The American Journal of Surgery (2013) 206, 400-409

Review

Up and down or side to side? A systematic reviewand meta-analysis examining the impact of incisionon outcomes after abdominal surgery

Kai A. Bickenbach, M.D.a,*, Paul J. Karanicolas, M.D., Ph.D.b, John B. Ammori, M.D.b,Shiva Jayaraman, M.D., M.E.S.C.b, Jordan M. Winter, M.D.b, Ryan C. Fields, M.D.b,Anand Govindarajan, M.D., M.S.C.b, Itzhak Nir, M.D.b, Flavio G. Rocha, M.D.b,Murray F. Brennan, M.D.b

aUniversity of Medicine and Dentistry of New Jersey, New Jeb

rsey School of Medicine, 205 South Orange Avenue, G-1222,Newark, NJ 07103, USA; Department of Surgery, Memorial Sloan-Kettering Cancer Center, New York, NY, USA

KEYWORDS:Incision;Hernia;Midline incision;Transverse incision;Paramedian incision;Meta-analysis

The authors declare no conflicts of i

* Corresponding author. Tel.:11-973

E-mail address: bickenka@umdnj.ed

Manuscript received November 29, 2

19, 2012

0002-9610/$ - see front matter � 2013

http://dx.doi.org/10.1016/j.amjsurg.20

AbstractBACKGROUND: The aim of this study was to examine whether midline, paramedian, or transverse

incisions offer potential advantages for abdominal surgery.DATA SOURCES: We searched MEDLINE, Embase, Web of Science, and The Cochrane Central

Register of Controlled Trials from 1966 to 2009 for randomized controlled trials comparing incisionchoice.

METHODS: We systematically assessed trials for eligibility and validity and extracted data in dupli-cate. We pooled data using a random-effects model.

RESULTS: Twenty-four studies were included. Transverse incisions required less narcotics than mid-line incisions (weighted mean difference 5 23.4 mg morphine; 95% confidence interval [CI], 6.9 to39.9) and resulted in a smaller change in the forced expiratory volume in 1 second on postoperativeday 1 (weighted mean difference 5 26.94%; 95% CI, 210.74 to 23.13). Midline incisions resultedin higher hernia rates compared with both transverse incisions (relative risk 5 1.77; 95% CI, 1.09 to2.87) and paramedian incisions (relative risk 5 3.41; 95% CI, 1.02 to 11.45).

CONCLUSIONS: Both transverse and paramedian incisions are associated with a lower hernia ratethan midline incisions and should be considered when exposure is equivalent.� 2013 Elsevier Inc. All rights reserved.

Surgical access to the abdomen can be achieved throughmultiple incision types. The type of abdominal incision canaffect multiple outcomes including operative time, inci-dence of complications, postoperative pain, pulmonary

nterest.

-972-3115; fax:11-973-972-3730.

u

011; revised manuscript October

Elsevier Inc. All rights reserved.

12.11.008

function, and length of stay.1 With more than 5 million lap-arotomies performed each year across the United States,defining the optimal incision is of great importance.2

Midline incisions are generally preferred by most sur-geons because of their ease, speed, and exposure.3 Thereare several potential disadvantages with these incisions.First, these incisions are made along the avascular lineaalba, which may impair wound healing. Second, the con-traction of the abdominal wall muscles pulls laterally on

Table 1 Inclusion criteria

Targetpopulation

Patients undergoing abdominal surgerythrough transperitoneal approach

Intervention Midline incisionTransverse (including oblique)Paramedian

Outcomemeasure

1 or more of:PainNarcotic useWound complicationsRespiratory complicationsIncisional herniaQuality of lifeHospital length of stayReturn to workPainPostoperative complications

Methodologicalcriteria

Randomized controlled trialorQuasi-randomized controlled trial(patients allocated according to knowncharacteristics)

K.A. Bickenbach et al. Incisions for abdominal surgery 401

the incision, resulting in tension on the closure. Addition-ally, because the fibers of the aponeurosis cross the midlineobliquely, these fibers are usually cut perpendicularly by avertical incision. Transverse (including oblique) incisionsare well vascularized by the muscle bed upon which theyare situated, unlike the avascular tissue of the midline inci-sion.3 Because of the direction of force from the contractionof the oblique muscles, less tension is placed on the woundcompared with the midline incision, which may result in alower hernia rate and less pain.4–6 Pain may also be re-duced because the nerve fibers run parallel to the incisionand are divided less. Finally, the fibers of the aponeurosisof the abdominal wall musculature are not cut perpendicu-larly because they are in a midline incision. Paramedianincisions split the rectus and offer advantages in that theyare situated on a vascular-rich muscle bed, and they alsooffer a theoretic shutter mechanism that may contribute toa lower rate of incisional hernias.7,8 Several studies, includ-ing randomized trials, have addressed outcomes after lapa-rotomy with different incisions.3 The majority of thesetrials compared midline with transverse incisions. Few trialshave directly compared paramedian with either midline ortransverse incisions. Before embarking on this study, ourhypothesis was that midline incisions would be associatedwith an increased risk of hernia but would be less painful.In this study, we sought to systematically review random-ized controlled trials that addressed the optimal incisionfor laparotomy. We specifically compared midline, trans-verse, and paramedian incisions. We report here that mid-line incisions are associated with an increased risk ofhernia compared with both transverse and paramedian inci-sions. Transverse incisions are associated less with a de-creased postoperative narcotic requirement compared withmidline incisions.

