat measurement
Odds ratio
what is attrition bias?
loss of patients to follow-up from a specific group, can skew results, is a form of selection bias
screening tests should have a high___ and confirmatory tests should have a high ____
(1) sensitivity
(2) specificity
Number Needed to harm when comparing two treatments
1/ (absolute risk increase)
so step 1: calculate the risk of harm in drug X, and then drug Y
step 2: subtract the risk of harm in X from Y = absolute risk increase
step 3: do the inverse fraction of that= 1/ARI
how do you choose the 2 groups in a case control study?
(1) pts that HAVE the disease
2) pts that dont HAVE the disease (and have NORMAL exposure risks
case fatality rate=
(# of fatal cases of X)/ (# of fatal and nonfatal cases of X)
more sensitive screening tests can detect disease EARLIER and while the prognosis is actually unchanged, there is an “apparent increase” in survival time .. what is this bias?
Lead time bias
confused about if something is statistically significant? what should you check?
Confidency interval!
If it crosses the NULL, than it is not statistically significant
who is eligible for MediCARE
elderly >65
disabled: ALS, ESRD, Cerebral Palsy
in either a positive or a negative skewed bell graph curve what variables are the most skewed to the least skewed
Mean» median»mode
if you are trying to compare the means of 3+ groups what test would you use?
ANOVA
analysis of variance
a “snapshot” of “POPULATION” data and prevelance/incidence of X disease in a POPULATION is what kind of study?
Ecological Study
note cross-sectional is also a “snapshot” like this but used for individual’s data
if you stratify something and there IS a significant difference in the strata, it is ___
and if there isn’t a significant difference in the strata
effect modification
confounding bias
how to determine correlation coefficient
(1) is it positive or negative ( direct or inverse relationship, respectively)
(2) are the plot points kind of close together/looking linear-ish (then it is closer to -1 or 1)
if the plot points are more scattered, it is farther from -1 or 1, more like 0.2 or -0.2
whats the SD rule?
68/95/99.7
how do you decide how many people to treat before one person is cured?
1/ ARR
ARR= control event rate- experimental event rate
how to compare two categorical variables
Chisquare test
how do you do an Odd’s ratio for the occurence of X with drug Y
(((X w/ drug)/ (X without drug))) / (((no X with drug)/ (no X without drug)))
