|
Most doctors in training have to sit exit examinations that enable them to train further as specialists such as physicians, surgeons, psychiatrists and gynaecologists or as GP’s. Increasingly these exit examinations include a section on statistics, critical appraisal and presentation.
This paper gives a format for presentation of a research publication such as a ‘systematic review article’ in examinations and routine clinical meetings.
How
to present a Systematic Review Article
Remember to score points with headings
Title: The
title of my paper is……………………………………………..
Study
design: This is a systematic review of observational studies. The
expectation is that the SR will show the available evidence on this topic and
guide further studies.
OR
This is a systematic review of Randomised Controlled Trials.
The expectation is that the SR will quantify the strength of the available
evidence and provide precise point estimates of the treatment effects on the
outcome [Note: this means
that the larger total sample size will give narrower confidence intervals
around the result]
OR
This is a systematic review of a mixture of observational
studies and RCTs.
1. The
expectation is that the SR of the observational studies will show the available
evidence on this topic and guide further studies.
2. The
expectation is that the SR of the RCTs will quantify the strength of the
available evidence and provide precise point estimates of the treatment effects
on the outcome [meaning tighter confidence intervals around the result].
3.
There
is no reliable expectation when the results of observational studies [Odds
ratio] have been combined with results of RCTs [Relative risk].
Setting:
The study was reported from…………
Hypothesis
testing:
For SR of RCTS –
The null hypothesis was that there were no differences in
the main outcomes between the compared treatments.
The SR looked at other outcomes which the individual studies
could not reliably do because of small sample sizes. The review had a better
sample size to look at these outcomes.
For SR of observational studies –
This review did not have a specific hypothesis, and it was
not expected to have one because it was meant to gather the available evidence
to formulate a hypothesis for a future study.
Methods:
The methodology was satisfactory because it clearly detailed
the ingredients of a systematic review:
1. Formulation
of the research question:
Inclusion,
exclusion, participants, interventions, comparisons, outcomes, study designs
and methodological quality.
2.
Locating and selecting
studies
Search
strategy, searching electronic databases, reference lists, hand searching and personal
communication.
3.
Critical appraisal of
studies
Eligibility
checks > 1, strategy to resolve disagreements, keeping a log of excluded studies
and reasons for exclusions.
4.
Analysing and
interpreting results
Data extraction, tabulation, forest plots,
heterogeneity, limitations, strength of evidence, applicability and publication.
OR
The methodology was NOT satisfactory because it DID NOT
clearly detail the ingredients of a systematic review such as:
1. Formulation
of the research question:
Inclusion,
exclusion, participants, interventions, comparisons, outcomes, study designs
and methodological quality.
2.
Locating and selecting
studies:
Search
strategy, searching electronic databases, reference lists, hand searching and personal
communication.
3.
Critical appraisal of
studies:
Eligibility
checks > 1, strategy to resolve disagreements, keeping a log of excluded studies
and reasons for exclusions.
4.
Analysing and
interpreting results:
Data extraction, tabulation, forest plots,
heterogeneity, limitations, strength of evidence, applicability and publication.
[Note: In reality the systematic review will detail some of these and fail
in others. You will have to make a judgement on whether it was satisfactory or
not. IT DOES NOT MATTER WHAT TO DECIDE – REMEMBER YOU JUST WANT TO SCORE POINTS
BY POINTNIG OUT WHAT SHOULD HAVE BEEN INCLUDED]
Results:
I shall present the findings in the main outcome.
The forest plot in Table X
1. The
studies that compared drug A to drug B.
2. The
outcome was ‘pregnancy rate’
3. There
were 5 studies, with a total sample size of .......
In the total population of 360 patients [170 in Group A and
190 in Group B]:
1. 66% of
patients on Drug A compared to 33% on Drug B became pregnant. The RR of
pregnancy was 1.65, 95% CI was 1.33 to 1.81.
a. As the
95% CI did not include 1, this result is statistically significant
OR
b. 66% of
patients on Drug A compared to 55% on Drug B became pregnant. The RR of
pregnancy was 1.35, 95% CI was 0.86 to 2.33.
As the 95% CI included 1, this
result was NOT statistically significant.
2. The
test of heterogeneity shows a p-value which is > 0.05, so the subjects in
the 5 studies were similar or homogenous
a. Therefore
the outcome effect, that is, pregnancy will only occur in the type of subjects
recruited into the studies
OR
b. The
test of heterogeneity shows a p-value which is < 0.05, so the subjects in
the 5 studies were dissimilar or heterogenous
i.
Therefore the outcome effect, that is, pregnancy
can be expected in a generalised group [or the results can be generalised]
What
are the weaknesses?
A SR of observational studies that has included a
meta-analysis:
This is a major weakness because it will have combined
observational studies which are prone to bias and confounding. Even if each of
the studies has attempted to minimise bias and confounding, META-ANALYSIS will
add ‘salt to injury’ when it attempts to combine the studies.
What are
the strengths?
A SR of RCTs that has included a meta-analysis:
1. This
would give a precise estimate of the treatment effect in the population.
Remember this population answer is UNKNOWN, and we are trying to estimate it. The
larger sample size helps towards this.
2. RCTs
are handicapped by the weakness that the study population is usually homogenous
by the clear selection criteria. This limitation of this is that the results
cannot be generalised.
When
RCTs are combined in a SR, this give s an opportunity to include a diverse
population, that is, a more heterogenous population. If the
treatment effect shows a benefit still, then the results can be generalised.
Did the
weakness affect the results?
A SR of observational studies that has included a
meta-analysis:
If the meta-analysis was presented, this is a major weakness
which can affect the result.
Apart from the above [WHICH IS A ‘NO NO’]…..
You can actually say anything you like at this stage because
you will have scored all the available points
Conclusion:
You can actually say anything you like at this stage because
you will have scored all the available points
|
