5. Systematic Review and Meta-Analysis

Question 1

Systematic review

Answer 1

Systematic review is seen as providing the strongest level of evidence in support of, or
refuting, a specific proposition, although independent reviews of the reviews suggest
that 30–50% of them have significant flaws. They may include meta-analyses. The oral
may include interpretation of a typical forest plot

Question 2

‘A systematic review define

Answer 2

aims to identify all relevant studies on a particular
subject and to then aggregate and interpret the data

‘A systematic review summarizes the results of carefully designed healthcare studies
and provides a high level of evidence on the effectiveness of healthcare intervention’.

Question 3

What makes it a systematic r.v

Answer 3

It must address a clearly formulated question;

must search extensively to identify relevant studies;

must assess the quality of those studies;

and must summarize the evidence,

using statistical methods including meta-analysis as appropriate,
either for the whole
review or for sub-groups.

Interpretation of the findings should characterize heterogeneity
and, if necessary, assign more weight to well-conducted studies

assessing the risk of any forms of bias.

Question 4

Meta-analysis

Answer 4

This is a technique that aggregates the data from a number of
individual RCTs with the aim of confirming or refuting an effect that the smaller studies have been unable to do.

results of individual studies and applies techniques of statistical analysis
to obtain a more accurate judgement of effect

Question 5

Difference

Answer 5

whereas a systematic review aims to answer a defined question by
summarizing the evidence that fits pre-determined eligibility criteria,

meta-analysis uses statistical tools to summarize those results

Question 6

Advantage Meta Analysis

Answer 6

meta-analysis can produce a conclusion (synthesis) from a number of
trials which may even have had contradictory findings. The power and significance
of the overview can be increased by this synthesis of the individual results,
and may allow a definite conclusion to be drawn even when individual studies
conflict.

Question 7

Problems: meta-analyses

Answer 7

‘publication bias’, as negative studies
are much less likely to be published than positive ones

by double counting, which may occur when the same data are incorporated into
more than one trial report

credibility is also tested severely if the populations
in the RCTs are different. Even if the populations are similar, the trial
designs may be very different, with matched subgroups being too small to permit
formal meta-analysis.

Question 8

Forest plots

Answer 8

graphical representation of the effect sizes and confidence intervals from a meta-analysis

Each individual trial
(which is usually listed on the left of the plot)
is represented as a line,

with the centre box representing
the mean effect size of the study.

size of this box represents the weight that is
given to the study. The ends of each line define the 95% confidence intervals

wider the confidence intervals, the less reliable are likely to be the results

Question 9

interpreting forest plots

Answer 9

to the left of which is the side favouring treatment, to
the right of which favours the contr

A summary effect size appears at the base of
the plot in the shape of a diamond.

If part of the diamond touches the
centre line, then the overall combined result is not statistically significant

Question 10

Cochrane Collaboration

Answer 10

produces systematic reviews
of primary research in human health care which are updated in the light of any new
research. Independent assessments of Cochrane reviews suggest that although up to
30% may have significant problems, this nonetheless means that they remain the
most reliable source of aggregated data that is available.

A, level I studies;
B, level II or III studies;
C, level IV studies; and
D, level V evidence or inconsistent or
inconclusive studies of any level.