Académique Documents
Professionnel Documents
Culture Documents
Background: This meta-analysis aimed to investigate whether preoperative biliary drainage (PBD) is
were extracted by two independent reviewers. Risk ratios, rate ratios or mean differences were calculated
with 95 per cent confidence intervals (c.i.), based on intention-to-treat analysis, whenever possible.
Results: Six trials (four using percutaneous transhepatic biliary drainage and two using endoscopic
sphincterotomy) including 520 patients with malignant or benign obstructive jaundice comparing PBD
(265 patients) with no PBD (255) were included in this review. All trials had a high risk of bias. There was
no significant difference in mortality (risk ratio 1.12, 95 per cent c.i. 073 to 171; P = 060) between the
two groups. Overall serious morbidity (grade III or IV, ClavienDindo classification) was higher in the
PBD group (599 complications per 1000 patients) than in the direct surgery group (361 complications
per 1000 patients) (rate ratio 166, 95 per cent c.i. 128 to 216; P < 0001). Quality of life was not
reported in any of the trials. There was no significant difference in length of hospital stay between the
two groups: mean difference 487 (95 per cent c.i. 128 to 1102) days (P = 012).
Conclusion: PBD in patients undergoing surgery for obstructive jaundice is associated with similar
mortality but increased serious morbidity compared with no PBD. Therefore, PBD should not be used
routinely.
Paper accepted 8 July 2013
Published online in Wiley Online Library (www.bjs.co.uk). DOI: 10.1002/bjs.9260
Introduction
1590
Two authors independently identied trials for inclusion, and two other authors extracted data related to the
above outcomes and assessed the risk of bias in the trials.
All differences of opinion were resolved by discussion.
The systematic review was conducted following the Preferred Reporting Items for Systematic Reviews and MetaAnalyses (PRISMA)15 and Cochrane Collaboration14
guidelines.
Statistical analysis
The software package RevMan 526 , provided by the
Cochrane Collaboration, was used for analysis. For
dichotomous variables, the risk ratio with 95 per cent condence interval (c.i.) was calculated. For count data variables
such as serious adverse events, the rate ratio with 95 per cent
c.i. was determined. For continuous variables, the mean difference with 95 per cent c.i. was calculated. If mean values
were not available for continuous outcomes, medians were
used for meta-analysis. If standard deviations were not
available for continuous outcomes, they were calculated
according to Cochrane Collaboration guidelines14 . This
involves assumptions that both groups have the same variance, which may not be true. In order to assess the impact
of such imputations, a sensitivity analysis was performed by
excluding such trials. Random-effects and xed-effect models were used27,28 . Where there was no difference between
the results of random-effects and xed-effect models that
would change interpretation of the results, the results of
the xed-effect model are reported; otherwise, both results
are reported. Heterogeneity was explored using the 2
test, with signicance set at P = 0100, and the amount of
heterogeneity29 was determined by means of I 2 .
All analyses were based on the intention-to-treat
principle30 , if possible. Otherwise, an available case
analysis14,19 was performed. Planned subgroup analyses
included: trials with low risk of bias versus trials with an
unclear or high risk of bias; PTBD or endoscopic method;
and trials with more than 95 per cent of patients having
malignant jaundice versus those with less than 95 per cent.
The rst subgroup analysis based on the risk of bias was not
carried out because of lack of trials. A sensitivity analysis
was performed by excluding trials in which median or
standard deviation was imputed for continuous outcomes.
www.bjs.co.uk
Safety and efficacy of biliary drainage before surgery for obstructive jaundice
1591
Records screened
n = 3783
Records excluded
n = 3768
Full-text articles assessed for eligibility
n = 15
Full-text article excluded (quasi-randomized study)
n=1
Trials included in qualitative synthesis
n = 6 (14 references)
Search results
A total of 4412 references were identied from medical
journal databases. One new reference was identied by
reference search. No new trials were identied through reference searches or the mRCT. In total, 14 references33 46
for six completed randomized trials fullled the inclusion
criteria (Fig. 1). When citing the references for a trial, both
the main publication and any secondary reports are shown.
A total of 520 patients randomized to PBD (265 patients)
or no PBD (255 patients) were included in the six trials.
Effect estimates
The effect estimates are summarized in Table 2.
Mortality
There was no signicant difference (risk ratio 112, 95
per cent c.i. 073 to 171; P = 060) in mortality between
www.bjs.co.uk
1592
Table 1
Allocation
concealment
Blinding
Incomplete
outcome
data
?
?
?
?
+
+
?
+
Reference
Postrandomization dropouts
(no. and reasons)
Selective
reporting
Vested
interest
bias*
?
?
+, Low risk of bias; , high risk of bias; ?, unclear risk of bias. *Present when a trial is conducted or funded by a party that stands to gain or lose
depending on the results of the trial.
