Brainscape's Knowledge GenomeTM

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


See full index

bds4 pmhp > Evidence Based Dentistry > Flashcards

Evidence Based Dentistry Flashcards

1
Q

What are the steps in PICO?

A

Look at method section:
- Population
- Intervention
- Comparison
- Outcome

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

What is the CASP tool used for?

A

used for Randomised Control trials
- Helps us determine whether results are valid and whether they apply to population you treat

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

Why are confidence intervals important?

A

Shows what happens in population not just study

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

What are the benefits of systematic review over single studies?

A

Saves reader time

Provides reliable evidence

Resolves inconsistencies

Identifies gaps

Identifies if question has been fully answered

Explores differences between studies

May be cheaper to review current RCTs than conduct new study

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

What are the key characteristics of a systematic review?

A
  1. Well formulated question
  2. Comprehensive data search
  3. Unbiased selection and abstraction process
  4. Assessment of papers
  5. Synthesis of data
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

What are the steps in hierarchy of evidence?

A

SR
RCT
Cohort study
Case-control study
Cross sectional study
Ecological study
Case series- case report
Ideas, editorials, opinions

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

Why are SR important?

A

Reduce large quantities of information into manageable portions

Formulate policy and develop guidelines

Efficient use of resources

Increased power/precision- pooling of data

Limit bias and improve accuracy

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

What are the steps in the process of systematic review?

A
  1. Authors
    -> Two or more
    -> Topic expert
    -> Methodological expert
  2. Study protocol
    In advance set out what they plan to do methodologically
  3. Specific Question- Using PICO
  4. Search strategy
    -> comprehensive and repeatable
    -> multiple electronic databases
    -> published and unpublished literature
    -> ideally without language restrictions
  5. Inclusion/Exclusion criteria
    -> Specific
    ->Agreed in advance
  6. Critical Appraisal
    -> Systematic and thorough
    -> Risk of Bias
  7. Synthesis
    -> Qualitative (narrative) synthesis
    -> Quantitative pooling of data in meta-analysis,
    relative precision and quality of the included studies
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

What tools are used to determine quality of systematic review?

A

AMSTAR 2/ROBIS

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

What is done to formulate a good quality question in SR?

A

Use PICO- should contain all aspects of this within question
- Population/Participants- who is the review interested in studying (be specific about patient type OR can be more generalised )
- Intervention (exposure)- what is the new thing that is being done or offered
- Comparison- what is current practice
- Outcome- look at data that tells us which intervention is more effective (must be clear)- measurement

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

What are the different aspects of comprehensive data search?

A

Requires multiple electronic databases (consider help of librarian):
- Look at the paper’s reference lists
- Hand searching- look at reference lists of books
- Non-English language- consider how data can be translated and included
- Unpublished studies- avoid publication bias (studies that are positive or sensational are more likely to be published)
- Grey literature- in public domain but may not be peer reviewed

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

What are the different types of reporting bias?

A

Statistically significant results between 2 groups- it’s more likely to be published

Studies with null effect are less likely to be published

Studies that are exciting/sensational are published rapidly (time lag bias)

Positive results are more likely to be cited by others- less cited studies are less easy to find but may be as robust (citation bias)

More likely to be published if in English (language bias)

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

What can happen when you uncover hidden or unpublished data?

A

You may find balance between positive and negative results
-> Consider if motivation for supressing certain results
-> Ethics issue

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

How is unbiased selection achieved in an SR?

A

Select relevant papers from all that search has found

Authors sift through all papers using screening form which corresponds to criteria of SR

Use data extraction form (published and duplicated)

Be clear about why you have not included a paper- not relevant, doesn’t apply (publish this)- authors will have discussed this

Describe why the studies that were selected were included

Detail who studies were funded by

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

How is assessment of papers in a systematic review achieved?

A
  • Look at how well conducted and designed studies are (methodologically)
  • Must be done independently by at least 2 reviewers- 3rd reviewer will adjudicate if they do not agree
  • If studies are poor or at high risk of bias-discuss this in analysis
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

What is a composite scale?

A

Give numerical value of quality to study
-> High score through computer generated randomisation, or if both patient and assessor are blinded (problematic- papers can achieve the same score for different reasons)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

How is the component approach carried out?

A

Assess individual parts and decide how well they were done (preferred)
 Randomisation using computer
 Blinding
 Drop-outs
Can produce graphic which helps us show quality overall

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

What must be considered when determining the level of bias in a study?

A

 Good studies can still be biased
 If bias exists- is it likely to lead to over/underestimation of effect (consider this in discussion phase of SR- suggest direction the bias is pushing results)
 Magnitude of bias- use own judgement

19
Q

What is the issue with risk of bias?

A

It is impossible to determine quantitatively

20
Q

How is risk of bias ascertained?

A
  • If computer generated sequence in remote location- lower risk
  • Allocation concealment- at point of randomisation, can next allocation be concealed from patient or researcher (ideally, they would have no idea so they cannot influence)
  • Blinding- if this is not possible or done properly (acknowledge level of bias)
  • Outcome data- is it reported properly and completely (or is it selective- increasing risk of bias)
  • Selective outcome reporting- are studies complying with question they set out to answer
  • Who funded study
21
Q

When may risk of bias be unclear?

