Bias - PowerPoint PPT Presentation


Title: Bias


1
Bias
M.Valenciano, 2006 A. Bosman, 2005
T. Grein, 2001- 2004
2
  • Every epidemiological study should be viewed as a
    measurement exercise
  • Kenneth J. Rothman, 2002

.. in order to understand the truth
3
What epidemiologists measure
  • Rates, risks
  • Effect measures
  • Rate Ratio
  • Odds ratio
  • ....... yet these are just estimates
    of the true value
  • the amount of error cannot be determined

4
Objective of this session
  • Define bias
  • Present type of bias and influence in estimates
  • Identify methods to prevent bias

5
Should I believe my measurement?
Mayonnaise Salmonella
RR 4.3
6
Errors
  • Two broad types of error
  • Random error reflects amount of variability
  • Chance?
  • Systematic error (Bias)

Definition of bias Any systematic error in
an epidemiological study resulting in an
incorrect estimate
of association between exposure and risk of
disease
7
Errors in epidemiological studies
Error
Random error (chance)
Systematic error (bias)
Study size
Source Rothman, 2002
8
Categories of bias
  • Selection bias
  • Information bias
  • Confounding

9
Selection bias
  • Errors in the process of identifying
    the study population
  • When ?
  • Inclusion in the study
  • How ?
  • Preferential selection of subjects
    related to their
  • Disease status cohort
  • Exposure status case control

10
Selection bias
  • When?
  • How?
  • Consequences?

frequency of disease (cohort) frequency of
exposure (case control) different among
- those included in the study - those
eligible
11
Types of selection bias
  • Sampling bias
  • Ascertainment bias
  • surveillance
  • referral, admission
  • diagnostic
  • Participation bias
  • self-selection (volunteerism)
  • non-response, refusal
  • healthy worker effect, survival

12
Selection bias in case-control studies
13
Selection bias
e.g alcohol and cirrhosis?
OR 6
  • How representative are hospitalised trauma
    patients of the
    population which gave rise to the cases?

14
Selection bias
OR 6 OR 36
  • Higher proportion of controls drinking alcohol
    in trauma ward
  • than in non-trauma

15
SB Diagnostic bias
Diagnostic approach related to knowing exposure
status e.g OC and uterine cancer?
  • OC use ? breakthrough bleeding ? increased chance
    of detecting uterine cancer

16
SB Admission bias
Exposed cases different chance of admission
than controls e.g asbestos and lung
cancer?
  • Prof. Pulmo, head respiratory department,
    145 publications on
    asbestos/lung cancer
  • Lung cancer cases exposed to asbestos
    not representative of lung
    cancer cases

17
SB Survival bias
Only survivors of a highly lethal disease enter
study e.g. Hospital and haemorrhagic fever?
  • Contact with risk hospital leads to rapid death

18
SB Non-response bias
  • Controls chosen among women at home
    13000 homes contacted ?1060 controls
  • Controls mainly housewives with lower chance of
    test

19
Selection bias in cohort studies
20
SB Healthy worker effect
Source Rothman, 2002
21
Healthy worker effect
Source Rothman, 2002
22
Non-response bias
lung cancer yes no
Smoker 90 910
1000 Non-smoker 10 990
1000
23
SB Non-response bias
lung cancer yes no
10 of smokers dare to respond
Smoker 9 91
100 Non-smoker 10 990
1000
No bias !
24
Non-response bias
lung cancer yes no
Smoker 45 910
955 Non-smoker 10
990 1000
50 of cases that smokedlost to follow up
25
SB Loss to follow-up
  • Difference in completeness of follow-up between
    comparison groups
  • Example
  • study of disease risk in migrants
  • migrants more likely to return to place of origin
    when having disease ? lost to follow-up?
    lower disease rate among exposed (migrant)

26
Minimising selection bias
  • Clear definition of study population
  • Explicit case and control definitions
  • Cases and controls from same population
  • Selection independent of exposure
  • Selection of exposed and non-exposed without
    knowing disease status

27
Categories of bias
  • Selection bias
  • Information bias

28
Information bias
  • Systematic error in the measurement
    of information on exposure or outcome
  • When?
  • During data collection
  • How?
  • Differences in accuracy
  • of exposure data between cases and controls
  • of outcome data between exposed and unexposed

29
Information bias
  • When?
  • How?
  • Consequences?
  • Misclassification
  • Study subjects are classified
    in the wrong category
  • Cases / controls
  • Exposed / unexposed