Methods

Study selection

We searched MEDLINE, Embase, Web of Science, andThe Cochrane Central Register of Controlled Trials from1966 to 2009 for potentially relevant randomized controlledtrials using the following search terms: transverse, midline,vertical, oblique, paramedian, abdominal, incision, laparot-omy, and hernia. We reviewed the reference lists of allarticles obtained to identify any other missed articles. Touncover unpublished potentially relevant trials, we searchedthe database of registered trials at www.clinicaltrials.gov.Two reviewers assessed all the abstracts from the articlesobtained to determine whether the articles met our inclu-sion criteria (Table 1). Full-text articles were obtainedfrom those abstracts selected and assessed by 2 reviewersto determine if they met our inclusion criteria. Agreementbetween reviewers was measured using the kappa statistic.The reviewers discussed any disagreements, and a consen-sus was obtained.

Validity assessment

Two investigators conducted a review of articles meetingthe eligibility criteria, rating the methodologic quality ofthe primary research. Concealment of allocation, patientblinding, clinician blinding, outcome assessor blinding, andcompleteness of follow-up were assessed to determine thevalidity of articles meeting our eligibility requirements. Anagreement between reviewers was measured using the kstatistic. The reviewers discussed any disagreements, and aconsensus was obtained.

Data extraction

Two investigators collected relevant information induplicate regarding the patient population, types of surgery,intervention, and outcomes from each selected article usinga computerized standardized data extraction form. Whenpossible, we contacted the investigators for any missingdata. When studies reported separate data for differenttypes of operations, we extracted those datasets separatelyrather than combining them.

Analysis

When studies reported the median and range, we esti-mated the mean and standard deviation using the methoddescribed by Hozo et al.9 Data were analyzed using Rev-Man version 5 (The Cochrane Collaboration, Copenhagen).For dichotomous outcomes, data were analyzed usinga random-effects model to calculate relative risks (RRs).For continuous outcomes with the same measure, arandom-effects model was used to provide weighted mean

Figure 1 A summary of the trial search and eligibility process.

402 The American Journal of Surgery, Vol 206, No 3, September 2013

differences (WMDs). A P value ,.05 was considered sig-nificant. Estimates of the extent of heterogeneity were ob-tained using the Breslow-Day test and the I2 statistic. Weexplored heterogeneity (I2 . 0) with subgroup andsensitivity analyses based on predetermined hypotheses(Table 2).

To compare the total amount of analgesics required forthe interventions, we converted the units reported tomilligrams of morphine. This required the assumptionsthat 1 mg ketobemidon equals 1 mg morphine and that1 mg meperidine equals 5 mg morphine. Additionally, forthose studies that reported data as dose per kg of bodyweight, doses were calculated by multiplying by the meanbody weight if reported and 70 kg if not reported.

Results

Characteristics and methodologic quality ofincluded studies

Our initial search yielded 1,141 abstracts (Fig. 1); 70 ofthese articles were selected for full-text review. Two ofthese were not available in English, 1 was only availablein abstract form, 1 was duplicated, and 42 articles did notmeet our inclusion criteria. This left 24 randomized con-trolled trials that were included in our analysis. The re-viewers achieved good agreement in the application ofthe eligibility criteria (k 5 .71). The characteristics of thetrials are described in Table 3.

The methodologic quality in these studies was generallypoor (Table 4). In only 9 trials was the method of conceal-ment reported as adequate. The majority of these trials werenot blinded, with only 2 trials blinding patients, medicalstaff, and outcome assessors. This was accomplished bythe use of dressings designed to cover the entire abdomen,which were not removed until discharge.

Table 2 Possible sources of heterogeneity

Population AgeSexASABMIComorbiditiesNature of abdominal operation

Intervention Method of fascial closureMethod of skin closureCointerventions

Outcomes Definition/measurement of wound infectionDefinition/measurement of narcotic useDefinition/measurement of painDefinition/measurement of quality of lifeDefinition/measurement of herniaDefinition/measurement of complication

Methodology Components of validity assessment

ASA 5 American Society of Anesthesiologists; BMI 5 body mass

index.

