Professional Documents
Culture Documents
Nutrition http://pen.sagepub.com/
Comparison of the Safety of Early Enteral vs Parenteral Nutrition in Mild Acute Pancreatitis
Stephen A. McClave, Lisa M. Greene, Harvy L. Snider, Laszlo J.K. Makk, William G. Cheadle, Nancy A. Owens,
Larry G. Dukes and Linda J. Goldsmith
JPEN J Parenter Enteral Nutr 1997 21: 14
DOI: 10.1177/014860719702100114
Published by:
http://www.sagepublications.com
On behalf of:
Additional services and information for Journal of Parenteral and Enteral Nutrition can be found at:
Subscriptions: http://pen.sagepub.com/subscriptions
Reprints: http://www.sagepub.com/journalsReprints.nav
Permissions: http://www.sagepub.com/journalsPermissions.nav
Citations: http://pen.sagepub.com/content/21/1/14.refs.html
What is This?
STEPHEN A. McCLAVE, MD; LISA M. GREENE, RN; HARVY L. SNIDER, MD; LASZLO J. K. MAKK, MD; WILLIAM G. CHEADLE, MD;
NANCY A. OWENS, RD; LARRY G. DUKES, RPH; AND LINDA J. GOLDSMITH, PHD
From the Departments of Medicine and Surgery, University of Louisville School of Medicine and Veterans’ Affairs Medical Center, Louisville, Kentucky
ABSTRACT. Background: This prospective study was designed to than that for TEN vs $761, respectively,
($3294 p < .001). Mean serial
compare the safety, efficacy, cost, and impact on patient outcome of Ranson criteria, APACHE III, and MOF scores recorded every 2 to 3
early total enteral nutrition (TEN)
vs total parenteral nutrition (TPN) days decreased in the TEN group, whereas those in the TPN group
in acute pancreatitis. Methods: Patients admitted with acute pancreati- increased. Only the difference in the third Ranson criteria (mean 6.3
tis or an acute flare of chronic pancreatitis, characterized by abdomi- days after admission) for the TEN and TPN groups (0.5 vs 2.8, respec-
nal pain and elevated serum amylase and lipase, were randomized to tively) reached statistical significance (p .002). Stress-induced hy-
=
receive either isocaloric and isonitrogenous TEN (via a nasojejunal perglycemia was worse in the TPN group, as serum glucose levels
feeding tube placed endoscopically) or TPN (via a central or periph- increased significantly over the first 5 days of hospitalization (p < .02),
eral line) started within 48 hours of admission. Results: Thirty patients whereas those in the TEN group showed no significant change. An
were studied over 32 admissions (TEN given on 16 and TPN on 16) for exacerbation of pancreatitis, occurring in one TEN patient when the
acute pancreatitis. There were no differences on admission in mean nasojejunal tube was dislodged into the stomach, resolved after place-
age, Ranson criteria, multiple organ failure score (MOF), or APACHE ment back in the jejunum. Three patients who became asymptomatic
III score between TEN and TPN groups. Although slower to approach and normalized amylase on TEN flared upon advancing to diet by
goal feeding over the first 72 hours of admission, TEN patients re- mouth. Conclusions: TEN for acute pancreatitis is as safe and effective,
ceived 71.3% goal calories by day 4 vs 85.2% for TPN patients (not sig- but is significantly less costly than TPN. Compared with TPN, TEN
nificant). There were no deaths and no differences between groups in may promote more rapid resolution of the toxicity and stress response
serial pain scores, days to normalization of amylase, days to diet by to pancreatitis. TEN via jejunal feeding should be used preferentially
mouth, serum albumin levels, or percent nosocomial infection. How- in this disease setting. (Journal of Parenteral and Enteral Nutrition
ever, the mean cost of TPN per patient was over four times greater 014-020, 1997)
21:
Acute pancreatitis is a hypermetabolic, hyperdynamic Benefits from the use of total enteral nutrition (TEN)
disease process that creates a catabolic stress state pro- have been noted in a number of other disease states, such
moting a systemic inflammatory response and nutritional as burns, trauma, and sepsis.11 In comparison with TPN,
deterioration. Nutrition support plays an important role use of TEN reduces nosocomial infection, multiple organ
in the adjunctive management of these patients. Total failure (MOF), and length of hospitalization .5, 1,11 Recent
parenteral nutrition (TPN) has been the standard of prac- evidence suggests that use of TEN in pancreatitis may be
tice for providing exogenous nutrients, while avoiding feasible. In a dog model, Ragins et a19 showed that the site
pancreatic stimulation. Two prospective nonrandomized in the gastrointestinal (GI) tract to which feedings are
trials in pancreatitis suggested a benefit from TPN given delivered determines whether the pancreas is stimulated
early (within 72 hours of admission) to maintain positive and that jejunal feedings result in negligible increases in
nitrogen balance.2,3 However, the only prospective random- enzyme, bicarbonate, and volume output from the pan-