30
Information bias misclassification
  • Measurement error leads to assigning wrong
    exposure or outcome category
  • Non-differential
  • Random error
  • Missclassifcation exposure EQUAL
    between cases and controls
  • Missclassification outcome EQUAL
    between exposed nonexp.
  • gt Weakness measure
    of association
  • Differential
  • Systematic error
  • Missclassification exposure DIFFERS
    between cases and controls
  • Missclassification outcome DIFFERS
    between exposed nonexposed
  • gt Measure association
    distorted in any direction

31
Two main types of information bias
  • Reporting bias
  • Recall bias
  • Prevarication
  • Observer bias
  • Interviewer bias
  • Biased follow-up

32
IB Recall bias
Cases remember exposure differently than
controls e.g. risk of malformation
  • Mothers of children with malformations remember
    past exposures better than
    mothers with healthy children

33
IB Prevarication bias
Cases report exposure differently than
controls e.g. isolation and heat related death
  • Relatives of dead elderly may deny isolation
  • Underestimation a ? underestimation of OR

34
IB Interviewer bias
Investigator asks cases and controls differently
about exposure e.g soft cheese and
listeriosis
Cases of
Controls
listeriosis
Eats soft cheese
a
b
Does not eat
c
d
soft cheese
  • Investigator may probe listeriosis cases
    about consumption of
    soft cheese

35
IB Biased follow-up
  • Unexposed less likely diagnosed
  • for disease than exposed
  • Cohort study risk factors for mesothelioma
  • Difficult histological diagnosis
  • gt Histologist more likely
    to diagnose specimen as
    mesothelioma
  • if asbestos exposure kown

36
Nondifferential misclassification
  • Misclassification does not depend
    on values of other variables
  • Exposure classification NOT related to disease
    status
  • Disease classification NOT related to exposure
    status
  • Consequence
  • if there is an association,
  • weakening of measure of association
  • bias towards the null

37
Nondifferential misclassification
  • Cohort study Alcohol ? laryngeal cancer

38
Nondifferential misclassification
  • Cohort study Alcohol ? laryngeal cancer

39
Minimising information bias
  • Standardise measurement instruments
  • Administer instruments equally to
  • cases and controls
  • exposed / unexposed
  • Use multiple sources of information
  • questionnaires
  • direct measurements
  • registries
  • case records
  • Use multiple controls

40
Questionnaire (tomorrow)
  • Favour closed, precise questions
    minimise open-ended questions
  • Seek information on hypothesis through
    different questions
  • Disguise questions on hypothesis
    in range of unrelated questions
  • Field test and refine
  • Standardise interviewers technique through
    training with questionnaire

41
Bias
  • Should be prevented !!!!
  • protocol
  • If bias
  • incorrect measure of association
  • should be taken into account
    in the interpretation of the results
  • magnitude?
  • overestimation? underestimation?

42
References
Rothman KJ Epidemiology an introduction. Oxford
University Press 2002, 94-101 Smith (1984)
43
Bias in randomised controlled trials
  • Gold-standard randomised, placebo-controlled,
    double-blinded study
  • Least biased
  • Exposure randomly allocated to subjects -
    minimises selection bias
  • Masking of exposure status in subjects and study
    staff - minimises information bias

44
Bias in prospective cohort studies
  • Loss to follow up
  • The major source of bias in cohort studies
  • Assume that all do / do not develop outcome?
  • Ascertainment and interviewer bias
  • Some concern Knowing exposure may influence how
    outcome determined
  • Non-response, refusals
  • Little concern Bias arises only if related to
    both exposure and outcome
  • Recall bias
  • No problem Exposure determined at time of
    enrolment

45
Bias in retrospective cohort case-control
studies
  • Ascertainment bias, participation bias,
    interviewer bias
  • Exposure and disease have already occurred ?
    differential selection / interviewing of
    compared groups possible
  • Recall bias
  • Cases (or ill) may remember exposures differently
    than controls (or healthy)

46
Question to you
  • Suppose a computer error in data entry
  • Exposed group classified as unexposed
  • Unexposed group classified as exposed
  • What effect has this error on the result?
  • Is it bias?
  • If so what type
  • If not, what type of error?
View by Category
About This Presentation
Title:

Bias

Description:

Bias M.Valenciano, 2006 A. Bosman, 2005 T. Grein, 2001- 2004 Every epidemiological study should be viewed as a measurement exercise Kenneth J. Rothman, 2002 What ... – PowerPoint PPT presentation

