Enteral versus parenteral nutrition: a pragmatic study

APPLIED NUTRITIONAL INVESTIGATION

Enteral Versus Parenteral Nutrition: A Pragmatic Study Nicholas P. Woodcock, FRCS, Dietmar Zeigler, FRCS, M. Diane Palmer, BSc, RGN, Paul Buckley, FRCA, Charles J. Mitchell, FRCP, and John MacFie, MD, FRCS From the Combined Gastroenterology Unit, Scarborough Hospital, Scarborough, UK Controversy persists as to the optimal means of providing adjuvant nutritional support. The aim of this study was to compare enteral nutrition (EN) and parenteral nutrition (TPN) in terms of adequacy of nutritional intake, septic and nonseptic morbidity, and mortality. This was a prospective pragmatic study, whereby the route of delivery of nutritional support was determined by the attending clinician’s assessment of gastrointestinal function. Patients considered to have inadequate gastrointestinal function were given TPN (group 1), while those deemed to have a functioning gastrointestinal tract received EN (group 2). Patients in whom there was reasonable doubt as to the adequacy of intestinal function were randomized to receive either TPN (group 3) or EN (group 4). The trial setting was a large district general hospital with a dedicated nutrition team. A total of 562 patients were included in the study (331 males; median age 67 y). Gastrointestinal function on entry into the study was considered inadequate in 267 patients who were given TPN (group 1) and adequate in 231 whom received EN (group 2). There was clinical uncertainty about the adequacy of gut function in 64 patients (11.4%) who were randomized to receive either TPN (group 3, 32 patients) or EN (group 4, 32 patients). The incidence of inadequate nutritional intake was significantly higher in group 4 compared with group 3 (78.1% versus 25%, P ⬍ 0.001). Complications related to the delivery system and other feed-related morbidity were significantly more frequent in both EN groups compared with the respective TPN groups. EN was associated with a higher overall mortality in both nonrandomized and randomized patients. There were no significant differences observed in the incidences of septic morbidity between patients receiving TPN and those given EN. EN is associated with a higher incidence of inadequate nutritional intake, complications related to the delivery system, and other feed-related morbidity than TPN. There is no evidence from this study to support a difference between the two modalities in terms of septic morbidity. Patients in whom there is reasonable doubt as to the adequacy of gastrointestinal function should be fed by the parenteral route. Nutrition 2001;17:1–12. ©Elsevier Science Inc. 2001 Key words: enteral nutrition, parenteral nutrition, morbidity, nutritional support, gut function

INTRODUCTION There is now good evidence that perioperative nutritional support is of benefit, particularly to patients with severe malnutrition,1 but the optimal route of delivery remains controversial. Most authorities recommend the use of enteral, as opposed to parenteral, nutrition. Enteral nutrition (EN) is claimed to be less expensive, safer, and more physiologic in the sense that it preserves gut barrier function.2– 4 Parenteral nutrition (TPN), in contrast, is more expensive and may have deleterious effects on the gut barrier that may predispose to septic morbidity.5 There is a substantial amount of experimental data, mainly derived from animal studies, to support these views.6,7 The situation is, however, less clear in clinical practice. There are a number of possible explanations that may account for the discrepancy in the results of the various comparative

N. P. Woodcock was supported by a grant from Fresenius Kabi, Uppsala, Sweden. Correspondence to: John MacFie, MD, FRCS, Consultant Surgeon, Scarborough Hospital, Woodlands Drive, Scarborough YO12 6QL, UK. Email: [email protected] Date accepted: Sep. 13, 2000. Nutrition 17:1–12, 2001 ©Elsevier Science Inc., 2001. Printed in the United States. All rights reserved.

clinical studies published to date. First, trials comparing enteral with parenteral nutrition are bedevilled by the fact that they often do not compare equivalent patient groups. Patients receiving TPN who cannot tolerate enteral feeding have “intestinal failure,” which may itself predispose the patient to an increased risk of morbidity.8 Conversely, patients who are able to tolerate enteral nutrition do not need TPN. Such inappropriate use of TPN is a feature of some previous studies and must impact results. Second, it is now well recognized that overfeeding and hyperglycemia, which occur more frequently with TPN, are potent causes of nutrition-related morbidity.9,10 Third, many of the published studies do not take into account nonseptic morbidity, particularly that related to delivery systems. Finally, the potential confounding effects of different protein and energy intakes or the effects of specific substrates, such as glutamine, are often not taken into consideration. In an attempt to overcome these problems, we have applied the statistical principle of pragmatism in a prospective study. Pragmatic trials are characterized by the fact that they appraise management protocols and reflect the natural variations that occur between patients, thereby enabling measurement of the effectiveness of a treatment, this being the benefit it produces in routine clinical practice.11 Results are always analyzed on an intention-totreat basis. In this way clinicians are permitted to follow their usual practice and appropriate endpoints are selected that reflect the aims of therapy. In this study the objectives of adjuvant nutritional support were assessed in terms of adequacy of nutritional intake 0899-9007/01/$20.00 PII S0899-9007(00)00576-1

