week 3 Flashcards

1
Q

How can we decide if a risk factors causes a disease?

A

Solid body of evidence (strength of evidence)

checklist that helps us decide

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

Bradford hill criteria

A

provide ways of examining whether cause and effect is a reasonable inference

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

How many points does the bradford-hill criteria has?

A

9

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

what are the bradford hill criteria

A
  1. temporal relationship (essential)
  2. strength (effect size)
  3. consistency
  4. Analogy
  5. Specificity
  6. Reversibility
  7. Dose-response relationship (not essential)
  8. Plausibility
  9. Coherence
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

what is temporal relationship?

A

risk factor must occur/be present BEFORE the disease.

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

Strength - effect size

A

A strong association is more likely to be causal (but the reverse is NOT true)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

Consistency (reproducibility)

A

similar results in different populations with different study designs.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

Analogy

A

similarities with other well-established cause effect relationships

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

Specificity

A

t rarely occurs, as most diseases have multiple causes and most exposures, multiple effects

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

Reversibility

A

If the removal of a possible risk factor results in reduced risk of disease,

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

Dose-response relationship

A

e.g. Smoking, increased exposure increases risk of lung cancer

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

Plausibility

A

Must be consistent with knowledge from other sources (e.g. animal experiments) & should make
biological sense

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

Coherence

A

suggested cause-effect should be consistent with the natural history and biology of the disease

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

Causality

A

“Both practical and ethical considerations mean that causality cannot, in general, be proved in
human studies. Rather, it must be induced from demonstrated associations between and
exposure and health outcomes.

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

Challenges of causality

A

outcomes having multiple component causes

Distinguishing which of these are necessary or sufficient is central to preventive efforts

how much should it be pursue

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

When can links between exposures and outcomes/disease can be considered causal?

A
once full consideration has been taken of
epidemiological noise
-chance, 
-bias 
 -confounding
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

External validity

A

generalisation to the entire poplation

The degree to which the study conclusions can be applied to other samples

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

Internal validity

A

measurements and sample.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
19
Q

two underlying concepts of external validity

A
  1. generalisability

2. applicability (to a particular sample within a population)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
20
Q

charact. of external validity

A

inclusion and exclusion criteria for study (age, health…)

Individuals agreeing to participate in research studies are different from those who don’t

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
21
Q

Can recruited sample be 100% representative?

A

No, never!

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
22
Q

Judgements to consider for external validity

A
  • Age, sex, severity of disease, comorbid conditions
  • Similar drugs, other doses, timing, route of administration
  • Other outcomes (not assessed), different duration of treatment

“As the intervention was implemented for both sexes, all ages, all types of sports, and at different levels of
sports, the results indicate that the entire range of athletes,

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
23
Q

Internal validity

A

Degree to which the investigator draws the correct conclusion about what actually happened in
the study

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
24
Q

what could the design methods and conduct of the trial/study could do for the analysis?

A
  • reduce likelihood of errors and chance findings
  • eliminate possibility of bias
  • minimise impact of toher factors (cofounding , interaction)
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
25
Q

what can you do to minimise ‘chance’ findings? (false positive)

A
✓ Strong a priori rationale
✓ Plausible
✓ Adequate sample size
✓ Correct statistical analysis
✓ Replication of findings
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
26
Q

Bias

A

A systematic error in the design, recruitment, data collection or analysis that results in a mistaken
estimation of the true effect of the exposure and outcome

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
27
Q

What does bias does to studies?

A

limits validity and generalizability of study results

rarely eliminates during analysis.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
28
Q

Selection bias:

A

systematic error in the selection or retention of participants

29
Q

Information (misclassification) bias:

A

systematic error due to inaccurate measurement or

classification of disease, exposure or other variable

30
Q

Selection bias in case control studies

A

inherent in design ( non comparable between cases and controls)

problematic in hospital settings

31
Q

Why is selection bias particularly problematic in hospital settings?

A

Cases: hospital will fail to enrol severe cases that died before reaching hospital. ( not representative of all cases in hospital)

controls: overrepresentation in the control group could underestimate association.

32
Q

Selection bias in cohort studies.

A

less problematic.

exposure status identified prior to outcome occurring

can occur when loss to follow up / non-response rate differes between expose groups (e.g. heavy drinkers not responding the survey or family might be more likely to participate).

33
Q

Will the cohort prevalence and incidence rates will be the same from the general population?

A

No, they will be different mostly due to selection bias.

34
Q

when is it less likely to have selection bias in RCT?

A

If:

  • randomisation performed correctly
  • Sufficiently large sample
  • blinding to treatment allocation
35
Q

Types of information bias:

A
  1. Non-differential classification
  2. Differential classification
  3. observer bias
  4. interviewer bias
  5. reporting bias
  6. recall bias
36
Q

What is non differential classification>?

A

it referst o when misclassification is randon and all individuals have the same probability of being misclassified.

