World J Surg
DOI 10.1007/s00268-015-3252-9
SCIENTIFIC REVIEW
Mesh Location in Open Ventral Hernia Repair: A Systematic
Review and Network Meta-analysis
Julie L. Holihan1 • Duyen H. Nguyen1 • Mylan T. Nguyen1 • Jiandi Mo1
Lillian S. Kao1 • Mike K. Liang1
•
Ó Société Internationale de Chirurgie 2015
Abstract There is no consensus on the ideal location for mesh placement in open ventral hernia repair (OVHR).
We aim to identify the mesh location associated with the lowest rate of recurrence following OVHR using a
systematic review and meta-analysis. A search was performed for studies comparing at least two of four locations for
mesh placement during OVHR (onlay, inlay, sublay, and underlay). Outcomes assessed were hernia recurrence and
surgical site infection (SSI). Pairwise meta-analysis was performed to compare all direct treatment of mesh locations.
A multiple treatment meta-analysis was performed to compare all mesh locations in the Bayesian framework.
Sensitivity analyses were planned for the following: studies with a low risk of bias, incisional hernias, by hernia size,
and by mesh type (synthetic or biologic). Twenty-one studies were identified (n = 5,891). Sublay placement of mesh
was associated with the lowest risk for recurrence [OR 0.218 (95 % CI 0.06–0.47)] and was the best of the four
treatment modalities assessed [Prob (best) = 94.2 %]. Sublay was also associated with the lowest risk for SSI [OR
0.449 (95 % CI 0.12–1.16)] and was the best of the 4 treatment modalities assessed [Prob (best) = 77.3 %]. When
only assessing studies at low risk of bias, of incisional hernias, and using synthetic mesh, the probability that sublay
had the lowest rate of recurrence and SSI was high. Sublay mesh location has lower complication rates than other
mesh locations. While additional randomized controlled trials are needed to validate these findings, this network
meta-analysis suggests the probability of sublay being the best location for mesh placement is high.
Introduction
& Julie L. Holihan
[email protected]; [email protected]
Duyen H. Nguyen
[email protected]
Mylan T. Nguyen
[email protected]
Jiandi Mo
[email protected]
Lillian S. Kao
[email protected]
Mike K. Liang
[email protected]
1
Department of General Surgery, University of Texas Health
Science Center, 6431 Fannin St, Houston, TX 77030, USA
Repair of ventral hernias with mesh as opposed to suture
has substantially improved long-term outcomes and is
accepted as the standard of care [1–3]. However, many
studies demonstrate an increased risk for wound complications with mesh placement including infections, seromas,
and mesh erosions [3, 4]. The risks of these complications
are affected by where the mesh is placed. For example,
mesh exposed to intra-abdominal contents potentially
increases the risks of adhesions, bowel obstruction, and
fistula formation [5, 6]. While repair of ventral hernias with
mesh is considered routine, there is no consensus on the
best location to place the mesh.
For laparoscopic ventral hernia repair, the mesh is routinely placed in the intra-peritoneal position. However, for
123
World J Surg
load-bearing tissue in-growth from two directions. Underlay repair or intra-peritoneal repair was popularized with
the advent of laparoscopy. Placing mesh in this location
can be technically cumbersome requiring sutures to be
placed closely to prevent intra-abdominal contents from
sliding between the mesh and anterior abdominal wall.
However, while protected from superficial wound complications, the mesh is exposed to intra-peritoneal contents.
Meshes placed in this location must have an anti-adhesive
barrier or anti-adhesive properties on the peritoneal side [6,
8–13].
In this study, we conducted a systematic review and
performed a multiple treatment meta-analysis to identify
the best location for mesh placement with open ventral
hernia repair. We hypothesized that in open ventral hernia
repair, sublay mesh placement has the lowest recurrence
rate as compared to inlay, onlay, or underlay placement.
Fig. 1 Mesh location. a onlay repair, b inlay repair, c sublay repair,
d underlay; Key blue mesh red muscle black fascia gray hernia sac
open surgery, there are numerous options for mesh placement (Fig. 1). Onlay repair places the mesh on the anterior
fascia, which typically involves dissection of flaps and
primary closure of the fascia below the mesh. Inlay repair
places the mesh in the hernia defect and secures the mesh
circumferentially to the edges of the fascia. Sublay repair
refers to retro-rectus or preperitoneal mesh placement. It is
also commonly referred to as a Rives-Stoppa or retromuscular repair. Finally, underlay repair is when mesh is
placed in the intra-peritoneal position and secured to the
anterior abdominal wall.
Each mesh location has its theoretical risks and benefits.
With onlay repair, skin flaps must be created, which
increases the risk of wound complications and mesh
infection [7]. However, onlay repair is technically easy to
perform. In addition, for large complex hernias, this space
is often already dissected with excision of the hernia sac or
with myo-fascial release (i.e., anterior component separation). Inlay repair is technically easy. However, not only is
the mesh often exposed to the intra-peritoneal contents, it is
also vulnerable to superficial wound complications. Lack
of overlap precludes mesh-tissue integration and theoretically increases the risk of recurrence. Sublay repair is often
considered more challenging and complex to perform.
