22-28 lectures Flashcards
(138 cards)
1
Q
what are the 4 catagories that chance is divided into?
A
validity
sampling error
confidence intervals
p-values
2
Q
what is validity split into?
A
internal and external validity
3
Q
what is internal validity?
A
we need to consider chance, bias and confounding
its when we ask are there other explanations for the study findings, apart from them being right
4
Q
what is external validity?
A
the extent to which the study findings are applicable to a broader or different population
5
Q
how do we estimate the parameters of the actual prevalnce/incidence/RR/OR/RD?
A
we you the data from a study population parameters to take an estimate
6
Q
what is sampling errors?
A
when samples would be quite different to the others samples and this is up to chance and is refferend to as sampling error
7
Q
how do we reduce chance?
A
we can increase the number of samples in the study which reduces sample variablity and increases the likelihood of getting a represtative sample and increases the percision of the parameters
8
Q
what percentage is confidence intervals?
A
95%
9
Q
what does the 95% confidence intervals represent?
A
the range of values within which the parameter will lie 95% of the time if we continue to repeat the same study with new samples
10
Q
where are our odds ratio found?
A
between our 2 confidence intervals
11
Q
what can we use the 95% confidence intervals help us with?
A
to make predictions where our prevalence and OR will fall
tell us whether the study findinhgs are clinically important
12
Q
what are parameters?
A
the true value of the measure in the population that the study is trying to recover
13
Q
what is the point estimate?
A
the measure found in the sample study
14
Q
what are p-values?
A
the possability of getting estimate when there is no association just because of sampling error (chance)
15
Q
if the p-value possibility is really low then what?
A
it is unlikely to be that the estimate is due to sampling error (chance)
16
Q
what is H0?
A
the null hypothesis
17
Q
what is the null hypothesis?
A
there is no association in the population so the RR,OR = 1 and the RR = 0
18
Q
what is Ha?
A
alternative hypothesis
19
Q
what is the alternative hypothesis?
A
when the parameter does not equal the null value so the RR, OR doesnt equal 1 and RD doesnt equal 0
20
Q
what is statisical significance?
A
the threshold we set because of sampling error (type-1 error)
21
Q
what happens if the p less than 0.05?
A
then we reject H0 and accept Ha so association is not statistically significant
22
Q
what happens if the p is grater than 0.05?
A
then we fail to reject H0 and reject Ha so the association is significantly significant
23
Q
what can no association be beacause?
A
chance
24
Q
what if we dont have association of the true (unknown) and association of study results?
A
then we have a type-1 error
25
what if we have association of the true (unknown) and not the association of the study results?
then we have a type-ll error
26
what are type-ll errors?
incorrectly rejection of H0 when it shouldve been
27
what causes type-ll errors?
due to having too few people in the study
28
how to we not get type-ll errors?
by having a bigger sample size to have a smaller p- value
29
what happens if the CI include the null value?
the p-value is greater that 0.05 so it is not statistically significant
30
what happens if the CI doesnt include the null value?
then the p-value is less than 0.05 so it is statistically significant
31
what is CI?
confidence interval
32
what are limitations of the p-value?
arbitrary threshold
only about H0
nothing about importance
33
what is arbitrary threshold?
the statistical significance threshold is arbitrary and artificial
its useful for reporting p-values rather than just the statistical significance
34
what is only about H0?
just giving evidence about consistancy with the null hypothesis
not really percise
35
what is nothing about importance?
statistical significance is not clinical significance
does not say that the results of the study are valid useful or correct
36
what is bias?
a systematic error in an epidemiologic study that results in an incorrect estimate of the association between exposure and risk of a disease
37
what is random error?
something that doesnt really have a pattern with it
38
what is sytematic error?
iif there is a pattern that we cant change no matter the amount of samplinh that we do
39
what can systematic error be?
an over-estimation
an under estimation
not affected
40
when is selection and information bias collected?
can only be controlled during the design and data collection phases of a study
41
what are 3 potential sources of bias when collecting data?
selection bias
information bias
publication bias
42
What is selection bias?
who is going to be in the study
43
when does systematic bias occur?
when the systematic difference between the people who are included in the study and those who are not, or when study and comparison groups are selected inappropriately or using different criteria
44
where is there systematic difference?
between those who are included and those who are not included
45
what is recruitment?
how people can get recruited into the epidemiology study one way is an ad for people to notice
46
what is random selection?
better at giving us a better representation for there to be minimised bias course
47
what do people who dont agree to participate and can lead to?
systematuic error
48
why do we want to maximise the response rate?
because the difference of people who are in the investigation and who isnt can lead to a negative affect on the data
49
what can minimise the amount of people from dropping out?
following up
50
what are some ways to minimise lose to follow up?
