Professional Documents
Culture Documents
perioperative and postoperative analgesia in all hospitals SPSS 14.0 for Windows. Standard approximation to binomi-
throughout the trial. Nasogastric tubes were removed at the nal distribution by 2 analysis or Fisher exact test was used
end of surgery or at the latest on the first postoperative day when appropriate. Continuous variables were compared by
(POD 1) from when mobilization was encouraged. Student t test. Adjustments were by linear or binary logistic,
The patients randomly assigned to be allowed normal or Cox regression as appropriate. Differences were consid-
food at will were allowed to take ordinary hospital food as ered significant at P ⬍ 0.05.
they wanted from the first day after surgery. They were
encouraged to begin the oral intake carefully and adjust Assignment
according to tolerance. The nutritional intake was not mea- Computerized randomization lists were created and the
sured and there were no target values. results were put in sealed opaque envelopes by individuals not
The patients randomly assigned to enteral tube feeding by involved in the trial. The envelopes were stored in a central
needle-catheter jejunostomy (ETF group) had a feeding catheter randomization office in Tromsø, physically removed from wards
(FloCare Jejunocath, 5 Ch., external diameter 1.7 mm; Nutricia and theaters. Randomization was stratified for hospital and for
Norway AS, Oslo, Norway) passed through the abdominal wall high (esophageal and pancreatic resections) or medium risk
via a split cannula at a 45° angle to the skin and along the axis operations and blocked with a fixed but unknown block size.
of a targeted jejunal loop distal to any anastomosis. Another split After informed consent, the patients were randomized intraop-
cannula allowed a 3 cm intramural passage to be created before eratively if surgery fulfilling the criteria had been performed. A
entering the bowel lumen. The jejunal segment was fixed to the phone call was made to the central randomization office imme-
abdominal wall at the insertion site with absorbable sutures over diately before the closing of the abdominal incision, after the
a length of 3 to 6 cm to prevent kinking.17 The patients received intra-abdominal surgical procedure was completed.
isotonic saline by the catheter at 20 mL/h until the morning of Ethics and Consent
POD 1. Nutrition was then commenced (standard Fresubin, 1
The study was approved by the Regional Division of
kcal/mL; Fresenius Kabi AS, Halden, Norway) at 20 mL/h. The
the National Committee for Research Ethics (P-REK V 52/
rate was increased by 20 mL/h each day if tolerated, up to 80
2000) and registered at ClinicalTrials.gov on August 23, 2005
mL/h. In the event of abdominal pain, severe distension, or
(NCT00134407). It was carried out in accordance with the
vomiting, infusions were reduced or temporarily halted. A max-
imum of 450 mL of water per day was allowed by mouth. At
POD 6, the patients were allowed food at will and enteral
infusion halted.
Statistics
We estimated the minimum rate of patients with major
complications in this heterogeneous population at 25%. A
reduction of this by half (to 12.5%) was considered clinically
important. Detecting a difference of this magnitude or greater
at a level of statistical significance of 0.05 and a power of
0.90 with a two-tailed test of proportions required a total of
215 patients in each group. Correcting for the effect of one
interim analysis, a total number of 444 patients was needed.
The data were analyzed on an intention-to-treat basis, with FIGURE 1. Trial flowchart.
chart: Fig. 1). The groups were comparable by baseline with 100 major complications in the 220 patients allowed
characteristics (Table 2), and by procedures performed normal food at will (P ⫽ 0.012). In the ETF group, 36
(Table 3)—also within each hospital. (15.9%) patients were reoperated compared with 29 (13.2%)
Complications and Adverse Events in the group allowed normal food at will (P ⫽ 0.50) (Table
In the ETF group, 76 (33.5%) of 227 patients suffered 4). Thirty-day mortality was 4.4% in the ETF group and 5.0%
major complications compared with 62 (28.2%) of 220 in the in the group allowed normal food at will (P ⫽ 0.83). Total
group allowed normal food at will (P ⫽ 0.26). When adjusted mortality within the trial period was 8.4% in the ETF
for main baseline variables (gender, age, emergency surgery, group versus 5.9% in the group allowed normal food at
high-risk operations, and operating hospital) there was still no will (P ⫽ 0.36).
