Professional Documents
Culture Documents
Systematic Review and Meta-Analysis
Systematic Review and Meta-Analysis
. . 1, .. 2, .. 3
1
, 2
(Concept of Systematic Review)
-
Systematic Review
Systematic review
(systematic)
(quantitative)
Meta-analysis Meta-analysis systematic
review
-
2. Systematic review
(quantitative)
3. Systematic review
Systematic review
Relative risk
reduction (RRR)
18%
10%
16%
20%
10%
50%
type I error = 0.05 power 80%
()
12,280
2,994
438
1 (RRR =10% ,
20% 50%)
1
RRR 10 RRR 20
50 4 30
Systematic review Meta-analysis
80-90
10 10
10
Systematic review Meta-analysis
Meta-analysis The Lancet .
2547 observational Antioxidant beta
carotene, vitamins A, C, E
Systematic review Meta-analysis Controlled trial
Antioxidant
Antioxidant
(Question and inclusion criteria formulation for
selecting the primary studies)
(Population),
(Intervention/Exposure),
(Comparison)
(Outcome) PICO1
1. (therapy)
1. :
2. :
Transarterial therapy
3. :
4. :
1
Transarterial therapy
1
2. (risk factor/etiology)
1. :
2. :
(passive smoking)
3. :
4. :
(Inclusion criteria)
1. :
2
1.1.
1.1.1.
1.1.2.
(Imaging study
Ultrasonography, CT MRI scan) alpha
fetoprotein
1.2.
1.2.1.
2. :
beta blocker
2.1. Transarterial therapy
2.1.1. Transarterial chemoembolization
2.1.2. Transarterial embolization
2.1.3. Transarterial chemotherapy
2.2.
(co-intervention)
2.2.1.
( )
3. :
3.1.
4. :
4.1. 1 (one-year survival)
4.2. (tumor response)
4.3. (adverse effect)
5. : Randomized Controlled Trial (RCT)
systematic review
RCT
systematic review
Observational studies
case-control, cohort cross-sectional
systematic review
5.1. Randomized Controlled Trial
keywords
Publication Bias
Systematic review Meta-analysis
(positive trials)
(negative trials)
negative trial
negative trial
Sample Size
Pooled Estimate
of treatment effect
Favor Intervention
Favor Control
Sample Size
Biased estimate
(publication biased)
Favor Intervention
Unbiased estimate
Favor Control
negative trials,
Publication bias
(unpublished trial)
(Study Identification)
(Population, Intervention,
Comparison, Outcome PICO)
systematic review
2
-
(Existing reviews)
PUBMED
keyword methodology filters
PUBMED clinical queries Systematic reviews
o PUBMED
- Evidence Based Medicine Reviews
.
Cancer Care Ontario : Clinical Practice Guidelines
Shojania KG, Bero LA. Taking advantage of the explosion of systematic reviews: an
efficient MEDLINE search strategy. Eff Clin Pract. 2001 Jul-Aug;4(4):157-62.
1. Keyword methodology
filter ( systematic review, metaanalysis)
a. Keyword Hepatocellular Carcinoma,
- (PICO),
- index keyword
subject heading
-
index
b. Keyword methodology filter
2. PUBMED
clinical queries systematic reviews methodology filter
National Library of Medicine Dr.Haynes B et al,
Keyword
3. Evidence
Based Medicine Reviews Cochrane Library Database of
Abstracts of Reviews of Effects Clinical Evidence UPto date systematic review / meta-analysis
Keyword
4.
search engine web site
http://www.shef.ac.uk/scharr/ir/netting/ Cancer Care
Ontario Clinical Practice Guidelines American Society of Clinical
Oncology
Systematic review 3
1 2
From our review of review, currently there are three meta-analyses addressing this
method of treatment Simonetti (5) in 1997, Mathurin (6) in 1998 and Comma (7) in 2002. The
first and the second meta-analysis included 3 and 6 RCTs which compared TAE or TACE with no
active treatment or co-intervention in both treatment and control groups. Simonetti et al. reported
border-line significant therapeutic benefit (Odd ratio of 1-year survival rate = 2.0, 95% CI 1.1-3.6)
in the set of 3 RCTs where embolization was compared to no treatment or to intravenous 5-FU in
1997. In 1998, Mathurin et al reported the result from a meta-analysis of 6 RCTs, comparing
TACE or TAE with no active treatment or tamoxifen or intravenous 5-FU, which failed to show a
benefit of TACE on 1-year survival (Risk difference=0.3%, 95%CI = -11 to 12 %). In this year,
2002, instead of reporting the result of 1-year survival, Comma reported the benefit of 2-years
survival of 5 RCTs meta-analysis (OR = 0.54, 95% CI = 0.33-0.89).
