Brainscape's Knowledge GenomeTM

Browse over 1 million classes created by top students, professors, publishers, and experts.


See full index

EBM > Bias + Confounders > Flashcards

Bias + Confounders Flashcards

1
Q

What is selection bias? What are examples of selection bias? What can you look for in study to indicate selection bias?

A

Systematic error due to difference in characteristics between those selected into the study compared to those than have not been

Examples:

  • Allocation bias
  • Sampling bias
  • Spectrum bias
  • Responder bias
  • Attrition/loss to follow up bias
  • Healthy worker effect

Look for:

  • whether group participants are selected from same population of patients
  • whether group of participants are presentative of wider population
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

What is sampling bias?
What study types are prone to this kind of bias?
What can you look for it a study to indicate possible bias? How can you minimise it?

A

Each potential member of population does not have equal chance of selection meaning that there is systematic difference between those selected into study and those not

Studies:

  • cohort
  • case-control
  • RCT

Look for:
-what the exclusion/inclusion criteria are
I.e. age or sex etc

Minimise:
-Don’t exclude portion of population to give all members of population equal chance of being selected

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

What is allocation bias?
What study types are prone to this kind of bias?
What can you look for it a study to indicate possible bias? How can you minimise it?

A

Systematic difference between characteristics of study arms i.e. allocation should be independent of characteristics

Studies:
-RCT

Look for:
-process of allocation i.e. was there random allocation or concealment of allocation

Minimise:

  • randomisation i.e. random sequence generation to form groups
  • allocation concealment
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

What is spectrum bias?
What study types are prone to this kind of bias?
What can you look for it a study to indicate possible bias? How can you minimise it?

A

Where “hard to diagnose” patients are excluded from having index test in DTA study leading to index test appearing more accurate than it is

Could also be exclusion of people from test due researcher thinking they will not be able to comply with test

Study= DTA

Look for:
-researchers stating that they are a purposefully excluding certain population members from having test

Minimise
-include entire population that test will need to be used on because want to represent true accuracy rather than cherry picking patients who will give you the best results

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

What is response bias?
What study types are prone to this kind of bias?
What can you look for it a study to indicate possible bias? How can you minimise it?

A

Difference in characteristics of people who volunteer to be part of studies and those who don’t

Studies
-any study where patient population ask to be recruited

Look for:

  • how they asked participants to get involved i.e. was there monetary compensation for time and transport etc
  • barriers which might prevent people from taking part in study

Minimise:
-try to over come possible barriers
I.e. cost or time or travel

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

What is attrition bias?
What study types are prone to this kind of bias?
What can you look for it a study to indicate possible bias? How can you minimise it?

A

Difference between people who remain in study and those who are lost to follow up based on specific characteristics
I.e. elderly people might be more likely to drop out of RCT when intervention has adverse affects which might effect elderly people more leading to healthier younger people remaining in the study= intervention appears to be more affective

Studies:

  • cohort
  • RCT

Look for

  • number of people recruited/at start of study vs number at end
  • whether study has explicitly stated the number of people who withdrew
  • characteristics of people who withdrew
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

What is the healthy worker effect?
What study types are prone to this kind of bias?
What can you look for it a study to indicate possible bias? How can you minimise it?

A

Workers have lower morbidity and mortality compared with general population as have to have certain level of health in order to work. Can lead to weakening of association between the exposure and outcome

Studies:

  • cohort
  • RCT

Look for:
-population which has been sampled from i.e. need to be conscious of fact that having higher proportion of workers might lead to bias

Minimise:

  • consider when evaluating
  • try and select from wider population representative of wider population unless wanting to look at specific working population for exposure
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

How can selection bias be minimised?

A

Random sequence generation and allocation concealment in RCT
Inclusion of hard to diagnose patients in DTA
Ensure baseline characteristics are comparable

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

What is the definition of information bias? What are the different types?

A

Systematic difference in way the data is gathered from comparison groups leading to different quality of data between the groups

Interviewer bias 
Recall bias 
Recording bias 
Social acceptability bias 
Detection bias 
Performance bias 
Reporting bias 
Verification bias 
Review bias
Reflexive stance of researcher
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

What is interviewer bias?
What study types are prone to this kind of bias?
What can you look for it a study to indicate possible bias? How can you minimise it?