2

Woodcock et al.

together with any associated morbidity. The options for providing nutritional support were enteral or parenteral nutrition. The technique used for either mode of delivery of nutrition was that which the attending clinician considered most appropriate for each individual patient. The indication for nutrition support was always 7 days of no or negligible oral intake, actual or anticipated. The decision as to whether or not a patient received enteral or parenteral nutrition was based entirely on a clinical assessment of gastrointestinal function. If gastrointestinal function was deemed to be adequate patients received EN, whereas if it was considered inadequate patients were given TPN. At the outset of this study we recognized that there would be occasions when the clinician would be genuinely uncertain as to the adequacy of gut function and therefore unable to decide between EN and TPN. This subgroup of patients all had in common the facts that they required adjuvant nutritional support and had doubtful gastrointestinal function. As such they constituted a homogeneous group in whom we felt it was both ethically and statistically valid to submit to prospective randomization. This study, therefore, comprises two distinct parts: the first and larger part is based on pragmatism and compares the outcomes of different feeding strategies, EN and TPN, both of which have the same aim. The second is a prospective randomized study, which also compares EN and TPN but, in contrast to many previous studies, seeks to do so in two homogeneous groups of patients. The primary objectives of this study were to compare EN and TPN in terms of adequacy of nutritional intake and morbidity, both septic and nonseptic. No attempt was made in this study to assess the efficacy of nutritional support with respect to alterations in parameters of nutritional status, nor its effects on gut barrier function.

PATIENTS AND METHODS Between November 1995 and July 1999 all patients aged 18 years or over who required adjuvant nutritional support at Scarborough Hospital were entered into the study. All nutritional support was provided under the auspices of the Combined Gastroenterology Unit. The indication for nutritional support in our unit is actual or anticipated inadequate oral nutritional intake for 7 days or more. There were no clinical exclusion criteria for study entry. Approval for the study was obtained from the Scarborough Locally Organised Ethical Committee. The recommendation of this committee was that written consent should not be a prerequisite for entry into the study as the therapeutic interventions accorded with routine clinical practice, and instigation of the study protocol did not influence clinical decision making. Patient Groups This study was based on the principle of clinical pragmatism, whereby the consultant physician or surgeon looking after the patient was permitted to instigate that mode of nutritional support that he or she considered appropriate. This decision was based on a clinical assessment of intestinal function, which in all cases included inquiry as to bowel habit and passage of flatus together with clinical examination. No strict criteria were imposed on this decision, which was left entirely to the individual clinician’s discretion. As such, other factors such as the patient’s underlying condition or anticipated speed of recovery from surgery or other intervention were taken into account. Those patients deemed to have permanent or temporary intestinal failure or in whom bowel rest was required were given TPN (group 1). If the clinician felt that the patient had a functioning gastrointestinal tract, the patient was given EN (group 2). Patients in whom there was reasonable doubt as to the adequacy of intestinal function were randomized to receive either TPN (group 3) or EN (group 4). Randomization was achieved by the drawing of a numbered sealed envelope, with the designated treatment groups generated by random order sequence.

