Lecture 13 - Reviews Of Evidence (Systemic Reviews And Meta-analyses)

Question 1

What are the 2 types of descriptive studies?

Answer 1

Ecological studies
Cross-sectional surveys

Question 2

What are the 2 types of analytical studies?

Answer 2

Case control studies
Cohort studies

Question 3

What are the strongest type of scientific evidence?

Answer 3

Meta-analyses
Systematic reviews

Question 4

What is primary research?

Answer 4

One study done like an RCT

Question 5

What is secondary research?

Answer 5

Reviews (so research on research)

Question 6

What is a narrative review?

Answer 6

Implicit assumptions, opaque methodology and is not reproducible

Its biased and subjective

Question 7

What are systematic reviews?

Answer 7

Explicit assumption, transparent ,methodology and IS Reproducible

A summary of medical literature on a topic conducted using explicit methods

Its unbiased and objective so BETTER THAN NARRATIVE REVIEWS

Question 8

What is meta-analysis?

Answer 8

A quantitative synthesis of the results of two or more primary studies (lots of studies) that addressed the same hypothesis in the Same way

Question 9

What are the criteria for a systematic review?

Answer 9

Explicitly states inclusion and exclusion criteria:

-types of study
-types of participants
-types of interventions
-types of outcome measures

So a clearly focused research question

Question 10

What is the purpose of a meta-analysis?

Answer 10

Systematically collates study results and to facilitate the synthesis of a large number of study results

This reduces problems of interpretation due to variations in sampling

Question 11

What is the quality criteria for a meta-analysis?

Answer 11

Complication of complete set of studies
Identification. Of common variable or category definition
Standardised data extraction
Analysis incorporation for sources of variation

Question 12

How are odds ratios and confidence intervals used in meta analysis?

Answer 12

In meta analysis, odds ratios and their 95% confidence intervals are calculated for all studies and are combined to give a pooled estimated odds ratio

The studies are WEIGHTED depending on their size and uncertainty of their odds ratio (narrower Confidence interval = greater weight to result)

Question 13

Look at slide 34 at the forest plot,

What does the size of the square represent?
What does the solid line indicate?
What does the diamond and its width indicate?

Answer 13

Size of square = proportional to the weight given to the study

Solid line = null value (1 in this case cuz its risk ratio, would be 0 in mean absolute difference)

Diamond width = pooled 95% confidence interval
Diamond = pooled estimate/dotted line = pled odds ratio

Question 14

Looking at slide 33 and 34 which test was statistically significant and why?

Answer 14

7 and the meta analysis pooled estimate

The horizontal line for study 7 (which indicates the confidence interval doesn’t cross the null value of 1, so null value isn’t in the confidence interval)

Meta analysis pooled estimates confidence interval is represented by the width of the diamond and it doesn’t cross the null value of 1

Question 15

What are the 3 key problems in meta-analaysis?

Answer 15

Heterogenity between studies
Variable quality of the studies
Publication bias in selection of studies

Question 16

What is publication bias?

Answer 16

when you only use studies that support what you want the meta-analysis to support.

Question 17

What is meant by heterogenity between studies?

Answer 17

The observed effects in the studies are more different than you’d expect by chance

Question 18

What causes heterogenity between studies?

Answer 18

Differences in methods (study design) methodological heterogenity

Difference in patients, interventions and outcomes (clinical heterogenity)

Question 19

How does a homogenous forest plot differ to a more heterogenous forest plot?

(Slide 43)

Answer 19

Homogenous = all studies cross the null value solid line

Heterogenous = not all studies cross the null value solid line

Question 20

How can we count for heterogenity between studies?

What 2 models?

Answer 20

Fixed effect model

Random effects model

Question 21

What is the fixed effect model?

Answer 21

Assumes that studies are estimating exactly the same true size effect

Question 22

What is the random effects model?

Answer 22

Incorporates heterogenity by assuming that studies are estimating similar but not the same true effect size

Question 23

What model should you always use off theres heterogenity between studies?

Answer 23

RANDOM EFFECTS MODEL

Question 24

How is the confidence interval affected with heterogenity in random effects model than in fixed model?

Answer 24

Confidence interval slightly wider in Random Effects model

Question 25

What is the problem of the random effects model with heterogenity?

Answer 25

Only incorporates heterogenity and DOESNT EXPLAIN it

Question 26

What are the issues with the variable quality of the studies in meta-analyyysis?

Answer 26

Poor study design
Poor design protocol
Poor protocol implemtation

Question 27

What type of studies are more susceptible to bias and confounding?

What types of studies are less susceptible to bias and confounding?

Answer 27

More susceptible = case-control, cohort

Less susceptible = randomised control trials, non-randomised controlled trials

Question 28

How do you decide what. Studies you’re going to combine in meta—analysis?

Answer 28

Define a basic quality standard and only used them

Score each study for its quality and then incorporate the quality score and allocate it a weight

Question 29

How can you decide a meta-analysis has been subject to publication bias?

Answer 29

A funnel plot

Question 30

Loo at slide 66:

Why does the funnel plot on the right side show evidence of publication bias?

Answer 30

There lacking in smaller studies like the other funnel plot

Question 31

How do you interpret a funnel plot?

Answer 31

Measures study size against measure of effect

If no publication bias then the plot will be balanced if there is publication bias there are going to be fewer smaller studies with results indicating small or negatives measures of effect