ized trial that evaluated the impact of early TPN in acute creas. In both animal and human studies, elemental or
pancreatitis showed that the group receiving early TPN semi-elemental small peptide formulas have been shown
had significantly longer hospitalization (16 vs 10 days, P < to be well tolerated and efficiently absorbed in a gut lu-
.04) and a higher rate of catheter sepsis (10.5% vs 1.5%, p = menal environment with little or no pancreatic enzyme
.003) compared with the control group, which received secretion. 10-13 Case series have documented that TEN could
only analgesia and IV fluid resuscitation.’ The results of be used safely as transitional feeding in patients who had
this study suggested that the widespread use of TPN might peaked and had begun to resolve the inflammatory re-
not be warranted.~4 sponse associated with pancreatitis. 14-21
The potential advantages of TEN over TPN and the fea-
sibility that TEN administered via jejunal feedings might
be well tolerated prompted us to set up this prospective
Received for publication, May 20, 1996.
randomized trial comparing early TEN to TPN in acute
Accepted for publication, September 3, 1996.
Correspondence: Stephen A. McClave, MD, Division of Gastroenterology/ pancreatitis. The objectives of this study were to deter-
Hepatology, University of Louisville School of Medicine, Louisville, KY mine the safety of TEN in a setting of acute pancreatic
40292. inflammation and to determine any benefit of TEN over
14
TPN with regard to cost, efficacy, and impact on patient central line placement was assigned to each TPN patient
outcome in pancreatitis patients. regardless of whether a peripheral or central route was
used. Each time a nasoenteric tube or IV line had to be
MATERIALS AND METHODS replaced for nutrition support, the $600 charge for endo-
Patients admitted to the hospital with acute pancreati- scopic tube placement or $500 charge for central line place-
ment was added to the patient’s cost of nutrition support.
tis or an acute flare of chronic pancreatitis, characterized
The initial APACHE III score calculated on admission was
by abdominal pain with elevated amylase and lipase, were determined by adding the scores for neurologic abnormali-
entered in this study. Patients were excluded if they had
any evidence of short bowel syndrome, Crohn’s disease,
ties, acid-base disturbances, and age and chronic health
evaluation to the acute physiological score.~-’3 Serial
major pancreatic resection, or failure to start TEN or TPN APACHE III scores were determined using only the acute
within 48 hours of admission. After entry into the study,
patients were excluded if they failed to adhere to dietary physiological score.23 Days to normalization of amylase was
a clinical endpoint irrespective of lipase (lipase could still
restrictions or to the protocol terms for enteral tube place-
be elevated). Thirteen specific nosocomial infections were
ment.
defined (see Appendix B). No patient received empiric
Upon admission patients were prospectively random- antibiotics upon admission for acute pancreatitis. Any an-
ized to receive either TEN or TPN. Both groups were
tibiotics given were chosen to treat specific nosocomial
placed on isocaloric-isonitrogenous feedings (with simi- infections according to bacterial cultures and sensitivities.
lar carbohydrate-to-fat ratio) designed to provide a goal
For purposes of statistical analysis, chi-square and t tests
amount of 25 kcal and 1.2 g protein/kg per day. The TPN
were used to verify random assignment with regard to age
was infused through a central or peripheral line. The TEN
and sex. For comparisons of Ranson criteria, MOF scores,
group received full strength Peptamen (Clintec Nutrition APACHE III scores, length of hospitalization, and days to
Company, Deerfield, IL) infused through a nasojejunal tube normalization of amylase, the two-group independent
placed endoscopically down into the jejunum. Routine
endoscopic placement was used instead of fluoroscopic sample t test was used. Kolmogorof-Smirnof test did not
placement to assure placement of the feeding tube at or reject the hypothesis of normality in these data. Compari-
sons were made using the Mann-Whitney U tests for the
below the level of the ligament of Treitz. The study period
was defined as that time interval from admission to the following measures: days to diet by mouth, days in the in-
tensive care unit, percent of goal calories achieved, initial
emergency room to advancement to diet by mouth. Spe- APACHE III score, and cost of nutrition support. These
cifically, days to diet by mouth was defined by the point at variables were judged not normal through the Kolmogorof-
which the patient showed signs of clinical resolution (no
Smirnof test of the data. Repeated-measures analysis of
pain or nausea and vomiting with amylase decreasing for variance was used to test for overall differences or differ-
48 hours), and the patient was advanced to clear liquids.