A

Subconscious/conscious difference in way interviewer gathers data from cases and controls
I.e. might ask more indepth questions to cases than controls

Studies:

  • case-control
  • RCT

Look for:

  • whether same set questions used throughout
  • whether the interviewer was concealed from knowing which group patient allocated to

Minimise:
-Blinding so that interviewer does not know which group patient has been assigned to

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

What is recall bias?
What study types are prone to this kind of bias?
What can you look for it a study to indicate possible bias? How can you minimise it?

A

Difference in accuracy and completeness of recall of events between cases and controls
I.e. might occur due to cases being more informed about their health or have been made more aware

Studies:

  • case-control
  • retrospective cohort

Look for:
-difference in detail of information provided between cases and control regarding previous exposures

Minimise:
-not sure

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

What is recording bias?
What study types are prone to this kind of bias?
What can you look for it a study to indicate possible bias? How can you minimise it?

A

Information systematically better recorded in cases compared with controls i.e. might be because of increased contact with healthcare

Need to consider if there is possibility of other health records outside the ones used for the study
I.e. private hospitals or community pharmacy
(Important to consider alternate records in context of prescription)

Studies:
-case-control

Look for:

  • difference in level of detail of notes between cases and controls
  • sources of information used in study

Minimise:
-try and include multiple sources to try and get as much detail as possible i.e. community prescriptions or private hopsitals etc

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

What is social acceptability bias?
What study types are prone to this kind of bias?
What can you look for it a study to indicate possible bias? How can you minimise it?

A

Suppression or selective revealing of information relating to exposure of interest leading to subject giving more social acceptable answer

Bias to be aware of in questionnaires as people might alter their answers

Studies:

  • RCT
  • Case-control
  • Qualitative

Look for:
-conditions or scenarios where there might be social stigma
I.e. alcohol intake/breast feeding/mental healthy

Minimise:
-create non-judgmental environment and encourage participants to be honest to try and get the most representative results

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

What is detection bias?
What study types are prone to this kind of bias?
What can you look for it a study to indicate possible bias? How can you minimise it?

A

Differences in how outcomes are determined

Studies:
-RCT

Look for:

  • blinding of outcomes
  • whether person who allocated participants to groups is same person who is analysing

Minimise:
-Blind the outcome assessors so don’t know whether intervention or control so that this knowledge cannot affect the assessment of outcomes

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

What is performance bias?
What study types are prone to this kind of bias?
What can you look for it a study to indicate possible bias? How can you minimise it?

A

Difference care or attention or exposure to factors other than the intervention between the 2 groups

Studies:
-RCT

Look for:

  • whether there was allocation concealment or blinding
  • whether there was difference in interaction between people conducting study and participants in intervention groups i.e. more frequent follow ups etc

Minimise:

  • Blinding so that unaware whether case or control to ensure that they are not treated differently because of the group they are in.
  • Should get the same treatment outside the intervention
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

What is reporting bias?
What study types are prone to this kind of bias?
What can you look for it a study to indicate possible bias? How can you minimise it?

A

Difference between the reported and unreported finding due to researchers being more likely to report the statistically significant findings rather than the not statistically significant findings

Studies:
-any study where statistical analysis between groups done

Look for
-whether statistically significant findings and non-statistically significant findings have been explored

Minimise:
-??

17
Q

What is verification bias?
What study types are prone to this kind of bias?
What can you look for it a study to indicate possible bias? How can you minimise it?

A

When not all the participants receive the index test and reference standard meaning the results cannot be verified

More likely to occur when the result for the index test is negative so reference standard is not performed

Study= DTA

Look for:

  • whether all participants receive index and reference standard test
  • whether there is specific criteria/Boolean for participant to have test

Minimise:
-ensure that all participants get tests regardless of index test outcome

18
Q

What is review bias?
What study types are prone to this kind of bias?
What can you look for it a study to indicate possible bias? How can you minimise it?

A

When the index and reference standard are not reviewed independently of each other
I.e. can lead to the index test appearing more accurate if already know the reference standard because it might influence interpretation (Especially when the index test is subjective)

Study= DTA

Look for:

  • whether the person reviewing the index test/reference standard was blinding from the other test results
  • whether person reviewing was independent of the people carrying out the tests

Minimise:
-blind the people interpreting the tests so unable to know the results of the other test before interpreting allocated test

19
Q

What is the bias associated with the reflexive stance of researchers?

A

In qualitative studies, researches need to explicitly reflect on how their background, professional role or personality might influence their relationship with participants

20
Q

What bias are associated with diagnostic test accuracy studies? How can these types of bias be reduced?