significant difference (P ⫽ 0.46). The 227 patients in the ETF Subgroup analysis of the different operations showed
group suffered a total of 165 major complications compared no significant difference in rates of major complications
between groups (data not shown) except for total gastrecto- high-risk operations, and operating hospital) the difference
mies, where 6 (15.8%) of 38 patients in the ETF group had an remained significant (P ⫽ 0.03). The time to first bowel
intra-abdominal abscess compared with none of the 39 pa- movement was not significantly different (mean 4.3 days in
tients in the group allowed normal food at will (P ⫽ 0.012). the ETF group and 4.0 days for those allowed normal food at
Adjusting for the presence or lack of an upper GI anastomosis will, P ⫽ 0.112). The length of stay was longer in the ETF
did not result in any significant differences between treatment group than in the group allowed normal food at will (mean
groups regarding anastomotic leak rate, major infectious 16.7 vs. 13.5 days, P ⫽ 0.046, 95% CI for difference in
complications, or number of patients with any major compli- means from 0.1 to 6.3).
cation.
The rates of minor complications or adverse events Follow-up
were not significantly different (Table 4), nor was the need At follow-up, 40 (19.0%) of 211 patients in the ETF
for parenteral nutrition or selective use of nasogastric tube. group who were discharged alive were found to have suffered
a late complication compared with 24 (11.5%) of 209 in the
Enteral Tube Feeding group allowed normal food at will (P ⫽ 0.04) (Table 6).
Of 224 attempts, 223 NCJ catheters were inserted. Four There were significantly more wound infections in the ETF
patients died within the first 5 postoperative days and another group, 17 of 211 versus 5 of 209 patients in the group allowed
8 patients with functioning catheters requested crossover to normal food at will (P ⫽ 0.01). The patients in the ETF group
food. Thirty-one (13.9%) catheters were prematurely re- had a higher postoperative weight loss (mean 5.2 kg reduc-
moved (Table 5). The remaining 181 patients received a tion in 144 patients) than those in the group allowed normal
median of 4.800 mL enteral feed during the first 5 postoper- food at will (mean 4.0 kg reduction in 152 patients, P ⫽
ative days. Three (1.3%) of 224 patients were reoperated as a 0.06). One of the patients in the ETF group developed a
direct result of catheter complications (two intra-abdominal fistula at the site of the catheter insertion.
leaks and one subcutaneous abscess).
Bowel Function and Recovery DISCUSSION
Time to first flatus was significantly longer in the ETF This randomized trial shows that allowing normal food
group (mean 3.0 days) compared with the group allowed at will immediately after major upper GI surgery does not
normal food at will (2.6 days, P ⫽ 0.01, 95% CI for increase the rate of major complications compared with a
difference in means from ⫺0.7 to ⫺0.2). When adjusted for standard regimen of nil-by-mouth and enteral feeding for 5
main baseline variables (gender, age, emergency surgery, days. Furthermore, both time to resumed bowel function and
postdischarge rate of complications were significantly re-
duced for the group allowed normal food at will.
TABLE 5. NCJ-related Outcome Main Outcome
Enteral Tube Feeding The benefit of allowing normal food at will, as opposed
n ⴝ 227, n (%) to nil-by-mouth, with or without enteral tube feeding, has
Insertion not attempted 3 (1.3) been shown repeatedly for colorectal and major gynecologic
Insertion aborted after difficult 1 (0.4) surgery.1– 4 It does not seem to be dependent either on the
dissection kind or quantity of food taken, and high-caloric sip-feeds
Catheter-related complications Proportion of inserted within the first 5 postoperative days have not been shown to
catheters n ⫽ 223 improve clinical recovery.23 Our aim was to investigate a
Catheter-related infection 3 (1.3) regimen of allowing early food at will in major upper GI
Leakage around insertion site 3 (1.3) surgery as it is practiced in colorectal surgery. The amount of
Intra-abdominal leak verified 2 (0.9) food taken during the first 5 days was not an issue in this trial
Intra-abdominal leak suspected 1 (0.4) as we wanted to investigate the assumed hazards of allowing
Problematic removal 3 (1.3) any food at all, ie, to challenge the tradition of nil-by-mouth.