Number of Views:911
Avg rating:3.0/5.0
Slides: 47
Provided by: Gre113
Category:
Tags: bias | mayonnaise

less

Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: Bias


1
Bias
M.Valenciano, 2006 A. Bosman, 2005
T. Grein, 2001- 2004
2
  • Every epidemiological study should be viewed as a
    measurement exercise
  • Kenneth J. Rothman, 2002

.. in order to understand the truth
3
What epidemiologists measure
  • Rates, risks
  • Effect measures
  • Rate Ratio
  • Odds ratio
  • ....... yet these are just estimates
    of the true value
  • the amount of error cannot be determined

4
Objective of this session
  • Define bias
  • Present type of bias and influence in estimates
  • Identify methods to prevent bias

5
Should I believe my measurement?
Mayonnaise Salmonella
RR 4.3
6
Errors
  • Two broad types of error
  • Random error reflects amount of variability
  • Chance?
  • Systematic error (Bias)

Definition of bias Any systematic error in
an epidemiological study resulting in an
incorrect estimate
of association between exposure and risk of
disease
7
Errors in epidemiological studies
Error
Random error (chance)
Systematic error (bias)
Study size
Source Rothman, 2002
8
Categories of bias
  • Selection bias
  • Information bias
  • Confounding

9
Selection bias
  • Errors in the process of identifying
    the study population
  • When ?
  • Inclusion in the study
  • How ?
  • Preferential selection of subjects
    related to their
  • Disease status cohort
  • Exposure status case control

10
Selection bias
  • When?
  • How?
  • Consequences?

frequency of disease (cohort) frequency of
exposure (case control) different among
- those included in the study - those
eligible
11
Types of selection bias
  • Sampling bias
  • Ascertainment bias
  • surveillance
  • referral, admission
  • diagnostic
  • Participation bias
  • self-selection (volunteerism)
  • non-response, refusal
  • healthy worker effect, survival

12
Selection bias in case-control studies
13
Selection bias
e.g alcohol and cirrhosis?
OR 6
  • How representative are hospitalised trauma
    patients of the
    population which gave rise to the cases?

14
Selection bias
OR 6 OR 36
  • Higher proportion of controls drinking alcohol
    in trauma ward
  • than in non-trauma

15
SB Diagnostic bias
Diagnostic approach related to knowing exposure
status e.g OC and uterine cancer?
  • OC use ? breakthrough bleeding ? increased chance
    of detecting uterine cancer

16
SB Admission bias
Exposed cases different chance of admission
than controls e.g asbestos and lung
cancer?
  • Prof. Pulmo, head respiratory department,
    145 publications on
    asbestos/lung cancer
  • Lung cancer cases exposed to asbestos
    not representative of lung
    cancer cases

17
SB Survival bias
Only survivors of a highly lethal disease enter
study e.g. Hospital and haemorrhagic fever?
  • Contact with risk hospital leads to rapid death

18
SB Non-response bias
  • Controls chosen among women at home
    13000 homes contacted ?1060 controls
  • Controls mainly housewives with lower chance of
    test

19
Selection bias in cohort studies
20
SB Healthy worker effect
Source Rothman, 2002
21
Healthy worker effect
Source Rothman, 2002
22
Non-response bias
lung cancer yes no
Smoker 90 910
1000 Non-smoker 10 990
1000
23
SB Non-response bias
lung cancer yes no
10 of smokers dare to respond
Smoker 9 91
100 Non-smoker 10 990
1000
No bias !
24
Non-response bias
lung cancer yes no
Smoker 45 910
955 Non-smoker 10
990 1000
50 of cases that smokedlost to follow up
25
SB Loss to follow-up
  • Difference in completeness of follow-up between
    comparison groups
  • Example
  • study of disease risk in migrants
  • migrants more likely to return to place of origin
    when having disease ? lost to follow-up?
    lower disease rate among exposed (migrant)

26
Minimising selection bias
  • Clear definition of study population
  • Explicit case and control definitions
  • Cases and controls from same population
  • Selection independent of exposure
  • Selection of exposed and non-exposed without
    knowing disease status

27
Categories of bias
  • Selection bias
  • Information bias

28
Information bias
  • Systematic error in the measurement
    of information on exposure or outcome
  • When?
  • During data collection
  • How?
  • Differences in accuracy
  • of exposure data between cases and controls
  • of outcome data between exposed and unexposed