ences in trends between groups. Measures tested were
Patients were then followed peripherally after that until
time of discharge. amylase levels, lipase levels, glucose levels, albumin lev-
Throughout the study patients were monitored by a va- els, and pain scores over time.
riety of parameters. Safety parameters included serial Sample size was calculated before the study based on
amylase, lipase, serum glucose, serum albumin, subjec- achieving 80% power with a significance level of .05. A
tive pain score, Ranson criteria, APACHE III score, and a planned sample size of 20 per group afforded an 80% power
to detect a difference of 1.8 days in length of hospitaliza-
MOF score. The Ranson criteria, APACHE III criteria, and
MOF score were calculated within 48 hours of admission tion, assuming a standard deviation of 2.0. Similar calcula-
tions were completed for days to normalization of amy-
and then were repeated every 2 to 3 days through the study
period. The subjective pain score (determined daily) was lase-lipase and days to diet by mouth. This sample size also
afforded an 80% power to detect a $1000 difference in cost,
defined by the following scale: 0, no pain; 1-2, tolerable
pain; 3-4, mild pain; 5-6, moderate pain; 7-9, severe pain; assuming a standard deviation of 1100.
This study was approved by the Human Studies Com-
10 worst pain imaginable. The MOF score was adapted
mittee at the University of Louisville School of Medicine,
from that of Marshall et all’ (see Appendix A). Additional
and the Human Studies Subcommittee at the Veterans’
safety parameters included incidence of nosocomial in- Affairs Medical Center, Louisville, Kentucky.
fection and mortality. Efficacy parameters included per-
cent of goal calories (25 kcal/kg/d) achieved, days to ad-
RESULTS
vancement to diet by mouth, and length of hospitalization.
Because true cost of raw materials, cost to third-party .
TABLE I
Demographic data on study patients at entry into the study
Values are means ± SEM. TEN, total enteral nutrition; TPN, total
parenteral nutrition; EtOH, ethyl alcohol; APACHE IH; MOF, multiple
organ failure. FiG. 2. Serial mean amylase and lipase levels (-SEM).
Downloaded from pen.sagepub.com at UNIVERSITE LAVAL on June 24, 2014
17
TABLE II
Clinical study endpoin£
Values are means ± SEM. For abbreviations, see Table I. LOH, length
of hospitalization; *p < .005.
decreased steadily over time for the TEN patients. For the
TPN patients, slower decreases in actual Ranson criteria
for some patients caused the mean value to actually in- FIG. 3. Serial mean pain scores (±SEM).
crease over time, such that the third mean Ranson criteria
determined for the TPN group was significantly higher than
that for the TEN group (2.8 vs 0.5, p .002). Although a
=
TABLE III
Specific nosocomial infections per patient basis for each group
-
F!G. 4. Serial scores over time for Ranson criteria (RC), analyzed by the
For abbreviations, see Table I. two-group independent sample t test.
Downloaded from pen.sagepub.com at UNIVERSITE LAVAL on June 24, 2014
18
I~G. 5. Serial mean serum glucose levels (±SEM), as evaluated by repeated- FIG. 7. Effect of nasoenteric tube displacement from jejunum to stomach
measures analysis of variance. (SB, small bowel; STOM, stomach).