Other complaints 4 (1.8) Hence, we are not claiming that allowing normal food at will
Total 16 (7.2) confers nutritional benefits in the early postoperative period
Reoperation caused by catheter 3 (1.3) or that selective use of nutritional support is never needed. On
Unscheduled removal (within 31/223 (13.9) the other hand, our results indicate that allowing these pa-
first 5 postoperative days) tients to eat normal food is beneficial and that routine enteral
Clogged 5 feeding by NCJ, although probably providing more calories,
Accidentally pulled or fallen-out 13 is associated with a poorer overall outcome.
Caused by catheter-related 6 Parenteral nutrition support was used selectively in
complaint/complication
both study groups at the discretion of the surgeon based on a
Not caused by catheter 7
clinical judgment of inadequate nutritional intake. The use of
Patient-requested removal 3
parenteral nutrition did not indicate that the group allowed
Removed at reoperation for 3
other reason
food at will was clinically underfed compared with the ETF
Unknown 1
group. Furthermore, we believe that eating food, through the
activation of normal digestive reflexes, has an important
impact on gut recovery, which is central in overall recovery had postdischarge complications and they tended to lose less
after GI surgery.24,25 A multimodal care regimen as used in weight. Length of stay, in an unblinded trial without pre-
this trial (no tubes, thoracic epidural analgesia, avoidance of defined discharge criteria, is a poor outcome parameter and
parenteral opioids, early mobilization, etc) is probably crucial may not reflect recovery. However, alongside the data above,
to exploit the benefits of allowing early food.24,25 In our it might indicate that allowing normal food at will enhances
opinion, the literature somewhat erroneously focuses on the recovery.
different enteral feeding formulas and unjustly disregards the
benefits of eating food as such.
The total number of major complications was signifi- Limitations
cantly lower in the group allowed normal food at will. We Separate analysis of the subgroups in this trial did not
consider such pooling of complications, although widely used reveal any significant increase in morbidity for those allowed
in the medical literature, to be less robust than our primary normal food at will. Although the trial was not powered for
end point: the rate of patients suffering one or more major such stratification, the results of the largest subgroups
complications. We therefore refrain from emphasizing this (Whipple procedures, total and subtotal gastrectomies, and
further. On the other hand, it is important to note that the hepatic resections) accorded with this. Because the group of
catheter-related complications in this trial were similar or operations was heterogeneous, our conclusions may still not
fewer than in most of the recently reported series.14 –16,26 This be automatically valid for each subgroup. Further trials are
indicates that our results are not attributable to errors of needed for each type of surgery to assess the optimal post-
catheter use or technique of insertion. operative practice and there may be subgroups that would
benefit from nasogastric tubes and nil-by-mouth for some
Recovery days. The paucity of esophageal resections in our material
Allowing normal food at will improved recovery of gut reflects the reluctance of the participating surgeons to include
function as indicated by the time to first flatus, which was these patients in the trial. Particular features of these opera-
significantly shorter for those who were allowed to eat. Also, tions might indicate that they should be treated differently
significantly fewer patients in the group allowed normal food from the others.13
Approximately 17 percent of the patients were lost to Bjørk and Atle Bernstein (Hospital of Arendal), Jon Sen (now
postdischarge follow-up at 8 weeks. These were evenly East Iceland Regional Hospital, Iceland), Jorunn Sandvik
distributed between the treatment groups, and there were no (now Hospital of Ålesund), and Kristian Storli (now St. Olavs
deaths among them. Major surgical complications would Hospital, Trondheim University Hospital). In addition the
have prompted immediate readmittance to the operating trial- authors thank all the doctors and nurses at the 5 participating
hospital. We cannot exclude the possibility of minor nonsur- centers that made the trial possible by including patients and
gical complications among those lost to follow-up. registering outcomes. Bjørn-Odvar Eriksen, MD, PhD and
Our pretrial estimated rate of major complications was the staff at the Clinical Research Centre, University Hospital
well below that of the trial population. This reduces the risk Northern Norway, Tromsø, provided important assistance in
of type-II error as does our pretrial power set at 0.90. It may methodological discussions, the construction of the database,
be argued that our target reduction (by 50%) is large and that and in pretrial and post-trial statistical calculations. Profes-
a smaller difference may well be clinically important. Clearly, a sors Olle Ljungqvist (Ersta Hospital and Karolinska Institu-
type-II error for a smaller but clinically important difference tet, Stockholm, Sweden), Barthold Vonen (University Hospi-
cannot be excluded. The tendency of our results, however, is tal Northern Norway), and senior consultant Cornelis H.C.
