Midline versus transverse incision in major abdominal surgery: A randomized, double-blind equivalence trial

Authors: Seiler CM, Deckert A, Diener MK, et alJournal: Annals of Surgery 2009; 249: 913–920 Centre: University of Heidelberg, Heidelberg, Germany

BACKGROUND

Midline and transverse incisions are widely used in abdominal surgery; the choice is usually based on the surgeon's preference. Relevant outcomes that may guide the choice of incision are the incidence of postoperative complications (burst abdomen, pulmonary complications, wound infection, incisional hernia) and the time taken to open and close the abdomen. Pain and return to normal activities in the short-term, and cosmetics in the long-term, may also be of interest.

IN SUMMARY Midline vs transverse laparotomy incisions

Midline Transverse

Number randomized 101 99

Primary end points (pts) 70 64

Pain intensity (VAS) 13 17

Analgesic usage (mg/hr) 1.08 1.04

Secondary end points (pts) 96 97

Wound infection 5 15 *

Burst abdomen 0 1

Pulmonary complications 13 17

Procedure time (hrs) 4.2 4.1

Mortality 2 2

Incisional hernia at 1 yr 13 (of 79) 8 (of 69)* p = 0.02

None of the other differences were statistically significantAuthors' claim(s): “... According to the results of [this trial]... both incision types can be used without any relevant consequences for the patient.”

THE TISSUE REPORT There are too many flaws in this trial. The drop off in numbers between the accessible, intended (after applying inclusion/ exclusion criteria) and final study population is large. The decision to use a single measurement on the 2nd postoperative day (and the total analgesic usage as the dosage used within 24 hours of the VAS measurement) as the sole index of pain intensity seems very arbitrary. The study is peppered with misuse of terms and liberal interpretations of methodological concepts. There is too much post-hoc slicing of the pie in presenting study numbers. The intention to treat process excludes drop outs from the original cohort - a clear contravention of definitions. To cap it all, an incisional hernia rate of 12-16% at one year is alarming. And, oh by the way, how do you double blind surgical procedures like the title claims?

The devil is in the details (more on the paper) ...

© Dr Arjun Rajagopalan

EBM-O-METER

Evidence level Overall rating Bias levelsDouble blind RCT

TrashLife's too

short for this

Swiss cheese

Full of holes

SafeHolds water

News-worthy

“Just do it”

Sampling

Randomized controlled trial (RCT) Comparison

Prospective cohort study - not randomized Measurement

Case controlled study Interesting l | Novel l | Feasible l Ethical l | Resource saving lCase series - retrospective

RESEARCH QUESTION

Population

A series of patients undergoing elective primary laparotomy for stomach, colonic and pancreatic procedures. (The Postsurgical Pain Outcome of Vertical and Transverse Abdominal Incision (POVATI) trial )

Indicator variable

Midline and transverse incisions.

Outcome variable

Primary: Analgesic usage and pain intensity. Secondary: frequencies of early-onset complications (burst abdomen, pulmonary events, wound infections) and late-onset complications (incisional hernias after one year).

Comparison

As described above.

INTERVENTIONAL

6 July 2009DissectionsDissectionsEvidence-based Medicine for Surgeons

SAMPLING Sample type Inclusion criteria Exclusion criteria Final score card

Simple random Elective primary laparotomy for stomach, colonic and pancreatic procedures suitable for both incisions Life expectancy > than one year

Recurrent opening of the abdominal wall ? Certain procedures (liver, gallbladder, spleen, and rectum)

Midline Transverse

Stratified random Target ? ?

Cluster Accessible 1188

Consecutive Intended 101 99

Convenience Drop outs See below

Judgmental StudyPrimary end point20 end point - 30 days20 end point - 1 year

709779

649669

= Reasonable | ? = Arguable | = QuestionableDuration of the study: February 2007 to October 2007

Sampling bias: The sampling process is seriously flawed. The drop off in numbers between the accessible and intended population (after applying inclusion/ exclusion) criteria and the final study population is large. Excluding "certain procedures (liver, gallbladder, spleen and rectum)" is cause for concern. The authors perform a post-hoc pie slicing of the study population that is dubious. Their "intention to treat" methodology is unconventional: it excludes drop-outs after randomization.

© Dr Arjun Rajagopalan

COMPARISON Randomized Case-control Non-random Historical None

Controls - detailsAllocation details The closure method was standardized with a continuous, mass, all-layer closure using 2 Mono

Plus loops starting at both edges of the incision, with a 4:1 ratio (suture:incision length). Both rectus muscles were divided uniformly in the transverse group. Neither a subcutaneous drain nor a subcutaneous suture was used. The skin was closed with staples.

Postoperative analgesia was standardized with a dosage of 1mg/mL Piritramide through a patient-controlled pump.

Comparability Study groups were comparable for all patient and procedure characteristics and also within the different surgical categories

Disparity -

Comparison bias: The groups appear well matched and comparable.

MEASUREMENT Measurement error

Device used Device error Observer error

Device suited to task

Y ? N

Rep

etitio

n

Gol

d st

d.

Trai

ning

Prot

ocol

s

Sco

ring

Blin

ding

1.Pain intensity (Visual Analogue Scale - 1-100) Y N Y Y Y Y Y

Abdominal pain intensity at was rest measured on a scale from 0 to 100 with the Visual Analogue Scale (VAS) at day 2 postoperatively and the total amount of analgesics required (Piritramide mg/L) within 24 hours before VAS measurement after surgery.

At least an abdominal wall ultrasonography (or computer tomography or magnetic resonance imaging if performed for reevaluation of underlying disease) was used for assessment of incisional hernias.

Measurement bias: The decision to use a single measurement on the 2nd postoperative day and the total analgesic usage as the dosage used within 24 hours of the VAS measurement seems very arbitrary. There is no satisfying explanation offered for this specific choice rather than a mean or average of some sort from the whole postoperative period.