Dissection of this plane can risk damaging the muscle,
blood supply, and nerves to the rectus abdominus. In
addition, this mesh location may not be appropriate for offmidline defects. However, this space potentially protects
the mesh from both superficial wound complications and
intra-peritoneal contents. In addition, it also allows for
123
Methods
Search strategy
This systematic review and meta-analysis was registered
with PROSPERO, registration number CRD42015019722.
In accordance with preferred reporting items for systematic
reviews and meta-analyses (PRISMA), a search of
PubMed, Cochrane Central Register of Controlled Trials,
and Embase was performed to identify the articles for this
review [14]. Clinicaltrials.gov was also searched for
ongoing trials. The search included publications from
January 1990 through April 2015. The search strategies
were (ventral or incisional) AND hernia AND (onlay OR
inlay OR retrorectus OR retro-rectus OR retromuscular OR
retro-muscular OR preperitoneal OR Rives Stoppa OR
sublay OR underlay OR intraperitoneal OR intra-peritoneal
OR IPOM). Reference lists of selected articles were
reviewed for additional articles. Only the most recent
report from overlapping data was included. Studies were
included if they were human-related and comparative
studies evaluating two or more mesh locations in open
ventral hernia repair. Systematic reviews, meta-analyses,
letters, pediatric studies, laparoscopic repairs, techniques
manuscripts, and non-ventral hernia repairs (i.e., inguinal,
hiatal, etc.) were excluded. Non-comparative studies and
reports of less than 10 cases were also excluded.
Evaluation of articles
The methodological index of nonrandomized studies
(MINORS) was used to evaluate methodological quality
and potential bias of the articles selected for this review
[15]. The first eight of the twelve items evaluated non-
World J Surg
comparative and comparative studies. The last four items
applied only to comparative studies. Items were scored on
a 3-point scale of 0 (not reported), 1 (reported but inadequate), or 2 (reported and adequate). The global ideal score
is 24 for comparative studies. Two independent reviewers
scored the articles based on the criteria listed by the
assessment. Any disagreement or discrepancies were
resolved by consensus among the authors. Only studies
with MINORS C10 were included in the final analysis
[16].
Primary and secondary outcomes
The primary outcome measure was recurrence rate of
ventral hernia repairs. Secondary outcome was surgical site
infection (SSI).
Data extraction
The selected articles were reviewed and the primary and
secondary outcome data were extracted. Two reviewers
independently extracted the data. Study designs, number of
patients, patients’ age, patients’ BMI, percent of incisional
hernias included, mesh location, mesh type, mesh overlap,
mesh fixation method, and follow-up period were recorded.
Outcome measures recorded were recurrence, SSI, mesh
infection, and mesh explantation.
Statistical analysis
Pairwise meta-analysis was performed to compare all
direct treatment of mesh locations using fixed-effects and
random-effects models. A multiple treatment meta-analysis
was adapted from the guidelines developed by the International Society for Pharmacoeconomics and outcomes
research task force to compare the four different locations
for mesh placement in open ventral hernia repair [17, 18].
Unlike a traditional meta-analysis which only considers
studies with direct comparisons, a multiple treatment metaanalysis, or network meta-analysis, considers studies with
multiple competing treatments and combines direct and
indirect evidence to determine the most effective treatment
[17]. A network of treatment was created to map the
relationship between available studies. Heterogeneity was
assessed using a Mantel–Haenszel method that provided
pooled odds ratios and associated confidence intervals for
both fixed effect and random effect. It also produces test
statistics for heterogeneity utilizing the Higgins I2, and
reports the results as a p value. For a p value \0.05
(I2 [ 50 %) significant statistical heterogeneity exists and
the results of the random-effects model reported; otherwise
the results fixed-effects model was reported [17, 18]. A
multiple treatment meta-analysis was performed using
onlay as the reference treatment to compare all mesh
locations by estimating the fixed-effects and random-effects models in the Bayesian framework. The Bayesian
approach estimates the odds ratio and the 95 % credible
interval and the probability that each location is the best
using Markov chain Monte Carlo technique [18, 19]. The
credible interval is the Bayesian probability interval that is
similar in use to a frequentist confidence interval; however
the credible interval also incorporates information from the
prior distribution into the estimate while the confidence
intervals are solely based on the data. Model fit was
assessed by the deviance information criterion and was
used to select whether to use the fixed-effects or randomeffects model within the Bayesian framework [20].
Inconsistency of the network, also known as incoherence,
was assessed for each model and demonstrated that the
indirect evidence was not inconsistent with the direct evidence on the same treatment comparisons beyond chance
[17, 18]. Statistical analysis was performed within the
statistical software R version 2.15.0 [20].
Sensitivity analysis
Sensitivity analysis were determined prior to analyses and
utilized as a second assessment of heterogeneity. The following sensitivity analyses were planned: studies with a
low likelihood of bias based upon MINORS score ([17),
studies of incisional hernias only, based on size (large
median defect size C 100 cm2 or width C 10 cm; medium: median size between 36 and 100 cm2 or width
between 6 and 10 cm; and small: median defect
size B 36 cm2 or width B 6 cm), and studies reporting the
use of only synthetic mesh and use of only biological mesh
were evaluated separately [16, 21].
