All the biostat stuff Flashcards
1
Q
class of statistical methods for studying the occurrence and timing of events
A
survival analysis
2
Q
survival analysis can also be applied to
A
development of disease
response of treatment
relapse of disease
rehospitalization
quitting smoking
3
Q
percent of patient alive 5 years alfter treatment begins or 5 years after diagnosis
A
5-year survival
measure of prognosis
4
Q
probability of remaining alive for a specific length of time
A
survival
5
Q
subjects who are sensored when the information about survival time is incomplete
A
-lost to follow up
-still alive at the study termination date
6
Q
the set is said to be complete when there are
A
no censored survival times
7
Q
probability that an individual survives longer than some time
A
survival function
S(t)=Pr
kaplan-meier method estimates S(t)
8
Q
the hazard function describes
A
the conditional failure rate
9
Q
the probability of dying during a very small time interval, assuming the individual has survived to the beginning of the interval
A
hazard time
10
Q
hazard function
A
h(t)= # of patient dying near t/
total # of patients surviving near t
11
Q
describes the probability that an event has occurred by time t
A
cumulative hazard
the greater the value at time t, the greater risk of death by time t
12
Q
represent how much more or less likely the event is to occur in one group relative to a control
A
hazard ratio
13
Q
methods for comparing (S)t
A
log rank test
cox regression
14
Q
the time at which 50% of the subjects have had the event
A
median survival
estimated using Kaplan meier curve
15
Q
generated by even analyis
A
risk ratio
16
Q
risk ratio
A
probability of event in treatment group/probability of event in control group
17
Q
HR=1
A
rate of the event is the same in both groups
18
Q
HR>1
A
rate is higher in treatment group
19
Q
HR<1
A
rate is lower in treatment group
20
Q
RR=1
A
probability of the even is the same in both groups
21
Q
RR>1
A
probability is higher in treatment group
22
Q
RR<1
A
probability is lower in treatment group
23
Q
alternative hypothesis
A
H1 is generally a statement of effect, association, or difference
24
Q
null hypothesis
A
H0 is generally a statement of no effect, no association, no difference
25
provide standardized evidence of an effect, association, or difference
test statistics
26
t tests are appropriate for
comparing means when the outcome is continuous
27
chi-square tests are appropriate for
comparing frequencies between two groups when the outcome is categorical
28
measure of premature mortality
years of potential life lost
29
combines quantity + quality of life
quality adjusted life years
30
average number of additional years of life gained from intervention, multiplied by quality of life "weight" in each of those years
quality adjusted life years
31
QALYs: weight of 1
perfect health
32
QALYs: weight of 0
equivalent to death
33
used for evaluating new therapy/intervention on population basis
QALYs
34
measure of overall disease burden, expressed a cumulative number of years lost due to ill health, disability, early death
disability adjusted life years
35
risk factors for hypertension
increased age
race (black)
increased risk for men <45 over women
36
review study data at pre set time points to ensure participant safety
data safety and monitoring boards
37
when might the data safety and monitoring board stop a study early?
intervention too risky
benefit of intervention is clearly demonstrated
38
factors to minimize selection bias
clearly defined inclusion/exclusion criteria
randomized to study assignment groups
little loss to follow up
39
ways to minimize confunding bias
randomization
intention-to-treat analysis
40
determin whether a clinically relevant difference exists between two interventions
superiority study
41
determined whether a new treatment is neither worse nor better than another established treatment
equivalence trial
42
determine whether a new treatment is not inferior to another established treatment
non-inferiority study
43
points to consider for study generalizability
inclusion/exclusion criteria
biologic mechanism
strength of findings
44
results/data made publically available prior to author analysis and peer review
early termination
45
type I error
when you reject the null hypothesis but you shouldnt have
| major reason for replication
46
probability of making type I error
Alpha
47
type II error
when you dont reject the null hypothesis but you should have
| often occurs due to inadequate study power
48
Beta
probability of making type II error
49
how to avoid type I and II errors
careful study design and adherence
50
alpha=0.05
investigator is willing to accept a 5% risk of falsely concluding groups differ
| type I error
51
beta = 0.20
investigator accepts a 1 in 5 chance of missing a true difference between groups
| type II error
52
probability that a study can detect a true difference between groups
power
| 1-beta
53
how to calculate sample size
1. determine study design
2. set acceptable levels of type I and II error
3. determine magnitude of difference in outcomes between 2 groups
4. calculate size requirement
54
larger differences require [larger/smaller] sample sizes in order to be detected
smaller
55
reduced risk of type I error requires [larger/smaller] sample size
larger
56
reduced power results in [larger/smaller] sample size requirement
smaller
57
therapeutic goal of warfarin