Table 2
Effect estimates for preoperative biliary drainage versus direct surgery in patients with obstructive jaundice
Illustrative comparative risks*
Outcome
Mortality
Long-term survival
(follow-up 24 months)
Serious adverse events
Hospital stay (days)
Assumed risk
in control
groups
Relative effect
No. of patients
Quality of
evidence
(GRADE)
520 (6 studies)
185 (1 study)
Very low
Very low
503 (6 studies)
271 (2 studies)
Very low#
Very low# **
Values in parentheses are 95 per cent condence intervals (c.i.) unless indicated otherwise. *The basis for the assumed risk is the mean control group risk
across studies; the corresponding risk (and its 95 per cent c.i.) is based on the assumed risk in the comparison group and the relative effect of the
intervention (and its 95 per cent c.i.). Grading of Recommendations Assessment, Development and Evaluation (GRADE) Working Group grades of
evidence: high quality further research is very unlikely to change our condence in the estimate of effect; moderate quality further research is likely to
have an important impact on our condence in the estimate of effect and may change the estimate; low quality further research is very likely to have an
important impact on our condence in the estimate of effect and is likely to change the estimate; very low quality we are very uncertain about the
estimate. All trial(s) had a high risk of bias. Overlaps 1 and 075 or 125. There were few trials to assess publication bias. #I 2 value greater than 40 per
cent; there was also minimal overlap of condence intervals between some of the trials. **Overlaps 0 and minimal clinically important difference.
the two groups (Fig. 2). The mortality rate was 149
per cent (adjusted proportion calculated by multiplying the
risk ratio by the control group proportion; crude rate 40 of
265, 151 per cent) in the PBD group and 133 per cent (34
of 255) in the direct surgery (no PBD) group. There was
no change in the results when a random-effects model was
used. Heterogeneity was not statistically signicant (I 2 = 0
per cent; P = 070). Tests for subgroup differences were not
statistically signicant for the subgroup analyses. Exclusion
of the trial by van der Gaag and colleagues40 42,44 46 , which
measured mortality for up to 120 days (whereas the other
trials measured 30-day or in-hospital mortality), did not
affect the results. The only trial40 42,44 46 that reported
on long-term survival in patients who had successful
pancreatoduodenectomy for cancer found no signicant
difference between the two groups (hazard ratio 090, 95
per cent c.i. 065 to 124).
2013 British Journal of Surgery Society Ltd
Published by John Wiley & Sons Ltd
Safety and efficacy of biliary drainage before surgery for obstructive jaundice
1593
Mortality rate
Reference
PBD
33
Hatfield et al.
Lai et al.34,35
McPherson et al.3638
Pitt et al.39
van der Gaag et al.4042,4446
Wig et al.43
4 of 29
6 of 43
11 of 34
3 of 37
15 of 102
1 of 20
Total
40 of 265
Risk ratio
Risk ratio
117
171
181
57
360
115
1000
001
01
Favours PBD
10
100
Forest plot comparing mortality in trials that used preoperative biliary drainage (PBD) before surgery and those that did not
(direct surgery). A MantelHaenszel xed-effect model was used for meta-analysis. Risk ratios are shown with 95 per cent condence
intervals
Fig. 2
Weight (%)
s.e.
Hatfield et al.33
192
062
47
Lai et al.34,35
039
033
165
McPherson et al.3638
076
047
81
Pitt et al.39
van der Gaag et al.4042,4446
003
062
039
020
118
450
Wig et al.43
006
036
139
1000
Reference
Total
Rate ratio
Rate ratio
005
02
Favours PBD
20
Forest plot comparing serious adverse events in trials that used preoperative biliary drainage (PBD) before surgery and those that
did not (direct surgery). Data are shown in a logarithmic scale. An inverse-variance xed-effect model was used for meta-analysis. Rate
ratios are shown with 95 per cent condence intervals
Fig. 3
Quality of life
Costs
Hospital stay
Meta-analysis of two trials39 42,44 46 showed a signicantly
longer hospital stay in the PBD group in the xed-effect
model (mean difference 426 (95 per cent c.i. 092 to
761) days; P = 001) but not in the random-effects model
(mean difference 487 (128 to 1102) days; P = 012).
Heterogeneity was statistically signicant (I 2 = 68 per cent,
P = 008). No subgroup analysis was performed because
there were only two trials for this outcome. Sensitivity
analysis, excluding the trial of van der Gaag et al.40 42,44 46
in which the mean and standard deviation were imputed
from median and interquartile range values, resulted in
the inclusion of only one trial39 . This trial39 showed a
signicantly longer hospital stay in the PBD group.
2013 British Journal of Surgery Society Ltd
Published by John Wiley & Sons Ltd
Discussion
1594
www.bjs.co.uk
Safety and efficacy of biliary drainage before surgery for obstructive jaundice
1595
www.bjs.co.uk
1596
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
www.bjs.co.uk