Comparisons: midline versus transverseincisions

Fourteen trials compared midline with transverse inci-sions (Table 5). Pain was assessed in 12 of these trials. Theconversion of narcotics to morphine equivalents was possi-ble in 6 trials. The combined data show that patients whounderwent transverse incisions received significantly lessnarcotics than those who had midline incisions (WMD 523.4 mg morphine per admission; 95% confidence interval[CI], 6.9 to 39.9). However, there was significant heteroge-neity between trials (I2 5 87%) (Fig. 2). The Garcia-Valdecasas trial contained both emergency and electivecases.10 If this trial is excluded, the combination of the re-maining trials reduces the statistical heterogeneity and stillfavors transverse incisions (WMD 5 34 mg morphine peradmission; 95% CI, 29 to 39.1; I2 5 0) (Fig. 3). Combiningdata from the 5 trials that reported visual analog scale(VAS) pain scores on postoperative day (POD) 1 showedno significant difference (WMD 5 .4; 95% CI, 2.9 to1.8). Heterogeneity was again significant (I2 5 98%). Sen-sitivity analyses based on elective or emergency surgery orthe regimen of pain control did not change the resultsmaterially.

The effects of incision choice on pulmonary functionwere assessed in 10 trials by various methods. The timingand method of evaluation were variable between differenttrials, making comparisons difficult. The pooling of data onvital capacity on POD 1 from 5 trials showed no significantdifference (WMD 5 4.7% change from baseline; 95% CI,210.5 to 1.1). Data from the 4 trials, which evaluatedforced expiratory volume in 1 second (FEV1) on POD 1,showed that midline incisions had a significantly greaterdeterioration of pulmonary function (WMD5 6.9% changefrom baseline; 95% CI, 210.7 to 23.1). Pulmonary com-plications were similar in the 10 trials that reported them(RR 5 1.11; 95% CI, .74 to 1.66).

Seven trials reported incisional hernia rates with amedian follow-up from 4 months to 4.4 years. Combining

Table 3 Study characteristics

Study

Totalsamplesize

Elective oremergency Operation Comparison Outcomes

Ali23 19 Elective Cholecystectomy Midline vs transverse Pulmonary complications,spirometry, PaO2

Armstrong24 60 Elective Cholecystectomy Midline vs transverse Pain, spirometry, LOSGarcia-Valdecasas10

129 Both Cholecystectomy Midline vs transverse Pain, pulmonary complications,spirometry

Halm19 123 Elective Cholecystectomy Midline vs transverse Incision length, OR time, pain,complications, LOS, cosmesis

Seenu25 181 Elective Cholecystectomy Midline vs transverse OR time, LOS, complications, lengthof incision

Becquemin17 26 Elective Aortoiliac surgery Midline vs transverse Pulmonary complications, spirometry,surgical convenience

Fassiadis11 37 Elective Aortoiliac surgery Midline vs transverse Hernia rateMassucci26 32 Elective Aortoiliac surgery Midline vs transverse Spirometry, PaO2Lacy27 50 Elective Aortoiliac surgery Midline vs transverse Pulmonary complications, OR time,

ICU stay, ventilatory time, painBrown28 28 Elective Right hemicolectomy Midline vs transverse Pain, LOS, complications, length of

incisionLindgren29 40 Elective Right hemicolectomy Midline vs transverse Pain, spirometryInaba30 395 Elective Gastrectomy Midline vs transverse Pain, blood loss, OR time

Complications, incision lengthProske18 94 Elective Gastric or pancreas Midline vs transverse OR time, pain, LOS, mortality,

spirometry, wound complications,incision length, cosmesis

Greenall31,32 572 Both Abdominal surgery Midline vs transverse Blood loss, pulmonary complications,dehiscence, wound infection, herniarate

Seiler33 191 Elective Abdominal surgery Midline vs transverse Pain, pulmonary complications, Woundinfection, dehiscence, hernia rate,LOS, spirometry

Stone34 561 Both Abdominal surgery Midline vs transverse Wound infection, OR timeSalonia35 69 Elective Radical prostatectomy Midline vs transverse Pain, OR time, PaO2, complications,

Incision lengthCox36 329 Unclear Abdominal surgery Midline vs paramedian Dehiscence, hernia rate, pulmonary

complications, wound infection,OR time, incision length

Guillou37 207 Both Abdominal surgery Midline vs paramedian OR time, pulmonary complications,wound infection, dehiscence, herniarate

Kendall38 249 Both Abdominal surgery Midline vs paramedian Dehiscence, hernia rate, woundinfection, OR time, mortality

Ellis39 125 Elective Abdominal Surgery Midline vs ParamedianParamedian vs Transverse

Hernia rate, dehiscence, woundinfection

Chan40 50 Elective Peritoneal dialysiscatheter

Midline vs paramedian Wound infection, mortality, herniarate, pulmonary complications

Halasz41 100 Both Cholecystectomy Paramedian vs transverse Pulmonary complications, woundinfection, hernia rate, pain,pulmonary complications

Talwar16 56 Emergency Emergency laparotomy Paramedian vs transverse Wound infection, dehiscence,pulmonary complications, hernia

ICU 5 intensive care unit; LOS 5 length of stay; OR 5 operating room.

K.A. Bickenbach et al. Incisions for abdominal surgery 403

the data from these studies shows a significant benefit infavor of transverse incisions (RR 5 1.77; 95% CI, 1.09 to2.87; Fig. 4). Subgroup analysis including only elective sur-gery still shows a significant advantage for transverse

incisions (RR 5 2.13; 95% CI, 1.13 to 4.05). There wasno significant difference in incision length, operativetime, wound infections, wound dehiscences, or postopera-tive length of stay between the 2 groups.