37
Q

what is differential classification?

A

misclassification of disease status is dependent upon risk factors (or vice versa).

Includes instrumentation errors, misdiagnosis, missing data.

38
Q

observer bias

A

prior knowledge of expected outcomes influences the way inofmration is collected, measured or interpreted.

39
Q

Interviewer bias

A

leasing questions

40
Q

Reporting bias

A

: individuals may selectively suppress or reveal information

41
Q

Recall bias

A

when the information provided on exposure differs between exposed & unexposed

42
Q

when is Recall bias particularly problematic?

A

in case -control and retrospective cohort studies.

e.g. individuals with cancer may be more likely to recall exposure to toxic chemicals than controls

43
Q

Hawthorne effect

A

Individual’s change in behaviour due to awareness of being observed. (e.g. with placebo)

44
Q

Why could misclassified information happen?

A

ambiguous questions

instruments developed in one setting not appropriate to another setting.

Incriminating or personal embarrassing questions
multiple interviewers

self-reported, telephone and face to face survey can provide different results

45
Q

Key questions to identify selection bias

A

Is the study population defined?
▪ Is it representative of the target population?
▪ Are the definitions of disease and exposure clear?
▪ Is the case definition precise?
▪ What are the criteria for inclusion/exclusion of participants?
▪ In case controls studies, are the controls representative of the population from which the cases
came?
▪ Could exposure status have influenced the selection of cases or controls?
▪ Are the cohorts comparable except for exposure/intervention status?
▪ Were losses to follow up kept to a minimum?

46
Q

key questions to identify information bias:

A

Are the measurements as objective as possible?
▪ Were the observers/interviewers rigorously trained?
▪ Is the study blinded as far as possible?
▪ Are levels of follow up adequate & is it equal for all cohorts?
▪ Was the appropriate analysis performed?
▪ Were the variable groups defined a priori
▪ Is the interpretation supported by evidence?
▪ Were clearly written procedures used to standardise procedures?
▪ Were study participants randomised to observers / interviewers?
▪ Was self-reported information validated against any existing records?

47
Q

what is a confounder

A

An extraneous variable that wholly or partially accounts for the observed effect of a risk factor
on disease status

Masks the true effect of an exposure on an outcome

48
Q

Effect modifier (interaction)

A

A variable that differentially (positively or negatively) modifies the observed effect of a risk
factor on disease status

The effect of an exposure on an outcome is different for different groups

49
Q

What is mediator?

A

on the causal pathway between exposure and outcome.

50
Q

Confounding and interaction

A

NOT the same

The influence of 3rd factor which leads to an incorrect estimate of the association
between the exposure and outcome

51
Q

what could confounding be?

A

an explanation for an association between exposure & outcome.

52
Q

What happen to the association between two variables with a 3rd factors?

A

the association is distorted because of the third factor.

it can over or under estimate a true association.

53
Q

where does the confounder must be associated and where not?

A

associated with both exposure and outcome and not on the causal pathway.

54
Q

How to control confounding at the design stage?

A
  • randomisation:
  • restriction
  • matching
  • identified at design based on previous knowledge.
55
Q

Where is randomisation possible?

A

only in clinical trial

56
Q

what is restriction when controlling cofounders?

A

limits study participants to those with similar confounders ( e.g. only smorker) but limits generalibilizity.

57
Q

What is matching?

A

refers to individuals in casee and control groups are as similar as possible. (mostly in case control studies).

must be decided in advace.

58
Q

How do you control for confounding at the analysis stage?

A
  • stratification
  • multivariable analysis
  • standarisation
  • residual confounding
59
Q

what is stratification?

A

examines the association between exposure and outcome withing different strata of the cofounding factors (e.g. seperate in smokers and non smokers).

60
Q

Multivariable analysis

A

statistical modelling to control for 1+ cofounders

61
Q

which method is the most common one used for cohort studies?

A

Multivariable analysis.

62
Q

what does standardisation conveys?

A

using a standard population for reference to help negate the effect of cofounders.

63
Q

what is residual confounding?

A

distortion that remains after controlling for confounding in the design and/or analysis of a study.

only adjusts for things we know about and have measured.

64
Q

when does the 3rd variable ‘modifies’ a relationship?

A

When the association between exposure & outcome differs by a ‘level’ of another (3rd) factor

65
Q

Does modifies a relationship means causality?

A

NO

66
Q

what happens When the association between exposure & outcome differs by a ‘level’ of another (3rd) factor?

A

exposure has a different effect on outcome in different subgroups (stratified analysis)/

the overall estimate of the association between outcome and exposure is misleading

67
Q

In the association between the flu and dying what factor can be an effect modifier?

A

age, health status, gender?

68
Q

Is an association or effect existent if the ffect is statistically significant?

A

not necessarily.

69
Q

Should you conclude about the scientific importance based on statistical significant?

A

NO :)