A

If not clear from paper how aspect of paper you are looking for bias in was conducted
-> Example- if they did not include method used for sequence generation

22
Q

What is attrition bias?

A

How many patients dropped out of trial
 If only a few it is unlikely to affect outcome

23
Q

How is risk of bias presented?

A

In risk of bias table:
 Each component is given low (green), high (red) or unclear score (yellow)
 If many red dots, be cautious about interpretation and conclusion

24
Q

How is data synthesised in SRs?

A
  • Appropriate pooling- look at 10-20 RCTs that look at same thing (results in risk differences or risk ratios)
  • Qualitative- narratively
  • Quantitive- take data out (meta-analysis)
25
Q

What is meta-analysis?

A

Use statistical methods to combine results- integrate, pool, identify trends
- Calculates treatment effect based on pooled data from a group of studies (i.e may look at failure rates)
- Estimates common treatment effect
- Improves precision of a point- brings all available data together

*** May not be appropriate if lack of data or heterogeneity

26
Q

What are the ADV and DIS of meta-analysis?

A

ADV
- Increase in power/precision
- Ability to answer questions not possible to answer in single studies (can look at efficacy of different dosages and compare)

DIS
- Misleading- only include MA if there is minimal differences, outcome measure is the same (for fair comparison) and data is available from the study (extractable)

27
Q

What is dichotomous data?

A

Binary data
- Odds ratio
- Risk ratio
- Risk reduction
- Risk difference
- Number needed to treat

28
Q

What is continuous data?

A

Quantitive data measured precisely
- Weight, BP, pain threshold
- Weighted mean difference/standardised mean difference (taking control group away from intervention group)

29
Q

What is weighting of studies?

A

More weight given to studies with better information (more participants, more events, lower variance/standard deviation)

30
Q

What are the components of a forrest plot?

A

Horizontal line at bottom- scale measuring treatment effect

Left of line- treatment outcome is less likely
Right of line- treatment outcome is more likely
- These can be switched- so read carefully

Vertical line- treatment and control have same effect (line of no effect)

Each individual study has an ID, data divided into experimental/control group, % weight given to study, statistic being used (relative risk etc), part shows the data in numbers and graph format

31
Q

How do data point appear on forrest plot?

A

Blob placed where data measures effect

Horizontal line- confidence interval (wider- less confident about observed effect)

Diamond- pooled estimate (middle of diamond), horizontal width (CI)

-> If confidence interval crosses/overlaps line of no effect- insufficient evidence between intervention and control

32
Q

What is heterogeneity?

A

If too much variation between studies- we may not pool results together

33
Q

What are the types of heterogeneity?

A

Clinical- studies conducted/designed in different ways, variation in participants/interventions/outcomes or measurements of success

Methodological- variation in methods used (how patients were randomised/blinded or how allocation was concealed)

Statistical- excessive variation in results of studies, variation of treatment effects above that expected by chance

34
Q

How is heterogeneity identified?

A

Seen as outlier (CI not overlapping)

Test using chi-squared

If P value is <0.1 there is statistically significant heterogeneity (may not be able to pool data)

Over 50% on I squared

If the studies do not overlap- more heterogeneity

*** No heterogeneity- if you can draw a line that goes through all confidence intervals

35
Q

What is subgroup analysis?

A

Evaluation of things you expect in advance may alter effect of intervention (predetermined)
- Gender, age, disease subtypes (mild, moderate, severe)

36
Q

What is done when carrying out sensitivity analysis?

A

Consider if results change with small variations in data and methods
- Look at risk difference and risk ratio
- If unsure about data exclude
- May choose to exclude high risk of bias studies or analyse separately

37
Q

How are studies labelled if they have high/low risk of bias?

A

B- high risk of bias
A- Low risk

38
Q

What is a common sensitivity analysis?

A

Rerun meta-analysis with studies that have low risk of bias (based on quality)

39
Q

What are fixed effects?

A

All pooled studies are so similar that it is like they have come from same study, true answer is the same
- Narrower confidence interval

40
Q

What are random effects?

A

Assumes all studies are slightly different (can be similar), true answer is slightly different
- More conservative and wider CI (preferred)
- Should use this method unless there is good reason to believe that they can be considered in fixed effects model

41
Q

What is GRADE? How is it used?

A

How confident are you in results of meta-analysis
- Evaluation of quality of body of literature
- High if certainty of evidence is high- low risk of bias, little heterogeneity, high level of confidence that point estimate is correct
- Moderate, low, very low
- Gives feel for what the body of evidence is like
- How reliable is estimate

42
Q

What factors can lead to downgrading of studies?

A
  • High or unclear risk of bias (look at risk of bias table- reds and ambers)- poor design and conduction
  • Lack of consistency between studies- heterogeneity (can draw line down confidence intervals, I2 value is high)
  • Indirectness- were all studies the same in terms of PICO (if different then down grade)
  • Imprecision- if very wide confidence intervals (less certainty about estimate), wider diamond
  • Publication bias- if it is likely that negative or null results were not published (tested using funnel plot)
43
Q

What is contained in a summary of findings tables? (found at beginning)

A
  • Includes key findings- results, number of studies/participants, relative and absolute effects
  • Gives quality of evidence, magnitude of effect and reasons behind judgements
  • Takes format of PICO
  • GRADE

bds4 pmhp (3 decks)

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