29
Information bias
  • When?
  • How?
  • Consequences?
  • Misclassification
  • Study subjects are classified
    in the wrong category
  • Cases / controls
  • Exposed / unexposed

30
Information bias misclassification
  • Measurement error leads to assigning wrong
    exposure or outcome category
  • Non-differential
  • Random error
  • Missclassifcation exposure EQUAL
    between cases and controls
  • Missclassification outcome EQUAL
    between exposed nonexp.
  • gt Weakness measure
    of association
  • Differential
  • Systematic error
  • Missclassification exposure DIFFERS
    between cases and controls
  • Missclassification outcome DIFFERS
    between exposed nonexposed
  • gt Measure association
    distorted in any direction

31
Two main types of information bias
  • Reporting bias
  • Recall bias
  • Prevarication
  • Observer bias
  • Interviewer bias
  • Biased follow-up

32
IB Recall bias
Cases remember exposure differently than
controls e.g. risk of malformation
  • Mothers of children with malformations remember
    past exposures better than
    mothers with healthy children

33
IB Prevarication bias
Cases report exposure differently than
controls e.g. isolation and heat related death
  • Relatives of dead elderly may deny isolation
  • Underestimation a ? underestimation of OR

34
IB Interviewer bias
Investigator asks cases and controls differently
about exposure e.g soft cheese and
listeriosis
Cases of
Controls
listeriosis
Eats soft cheese
a
b
Does not eat
c
d
soft cheese
  • Investigator may probe listeriosis cases
    about consumption of
    soft cheese

35
IB Biased follow-up
  • Unexposed less likely diagnosed
  • for disease than exposed
  • Cohort study risk factors for mesothelioma
  • Difficult histological diagnosis
  • gt Histologist more likely
    to diagnose specimen as
    mesothelioma
  • if asbestos exposure kown

36
Nondifferential misclassification
  • Misclassification does not depend
    on values of other variables
  • Exposure classification NOT related to disease
    status
  • Disease classification NOT related to exposure
    status
  • Consequence
  • if there is an association,
  • weakening of measure of association
  • bias towards the null

37
Nondifferential misclassification
  • Cohort study Alcohol ? laryngeal cancer

38
Nondifferential misclassification
  • Cohort study Alcohol ? laryngeal cancer

39
Minimising information bias
  • Standardise measurement instruments
  • Administer instruments equally to
  • cases and controls
  • exposed / unexposed
  • Use multiple sources of information
  • questionnaires
  • direct measurements
  • registries
  • case records
  • Use multiple controls

40
Questionnaire (tomorrow)
  • Favour closed, precise questions
    minimise open-ended questions
  • Seek information on hypothesis through
    different questions
  • Disguise questions on hypothesis
    in range of unrelated questions
  • Field test and refine
  • Standardise interviewers technique through
    training with questionnaire

41
Bias
  • Should be prevented !!!!
  • protocol
  • If bias
  • incorrect measure of association
  • should be taken into account
    in the interpretation of the results
  • magnitude?
  • overestimation? underestimation?

42
References
Rothman KJ Epidemiology an introduction. Oxford
University Press 2002, 94-101 Smith (1984)
43
Bias in randomised controlled trials
  • Gold-standard randomised, placebo-controlled,
    double-blinded study
  • Least biased
  • Exposure randomly allocated to subjects -
    minimises selection bias
  • Masking of exposure status in subjects and study
    staff - minimises information bias

44
Bias in prospective cohort studies
  • Loss to follow up
  • The major source of bias in cohort studies
  • Assume that all do / do not develop outcome?
  • Ascertainment and interviewer bias
  • Some concern Knowing exposure may influence how
    outcome determined
  • Non-response, refusals
  • Little concern Bias arises only if related to
    both exposure and outcome
  • Recall bias
  • No problem Exposure determined at time of
    enrolment

45
Bias in retrospective cohort case-control
studies
  • Ascertainment bias, participation bias,
    interviewer bias
  • Exposure and disease have already occurred ?
    differential selection / interviewing of
    compared groups possible
  • Recall bias
  • Cases (or ill) may remember exposures differently
    than controls (or healthy)

46
Question to you
  • Suppose a computer error in data entry
  • Exposed group classified as unexposed
  • Unexposed group classified as exposed
  • What effect has this error on the result?
  • Is it bias?
  • If so what type
  • If not, what type of error?
About PowerShow.com