Results
The search resulted in a total of 957 titles and abstracts
(Fig. 2). After de-duplication, 472 titles and abstracts were
reviewed. An additional 431 articles were excluded for
reasons including pediatric only or non-human studies,
reviews, meta-analyses, letters, non-ventral hernia related,
studies not related to open ventral hernia repair, noncomparative, and case reports of fewer than 10 patients.
After screening, 41 full-text articles were reviewed and 19
articles were excluded. A total of 21 articles were subjected
to MINORS for assessment and included in the systematic
review. The studies included randomized controlled trials
(n = 3), retrospective chart reviews (n = 13), and
prospective cohort studies (n = 5) (Tables 1, 2).
123
World J Surg
Fig. 2 Search flow-chart
Quality of and heterogeneity between the studies
The median (range) MINORS score was 15 (10–21). The
studies suffered the greatest weakness in criteria: (5)
unbiased assessment of study endpoint, (8) calculation of
sample size, (7) loss to follow-up of less than 5 %, (6)
follow-up period appropriate to the aim of the study, (11)
baseline equivalence of groups, and (12) adequate statistical analysis. The three randomized studies had the highest
minor scores (Table 2) [22–24]. Of the remaining studies,
there were 13 retrospective chart reviews and 5 prospective
observational studies of variable quality.
While the studies evaluated patients of similar age and
BMI and treated patients with similar amounts of mesh
overlap and mesh fixation, there was heterogeneity in the
type of hernia included (incisional vs. primary), the type of
mesh used (synthetic vs. biologic), as well as hernia defect
size (Table 1). Recurrence reporting was based upon
clinical exam in most studies. Follow-up duration ranged
from mean 5–60 months (Table 1). There was sufficient
123
clinical homogeneity between studies to perform a quantitative synthesis of results.
Pooled results and analysis
Twenty-one studies reported on recurrence rates for ventral
hernia repair (Fig. 3). The pooled recurrence rate (range n)
was 16.5 % (0–36 %, 1267) for onlay, 30.2 % (0–80.0 %,
159) for inlay, 7.0 % (0–48.0 %, 1719) for sublay, and
14.7 % (0–56.0 %, 2746) for underlay. Ten studies reported on SSI. The pooled SSI rate (range n) was 16.9 %
(0–33.0 %, 302) for onlay, 31.3 % (8.7–37.1 %, 112) for
inlay, 3.7 % (0–21.0 %, 702) for sublay, and 16.7 %
(0–34.3 %, 402) for underlay.
Onlay mesh placement
Seventeen of 21 studies included onlay mesh placement. In
pairwise meta-analysis, onlay was associated with higher
odds for recurrence and SSI compared to sublay and
2012
2012
2012
2011
2011
2011
2009
2009
2010
2010
2009
Kumar et al.
Li et al.
Rosen et al.
Forte et al.
Prasad et al.
Scheuerlein
et al.
Diaz et al.
Lin et al.
Abdollahi et al.
Berrevoet et al.
Gleysteen et al.
2013
2010
Venclauskas
et al.
2012
2010
Weber et al.
Helgstrand et al.
2005
Afifi et al.
Hope et al.
Year
Author
-
Sublay:
279.6 ± 163.1 cm2
12
Sublay: 55
–
Underlay: 29.4
Underlay: 48.1
Onlay: 57 (mean)
Sublay: 28.2
-
35.26
30.7 (9.6 SD)
Sublay: 54.8
52 (2–85)
55 (23–84)
52.2 (15.2)
[17–87]
Underlay: 59
[median, range]
Sublay: 62[3-116]
Onlay: 66[1–204]
Underlay: 31[24–42]
[median, range]
Sublay: 48.8[42–56]
98 (48–174)
5
10.3
20 [4–35] (median, range)
Sublay: 9(3-20) cm
Onlay: 10 (3-30) cm
Sublay: 15
(30)
Onlay: 29
(39)
0
Underlay: 2.3 cm2
Sublay: 2.6 cm2
Biological
Synthetic
Synthetic
Synthetic
–
Underlay: C 2.5 cm
Sublay: C 3 cm
-
-
140 (100)
354 (100)
-
C25 cm2
Biological
3–5 cm
Onlay &
sublay: C 3 cm
208 (100)
Synthetic
201 cm2
29 (100)
C5 cm
3–5 cm
-
-
-
5 cm
5 cm
C5 cm
C4 cm
Mesh overlap
Underlay: C5 cm
Onlay & sublay: 57
(3–314) cm2
-
Synthetic
Synthetic,
biological
246 (100)
195 (70)
Biological
-
-
Synthetic,
biological
-
synthetic
synthetic
Synthetic
Mesh type
Underlay: 37 (3–158)
cm2
Underlay: 29.9 ± 22.0
cm2
Underlay:
30.9 ± 2.6
Underlay:
49.2 ± 11.4
Underlay: 22.5 ± 11.9
Sublay: 30.8 ± 24.4
cm2
12
Sublay: 22.7 ± 13.4
Sublay:
30.2 ± 3.0
Sublay:
51.1 ± 11.1