Table 4 Validity assessment

First author Concealment of allocationPatientblinding

Clinicianblinding

Outcomeassessorblinding

Loss tofollow-up(%)

Ali23 Unclear No No No NoneArmstrong24 Unclear No No No NoneBecquemin17 Unclear No No No NoneBrown28 Yes: draw from bag No Some Yes NoneChan40 Unclear No No No NoneCox36 Yes: masked from operating surgeon

determined by random numbersNo No No .20

Ellis39 Unclear No No No 5–20Fassiadis11 Yes: numbered, opaque envelopes No No No NoneGarcia-Valdecasas10 Unclear No No No NoneGreenall31,32 Unclear No No No ,5Guillou37 Yes: blind card No No No 5–20Halasz41 Yes: drawing cards No No No NoneHalm19 Yes: envelopes No No No 5–20Inaba30 Unclear No No No 5–20Kendall38 Yes: blind card system No No No .20Lacy27 Unclear No No No NoneLindgren29 Yes: envelopes No No No NoneMassucci26 Unclear No No No NoneProske18 Unclear No No No NoneSalonia35 Unclear No No No NoneSeenu25 Unclear No No No NoneSeiler33 Yes: envelopes Yes No Yes 5–20Stone34 Unclear No No No UnclearTalwar16 Unclear No No No Unclear

404 The American Journal of Surgery, Vol 206, No 3, September 2013

Comparisons: midline versus paramedianincision

Five trials reported rates of incisional hernia betweenpatients who underwent midline or paramedian incisions(Table 6). The follow-up in all 5 trials was 12 months.Combining the data showed a lower rate of incisional her-nia in paramedian incisions (RR 5 3.41; 95% CI, 1.02 to11.45); however, there was significant heterogeneity be-tween studies (I2 5 68%, Fig. 4). There were insufficientdata to analyze patients as subgroups.

No trials provided data to make comparisons of opera-tive time, incision length, length of stay, pain control, orpulmonary function. There were no significant differencesin pulmonary complications, wound infections, or rates ofdehiscence.

Comparisons: paramedian versus transverseincision

We identified 3 trials that compared paramedian withtransverse incisions (Table 6). The available data only al-lowed comparisons of postoperative complications. Therewere no significant differences in the rates of pulmonarycomplications, wound infections, or hernia rates althoughthe follow-up was short (ie, 3 to 12 months). Two trials

reported rates of wound dehiscence, with the combineddata favoring transverse incisions (RR 5 4.29; 95% CI,1.18 to 15.62).

Comments

This systematic review was performed to determine theoptimal incision choice for abdominal operations. Theresults of this review show that both transverse andparamedian incisions offer superior results compared withmidline incisions. Transverse incisions require less narcoticuse than midline incisions. Both transverse and paramedianincisions have a lower incidence of incisional hernias whencompared with midline incisions.

The combined data suggest that there is lower narcoticuse associated with a transverse incision. Patients withmidline incisions required 23.4 mg morphine more overtheir hospitalization period than patients with transverseincisions. There was significant heterogeneity with thisresult. Sensitivity results showed that if emergency surgerywas excluded from the analysis, the significance remainedbut the heterogeneity decreased. We believe this is clini-cally sensible because the disease process in emergencysurgery may have created more pain to mask that of theincision. For example, it is feasible that the pain ofperitonitis may be greater than the incisional pain. The

Table 5 The effectiveness of midline compared with transverse incisions

Outcome or subgroup

No. ofparticipants(trials)

Midlinemean/proportion

Transversemean/proportion

Relativeriskx (95% CI)

Absolute(per 100 patients)(95% CI)k

Quality ofevidence‡

Hernia rate 1,368 (7) 61/697 (9.9) 30/671 (4.5) 1.77 (1.09–2.87)* 3 more (0 to 8 more) Moderate{

Pulmonarycomplications

1,770 (10) 130/888 (14.6) 129/882 (14.6) 1.11 (.74–1.66) 2 more (4 less to 10 more) Low{,#

Wound infection 2,395 (11) 124/1,207 (10.2) 136/1,188 (11.4) .96 (.67–1.38) 0 less (4 less to 4 more) Moderate{

Wound dehiscence 1,903 (9) 15/959 (1.6) 7/944 (.7) 1.71 (.75–3.89) 0 more (0 less to 2 more) Low{,#

Total mg morphine 326 (6) 87.6 65.3 d WMD 5 23.4 (6.9–39.9)* Low{,**VAS pain scorePOD 1†

309 (6) 4.2 3.8 d WMD 5 .4 (2.9 to 1.8) Low{,**

% VC of original onPOD 1

263 (5) 50.0 58.4 d WMD 5 4.7 (21.1 to 10.5) Low{,**

% FEV1 of original onPOD 1

244 (4) 43.2 53.0 d WMD 5 6.9 (3.1–10.7)* Moderate4

OR time (min)Cholecystectomy 409 (4) 68 68 d WMD 5 1.3 (25.2 to 7.9) Moderate{

Right hemicolectomy 68 (2) 95 102 d WMD 5 26.2 (218.2 to 5.8) Low{,#Aortoiliac surgery 76 (2) 244 240 d WMD 5 6.5 (237.3 to 50.2) Low{,#