inflamed to mount a response). No antibiotics were ever put from the pancreas, duodenal infusions increased only
started, and the patient was ultimately discharged on day 16. the volume of pancreatic secretion. Jejunal feedings
caused no significant change in any aspect of pancreatic
DISCUSSION secretion compared with saline control.9 Other studies in-
Considerable evidence exists that the degree to which volving either animals25 or humanslo,l2°26 have all shown de-
creases in pancreatic secretion in response to jejunal
the pancreas is stimulated by enteral feedings is deter-
mined by the site in the GI tract at which the feedings are feedings. Only one study in humans27 showed that a high
rate of amino acid infusion into the jejunum increased
infused. 14 Feeding by mouth increases pancreatic secre-
tion by invoking three levels of stimulation via the cepha- pancreatic stimulation. A study in dogs by Cassim2S showed
that jejunal feedings increased volume and bicarbonate
lic, gastric, and intestinal phases. Multiple factors at each output from the pancreas, but caused no increase in pro-
of these levels are involved in the stimulation of the pan-
tein enzyme output. At least eight studies are reported in
creas. These factors include vagal stimuli, chemical stimuli
the literature in which the infusion of enteral nutrients
(gastric acid, gastric distention, intestinal proteins, and into the jejunum were used as transition feeding in pan-
fat), and hormonal stimuli (gastrin, secretin, vasoactive creatitis patients in whom the inflammation had clearly
intestinal polypeptide, and cholecystokinin). The more
distal the site of feeding in the GI tract, the fewer the num- peaked and begun to subside. 14-21
In our study, there was concern that even if jejunal
ber of these factors or levels of stimulation that are in-
volved. A lower limit of stimulation appears to exist at the feedings did cause some pancreatic stimulation, that an
adverse clinical response might not be evident simply be-
level of the duodenum, such that feedings infused into the
cause the patients had only a mild degree of predominantly
jejunum result in minimal or negligible stimulation.24 In a alcoholic pancreatitis. Anecdotal experience in the three
1973 animal study,9 Ragins compared an elemental prod-
TEN patients who failed to tolerate advancement to diet
uct (Vivonex) to saline control infused into the stomach,
by mouth and the one TEN patient in whom the jejunal
duodenum, and jejunum of dogs. Whereas gastric feedings tube became displaced into the stomach provided reas-
increased volume, bicarbonate, and protein enzyme out-
surance against such concern. Clearly in these four pa-
tients, jejunal feedings resulted in negligible stimulation
and appeared to put the pancreas at rest, whereas feedings
either into the mouth or stomach led to obvious stimula-
tion of a potentially inflammed gland and clear-cut exac-
erbation of pancreatitis.
In a number of disease processes, use of TEN has been
shown to reduce or contain the stress response when com-
pared with use of TPN in patients with equivalent injury
or degree of critical illness. In disease processes includ-
ing burns, trauma, and major surgery, use of TEN in com-
parison with TPN results in higher visceral protein lev-
els,19 reduction in stress-induced hyperglycemia (or the
amount of insulin required to keep blood glucose levels
below a certain level),29 reduction in energy expendi-
ture,29,30 and a more rapid return of elevated cytokinins to
baseline levels.&dquo; Although both groups on entry into our
FiG. 6. Failure to tolerate progression to oral diet. study had a similar degree of severity of pancreatitis, the
Downloaded from pen.sagepub.com at UNIVERSITE LAVAL on June 24, 2014
19
biliary secretions in man. Gastroenterology 61:686-92, 1971 following multiple trauma. JPEN (submitted)
28. Cassim MM, Allardyce DB: Pancreatic secretion in response to jeju- 32. Moore EE, Moore FA: The role of the gut in provoking the systemic
nal feeding of elemental diet. Ann Surg 180:228, 1974 inflammatory response. J Crit Care Nutr 2:9-15, 1994
29. Moore FA, Moore EE, Jones TN, et al: TEN versus TPN following 33. De Beaux, AC, Plester C, Fearon KCH: Flexible approach to nutri-
major torso trauma—reduced septic morbidity. J Trauma 29:916-923, tional support in severe acute pancreatitis. Nutrition 10:246-249, 1994
1989 34. IhseI, Andersson R, Andren-Sandberg A, et al: Conservative treat-
30. Saito H, Trocki O, Alexander JW, et al: The effect of route of nutrient ment in acute pancreatitis. Ann Ital Chir 66:181-185, 1995
administration on the nutritional state, catabolic hormone secretion, 35. Povoski SP, Nussbaum MS: Nutrition support in pancreatitis: Fertile
and gut mucosal integrity after burn injury. JPEN 11:1-7, 1987 ground for prospective clinical investigation. NCP 10:43-44, 1995
31. Charash WE, Kearney PA, Annis KA, et al: Early enteral feeding is 36. Bodoky G, Harsanyi L, Pap A, et al: Effect of enteral nutrition on
associated with an attenuation of the acute phase/cytokine response exocrine pancreatic function. Am J Surg 161:144-148, 1991
APPENDIX A
Multiple organ failure scoring
Adapted from Marshall.22 CNS, central nervous system; DTH, delayed-type hypersensitivity.
APPENDIX B