Underlay:
31.1 ± 4.8
Underlay:
183.2 ± 125 cm2
Underlay: 50.3 ± 12.9
cm2
Underlay: 39 (2–140)
(median, range)
49 (100)
92 (69)
Underlay:
31.2 ± 0.72
-
Onlay: 56.4 ± 9.4 cm2
60
Underlay:
58.2 ± 13.4
Onlay & Sublay:
57
593 (18)
49 (84)
Onlay: 77(1–143)
Sublay:
30.8 ± 5.7
62 (34–84 range)
7 cm (median)
-
107 (100)
-
13
Sublay: 110 ± 83.8
Onlay: 114.5 ± 90.9
-
41 (100)
[10 cm diameter
[5 cm2
Incisional,
n (%)
Size of defect
Onlay:
32.3 ± 0.69
-
21 (10–35) (median, IQR)
12
Onlay: 30.5
Sublay: 28
60
30 (median)
–
-
F/Ua (months)
BMI (kg/m2)
Sublay:
58.6 ± 13.5
Underlay:
59.4 ± 1.2
Onlay: 57.6 ± 1.1
-
60 (24–89) [mean
(range)]
60 (49–69 IQR)
Sublay: 53 ± 11.6
Onlay:
56.9 ± 11.5
–
Underlay:
52.15 ± 10.59
Onlay: 52.8 ± 8.8
Age
Table 1 Characteristics of mesh repairs
Sutures
sutures
-
-
-
Underlay: sutures ? clips
Onlay & sublay: sutures
Suture, tacks
Sutures
Sutures
-
-
-
sutures, tacks, glue, clips
sutures
sutures
Onlay: suture underlay:
suture, staplers
Mesh fixation
World J Surg
123
Underlay: suture
Inlay mesh placement
Sutures
6–8 cm
50 (100)
[5 cm diameter
Synthetic
Inlay: sutures
Onlay: staples
C2 cm
53 (100)
[10 cm diameter
Synthetic
Underlay:
sutures ? staplers
3–5 cm
Synthetic
Sublay: 32
(64)
Sublay: 55.88 cm2
Underlay:
36 (72)
–
–
–
296 (100)
[3 cm diameter
Underlay: 93.96 cm2
–
–
Synthetic
–
95 (100)
Incisional,
n (%)
Size of defect
Mesh overlap
Sublay: sutures
underlay but lower odds compared to inlay (Table 3). On
multiple treatment meta-analysis, onlay was one of the
worst mesh placement options with a very low probability
(\0.001) of being the best treatment (Table 4).
Mesh type
Mesh fixation
World J Surg
Of the 21 studies, six reported on inlay. In pairwise metaanalysis, inlay was associated with higher odds for recurrence and SSI than onlay, sublay, and underlay (Table 3).
On multiple treatment meta-analysis, inlay had higher odds
for recurrence compared to onlay [3.946 (0.487–13.256)]
(Table 4). Inlay also had a higher odds for SSI compared to
onlay [1.113 (0.088–3.833)] (Table 3). Inlay was one of the
worst mesh placement options with a very low probability
(\0.001) of being the best treatment (Table 4).
Sublay mesh placement
Fourteen of the 21 studies reported on sublay. In pairwise
meta-analysis, sublay was associated with lower risk of
recurrence and SSI compared to onlay, inlay, and underlay
(Table 3). On multiple treatment meta-analysis, sublay had
a lower risk for recurrence [0.218 (0.061–0.465)] and SSI
[0.449 (0.118–1.155)] compared to onlay (Table 4). Sublay
was ranked the best mesh placement option with a high
probability of being the best treatment. Sublay had a
94.2 % probability of having the lowest odds of recurrence
and 77.3 % probability of having the lowest odds for SSI.
a
reported as mean unless otherwise stated
6–72 (range)
–
57.8 (33–79)
2004
Kingsnorth
et al.
Underlay: 33.9
Inlay: 33.2
Onlay: 19.4
–
60.4 (28–94)
2004
de Vries
Reilingh
et al.
Underlay: 19.6
Sublay: 21.9
29
–
61 (22–94)
55.25 (30–83)
2006
2006
Israelsson et al.
–
123
Lomanto et al.
–
–
2009
Demetrashvili
et al.
–
F/Ua (months)
Year
Author
Table 1 continued
Age
BMI (kg/m2)
Underlay mesh placement
Fifteen of the 21 studies included underlay. In pairwise
meta-analysis, underlay was associated with lower risk for
recurrence and SSI than onlay and inlay but had higher risk
compared to sublay (Table 3). On multiple treatment metaanalysis, underlay had a lower risk for recurrence [0.59
(0.069–1.504)] and SSI [0.878 (0.291–1.985)] compared to
onlay (Table 4). While underlay had the second highest
probability of being the best treatment, it remained a distant second. The probability of underlay having the lowest
odds of recurrence was only 5.8 % and for SSI was 11.4 %.
Sensitivity analysis
Sensitivity analysis was performed for studies at low risk
of bias (MINORS \ 17), studies of only incisional hernias,
and studies using only synthetic mesh. For all of these, the
probability that sublay repair was associated with the
lowest risk of recurrence and SSI remained high (Table 5).