uniform. This might indicate that a nonshowing, smaller differ- Dejong, MD, PhD (University Hospital Maastricht, The
ence would be that of a decrease, not an increase, in major Netherlands) kindly read the manuscript and offered con-
morbidity associated with allowing normal food at will. structive comments. The authors also thank Fresenius Kabi
AS, Norway for providing the enteral nutrition formulas.
Implications and Recommendations
This is the first large randomized trial to challenge the REFERENCES
existing routine of nil-by-mouth in upper GI surgery. Periop- 1. Lewis SJ, Egger M, Sylvester PA, et al. Early enteral feeding versus “nil
erative practice for hepatic and pancreatic surgery and other by mouth” after gastrointestinal surgery: systematic review and meta-
major upper GI procedures may include early removal of the analysis of controlled trials. BMJ. 2001;323:773–776.
nasogastric tube and food at will at some centers, but the 2. Han-Geurts IJ, Jeekel J, Tilanus HW, et al. Randomized clinical trial of
patient-controlled versus fixed regimen feeding after elective abdominal
literature in English is scarce.13 Thus, our results may have surgery. Br J Surg. 2001;88:1578 –1582.
important clinical implications for a large group of patients. 3. Schilder JM, Hurteau JA, Look KY, et al. A prospective controlled trial
We believe that the customary regimen of nil-by-mouth after of early postoperative oral intake following major abdominal gyneco-
major upper GI surgery is unnecessary in most cases. Future logic surgery. Gynecol Oncol. 1997;67:235–240. 关See comments兴.
trials are needed to establish the optimal clinical pathway for 4. Steed HL, Capstick V, Flood C, et al. A randomized controlled trial of
early versus “traditional” postoperative oral intake after major abdomi-
each of the main types of operations reported in this study, nal gynecologic surgery. Am J Obstet Gynecol. 2002;186:861– 865.
but we believe that the majority of these patients should be 5. Brodner G, Van Aken H, Hertle L, et al. Multimodal perioperative
allowed normal food at will. There is no evidence that it is management– combining thoracic epidural analgesia, forced mobiliza-
inferior to the current routines, it may enhance recovery, and tion, and oral nutrition–reduces hormonal and metabolic stress and
improves convalescence after major urologic surgery. Anesth Analg.
there is a potential for reducing costs through less expenditure 2001;92:1594 –1600.
on catheters and feeding formulas, and possibly through 6. Mukherjee D. “Fast-track” abdominal aortic aneurysm repair. Vasc
shorter lengths of stay. Future trials should focus on the role Endovascular Surg. 2003;37:329 –334.
of nutrition support as an adjunct to food at will, not as an 7. Lassen K, Dejong CH, Ljungqvist O, et al. Nutritional support and oral
alternative. intake after gastric resection in five northern European Countries. Dig
Surg. 2005;22:346 –352.
8. Tanaka M. Gastroparesis after a pylorus-preserving pancreatoduodenec-
CONCLUSIONS tomy. Surg Today. 2005;35:345–350.
9. Suehiro T, Matsumata T, Shikada Y, et al. Accelerated rehabilitation
This trial has shown that allowing patients to eat normal with early postoperative oral feeding following gastrectomy. Hepato-
food at will from the first day after major upper GI surgery gastroenterology. 2004;51:1852–1855.
does not increase morbidity compared with traditional care 10. Hirao M, Tsujinaka T, Takeno A, et al. Patient-controlled dietary
with nil-by-mouth and enteral feeding. It is the first large schedule improves clinical outcome after gastrectomy for gastric cancer.