Since hepatocellular carcinoma is the fatal disease with only 20% survival rate at
one-year (8-10). For this reason, one year-survival rate would be the better information in the
decision making for both HCC patients and the treating physicians. Therefore, we conducted this
meta-analysis on the updated and comprehensive searched for the RCTs by using the overall oneyear survival as the end point.
Keyword methodology filter
( randomized controlled trial systematic review metaanalysis)
Dr.Haynes Keyword
o
PUBMED Cochrane Central Register
of Controlled Trials
o methodology filter
2.
3.
4.
random allocation.sh.
5.
6.
7.
or/1-6
8.
9.
7 not 8
15. placebo$.ti,ab.
16. random$.ti,ab.
17. research design.sh.
18. or/10-17
19. 18 not 8
20. 19 not 9
21. comparative study.sh.
22. exp evaluation studies/
23. follow up studies.sh.
24. prospective studies.sh.
25. (control$ or prospectiv$ or volunteer$).ti,ab.
26. or/21-25
27. 26 not 8
28. 27 not (9 or 20)
29. 9 or 20 or 28
30. exp asthma/
31. 29 and 30
Lines 30 and 31 for illustration only.
Search Strategies
1.
nasopharynx
2.
3. (hand searching)
(Reference lists)
search strategies
(Study selection)
systematic review
(review) (selection criteria) review
4
1. (study population)
2. (study intervention)
(risk factor) (prognostic factor)
(diagnostic test)
3. (outcomes)
4. (research methodology)
review review randomized controlled trials
review review
/
(selection criteria) 1
review
review
review
review
(
systematic review )
(bias)
review
systematic review (reliability
reproducibility) review
2 (
)
chance corrected agreement kappa ()
100
2 1
review 20 2 25 2
18 2
2 review 2
1
18
7
25
a b
a+b
2
2
73
75
c d
20
c+d
80
a+c b+d
100
n
2 2 ( ) 18
+ 73 = 91 100 91/100
(observed agreement raw agreement)
2 2
(chance)
1 20 100 20 % 2
25 100 25 % 2
( agreement ) 25 2
1 5 ( 20 % 25 ) 75
2 1 60 ( 80 %
75 ) 2
5 + 60 = 65 100 ( 3) 65/100
(chance agreement expected agreement)
3 review 2
2
1
5
20
25
2
15
60
75
20
80
100
2
2
kappa
observed agreement chance
agreement 1
chance agreement
kappa =
( 2)
a+d
n
(
a + c )(a + b ) + (b + d )(c + d )
chance agreement =
n2
(91 / 100) (65/100) = 0.74
kappa =
1 (65/100)
observed agreement =
kappa 2
kappa 1 kappa 1 2
review 2
kappa 0 2
2 kappa
2 kappa 1
review
2 (
)
2
(consensus)
2
review
(consensus)
review
systematic review
(bias)
review
systematic review 1 (duplicate
publication)
3 5
3
(
)
review review
review duplicate publication
(
)
duplicate publication review
controlled trial)
2. concealed randomization (
subject )
3.
( blinding)
4. outcome (completeness
of follow-up)
5. subject subject ( )
subject (
intention-to-treat principle)
(diagnostic test)
1. diagnostic test
2. diagnostic test (gold standard)
3. diagnostic test gold standard gold standard
diagnostic test (blind comparison)
(risk factor etiology)
(prognostic factor)
1. cohort study (bias)
case-control study
Guyatt H, Rennie D, editors. The Evidence-Based Midicine working Group. Users guides to the medical
literature: A manual for evidence-based medicine clinical practice. AMA Press, 2002.
2
2.
2
3. (exposure assessment)
case-control study
4. (outcome assessment)
5. (completeness of follow-up)
systematic review 2
scale
Jadad scale 3 Chalmers scale 4
4
systematic
review
review
heterogeneity ( )
(reliability
reproducibility)
2
kappa
(data extraction)
systematic review
systematic review
1.
2.
Jadad AR, Moore RA, Carroll D, Jenklinson C, Reynolds DJM, Gavaghan DJ, McQuay HJ. Assessing the quality of
reports of randomized clinical trials: Is blinding necessary?. Controlled Clinical Trials 1996; 17: 1 12.
4
Chalmers TC, Smith H Jr, Blackburn B, Silverman B, Schroeder B, Reitman D, Ambroz A. A method for assessing
the quality of a randomized control trial. Controlled Clinical Trials 1981; 2: 31 49.
3
3.
4.
5.
150 30 20 %
30 150 20 %
systematic review
intention-to-treat principle
subject
(
)
2
reliability reproducibility
(corresponding author)
(e-mail)
systematic review sensitivity analysis (
)
( Plan of Analysis of Systematic Reviews)
4 ( Cochrane
Reviewers Handbook 4.2.4 March 2004)
1. 2
transarterial chemotherapy
(supportive treatment)
- (direction of effect )
transarterial chemotherapy supportive treatment
- ( size of effect )
transarterial chemotherapy supportive care 10%, 20%
50%
-
(heterogeneity)
2.