Onlay, inlay, and underlay remained very unlikely to have
the best outcomes. It was not feasible to perform a sensitivity analysis of ventral hernia repairs stratified by hernia
2010
2010
2013
2012
2012
2012
2012
2011
2011
2011
2009
2009
2010
Weber et al.
Venclauskas et al.
Helgstrand et al.
Hope et al.
Kumar et al.
Li et al.
Rosen et al.
Forte et al.
Prasad et al.
Scheuerlein et al.
Diaz et al.
Lin et al.
Abdollahi et al.
2010
2005
Afifi et al.
Berrevoet et al.
Year
Author
Prospective
Retrospective
Retrospective
Retrospective
Retrospective
Retrospective
Retrospective
Prospective
Retrospective
Prospective
Retrospective
Prospective
PRCT
PRCT
PRCT
Study type
Table 2 Study characteristics and outcomes
19
11
10
15
13
16
12
18
14
13
12
17
21
20
20
MINORS
116
354
140
208
29
279
246
49
134
63
58
2798
107
550
41
N
Sublay: 2 (3.6)
Underlay: 5 (8.3)
Underlay (60)
Underlay: 0
Sublay (56)
Underlay (9)
Onlay: 2 (6.1)
Sublay: 2 (0.6)
Onlay (33)
Underlay: 24 (23.3)
Underlay (103)
Sublay (312)
Inlay: 12 (35.3)
Inlay (34)
Underlay: 17 (18.7)
Underlay (91)
Onlay: 0
Inlay: 18 (20.2)
Inlay (89)
Onlay (3)
Onlay: 4 (14.3)
Underlay: 1 (6.7)
Underlay (15)
Onlay (28)
Onlay: 7 (70.0)
Sublay: 0
Onlay (10)
Underlay: 7 (3.3)
Underlay (211)
Sublay (4)
Sublay: 2 (2.9)
Underlay: 0
Underlay (30)
Sublay (68)
Onlay: 3 (33.3)
Sublay: 1 (0.48)
Onlay (9)
Underlay: 7 (26.9)
Underlay (26)
Sublay (207)
Sublay: 2 (8.7)
Sublay (23)
Onlay: 12 (17.9)
Underlay: 7 (10.4)
Onlay (67)
Underlay (67)
Onlay: 4 (8.9)
Underlay: 1 (5.6)
Underlay (18)
Underlay: 5 (55.6)
Underlay (9)
Onlay (45)
Onlay: 14 (35.9)
Inlay: 8 (80.0)
Onlay (39)
Underlay: 328 (16.2)
Underlay (2021)
Inlay (10)
Onlay: 73 (16.1)
Sublay: 39 (12.1)
Onlay (454)
Sublay (323)
Onlay: 6 (10.5)
Sublay: 1 (2.0)
Onlay (57)
Sublay: 53 (14.4)
Sublay (369)
Sublay (50)
Onlay: 22 (12.2)
Onlay (181)
Onlay: 6 (27.3)
Underlay: 0
Onlay (22)
Recurrence N (%)
Underlay (19)
Mesh locations
Underlay: 0
Sublay: 0
Underlay: 0
Sublay: 7 (2.2)
Onlay: 1 (3.0)
–
Underlay: 31 (34.1)
Inlay: 33 (37.1)
Onlay: 8 (28.6)
Underlay: 0
Sublay: 0
Onlay: 0
–
Underlay: 0
Sublay: 9 (4.3)
Onlay: 3 (33.3)
Underlay: 8 (30.8)
Sublay: 5 (21.7)
Underlay: 23 (34.3)
Onlay: 20 (29.9)
Underlay: 2 (11.1)
Onlay: 6 (13.3)
–
–
–
–
Underlay: 1 (5.3)
Onlay: 1 (4.5)
SSI N (%)
–
–
–
–
Underlay: 0
Sublay: 0
Onlay: 0
Underlay: 0
Sublay: 0
Underlay: 0
Sublay:2 (0.9)
Onlay: 2 (22.2)
–
–
–
–
–
–
–
–
Mesh infection N (%)
–
Underlay: 0
Sublay: 1 (0.3)
Onlay: 1 (3.0)
–
–
–
–
–
–
Underlay: 3 (4.5)
Onlay: 5 (7.5)
–
–
–
–
–
Underlay: 0
Onlay: 1 (4.5)
Mesh explantation N (%)
World J Surg
123
–
–
–
P value
Heterogeneity
P value*
Sublay: 1 (3.0)
Inlay: 0
Onlay (reference) vs.
Inlay
2.189
1.51–3.17
\0.001
0.734
Sublay
Underlay
0.379
0.873
0.29–0.48
0.73–1.05
0.999
0.993
\0.001
0.030
Inlay (reference) vs.
Sublay
0.173
0.12–0.26
0.999
0.649
Underlay
0.399
0.28–0.57
0.999
0.295
1.86–2.85
\0.001
0.506
Sublay (33)
Sublay (reference) vs.
