Dissections INTERVENTIONAL

2 June 2009
Evidence-based Medicine for Surgeons

Does type of pancreaticojejunostomy after pancreaticoduodenectomy decrease rate of

pancreatic fistula? a randomized, prospective, dual-institution trial
Authors: Berger AC, Howard TJ, Kennedy EP, et al
Journal: J American College of Surgeons 2009; 208:738–74
Centre: Thomas Jefferson University, Philadelphia, PA; and Indiana University, Indianapolis, IN, USA
The pancreatic-enteric anastomosis is the “Achilles heel” of pancreaticoduodenectomy (PD).
Leakage or failure of healing, results in pancreatic fistula (PF), an important cause of morbidity
following PD. The estimated incidence of this complication is 10% to 25%. There are two widely
BACKGROUND used methods to accomplish an end to side pancreaticojejunostomy (PJ) after PD: invagination PJ
(“dunking” the pancreatic remnant into the jejunum) or duct to mucosa PJ. It has been suggested
that the latter is safer, but firm evidence is lacking.

RESEARCH QUESTION IN SUMMARY

Population Study results
All patients at two university, Hard gland Soft gland
referral hospitals undergoing
pancreaticoduodenectomy (PD). Duct to Invagination Duct to Invagination
mucosa PJ PJ mucosa PJ PJ
Indicator variable
Number 47 49 50 51
A two-layer end to side pancreatic
duct to jejunal mucosa Pancreatic fistula rate 17.8% (35 patients) - overall
anastomosis (duct to mucosa) or a By gland type 8.00% 27.00%
two-layer end to side invagination (p=0.007)
technique, after stratification into
"hard" and "soft" glands. By technique (#p=0.06) 11.00% 6.00% 36.00% # 18.00%

Outcome variable There were no substantial differences in overall complications, mortality, and
rates of reoperation between the duct to mucosa and the invagination groups.
Primary: Pancreatic fistula (PF)
rate. Secondary: length of Major complications Duct to mucosa-25%; invagination-12% (p=0.03)
hospitalization, percutaneous Interventional Duct to mucosa-11%; invagination-3% (p=0.03)
radiologic intervention rates, radiological procedures
reoperation rates, morbidity and
PJ = pancreaticojejunostomy
30-day or in-hospital mortality.
Authors' claim(s): “... considerably fewer fistulas with invagination
Comparison compared with duct to mucosa pancreaticojejunostomy after
pancreaticoduodenectomy. Results confirm increased PF rates in soft as
As described above. compared with hard glands.”

THE TISSUE REPORT

The randomized clinical trial (RCT) is the definitive method of resolving the efficacy of interventions in clinical practice.
Quite often, as this study shows, the results are counter-intuitive and run against common wisdom. There is little room for
discomfort in the methodology of the study. Considering the well known dictum that pancreaticoduodenectomy should be
offered only by centres that can deliver the procedure with mortality rates of 2% or less, any definitive recommendation that
reduces risk is a step forward.

EBM-O-METER
Evidence level Overall rating Bias levels
Double blind RCT Sampling
Randomized controlled trial (RCT) Comparison
Trash Swiss Safe News-
Prospective cohort study - not randomized cheese worthy Measurement
Life's too Holds water
short for this Full of holes “Just do it”
Case controlled study
Interestingl | Novel l | Feasible l
Case series - retrospective  Ethical l | Resource saving l

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

© Dr Arjun Rajagopalan
 SAMPLING
Sample type Inclusion criteria Exclusion criteria Final score card
Simple random All patients Locally Hard+d-m Hard+inv Soft+d-m Soft+inv
undergoing unresectable 
Stratified random Target ? ? ?
pancreatico- Inability to locate
Cluster duodenectomy  pancreatic duct  Accessible 272
Operation other
Consecutive than PD  Intended 47 49 50 51
Convenience Metastatic disease Drop outs 0 0 0 0

Judgmental Study 47 49 50 51

 = Reasonable | ? = Arguable |  = Questionable

Duration of the study: TJUH: August 2006 to May 2008; IUH: February 2007 to May 2008
With alpha set at 0.05, 95 patients per group (190 total) were needed to have 80% power to detect a
difference from 30 to 15%

Sampling bias: Although the study was done in two referral, university hospitals, it is only such centres that can
carry out studies of this nature and recruit the required number of patients.

COMPARISON
Randomized Case-control Non-random Historical None

Controls - details
Allocation details Patients were stratified into two groups: soft (normal) texture (predicted fistula rate of 20% to
40%) and hard (fibrotic) texture (predicted fistula rate of 0% to 10%). Patients were
randomized to one of two groups: a two-layer end to side pancreatic duct to jejunal mucosa
anastomosis (duct to mucosa) or a two-layer end to side invagination technique. All other
aspects of the intraoperative and postoperative management of the patients were not
influenced by this study. Prophylactic octreotide was not used in any patient.

(Details of the technique along with excellent, colour illustrations are provided in the original
article.)
Comparability Patient demographics, including age and gender,were comparable between the two groups.
Intraoperative parameters were not statistically different between the duct to mucosa and the
invagination groups. Median estimated blood loss and red blood cell transfusion requirements
were similar in the two groups. Diameter of the pancreatic duct was similar for the two
groups. 142 patients who underwent PD for malignant disease, and 55 patients had benign or
premalignant processes.
Disparity No significant differences.

Comparison bias: Minimal.

MEASUREMENT
Measurement error
Device used Device error Observer error
Gold std.
Repetition

Protocols

Device suited to task

Training

Scoring

Blinding

Y ? N

1.Pancreatic fistula (ISGPF definition) Y N Y N N N N

The International Study Group on Pancreatic Fistulas (ISGPF) defines a pancreatic fistula as “out-put via an operatively
placed drain (or a subsequently placed, percutaneous drain) of any measurable volume of drain fluid on or after
postoperative day 3, with an amylase content greater than 3 times the upper normal serum value.

Measurement bias: Nothing of note.

© Dr Arjun Rajagopalan