Meta-Analisis

Bersifat naratif dan tidak sistematis

•penelusuran dan pemilihan artikel yg

hendak digabungkan tdk dilakukan
dgn kreteria yg ditetapkan sebelumnya Systematic
review yg
•kurang dilakuan telaah kritis dan memakai analisis
evaluasi sistematis thd kualitas artikel statistika formal

Tinjauan Pustaka Systematic review META

Literature review ANALYSIS
Tinjauan pustaka yg
Review article dibuat scr sistematic dan
Overview terencana, tapi tdk
menggunakan analisis
State of the art review statistik formal
Meta-analysis
• Systematic review of all available data from
published studies in a specific area.

• Statistical re-assessment of these data and is

used to give a summary or overview of what is
known about the area.
• A researcher for example, find all of the
published research articles relating to the
treatment of depression with Hypericum. The
data from these papers are then extracted,
summarised, re-analysed and presented as an
overview of the effectiveness of the treatment.
• Strengths:
– Can provide a systematic overview of a research
problem
– Can address larger research questions
– Can provide methodological assessment of the
current research designs.
– Can extract a significant finding from a diverse
array of studies.
• Weaknesses:
– Can be obscure
– Can be bias and misleading
– Can be chosen simply because of laziness or lack
of rigour
– Can be unsystematic
– Can be unreliable and invalid
Evaluating research for evidence-based practice

• Four things to be considered when evaluating

research publications (NHMRC, 1998):
1. Level of evidence
2. Quality of evidence
3. Relevance of evidence
4. Strength of evidence
• Level of evidence
– The study design is used as an indicator of the
degree to which bias has been eliminated by the
design.

• Quality of evidence
– The quality of the methods used by investigators
to minimise bias in a study
• Relevance of evidence
– The closeness of the study question to the clinical
question, which is determined by the relevance of the
outcome measures used and the applicability of study
results to other treatments, settings and patients.

• Strength of evidence
– The magnitude, precision and reproducibility of the
intervention effect; includes the effect size, confidence
interval, p value, and the exclusion of clinically
unimportant effects. In the case of non randomized
studies, additional factors such as biological plausibility,
biological gradient, and temporality of associations may be
considered.
LEVELS OF EVIDENCE

I. Evidence obtained from a systematic review of all

relevant randomized controlled trails (meta-
analysis)

II. Evidence obtained from at least one properly

designed randomized controlled trial

III-1 Evidence obtained from well designed pseudo-

randomized controlled trials (alternate allocation
or some other method of participant allocation)
III-2 Evidence obtained from comparative studies with
concurrent controls and allocation not randomized
(cohort studies), case control studies, or
interrupted time series with a control group

III-3 Evidence obtained from comparative studies with

historical control, two or more single-arm studies,
or interrupted time series without a parallel group

IV Evidence obtained from a case series, either post-

test or pre-test and post-test.
The basis of meta-analysis
• Good quality meta-analysis is seen as the
highest level of evidence for establishing cause
and effect relationships in health research.

• The quality of the meta-analysis is the crucial

determinant in the strength of the findings. The
quality of the methods used by investigator to
minimize bias in the study design.
Protocol for undertaking a meta-analysis:
1. Identify relevant studies:
– Search the literature
– Consult experts in the field to determine the best source of
studies
– Include unpublished studies, because of the effect of
“publication bias” favouring those studies with positive,
significant results.

2. Inclusion/exclusion criteria:
– Consult experts in the field to determine sound criteria
– Determine which designs will be used. Determine which
language articles to be included. Determine which articles to
use when multiple articles report on the same data.
– Minimum sample sizes
– Extent of follow up
 – Nature of the treatment and/or exposures
– Type of measurement used
– Poor studies should be eliminated on the basis of clear criteria
– Rating scales may be used to evaluated studies for inclusion
– Studies may be included in groups according to their quality.

3. Data abstraction:
– Data are taken from each study in a way which makes the
process reliable, valid and free of bias.
– Blinding the abstractor is the best way to do this.

4. Statistical analysis and exploration of heterogeneity:

– Produce summary estimate of association and variance with
95% confidence limits.