Inlay (1)
Underlay: 0
95 % CI
–
OR
Onlay: 2 (12.5)
Underlay: 2 (11.8)
Recurrence
Onlay (16)
Underlay (17)
Inlay: 2 (8.7)
Inlay: 10 (43.5)
Inlay (23)
Table 3 Summary of all pairwise meta-analysis
50
Underlay:0)
–
Underlay: 1 (2.0)
Onlay: 3 (23.1)
Onlay: 3 (23.1)
Underlay: 1 (2.0)
Underlay (50)
Onlay (13)
53
Sublay: 2 (4.0)
–
Sublay: 5 (10.0)
Sublay: 5 (10.0)
Sublay (123)
Sublay (50)
100
Sublay: 9 (7.3)
–
Onlay: 33 (19.3)
Inlay: 0
Inlay (2)
Onlay (171)
296
Onlay: 3 (6.8)
–
–
–
Sublay (51)
Onlay (44)
95
Sublay: 1 (2.0)
–
–
–
–
Onlay: 9 (12.0)
Sublay: 2 (4.0)
Sublay: 2 (4.0)
Onlay: 15 (20.0)
Onlay (75)
125
Sublay (50)
Mesh infection N (%)
Fig. 3 Network analysis of studies included in multiple treatments
meta-analysis. Width of line represents number of studies. Size of
node represents sample size
Underlay
2.299
SSI
Onlay (reference) vs.
Inlay
6.730
4.21–10.75
\0.001
0.162
Sublay
0.366
0.23–0.59
0.999
0.243
Underlay
0.592
0.41–0.86
0.999
0.219
0.429
1.02–2.55
0.144
0.682
Retrospective
Prospective
Retrospective
Retrospective
2006
2006
2004
2004
Lomanto et al.
de Vries Reilingh et al.
Kingsnorth et al.
Discussion
123
Israelsson et al.
Retrospective
2009
Sublay (reference) vs.
Underlay
1.618
Demetrashvili et al.
17
\0.001
0.999
Retrospective
16
0.999
0.06–0.14
2009
19
0.03–0.09
0.088
Gleysteen et al.
14
0.054
Underlay
Study type
15
Sublay
Year
19
Inlay (reference) vs.
size or hernias repair only with biologic mesh only due to a
low number of direct comparisons.
Author
Table 2 continued
MINORS
N
Mesh locations
Recurrence N (%)
SSI N (%)
Mesh explantation N (%)
World J Surg
* Random-effects model reported for pairwise analysis with heterogeneity P value \ 0.05
In this network meta-analysis, sublay repair had the highest
probability of having the lowest rate of hernia recurrence
and SSI. Underlay repair had the second highest probability
World J Surg
Table 4 Multiple treatments meta-analysis
OR
95 % credible interval
Probability of being the
best among four treatments
Rank
Inlay
3.946
0.49–13.26
\0.001
3, 4
\0.001
Sublay
0.218
3, 4
0.06–0.47
0.942
Underlay
1
0.590
0.07–1.50
0.058
2
1.113
0.09–3.83
\0.001
0.113
4
3
Sublay
0.449
0.12–1.16
0.773
1
Underlay
0.878
0.29–1.99
0.114
2
Recurrencea
Onlay (reference)
SSIb
Onlay (Reference)
Inlay
a
DIC 108.3
b
DIC 59.6
Table 5 Sensitivity analysis
Probability of being the best among four treatments
All trials
Low risk of bias
Synthetic mesh only
Incisional hernias only
Recurrence
Onlay (reference)
\0.001
0.045
0.005
\0.001
Inlay
Sublay
\0.001
0.942
\0.001
0.839
0.001
0.744
\0.001
0.833
Underlay
0.058
0.116
0.25
0.167
0.001
SSI
Onlay (reference)
\0.001
0.054
\0.001
Inlay
\0.001
\0.001
0.106
0.019
Sublay
0.942
0.946
0.612
0.734
Underlay
0.058
\0.001
0.282
0.246
of having the lowest rate of hernia recurrence and SSI
while onlay and inlay had poor results. While there were
few high-quality studies, the results when only evaluating
studies at low risk for bias did not change. Furthermore,
these results are in line with other systematic reviews
addressing mesh location [5, 25, 26].
When considering the best location for mesh placement,
a number of features are important to consider. First, meshtissue integration may reduce long-term recurrence, with
theoretically improved rates with greater mesh-tissue
overlap [27]. Second, wound complications increase the
risk of recurrence. Thus, techniques that avoid of the
development of devascularizing flaps may be preferred
[15]. Third, the ideal mesh placement should have tissue
coverage to minimize exposure to superficial SSIs as well
as intra-peritoneal contents. Last, technical ease may affect
surgeon choice of procedure as well as risks for postoperative complications.