prolongation of PT time
58
INR 1.5-2
low intensity anticoagulation
| long term
59
INR 2-3
moderate intensity anticoagulation
| intitally
60
INR 2.3-3.5
high intensity anticoagulation
| mechanical prosthetic heart valves
61
therapeutic goal of warfarin is acheived in
about one week
62
factors affecting warfarin PK
diet
GI status
other drugs
genetics
63
warfarin dosing has
wide inter-individual and intra-individual variability
64
results show a difference between the intervention and control groups, but in reality, the two groups are the same with respect to the outcome of interest
type I error
65
researchers conclude the groups are the same, but in reality they are different
type 2 error
66
to calculate relative risk improvement
risk in the intervention group - risk in the control group
then, divide by risk in the control group
67
NNT calculation
1/ARR
| ARR= absolute risk reduction
68
calculate ARR
risk of event in control group - risk of event in treatment group
| event can be negative or positive
69
calculate ARR
risk of event in control group - risk of event in treatment group
| event can be negative or positive
70
p value <0.001 would indicate
if the null hypothesis is true, there is less than 0.1% probability of obtaining a test statistic equal to or more extreme than the one obtained
71
if the confidence interval contains 1
finding is non significant
corresponds to p value >.05
72
p value >.05
non significant
73
phase of trialing that examines efficacy as primary standpoint
phase 2
74
phase of trialing that focuses on efficacy AND safety
phase 3
75
why is human testing still required for new drugs?
differences in disease processes and stages between animals and humans
intervariability in humans that is not seen in inbred testing animals
76
if the confidence interval contains 1
no statistical significance between groups
77
if the CI does not contain 1
statistically significant
78
a hazard ratio of 1 means
no difference in the rate of outcome between 2 groups
79
a chi square test is used
to evaluate if two proportions are different
80
an ANOVA is used
to compare mean values of more than two groups
81
a t test is used
to compare means of continous variables between two groups
82
occurs when sample selected is not representative of target population
sampling error
83
when the results of the intervention are not different, but researchers conclude that they are
type I error
84
when the results of interventions are different, but researches conclude that they are not
type II error
85
ensures equal distribution of possible confunding variables between intervention and control groups
randomization
86
deviation from results of inferences from the truth or processess leading to such deviation
bias
87
any trend in the collection, analysis, interpretation, publication, or review of data that can lead to conclusions that are systematically different from the truth
bias
88
results from flawed procedures in collecting data or imperfect definitions of study variables
information bias
89
distortion of the estimated of the effect of an exposure of interest on an outcome because it is mixed with the effect of an extraneous factor
confunding
90
resemblence of study population to the larger population from which it was drawn
external validity
91
this is present when the study results are obtained in an unbiased manner
internal validity
92
individual studies are at risk of inaccurately estimating the exposure-outcome relationship due to
bias and sampling variability
93
individual studies may not generalize across
heterogeneous populations and settings
94
how can we determin the current state of evidence?
1. define a question
2. search literature
3. assess the studies
4. combine the results
5. put findings in context
95
heterogeneity
how consistent is the effect across studies?
96
in the absence of bias, funnel plots should
resemble a symmentrical, inverted funnel
97
sampling variety [increases/decreases] as the sample size [increase/decreases]
decreases; increases
98
smaller studies showing no statistically significant effects remain unpublished, contributing to
publication bias
99
small differences in inclusion/exclusion cirteria can produce large differences in the set of studies included in the analysis
selection bias
100
informations produced where publishing is not the primary activity of the producing body
grey literature
101
some examples of grey literature
thesis, disseratations
conferences papers/posters
clinical trial data
government documents
102
why use grey literature?
more complete review of topic
good for data
may include negative results
faster
103
Agency for Healthcare Research and Quality
NICHSR ONESearch
The Health and Medicine Division Reports
Agency for Healthcare research and quality
provide what info?
grey literature database resources
104
BMC
Web of Science
provide what info?
BMC
Web of Science
conference abstracts and proceedings
105
clinicaltrials.gov
cochran central registrar of controlled trials
IRSCTN registry
provide what info?
clinical trial info, policy info, regulatory data, health stats
106
what study design can demontrate a causal relationship?
cohort
107
framework for developing a clinical question
PICO
Patient/problem
Intervention
Comparison
Outcome
108
optimal study design for impacts of treatments and therapies
randomized control trial
109
framework for QI projects
Plan Study Do Act