World J Surg. 2005;29:853– 857.
randomized trial to challenge the existing routine of nil-by- 11. Atkins BZ, Shah AS, Hutcheson KA, et al. Reducing hospital morbidity
mouth in these patients. and mortality following esophagectomy. Ann Thorac Surg. 2004;78:
1170 –1176.
12. Bisgaard T, Kehlet H. Early oral feeding after elective abdominal
ACKNOWLEDGMENTS surgery–what are the issues? Nutrition. 2002;18:944 –948.
The authors thank the following individuals who par- 13. Lassen K, Revhaug A. Early oral nutrition after major upper gastroin-
ticipated in various phases of the trial: senior consultant Tore testinal surgery: why not? Curr Opin Clin Nutr Metab Care. 2006;9:
Gauperaa, MD, PhD, implemented the trial in Arendal and 613– 617.
14. Ryan AM, Rowley SP, Healy LA, et al. Post-oesophagectomy early
offered important comments on the manuscript. Implementa- enteral nutrition via a needle catheter jejunostomy: 8-year experience at
tion in the other centers was totally dependent on Professors a specialist unit. Clin Nutr. 2006;25:386 –393.
of Surgery Asgaut Viste (Haukeland University Hospital, 15. Chin KF, Townsend S, Wong W, et al. A prospective cohort study of
Bergen), Jon Arne Søreide (Stavanger University Hospital, feeding needle catheter jejunostomy in an upper gastrointestinal surgical
unit. Clin Nutr. 2004;23:691– 696.
Stavanger); and Helge Myrvold and Jon-Erik Grønbech (St. 16. De Gottardi A, Krahenbuhl L, Farhadi J, et al. Clinical experience of
Olavs Hospital, Trondheim University Hospital). For some feeding through a needle catheter jejunostomy after major abdominal
time, trial inclusion locally was done by surgeons Glenn operations. Eur J Surg. 1999;165:1055–1060.
17. Sarr MG. Appropriate use, complications and advantages demonstrated 22. Perioperative total parenteral nutrition in surgical patients. The Veterans
in 500 consecutive needle catheter jejunostomies. Br J Surg. 1999;86: Affairs Total Parenteral Nutrition Cooperative Study Group. N Engl
557–561. J Med 1991;325:525–532. 关see comments兴.
18. Myers JG, Page CP, Stewart RM, et al. Complications of needle catheter 23. MacFie J, Woodcock NP, Palmer MD, et al. Oral dietary supplements in
jejunostomy in 2,022 consecutive applications. Am J Surg. 1995;170: pre- and postoperative surgical patients: a prospective and randomized
547–550. clinical trial. Nutrition. 2000;16:723–728.
19. Han-Geurts IJ, Verhoef C, Tilanus HW. Relaparotomy following com- 24. Kehlet H, Holte K. Review of postoperative ileus. Am J Surg. 2001;
plications of feeding jejunostomy in esophageal surgery. Dig Surg. 182:3S–10S.
2004;21:192–196. 25. Gatt M, Anderson AD, Reddy BS, et al. Randomized clinical trial of
20. Bozzetti F, Braga M, Gianotti L, et al. Postoperative enteral versus multimodal optimization of surgical care in patients undergoing major
parenteral nutrition in malnourished patients with gastrointestinal can- colonic resection. Br J Surg. 2005;92:1354 –1362.
cer: a randomised multicentre trial. Lancet. 2001;358:1487–1492. 26. Lobo DN, Williams RN, Welch NT, et al. Early postoperative jejunostomy
21. Woodcock N, MacFie J. Optimal nutrition support (and the demise feeding with an immune modulating diet in patients undergoing resectional
of the enteral versus parenteral controversy). Nutrition. 2002;18: surgery for upper gastrointestinal cancer: a prospective, randomized, con-
523–524. trolled, double-blind study. Clin Nutr. 2006;25:716 –726.