2. (Effect measures)
5. (Sensitivity
Analysis)
survival data (
-disease free survival overall survival) Meta-analysis survival data
individual patient data
individual patient data assumption constant hazard ratio follow
up period literature-based data fixed data 1 year
survival Dichotomous binary data (
)
4 Odds ratio
RR RR
initial risk 20%
o Continuous data
Mean difference
Standardised mean difference
o Ordinal data
Proportional odds ratio ( Cochrane review
manager)
Continuous data Binary data
o Counts and rates
Rate ratio
Odds ratio Risk ratio
Meta-analysis
publication bias
Meta-analysis
o Meta-analysis
treatment effect
(pooled results)
model Random
effect model Fixed effect model Random effect model
(treatment effect)
(distribution)
heterogeneity clinical heterogeneity
statistical heterogeneity Fixed effect model
review manager Cochrane Library
model
heterogeneity intervention
Random effect model
review manager
software Cochrane Library Review Manager Review Manager
software Cochrane Library
, Protocol
(Version 4.2.2)
click add
next stage ( )
Heterogeneity
o Heterogeneity
Heterogeneity
clinical diversity
clinical heterogeneity
methodological diversity methodological heterogeneity statistical
heterogeneity
power p value 0.1 0.05 p
heterogeneity
clinical heterogeneity methodological heterogeneity
o Statistical heterogeneity
Meta-analysis
heterogeneity
(Priori hypothesis) (Post
hoc analysis) Statistical heterogeneity
clinical heterogeneity
methodological heterogeneity meta-analysis
heterogeneity subgroup analysis
meta-regression
Ignore heterogeneity Fixed effect model
Random effect model
heterogeneity
heterogeneity
effect measures
continous outcome
meta-analysis
(robust)
Fixed effect
model Random effect model
sensitivity analysis
poor prognostic factors, type of intervention- embolization or chemotherapy or both,
type of chemotherapy, methodological issues, best case and worst case scenario of
some studies
allocation concealment
sensitivity analysis Madden, 1993
worst case scenario
1.
2.
3.
While we are waiting for the complete one-year over all survival data of Madden (61), the sensitivity analyses
assuming three scenarios were also carried out. The scenarios are as follows:
Best case scenario; All of the surviving patients at 150 days in the TAC group were still alive, but all of the surviving
patients at 150 days in controlled group were dead at one year
Worst case scenario; All of the surviving patients at 150 days in the TAC group were dead, but all of the surviving
patients at 150 days in controlled group were alive at one year
The study did not contribute any weight for the pooled estimate i.e. all patients in TAC and controlled group were
dead at one year.
The significant difference in one-year overall survival in favor of patients who received the transarterial interventions
was not changed in the best case and unweighted scenario but disappeared in the worst case scenario.
Alderson P, Green S, Higgins JPT, editors. Cochrane Reviewers Handbook 4.2.2
[updated March 2004]. In: The Cochrane Library, Issue 1, 2004. Chichester, UK:
John Wiley & Sons, Ltd.
(Presenting Results of Systematic Reviews)
(QUOROM)
checklist 5 trial flow
5 Checklist
Moher D. Cook DJ, Eastwood S et al. Improving the quality of reports
of meta-analyses of randomized controlled trials: the QUOROM
statement. Lancet 1999;354:1896-900.
( Interpreting Results )
(Cochrane Reviewers handbook)
1. (What is the direction of effect ?)
(survival )
transarterial chemotherapy supportive care hepatocellular carcinoma
(RR=0.85; 95% CI = 0.74 to 0.99; p=0.04) :
transarterial
chemotherapy marginal
(heterogeneity)
2. (What is the size of the effect ?)
10% ,
20% 50%
transarterial chemotherapy
0.85 supportive treatment
(clinical importance) 0.99
3. (Is the effect
consistent across the studies ? )
test for heterogeneity, p =0.05
<0.1 (heterogeneity)
3 (study design)
(quality of the studies)
:
o heterogeneity
o ( 1 )
update
o effect measures survival data
survival outcome Individual patient data Meta-analysis
Literature-based Meta-analysis
1
6 1
5.
transarterial chemotherapy
(validity)
(Prolongation of overall survival, disease free survival or
progression free survival ) (more quality of life score)
(tolerable toxicity of chemotherapy)
patient, intervention and
disease variation biologic and cultural variation, variation in compliance,
variation in baseline risk
implication Cochrane
reviewers handbook 4.2.2
overall survival 1 marginal
disease free progression free
survival, quality of life score
transarterial chemotherapy
inclusion criteria
subgroup analysis TACE
TAE TAC
( systematic review
meta-analysis )
marginal significance
heterogeneity