Incision length (cm)Cholecystectomy 168 (2) 14.3 14.3 d WMD 5 1.2 (21.1 to 3.5) Low{,**Right hemicolectomy 69 (2) 14.2 13.0 d WMD 5 .7 (2.6 to 2.0) Low{,#

Gastric/pancreas 489 (2) 18.8 24.5 d WMD 5 25.7 (211.6 to .2) Low{,**Length of stay (d)Cholecystectomy 370 (3) 4.0 2.8 d WMD 5 .7 (2.9 to 2.3) Low{,#

Right hemicolectomy 68 (2) 8.5 7.7 d WMD 5 .9 (2.5 to 2.3) Low{,#

OR 5 operating time; POD 5 postoperative day; VAS 5 visual analog scale; VC 5 vital capacity; WMD 5 weighted mean differences.

*Statistically significant.†VAS pain scale 1 to 10.‡Quality rated from 1 (very low quality) to 4 (high quality).xValues .1 favor transverse incision.kValues are expected differences if patients undergo midline instead of transverse incisions.{Evidence limited by methodological quality of studies.#Evidence limited by imprecise data (small sample size or event rate).

**Evidence limited by heterogeneity between studies.

K.A. Bickenbach et al. Incisions for abdominal surgery 405

finding of decreased narcotic use with the transverseincision is contrary to what most surgeons believe. Mostbelieve that transverse incisions, especially in the upperquadrants, are more painful than midline incisions. How-ever, these results contradict that belief.

The present study did not show a significant differencein VAS pain scores on POD 1. However, there was again

Figure 2 A Forrest plot comparing milligrams of morphine per hoincision.

substantial heterogeneity (I2 5 98%). We were unable toexplain the heterogeneity based on subgroup and sensitivityanalysis from our predefined hypotheses (Table 2). The lackof a difference in VAS pain scores could be explained byreasoning that patients with midline incisions were admin-istered more narcotics to normalize their VAS pain scoresto those of transverse incisions.

spitalization in patients with midline incisions versus transverse

Figure 3 A Forrest plot comparing milligrams of morphine per hospitalization in patients with midline incisions versus transverse inci-sions excluding the Garcia-Valdecasas trial.

406 The American Journal of Surgery, Vol 206, No 3, September 2013

The combined data show that there is a greater decreasein FEV1 on POD 1 in patients who have midline incisions.There is no difference noted in vital capacity between the 2incision choices. Because of differences in how data werereported, we were unable to combine data for the later post-operative days, so it is unclear what the effects later in theadmission were. There was no difference in pulmonarycomplications with a fairly narrow CI. Therefore, althoughthere was a difference in pulmonary function, a 6.9%change in FEV1 on POD 1 may not be clinically significantin the general population. It is conceivable that this differ-ence could be significant in patients with poor initial pul-monary function.

Our results showed a higher rate of incisional hernia inmidline compared with transverse incisions (Fig. 5). Thesedata are mainly driven by the Fassiadis trial, which had thehighest rate of incisional hernias.11 However, this trial alsohad the longest follow-up of 4.4 years, which is of particularrelevance given the preponderance of data suggesting thatmost hernias develop after 1 year from surgery.12,13 Mudgeand Hughes12 showed that less than 50% of hernias developin 1 year. This finding was corroborated by Hoer et al,13 whoshowed that it takes 2 years for 75% of hernias to develop.The majority of the trials had a follow-up of 1 year or less,so it is likely that the Fassiadis trial has a more accuraterate of detection. The combined data show that paramedianincisions also have a lower rate of hernias than midline inci-sions. There was a 3-fold increase in the hernia rate for

Figure 4 A Forrest plot comparing the hernia rate i

patients who had midline incisions. The follow-up in all 5of these trials was 1 year, so it is possible that the differencemay be more pronounced with a longer follow-up. However,with the lack of long-term follow-up, it is difficult to drawconclusions on the long-term effects of incision choice. Irre-spective of the type of incision surgeons use, they should ad-here strictly to proven methods of incision closure.14,15

It is difficult to draw conclusions when comparingtransverse with paramedian incisions because this study isunderpowered. There was no difference observed in herniarates, wound infections, or pulmonary complications be-tween the 2 incisions. There was a higher rate of wounddehiscence with paramedian incisions when they werecompared with transverse incisions, suggesting that atransverse incision is superior. However, this finding wasdriven mainly by the Talwar study,16 which had an unusu-ally high rate of wound dehiscence (38%). Dehiscence ratesof paramedian incisions ranged from 0% to 1% in the stud-ies that compared paramedian with midline incisions. TheTalwar study was conducted on patients with intestinal per-forations from typhoid and had a high incidence of woundinfections (52% transverse and 59% paramedian). It couldbe theorized that the high rate of wound infections may in-crease the dehiscence rate, but there was no statistically sig-nificant difference in the wound infection rate between the2 incisions.