-alternative contract details obtained at the start of the study
-maintaining regular contact
-making several attempts to contact people
51
what are cross sectional study designs?
the exposures and outcomes are assessed at one point of time and we can measure the prevalence of the study
52
when is comes to bias in cross sectional study designs what do we have to consider?
who entered the study
sample represntative of source population
what is the response rate
52
what happenns if the selection is dependant on their outcome status in case controlled studies?
exposure and outcome have already occurred
cases and controls are selective separately
52
what are case controlled studies?
the cases and controls goes to the measure past exposure when considering bias
53
what is the MOA in a protective factor?
underestimated beacsue of bias the measure of association is under estimated and the value is towards the null but estimation is bias numerically upwards
53
what is the MOA in a harmful factor?
underestimated because of bias the measure of association is under estimated and is numerically downwards
53
what are participatants in cohort studies?
classified of exposure and followed up over time
54
what is the loss of follow up?
the person that leaves and loses all there data and can be called a bias data situation
54
what are causes bias in cohort studies?
lose to follow up of people who may be related to the to the outcome of the study
55
what is bias in RCT?
could be used if people who picked the people for this study and knew which group that the people would go into there exposure is assigned and not measured
lose of follow up can lead to bias in the experiment
56
what is selection bias?
is the difference between who enters and who leaves the study
57
what is the GATE study useful to see?
to see whats happening in a study when it comes to bias
58
what is information bias?
collected by people who are involved with the study
59
how does measurement error occur?
by participants providing inaccurate responses as they cant remember that clearly so could lead to being bias self reported information like this is called subjective
60
what is a subjective measure?
pain
61
what is objective measure?
using a peice of equipment like a faulty tool
62
what is measurement error?
random lack of percision
63
what is systematic error?
a lack of accuracy
64
what can measurement error lead to in a descriptive study?
over/underestimation of prevalence
65
what can measurement error lead to in analytic study?
misclassification which can be due to random or systematic error
66
what can misclassification split into?
non-differential and differential
67
what is non-differential misclassification?
when there is not different between the study groups
when the measurement error and any resulting misclassification occur equally in all groups being compared
68
what is differential misclassification?
when there is different between the study groups
69
which group does misclassification effect in thne outcome?
both groups
70
what does misclassification do to the null value?
moves it closer to the null value
71
how can recall bias occur in case control studies?
if they have to recall past exposures
72
what is a recall bias?
systematic error due to the differences in accuracy or complteness of recall to memory of past events or experiences
73
what does recall bias do to the OR?
incraeses the bias numerically upwards and the ratio value moves away from the null value
74
how can we minimise recall bias?
we use iobjective measures instead of self-reported subjective measure
75
what do we have to consider in cohort studies?
if they have been classified correctly
76
what has already happened in historical studies?
the outcome
77
what is the observer bias?
if the interveiwer knew the exposure of the group so they might ask more probing questions
78
how do we minimise interveiwer/observer bias?
we clearly define study protocols and measures
structured questionaire and standard prompts
we can train the interveiwer or we can use blinding
79
what can minimise RCT bias?
blindingh
80
what are 3 ways on how we can minimise information bias?
collecting information from participants
measurement instruments
collecting information
81
what do we do in collecting information from participants?
validating survey instruments
validate using objective measure such as looking at medical records
82
what are measuerment instruments used for?
use standardised equipment, should be the same for all participants
use calibrated equipment, everything is working equally
83
what is collecting information vai interveiws/observers?
blinding
use objective measures
use structured interviews and standard prompts
training of interviewers
84
what is publication bias?
the publisher reports things that are not statistically significant or positive results
85
what is confounding?
muddling the effects when the relationship we are interrested in is confused by the effect of something else
86
what is the third factor that changes the risk outcome?
confounding
87
what are the 3 properties of a potential confounder?
independently associated with the outcome
independently associated with the exposure
not on the casual pathway
88
what is independantly associated with the outcome?
a risk/protective factor for the outcome by itself
89
what is independently associated with the exposure?
different proportions of people with potential confounder across exposure groups
90
what is 'not on the casual pathway'?
not the mechanism by which the exposure affects the risk of the outcome
91
what does confounding effect?
over-estimation of the true association, under-estimation of the true association,
change direction of the true association,
give appearance of a association when not one
92
how do we identify the potential confounders?
we look for an imbalance between groups
93
what are the 3 ways to control confounding?
randomisation
restriction
matching
94
what does randomisation apply to?
anything we didnt know about and we dont have to know all the potential confounders in the study
95
what is the only study that randomisation can be used on?
