Lecture 31: Systematic Reviews

Question 1

Define review

Answer 1

an assessment of something with the intention where change is made if necessary

Question 2

What are the two main types of review?

Answer 2

Narrative and systematic review

Question 3

How does the question differ between a systematic and narrative review?

Answer 3

in a narrative review, the question is often broad in scope whereas in a systematic focused clinical question

Question 4

What are some requirements of a systematic review?

Answer 4

it must be rigorous, reproducible and have a replicable method

Question 5

What are the risks of a narrative review?

Answer 5

the writers only tell us what they want you to know so they are not good for research review.

Question 6

It is possible to address different questions, such as

Answer 6

• what’s the effectiveness of the treatment?
• what do we know about the likely prognosis if someone presents with this condition?
• is the test sensitive and specific to a particular condition?
• how common is the condition? (prevalence and incidence)
• what are some patient experiences?

Question 7

Systematic reviews of which trial type of the most common?

Answer 7

RCT

Question 8

What are Cochrane reviews?

How are they conducted?

Answer 8

They are systematic reviews of primary research in human health care and health policy.
Cochrane reviews find as many RCTs as possible and conduct systematic reviews of all the RCTs.

Question 9

Why are systematic reviews conducted?

Answer 9

They involve collating evidence and synthesising their results so that if they are done well, they could reduce bias

Question 10

What are the 7 steps to conducting a systematic review?

Answer 10

1. formulation of a clear question
2. writing a protocol for the review
3. searching for relevant studies
4. collecting data from studies
5. assessment of included studies
6. synthesis of findings
7. interpretation of results

Question 11

One of the steps to conducting a systematic review is forming a clear question. What four things do we need to include in the question?

Answer 11

1. population
2. intervention
3. comparison
4. outcome of interest

Question 12

One of the steps to conducting a systematic review is writing a protocol for the review. What is the purpose of the protocol and what 9 things need to be included?

Answer 12

It sets out exactly how you are planning to conduct the review. It needs to state:

1. what the question is
2. it’s relevance/why it’s important
3. the objectives
4. strategy for searching for trials
5. selection criteria
6. how you are going to screen for eligibility
7. the risk of bias
8. how you are going to extract the data from the original trials
9. data synthesis

Question 13

One of the steps to conducting a systematic review is searching for relevant studies. What are two ways we can search for relevant studies?

Answer 13

• electronically as they are indexed on electronic databases on medical research
• paper copies looking for every record of RCT present at a conference

Question 14

What are three problems with finding relevant studies?

Answer 14

1. there could be missing data
2. publication bias - the statistically significant findings are more likely to be published so we are unlikely to find ones that show no benefit
3. language bias - we could be missing studies that are not published in English

Question 15

One of the steps to conducting a systematic review is writing a protocol for the review. One of the things needed in a protocol is screening and data extraction. What does this involve?

Answer 15

we have to decide whether the studies found are included or not and so we have two different people doing it and then we see if they agree

Question 16

How do we do a risk of bias assessment as part of the protocol?

Answer 16

We have to look at whether there is a high risk of systematic error (bias) is high or low.

Question 17

What is GRADE?

Answer 17

The Grading of Recommendations Assessment, Development and Evaluation working group has developed a common, sensible and transparent approach to assessing the risk of bias to upgrade or downgrade the strength of recommendations for clinical practice on the basis of evidence.

Question 18

How can we synthesise data?

Answer 18

using meta-analysis to form a forest plot

Question 19

Define heterogeneity

Answer 19

when something varies across different groups

Question 20

Define homogeneity

Answer 20

when something is consistent across different groups

Question 21

What are the differences between a systematic review and a meta-analysis?

Answer 21

• the SR is a research method for the review as a whole

- meta-analysis is a method of pooling data from individual studies in a reviw

Question 22

How do SR and meta-analysis relate to one another?

Answer 22

SR may may not contain a meta-analysis

Question 23

How can you read a forest plot?

Answer 23

• the horizontal lines represent the lines of one study with the width showing the width of the 95% confidence interval and the dot representing the RR
• vertical line represents the line of no effect
• total 95% CI is the pooled estimate of the effect with the diamond the pooled RR
• if the horizontal line crosses the vertical line, the study was not statistically significant

Question 24

What do we write about in the interpretation?

Answer 24

• principle findings
• limitations of the evidence
• implications both in practice and in research

Question 25

What are 8 advantages of systematic reviews?

Answer 25

1. can do our best to reduce publication bias due to poor quality trials
2. can address heterogeneity
3. rigorous and reproducible
4. comprehensive
5. has transparent limits
6. addresses our gaps in knowledge
7. is the basis for decisions

Question 26

What are two disadvantages of systematic reviews?

Answer 26

we can have conflicting reviews and inconclusive results