Although we found that the transverse and paramedianincisions offered some advantages over midline incisions,

n midline incisions versus paramedian incisions.

Figure 5 A Forrest plot comparing the hernia rate in midline incisions versus transverse incisions.

K.A. Bickenbach et al. Incisions for abdominal surgery 407

the adequacy of the incision to expose the operative field isof crucial importance. The adequacy of exposure was onlyaddressed in 1 study. In this trial, surgeons subjectivelyevaluated the adequacy of the exposure by rating it asexcellent, good, or poor.17 The midline incision was foundto be excellent in 13 cases and good in 2. The transverseincision was found to be excellent in 7 cases, good in 2,and poor in only 1 case. This was related to the difficultyin exposure of the right iliac artery for aortoiliac surgery.However, there was no difference in operative time in this

Table 6 The effectiveness of paramedian incision

Outcome orsubgroup

No. ofparticipants(trials) Midline Paramedian

Midline compared with paramedianHernia rate 1014 (5) 48/491 (9.8) 19/523 (3.6)Pulmonarycomplications

586 (3) 54/240 (22.5) 71/346 (20.5)

Wound infections 935 (4) 43/452 (9.5) 49/483 (10.1)Wounddehiscence

1010 (4) 4/468 (0.6) 4/452 (.9)

Paramedian compared with transverseOutcome orSubgroup

No. ofparticipants(trials)

Transverse Paramedian

Hernia rate 292 (3) 14/127 (11) 29/165 (17.6)

Pulmonarycomplications

156 (2) 2/77 (2.6) 11/79 (13.9)

Wound infection 156 (2) 24/77 (31.2) 20/79 (25.3)

Wounddehiscence

192 (2) 2/77 (2.6) 12/115 (10.4)

*Statistically significant.†Quality rated from 1 (very low quality) to 4 (high quality).‡Values .1 favor paramedian.xValues are expected differences if patients undergo midline or transversekEvidence limited by quality of studies.{Evidence limited by imprecise data (small sample size or event rate).#Evidence limited by heterogeneity between studies.

study or any of the other studies, suggesting that the effectof exposure was minimal. Furthermore, although most sur-geons would acknowledge that midline incisions are easierto open and close than transverse incisions, the similar totaloperative times suggests that this difference is not clinicallysignificant. Future studies should address the difference inexposure between these incisions.

Another important factor in incision choice is cosmesis.Cosmesis was only examined in 2 trials. The trial by Proskeet al18 showed a significantly higher patient-rated score for

RR‡ (95% CI)

Absolute(per 100 patients)(95% CI)x

Quality ofevidence†

3.41 (1.02–11.45)* 9 more (0 to 38 more) Lowk,#

1.04 (.75– 1.42) 1 more (9 less to 5 more) Lowk,{

1.09 (.71– 1.69) 1 more (3 less to 7 more) Moderatek

.90 (.19– 4.21) 0 less (1 less to 3 more) Lowk,{

Relative risk‡

(95% CI)Absolute(per 100 patients)(95% CI)x

Quality ofevidence†

.7 (.25–2) 5 less(13 less to 18 more)

VeryLowk,{,#

.27 (.02– 37.96) 10 less(14 less to 26 more)

Lowk,{

1.28 (.83– 1.96) 7 more(4 less to 24 more)

Lowk,{

0.23 (.06–.85)* 8 less (2 to 10 less) Lowk,{

instead of paramedian incisions.

408 The American Journal of Surgery, Vol 206, No 3, September 2013

transverse incisions than for midline incisions. This differ-ence was present despite a larger incision length for trans-verse incisions. The Halm et al trial19 also showed that bothpatients and surgeons were significantly more satisfied withthe cosmetic result of transverse incisions compared withmidline incisions. Transverse incisions may offer bettercosmesis because they are situated along the Langer linesof the skin. This results in less tension on the incisionand may result in a thinner scar. The width of the scarwas only addressed in the Halm et al study, which showedthat midline incisions had significantly thicker scars thantransverse incisions (8.3 vs 3.3 mm, P , .0001).19

Conclusions

Approximately 5 million laparotomies are performedeach year in the United States. This results in approxi-mately 400,000 to 500,000 incisional hernias and close to200,000 repairs each year.2 Given that the average totalhospital cost for an open incisional hernia repair is US$7,197 and US $6,396 for a laparoscopic repair, this is asignificant burden on our economy.20 This review showsthat the use of either a paramedian or a transverse incisionas opposed to a midline incision results in a lower herniarate. This difference could translate into significant cost re-ductions for the health care industry.

Strengths of this study include the comprehensive searchfor all eligible studies, systematic and explicit applicationof the eligibility criteria by duplicate assessment, carefulconsideration of the study quality, generation and testing ofa priori hypotheses to explain heterogeneity of the data, andstandardized data extraction. We identified a large numberof studies with a large number of patients and can makerigorous conclusions.