Sublay repair allows for tissue integration from two
load-bearing tissues from both sides: posterior rectus
sheath and the anterior myo-fascial complex. In addition,
sublay mesh placement protects the mesh from exposure
from superficial wound complications, intra-abdominal
adhesions, and contamination. Creation of devascularizing
skin flaps is avoided. Alternatively, inlay repair does not
allow for any tissue integration, any wound complication
can potentially expose the mesh, and the mesh is exposed
to intra-abdominal contents. While onlay allows for tissue
ingrowth from two directions, the skin flaps are not loadbearing. Mesh placed in the onlay location is vulnerable
forcing the surgeon to create devascularizing skin flaps and
leaving the mesh susceptible to superficial wound complications. Underlay has the advantages of protecting the
mesh from superficial wound complications and avoiding
development of skin flaps. However, this position requires
mesh to be placed intra-abdominally which leaves the mesh
123
World J Surg
susceptible to organ/space infections. In addition, the mesh
must grow into peritoneum, the body’s natural anti-adhesion barrier. This may explain why the recurrence rates
with this repair are higher despite the potential for
increased mesh overlap.
Sublay repair is not without its own set of challenges.
The surgical approach can be perceived as more technically
challenging than other techniques, particularly in patients
who have had prior abdominal surgeries. Patients with
previous stomas, gynecologic procedures, or ventral hernia
repairs may have a damaged posterior sheath or damaged
rectus muscle. This may leave this space difficult to
develop, limited in size, or non-existent in rare circumstances. In addition, risks of damaging the blood supply,
muscle, or lateral penetrating nerves pose technical concerns. Furthermore, the semilunar lines limit the lateral
extent of the sublay repair and potentially limiting the
amount of mesh overlap. Off-midline incisions may not be
ideal hernias to approach with this technique. While those
new to sublay repair may find it technically daunting,
anecdotal experience has demonstrated ease in learning and
adopting this approach; however studies to evaluate the
learning curve are needed.
Given the limited data available but the enormous
clinical experience available, most surgeons no longer
consider the inlay repair appropriate for ventral hernia
repair. All studies demonstrate that inlay repair has the
highest recurrence and SSI rates. Increasingly, onlay repair
is losing favor with sublay and underlay repairs growing in
popularity. Most studies with low risk of bias indicated that
the recurrence rate of onlay repair is inferior to sublay and
underlay. Between the two studies that examined the SSI
rate of onlay repair, both studies demonstrate that onlay
repair yield inferior results compared to sublay and
underlay. This network analysis suggests that sublay repair
may have the best results with underlay repair as a distant
second best. A randomized controlled trial to compare
sublay and underlay would require a sample size (given
a = 0.05, b = 0.20) of 562 to show a difference in
recurrence (7.0 vs. 14.7 %) and 198 to show a difference in
SSI (3.7 vs. 16.7 %).
Network meta-analyses allow for utilization of a larger
amount of evidence, estimation of the relative effectiveness
among multiple interventions, and rank ordering of the
interventions [28]. In contrast to pairwise meta-analysis,
network meta-analysis can provide estimates of relative
efficacy between all interventions. For many comparisons,
the network meta-analysis may yield more reliable and
definitive results. However, there are a number of potential
limitations. Inadequate reporting of findings and inadequate evaluation of the required assumptions may impede
confidence in the findings and conclusions. We followed
the grading of recommendations assessment, development,
123
and evaluation (GRADE) Working Group guidelines to
allow readers to determine confidence in the results from
this network meta-analysis [20, 29].
This network meta-analysis is limited due to the quality
of literature that is available. The majority of literature
pertaining to mesh placement is of poor methodological
quality. In particular, we noted that blinded outcomes
assessment and improved follow-up is needed. However,
using sensitivity analysis and only assessing studies with a
low risk of bias, the results did not change. Furthermore,
there was some heterogeneity in follow-up duration
between studies, with many studies reporting follow-up as
little as 1 month. Outcomes from studies with great differences in follow-up may not be comparable; however, the
majority of the studies had a mean follow-up of at least
1 year. Hernia size could also affect outcomes, and a
sensitivity analysis by hernia size was planned. However,
this analysis was unable to be performed due to a low
number of direct comparisons between studies with similar
sizes. In addition, other differences in repair technique may
have accounted for some of the differences in outcomes,
including mesh type and method for mesh fixation. Future
studies should compare mesh location in hernias of similar
sizes with similar repair techniques to validate these
results.
Conclusion
Sublay placement of mesh demonstrates improved outcomes compared to onlay, inlay, and underlay repairs;
however the quality and level of data remains poor. Randomized controlled trials with longer follow-up may be
warranted to validate the results of this network metaanalysis; however, this network meta-analysis demonstrates a high probability that sublay repair is associated
with the fewest SSIs and hernia recurrences.
Disclosures Mike K. Liang this work was supported by the Center
for Clinical and Translational Sciences, which is funded by National
Institutes of Health Clinical and Translational Award UL1 TR000371
and KL2 TR000370 from the National Center for Advancing Translational Sciences. The National Center for Research Resources or the
National Institutes of Health was not responsible for the design and
conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication.
References
1. Luijendijk RW et al (2000) A comparison of suture repair with
mesh repair for incisional hernia. N Engl J Med 343(6):392–398
World J Surg
2. Burger JW et al (2004) Long-term follow-up of a randomized
controlled trial of suture versus mesh repair of incisional hernia.