A

Verification bias
-ensure all participants receive both tests regardless of the result of index test

Review bias
-need to ensure that people interpreting the individual tests are not made aware of the results of the other test prior to interpretation to minimise the influence i.e. interpret each test blind to the results of the other

Spectrum bias

  • need clear description of population tested on
  • need to have the study population be representative of the population the test will be used on
  • no exclusion from being tested due to specific characteristics
21
Q

What bias is associated with cohort studies?

A

Healthy worker effect
-leads to exclusion of people with severe illness or chronically disable which means that the results cannot be generalised to the general population as these people are part of the general population

Attrition bias/loss to follow up
-may be more or less likely to drop out due to exposure

Responder bias
-might be systematic difference in characteristics of people who volunteer for the study compared to those that don’t
I.e. might be due to barriers or socioeconomic situation etc

Possible sampling bias
-might be systematic difference in the way members selected to be part of study population
I.e. look for exclusion/inclusion criteria

22
Q

What bias is associated with case-control studies?

A

Recall bias
-people more likely to remember exposures or be more informed of exposures

Recording bias

  • cases medical notes might be more detailed
  • consider whether there is possibility of additional sources outside of those looked at in study

Interviewer bias
-might ask more in-depth questions to cases

Response bias

  • might differ between cases and controls i.e. who volunteers to be part of the study
  • need to be aware that even when looking at patient records they still need to consent so there might be difference in people who consent to have their records used

Sampling bias

  • if underlying cohort not clearly defined
  • look to exclusion criteria i.e. age and sex

Social acceptability bias
-if information about stigmatised exposure or if it is gathered through questionnaire

23
Q

What bias are associated with RCT? What can be done to minimise these bias?

A

Selection bias

  • differences in baseline characteristics of groups
  • minimise through randomisation and allocation concealment to try and randomise the possible confounding baseline characteristics so that any difference in groups is due to chance

Allocation bias

  • when participants don’t have the same chance of entering group and there is difference in characteristics between the study arms
  • minimise with allocation concealment and randomisation process

Performance bias

  • difference in care and attention given outside of the intervention and control
  • minimise with allocation concealment and blinding so that unable to alter care given based on which group participant is in

Detection bias

  • difference in how outcome determined
  • minimise by blinding people assessing outcomes to minimise the influence

Attrition bias
-difference in groups with have been lost to follow up or have withdrawn (can be seen through incomplete outcome data)

Reporting bias
-difference between reporting and unreported findings i.e. people might alter reporting of activity depending on whether in intervention or control group based on what they are expecting to do
-

24
Q

What bias is associated with qualitative studies? How can you minimise the bias?

A

Social acceptability bias

  • people might selectively reveal information if that would appear more socially acceptable
  • try to remain impartial and not give opinions that then might influence how the participant responds

Assessing for reflexive stance of researcher
-need to ensure that researcher addresses how their background, professional role or personality might influence the relationship with the participants

25
Q

What is a confounder?

A

Factor associated with the exposure and outcome independently which can lead to an apparent relationship between the exposure and outcome

26
Q

What are the possible consequences of confounders?

A

Can lead to relationship between exposure and disease being affected which can lead to false relationship/association

27
Q

What are examples of possible confounding factors?

A

Ethnicity
Socioeconomic status
Etc

28
Q

What are the 2 points when confounders can be adjusted for? What processes are involved to try and address the confounders?

A

Can be adjusted for in the study design phase and the analysis stage

STUDY DESIGN
RCT:
-most powerful study design for controlling confounders
I.e. randomisation process of large number of participants means there is likely to be similar distribution of confounders between groups and any observed difference at baseline can be said to be due to chance
RCT randomisation process leads to equal distribution of known and unknown confounders where as observational studies can only adjust for known confounders

ANALYSIS STAGE
Matching:
-cohort= done by ensuring equal distribution of possible confounders between exposed and unexposed group
-unmatched case-control= logistic regression
-matched case-control= conditional logistic regression

Stratification

Multivariate analysis (regression curves)

29
Q

Which bias are each of the studies most prone to?

A

RCT:

  • allocation
  • performance
  • attrition
  • detection
  • placebo effect

Cohort

  • response
  • attrition
  • healthy worker effect

Case-contril

  • responder
  • recall

DTA

  • spectrum
  • verification
  • review

Qualitative
-reflexive stance

EBM (9 decks)

Brainscape helps you reach your goals faster, through stronger study habits.
© 2024 Bold Learning Solutions. Terms and Conditions