Limitations to this study include the methodologicquality of many of the primary studies. In only 9 trialswas the method of concealment adequate. Additionally, inonly 2 trials were the patients, medical staff, and outcomeassessors blinded. This review was also limited by incom-plete data reporting in some of the eligible studies and bythe limited follow-up. Another limitation of the study is thehigh level of heterogeneity of the studies. A final limitationis that some of the trials included are dated and should betreated with caution. Some of the older trials includeoperations that are not commonly performed today, suchas open cholecystectomy, which has been replaced bylaparoscopic surgery. However, this should not impact theresults because the choice of incision is being studied andnot the operation performed. A large randomized controlledtrial should be performed to better evaluate incision choicewith modern operations.

The Grading of Recommendations Assessment, Devel-opment and Evaluation (GRADE) system was used toassess the overall quality of evidence.21,22 Under this sys-tem, trials are rated as high-quality evidence unless theyare limited by serious impairments to study quality,

important inconsistency, uncertainty about directness, im-precise or sparse data, or high probability of reportingbias. The quality of data was rated as moderate when com-paring midline incisions with transverse incisions for theoutcomes of hernia rates, pulmonary complications, andwound infections. The data were limited by the methodo-logic quality of the studies. Comparisons regarding painwere rated as low quality because of the limitations ofthe methodologic quality and the heterogeneity betweenstudies. For the comparison of the hernia rate between mid-line and paramedian incisions, the quality was rated low be-cause of the methodologic quality and the heterogeneity ofresults. The quality of data for all outcomes when compar-ing paramedian with transverse incisions was rated low be-cause of the limitations of the methodologic quality and thepoor sample sizes.

In summary, transverse incisions are associated with lesspostoperative narcotic use than midline incisions. Addi-tionally, both transverse incisions and paramedian incisionshave lower incisional hernia rates than midline incisions.Based on these results, we recommend that surgeons useeither a transverse or paramedian incision when they arecomfortable with the exposure provided. Future studiesshould focus on surgeons’ ease and adequacy of exposureand should address the cosmetic results from the patients’perspectives.

References

1. Brown SR, Goodfellow PB. Transverse verses midline incisions

for abdominal surgery. Cochrane Database Syst Rev 2005;4:CD005199.

2. Sajid MS, Bokhari SA, Mallick AS, et al. Laparoscopic versus open

repair of incisional/ventral hernia: a meta-analysis. Am J Surg 2009;

197:64–72.

3. Burger JW, van’t Riet M, Jeekel J. Abdominal incisions: techniques

and postoperative complications. Scand J Surg 2002;91:315–21.

4. Thompson JB, MacLean KF, Coller FA. Role of the transverse abdom-

inal incision and early ambulation in the reduction of postoperative

complications. Arch Surg 1949;59:1267–77.

5. Rees VL, Coller FA. Anatomic and clinical study of the transverse ab-

dominal incision. Arch Surg 1943;47:136–46.

6. Sloan GA. A new upper abdominal incision. Surg Gynecol Obstet

1927;45:678–87.

7. Donaldson DR, Hegarty JH, Brennan TG, et al. The lateral paramedian

incisiondexperience with 850 cases. Br J Surg 1982;69:630–2.

8. Donaldson DR, Hall TJ, Zoltowski JA. Does the type of suture mate-

rial contribute to the strength of the lateral paramedian incision? Br J

Surg 1982;69:163–5.

9. Hozo SP, Djulbegovic B, Hozo I. Estimating the mean and variance

from the median, range, and the size of a sample. BMC Med Res

Methodol 2005;5:13.

10. Garcia-Valdecasas JC, Almenara R, Cabrer C, et al. Subcostal incision

versus midline laparotomy in gallstone surgery: a prospective and ran-

domized trial. Br J Surg 1988;75:473–5.

11. Fassiadis N, Roidl M, Hennig M, et al. Randomized clinical trial of

vertical or transverse laparotomy for abdominal aortic aneurysm re-

pair. Br J Surg 2005;92:1208–11.

12. Mudge M, Hughes LE. Incisional hernia: a 10 year prospective study

of incidence and attitudes. Br J Surg 1985;72:70–1.

13. Hoer J, Stumpf M, Rosch R, et al. Prevention of incisional hernia.

Chirurg 2002;73:881–7.

K.A. Bickenbach et al. Incisions for abdominal surgery 409

14. Millbourn D, Cengiz Y, Israelsson LA. Effect of stitch length on

wound complications after closure of midline incisions: a randomized

controlled trial. Arch Surg 2009;144:1056–9.

15. Hodgson NC, Malthaner RA, Ostbye T. The search for an ideal method

of abdominal fascial closure: a meta-analysis. Ann Surg 2000;231:

436–42.

16. Talwar S, Laddha BL, Jain S, et al. Choice of incision in surgical man-

agement of small bowel perforations in enteric fever. Trop Gastroen-

terol 1997;18:78–9.

17. Becquemin JP, Piquet J, Becquemin MH, et al. Pulmonary function af-

ter transverse or midline incision in patients with obstructive pulmo-

nary disease. Intensive Care Med 1985;11:247–51.