Ann Surg 240(4):578–583
3. Nguyen MT et al (2014) Comparison of outcomes of synthetic
mesh vs suture repair of elective primary ventral herniorrhaphy: a
systematic review and meta-analysis. JAMA Surg 149(5):
415–421
4. Berger RL et al (2014) Suture versus preperitoneal polypropylene
mesh for elective umbilical hernia repairs. J Surg Res 192(2):
426–431
5. Albino FP et al (2013) Does mesh location matter in abdominal wall
reconstruction? A systematic review of the literature and a summary of recommendations. Plast Reconstr Surg 132(5):1295–1304
6. Snyder CW et al (2011) Effect of mesh type and position on
subsequent abdominal operations after incisional hernia repair.
J Am Coll Surg 212(4):496–502
7. Berger RL et al (2013) Development and validation of a riskstratification score for surgical site occurrence and surgical site
infection after open ventral hernia repair. J Am Coll Surg
217(6):974–982
8. Novitsky YW et al (2007) Comparative evaluation of adhesion
formation, strength of ingrowth, and textile properties of prosthetic meshes after long-term intra-abdominal implantation in a
rabbit. J Surg Res 140(1):6–11
9. Robinson TN et al (2005) Major mesh-related complications
following hernia repair: events reported to the Food and Drug
Administration. Surg Endosc 19(12):1556–1560
10. Losanoff JE, Richman BW, Jones JW (2002) Entero-colocutaneous fistula: a late consequence of polypropylene mesh
abdominal wall repair: case report and review of the literature.
Hernia 6(3):144–147
11. Miller K, Junger W (1997) Ileocutaneous fistula formation following laparoscopic polypropylene mesh hernia repair. Surg
Endosc 11(7):772–773
12. Cuccurullo D et al (2013) Laparoscopic ventral incisional hernia
repair: evidence-based guidelines of the first Italian Consensus
Conference. Hernia 17(5):557–566
13. Silecchia G et al (2015) Laparoscopic ventral/incisional hernia
repair: updated guidelines from the EAES and EHS endorsed
consensus development conference. Surg Endosc 29:2463–2484
14. Moher D et al (2009) Preferred reporting items for systematic
reviews and meta-analyses: the PRISMA statement. J Clin Epidemiol 62(10):1006–1012
15. Slim K et al (2003) Methodological index for non-randomized
studies (minors): development and validation of a new instrument. ANZ J Surg 73(9):712–716
16. Kissane NA, Itani KM (2012) A decade of ventral incisional
hernia repairs with biologic acellular dermal matrix: what have
we learned? Plast Reconstr Surg 130(5 Suppl 2):194S–202S
17. Jansen JP et al (2011) Interpreting indirect treatment comparisons
and network meta-analysis for health-care decision making:
report of the ISPOR task force on indirect treatment comparisons
good research practices: part 1. Value Health 14(4):417–428
18. Hoaglin DC et al (2011) Conducting indirect-treatment-comparison and network-meta-analysis studies: report of the ISPOR task
force on indirect treatment comparisons good research practices:
part 2. Value Health 14(4):429–437
19. Gelman A, Rubin DB (1996) Markov chain Monte Carlo methods
in biostatistics. Stat Methods Med Res 5(4):339–355
20. Salanti G, Marinho V, Higgins JP (2009) A case study of multiple-treatments meta-analysis demonstrates that covariates
should be considered. J Clin Epidemiol 62(8):857–864
21. Vinuela EF et al (2012) Laparoscopic versus open distal gastrectomy for gastric cancer: a meta-analysis of randomized controlled trials and high-quality nonrandomized studies. Ann Surg
255(3):446–456
22. Afifi RY (2005) A prospective study between two different
techniques for the repair of a large recurrent ventral hernia: a
double mesh intraperitoneal repair versus onlay mesh repair.
Hernia 9(4):310–315
23. Weber G, Baracs J, Horvath OP (2010) [‘‘Onlay’’ mesh provides
significantly better results than ‘‘sublay’’ reconstruction.
Prospective randomized multicenter study of abdominal wall
reconstruction with sutures only, or with surgical mesh-results of
a five-years follow-up]. Magy Seb 63(5):302–311
24. Venclauskas L, Maleckas A, Kiudelis M (2010) One-year followup after incisional hernia treatment: results of a prospective
randomized study. Hernia 14(6):575–582
25. Deerenberg EB et al (2015) A systematic review of the surgical
treatment of large incisional hernia. Hernia 19(1):89–101
26. Eriksson A, Rosenberg J, Bisgaard T (2014) Surgical treatment
for giant incisional hernia: a qualitative systematic review. Hernia
18(1):31–38
27. Guerin G, Turquier F (2013) Impact of the defect size, the mesh
overlap and the fixation depth on ventral hernia repairs: a combined
experimental and numerical approach. Hernia 17(5):647–655
28. Mills EJ, Thorlund K, Ioannidis JP (2013) Demystifying trial
networks and network meta-analysis. BMJ 346:f2914
29. Brozek JL et al (2009) Grading quality of evidence and strength
of recommendations in clinical practice guidelines. Part 1 of 3.
An overview of the GRADE approach and grading quality of
evidence about interventions. Allergy 64(5):669–677
123