Monday wk 4 Biostats 1 Flashcards
in normal distribution, mean, median, and mode should have __ values
in positive dist, inc number of observations __ than the mean
curve shifts to __, w mean affected most
then __ and minimal affect to __
thus, __ is best central estimate
In neg skewed distr, inc number of observations __ than the mean
curve shifts to __ w mean affecyed most
then __ and mode
therefore, __ is best central evaluation
equal larger right median mode median
less
left
median
median
O__ bias occurs when investigators missclassify data due to prior knowledge
occurs when outcomes are s__
prevent via b___
ie outside institutions are __ whereas in hospital groups are aware
__ bias is apparent prolongation of survival due to earlier screening w/out improved prognosis
__ rate is number of fatal cases divided by total number of cases
RR > 1 indicates \_\_ risk of dz RR < 1 indicates \_\_ risk only stat significant if 95% CI and \_\_ value are significant CI must not include \_\_ value for 95% CI, p is < \_\_ for 99% CI, p is < \_\_
observer
subjective
blinding
unaware
lead time
case fatality
inc dec p null .05 .01
__ ANOVA test determines if there were significant differences bw means of 2+ independent groups
compares __ of each group
tests __ hypothesis that all groups are random, and is rejected if 2 means are statisically different
mulitple logistic regression detects probability of binary __ against 1+ independent variable
two sample t test compares only __ means
case control must use __ ratio
__ risk has more utility
analysis of variance
means
null
outcome
2
odds
relative
adverse event rate aka __ risk or attributable risk
basically just I__
probability that an independent event will occur the same is multiplying the probability of __
1- P indicates one event that turns out __
time elapsed from initial exposure/patho insul to CM of dz is __ period in infectious dz
L__ period in noncommunicable dz
taking it to RF, some effect may occur __ time before outcome
or exposure to RF must occur __ over long periods to have an effect
absolute
incidence
each test
different
incubation
latent
long
continuously
Rare dz assumption states that OR is same as __ w low incidence of dz
For NPV to be applicable to pt, the pt pretest probability must be similar to dz __
w a given test result, probability is determined using __/__
C__ occurs when exposure disease relationship is muddled as it relates to exposure/disease
to minimize, perform S__
E__ modification is an external variable that pos/neg impacts observed effect of RF
should have sig difference in __ groups
RR
prevalence
NPV/PPV
confounding
stratification
effect
stratified
__ mortality is number of deaths due to dz over total pop
__ incidence takes into account new cases of disease over specific period divided by people at risk at beginning of period
does not include D__ or those who already had the dz
Woman cancer incidence: B, L, C, U, T
mortality L, B, C, P, O
Men cancer incidence, P, L, C, B, M
Men cancer death L, P, C, P, B
disease specific
cumulative
deaths
breast, lung, colon, uterus, thyroid
lung, breast, colon, pancreas, ovary
prostate, lung, colon, bladder, melanoma
lung, prostate, colon, panc, bladder
L__ bias occurs when those w rapidly prog forms of dz are less likely to be detected by screening compared to those w slow progressive dz
long term __ rates should be compared for screening study
degree of overlap on graph bw healthy/diseased pop limits S/S
if less overlap occurs, both values will __
a higher correlation coefficient indicates __ of assc
Small standard deviation indicates minimal __ as points are close to mean
__ of observations lie w/in 1 SD
__ observations lie w/in 2 SD
_ observations lie w/in 3 SD
for 2SD, means that 2.5% occurs __ or __ value
length
mortality
sensitivity, specificity
improve
strength
variation
68%
95%
99.7%
above/below
__ is percentage indicating actual difference in event rate bw control/tx group
eq is __
__ is percentage indicating relative reduction in tx event rate compared to control
eq is __
1-beta is P__, the study’s ability to detect a difference when it exists
ie rejecting null hypothesis when __
depends on size and difference in __
Beta is probability of type __ error, failing to reject null when it is false
will miss a __ relationship
Type 1 __ rejects null hypothesis when it is true
find a __ when one isnt there
similar to __, less than .05 is stat significant
ARR
CER-TER
RRR
CER-TER/CER
power
false
outcomes
2
true
error
outcome
p value
outliers least likely to affect __ or most common result
SD/variance measure __ and are thus sensitive to outliers
__ design takes people w or w/out dz and looks at RF
__ dz takes RF and looks for disease incidence
M__ is number that separates right half of data from left
if even number of observations, must use __
__ likelihood ratio is ratio representing the likelihood of having disease given positive resut
eq
__ likelihood ratio is ratio representing likelihood of having disease w neg result
eq
mode
dispersion
case control
cohort
median
half
Positive
Sensitivity/ (1-specificity)
1-sensitivity/specificity
LR >1 indicates that test result is assc w __
LR < 1 indicates result is assc w __ of dz
__ bias occurs in nonrandom way pt is directed to a group
__ bias is a rf that leds to more identification w inc probability of disease being detected
ie smoking
referral/admin rate bias is due to differance in _ practices
think __ of hospital
__ is ability to produce identical results
_ is ability to measure what it is supposed to
must be compared to __ test
if both are high, __ of the test is high
dz
absence
allocation
detection
admission
specialty
precision
accuracy
gold standard
reliability
__ study uses POPULATION data
generates a hypothesis but cannot make a conclusion aka ecologic __
cross sectional study takes pt w __ and looks for prevalence of outcome
__ effect is tendency of study subjects to change behavior as result of awareness they are being studied
usually affects B__ study
Berkson bias is choosing __ pt as control
P_ effect is that a researchers belief in efficacy of tx influence outcme
ecological
fallacy
RF
hawthorne
behavioral
hospitalized
pygmalion
__ study is one in which subjects randomly allocated to a sequence of 2 or more tx given consequatively
allows pt to be own __
must have __ phase to allow effect of drug to wear off
__ sample sizes have greater power to detect a difference
larger sample, greater chance to detect __
__ rate is ratio of indiv who became ill over number of indiv at risk
used for __ investigations
Item responsible usually has largest attack rate and __ in attack rates bw those who consumed and those who did not
crossover
control
washout
larger
a difference
attack
outbreak
difference
__ value determines whether result is +/-
depending on sensitivty/specifiity preference, value can be __
attributable risk percent in exposed is excess risk due to __
calcuulated as 100* (risk in __ - risk in unexposed)/risk in exposed
changes to 100* (RR-1/__)
__ is probability of developing dz over time period w exposure
divide affected subjects by total number of subjects in __ group
essentially just _
cutoff
adjusted
exposure
exposed
RR
Risk
exposed
sensitivity
in reality, normal distribution, 95% of observations lie w/in __ SDs of mean
99% lie w/in __ SD of mean
since __ is used, variability is present
must use standard deviation of a series of means aka standard __
estimates how far sample mean is from unknown population __
uses __ as it better accounts for variability by including SE in the calculation
eq is __ +/- z score x SE
z score is number of __ containing deisred percentage of obs
aka 95% or 99%
ex is mean +/- 1.6 x SD/SRn
as n inc, SE dec and CI becomes more __
as SD inc, SE inc and CI becomes less __
pay attention for CONDFIDENCE __
1.96
2.58
SAMPLES
ERROR
MEAN
95% ci
mean
precise
precise
interval