18. Proske JM, Zieren J, Muller JM. Transverse versus midline incision for

upper abdominal surgery. Surg Today 2005;35:117–21.

19. Halm JA, Lip H, Schmitz PI, et al. Incisional hernia after upper ab-

dominal surgery: a randomised controlled trial of midline versus trans-

verse incision. Hernia 2009;13:275–80.

20. Earle D, Seymour N, Fellinger E, et al. Laparoscopic versus open inci-

sional hernia repair: a single-institution analysis of hospital resource

utilization for 884 consecutive cases. Surg Endosc 2006;20:71–5.

21. Atkins D, Best D, Briss PA, et al. Grading quality of evidence and

strength of recommendations. BMJ 2004;328:1490.

22. Guyatt G, Gutterman D, Baumann MH, et al. Grading strength of rec-

ommendations and quality of evidence in clinical guidelines: report

from an american college of chest physicians task force. Chest

2006;129:174–81.

23. Ali J, Khan TA. The comparative effects of muscle transection and me-

dian upper abdominal incisions on postoperative pulmonary function.

Surg Gynecol Obstet 1979;148:863–6. Available at: http://www.mrw.

interscience.wiley.com/cochrane/clcentral/articles/556/CN-00196556/

frame.html. Accessed October 1, 2010.

24. Armstrong PJ, Burgess RW. Choice of incision and pain following

gallbladder surgery. Br J Surg 1990;77:746–8.

25. Seenu V, MisraMC.Mini-lap cholecystectomy–an attractive alternative

to conventional cholecystectomy. Trop Gastroenterol 1994;15:29–31.

26. Massucci M, Lauri D, Faraglia V, et al. Approach to the abdominal

aorta: impairment of respiratory function after supraumbilical trans-

verse and midline laparotomy. Ital J Surg Sci 1989;19:247–53.

27. Lacy PD, Burke PE, O’Regan M, et al. The comparison of type of in-

cision for transperitoneal abdominal aortic surgery based on postoper-

ative respiratory complications and morbidity. Eur J Vasc Surg 1994;8:

52–5.

28. Brown SR, Goodfellow PJ, Adam IJ, et al. A randomised controlled

trial of transverse skin crease vs vertical midline incision for right

hemicolectomy. Tech Coloproctol 2004;8:15–8.

29. Lindgren PG, Nordgren SR, Oresland T, et al. Midline or transverse

abdominal incision for right-sided colon cancerda randomized trial.

Colorectal Dis 2001;3:46–50. Available at: http://www.mrw.

interscience.wiley.com/cochrane/clcentral/articles/867/CN-00424867/

frame.html. Accessed October 1, 2010.

30. Inaba T, Okinaga K, Fukushima R, et al. Prospective randomized study

of two laparotomy incisions for gastrectomy: midline incision versus

transverse incision. Gastric Cancer 2004;7:167–71.

31. Greenall MJ, Evans M, Pollock AV. Midline or transverse laparotomy?

A random controlled clinical trial. Part II: influence on postoperative

pulmonary complications. Br J Surg 1980;67:191–4.

32. Greenall MJ, Evans M, Pollock AV. Midline or transverse laparotomy?

A random controlled clinical trial. Part I: influence on healing. Br J

Surg 1980;67:188–90.

33. Seiler CM, Deckert A, Diener MK, et al. Midline versus transverse in-

cision in major abdominal surgery: a randomized, double-blind equiv-

alence trial (POVATI: ISRCTN60734227). Ann Surg 2009;249:

913–20.

34. Stone HH, Hoefling SJ, Strom PR. Abdominal incisions: transverse vs

vertical placement and continuous vs interrupted closure. South Med J

1983;76:1106–8.

35. Salonia A, Suardi N, Crescenti A, et al. Pfannenstiel versus vertical

laparotomy in patients undergoing radical retropubic prostatectomy

with spinal anesthesia: results of a prospective, randomized trial. Eur

Urol 2005;47:202–8.

36. Cox PJ, Ausobsky JR, Ellis H, et al. Towards no incisional hernias: lat-

eral paramedian versus midline incisions. J R Soc Med 1986;79:711–2.

37. Guillou PJ, Hall TJ, Donaldson DR, et al. Vertical abdominal inci-

sions–a choice? Br J Surg 1980;67:395–9.

38. Kendall SW, Brennan TG, Guillou PJ. Suture length to wound length

ratio and the integrity of midline and lateral paramedian incisions. Br J

Surg 1991;78:705–7.

39. Ellis H, Coleridge-Smith PD, Joyce AD. Abdominal incisionsdverti-

cal or transverse? Postgrad Med J 1984;60:407–10.

40. Chan HM, Hsieh JS, Huang YS, et al. Prospective randomized trial of

midline incision and paramedian incision for prevention of incisional

hernia. J Surg Assoc ROC 1992;25:1332–6.

41. Halasz NA. Vertical vs horizontal laparotomies. I. Early postoperative

comparisons. Arch Surg 1964;88:911–4.