RCTs
96
what is the only technique that can be used for known and unknoiwn confounders?
randomisation
97
what does randomisation need?
equipoise and intention-to-treat
98
what is restriction?
taking a particular confounder which is called a stratum
99
what does restricting one confounder do?
we are able to make sure that they are alike.
can reduce generalisability and reduces the number of potential participants
100
what does restriction have the potential for?
residual confounding with impercisly measured confounders and usually only one potential confounder
101
what is matching?
when we choose people to make the control/comparison group have the same composition as the cases/exposed group reguarding the potential confounders
102
what are matching used for?
in case-controlled studies
103
what are the 2 groups that matching are split into?
individual and frequency
104
what is individual matching?
each case mained with one or more controls having the same confounding visable characteristics
105
what is frequency matching?
matching at the aggregated level
106
what are the positives of matching?
useful for diffeercult to measure potential confounders and can improve effciency of case controlled studies with small numbers
107
what are some negatives of matching?
individual matching can be differcult and limit potential numbers of potential participants
they also need special matched analysis for individual matching
108
what are 3 ways that we can control confounding?
stratification,
multivariable analysis,
standardisation
109
what is stratification?
calculating measure of association for each stratum of potnetial confounder and comparing them
110
how do we stratificate?
first calculate the measure of association between exposure and outcome
we then divid the potential confounder into strata
for each stratum we calculate the measure of association between the exposure and outcome
we then compare stratum specific measures of association
111
what are some pros of stratification?
easy for small numbers of potential confounders with limit strata
they can evaluate imopact of confounding and can identify effect modification
112
what are some cons of stratification?
can leave residual confounding and can not be feasible when dealing with lots of potential confounders with many strata
113
what is multivariable analysis?
statistical method for estimating measure of association whilst controlling for multiple potential confounders
114
what are variety of different techniques are recognisable the term what?
regression
115
what is standardisation?
used in measuring of disease occurance when we have 2 properties
116
what are the limitations of standardisation?
have similar issues as stratification with multiple potential confounders of strata
117
what are potential issues of controlling is study analysis?
result in residual confounding and can only control what youve measured
118
why do we need to research ethics oversight?
to stop people from consenting to things that are unacceptably harmful
119
what needs to be considered to research to be acceptable
-asses the benifits and harms and ensure the ratio is acceptable
-beaware of potential vulnerabilities of participants
-avoid or manage conflicts of intrest
-obtain informed consent from participants
-consider how the benefits and burdens of the research should be shared across society
120
what is beneficience?
refers to the obligations that we have to benefit others
121
what is non-maleficence?
refers to the obligations that we have not harmed others without a justifying reason
122
what do research ethics committees require applicants to show?
-an awareness of the various costs or harms to participants includiong time, resources, coercive factors, and any oppotinity costs
-stratagies to address these harms and costs
-an awareness of potential culture sensitivities or intrests including the implications for maori
-evidence of the scientific validity of the research
123
what is clinical equipoise?
the requirement that researchers only provide an experimental treatment if the evidence for the experimental treatment is equal to that available for the standard treatment
124
what is a conflict of interest?
a situation where a person holds 2 or more potentially incompatible intrests
these are of concern in research where the researcher have intrests that might compromise the values and standards of ethically appropriate research
125
what can arise conflict of intrests?
professionally, academically, financially, politically
126
how can we manage conflicts of interest?
through peer reveiws
blinding
open access to data
auditing
independant people to recruit participants
127
what is informed consent?
required when participants are enrolled in research studies
128
what does informed consent require?
disclosure of the purpose
risks and process of the study
resonable efforts from the researcher to explain this information
the person is competant to give consent
absence of any coercive factors
129
what must consent be in?
writing
130
what does justice require?
transparency
that all people are considered of equal worth
that efforts are made to make the society equitable
131
what does ACART mean?
advisory committee on assisted reproductive technologies
132
what does ECART mean?
ethics committee on assisted reproductive technologies
133
what are some enforcement processes?
prerequisite for finding public funding
prerequisite for publication in major health journals
clinical trial registry
professional obligation