Nutrition Volume 17, Number 1, 2001 Nutritional and Physiologic Status On entry into the study the body mass index (BMI) was calculated. In addition, serum albumin was recorded and the nutritional risk index (NRI) was calculated according to the equation (1.519 ⫻ serum albumin) ⫹ (0.417 ⫻ % usual body weight). According to this calculation, patients are categorized as well nourished (score ⬎ 98.5), mildly/moderately malnourished (83.5–98.5), or severely malnourished (⬍ 83.5).12 A physiologic score was also calculated using the POSSUM (Physiological and Operative Severity Score for the Enumeration of Morbidity and Mortality) scoring system,13 according to parameters measured at the time of commencement of feeding. This score has been validated as an accurate predictor of risk of morbidity and mortality following surgery. In the context of the present study it was used as a comparative guide to severity of illness. Feeding Methods Patients receiving TPN (groups 1 and 3) were fed by the peripheral route through an 18 F short Teflon catheter wherever possible. However, patients with a central venous catheter in situ or with poor peripheral access were fed by the central route. Our preferred method of peripheral feeding (PPN) involves cyclical infusion with rotation of venous access sites.14 PPN was administered over a period of 12 hours, usually overnight. Infusion sites were inspected twice daily. Central venous access was obtained in the majority of patients by insertion of an infraclavicular subclavian line using full asepsis in the operating room. One lumen of the line was used exclusively for the provision of nutritional support. A small number of patients were fed via an implantable venous access device (Port-a-Cath, SIMS Deltec Inc., St. Paul, MN, USA). Central feeding was administered as a continuous infusion over 24 hours in the majority of patients. A commercially prepared TPN preparation was used for all patients (KabiMix 9, Fresenius Kabi, Uppsala, Sweden). The aim with TPN, whether peripheral or central, was to provide 30 kcal 䡠 kg⫺1 䡠 d⫺1 non-protein energy and 9 gN/d. In addition, Intensive Care Unit (ICU) patients with a nasogastric tube in situ received small amounts (5–10 mL/h) of polymeric enteral feed, increased according to patient tolerance. The standard method of delivery of EN (groups 2 and 4) was by way of a fine bore silicone feeding tube (SILK, Corpaq, Wheeling, IL, USA). Where appropriate, as decided by the attending clinician, EN was alternatively administered via a percutaneous endoscopic gastrostomy (PEG), a surgically placed gastrostomy tube, or a feeding jejunostomy using a 16 Ch Foley catheter. Similar target intakes of non-protein energy (30 kcal 䡠 kg⫺1 䡠 d⫺1) and protein (9 gN/d) were provided using a commercially available polymeric feed (Osmolite, Abbott Laboratories Ltd., Kent, UK). The initial rate of delivery was full strength feed at 30 mL/h, increasing stepwise to full intake over a period of 24 – 48 hours according to patient tolerance. During this build-up period tubes were aspirated every 6 hours. If less than 100 mL was aspirated, the aspirate was replaced and feeding continued, whereas if the volume of aspirate exceeded 100 mL, the aspirate was not replaced and the feed slowed or stopped temporarily. Feeds were usually given as a continuous infusion using a volumetric pump over a 24-hour period. The provision of all adjuvant nutritional support was overseen and closely monitored by a dedicated nutrition team. A senior dietician, specialist nutrition nurse, and research fellow were responsible for prescribing, monitoring, and recording intakes and complications on a twice-daily basis. All access sites for nutritional support were inspected twice daily. Outcome Measures Duration of feeding was recorded for all patients, as was the reason for cessation of therapy in each case. The volume of feed actually

Nutrition Volume 17, Number 1, 2001

Enteral Versus Parenteral Nutrition

3

FIG. 1. Group allocation.

received was recorded daily and expressed as a percentage of the target volume prescribed. For the purposes of this study, receipt of less than 80% of the target intake was deemed “inadequate.” If clinically indicated, patients were converted to an alternative route of feeding; however, such patients were always analyzed in the group to which they were originally allocated (intention-to-treat analysis). The design of this study permitted an objective assessment of the accuracy with which gastrointestinal function was assessed when initially deciding the route of nutritional support. Assessment was deemed to have been incorrect if patients were converted to an alternative route of feeding within 5 days of entry into the study. Septic morbidity was defined as the presence of recognized pathogens in body tissues that are normally sterile, confirmed by the results of culture and supported by clinical, radiologic, or hematologic evidence of infection: Chest infection isolation of pathogens in purulent sputum, with or without evidence of consolidation on chest x-ray Wound infection isolation of pathogens in pus or discharge from wound UTI bacteremia at a concentration of ⬎105/mL together with other signs of infection in case of a catheter specimen Line sepsis positive culture from line tip or discharge from an inflamed exit site PEG/jejunostomy site infection isolation of pathogens from an inflamed exit site Intraabdominal abscess isolation of pathogens from a collection requiring percutaneous or open drainage Septicemia isolation of pathogens in peripheral venous blood in the absence of an overt focus of infection All septic complications occurring between the time of commencement of nutritional support and discharge from hospital were recorded. All morbidity related to the delivery systems was recorded. In patients receiving TPN this included inadvertent removal or blockage of the feeding line, pneumothorax secondary to subclavian line insertion, and peripheral vein phlebitis. In the EN patients delivery system-related morbidity included dislodgement or blockage of the feeding tube and failure of nasogastric tube or PEG insertion. In addition, all other feed-related morbidity was recorded, such as fluid overload and metabolic sequelae (eg, hyperglycemia, hyperkalemia) necessitating cessation, suspension, or slowing of feed in

the TPN patients, and diarrhea (three or more loose stools per day), large nasogastric aspirates (⬎500 mL/d), vomiting and abdominal bloating/discomfort in those patients receiving EN. All-cause mortality during the same hospital admission was recorded. Statistical Analysis All parametric data are expressed as means (SD) and nonparametric data as medians (interquartile range). Comparisons between groups were made using Student’s t test or Mann-Whitney U test for quantitative data and the chi-square test for qualitative data. A P value of 0.05 or less was taken to signify a statistically significant difference. A sample size calculation demonstrated that 32 patients would be required in each of the two randomized groups to show a significant difference in the incidence of septic morbidity at the 5% significance level with a power of 80%, assuming an incidence of 45% in the TPN patients and 15% in the EN patients.15

RESULTS A total of 562 patients were entered into the study. Gastrointestinal function was considered inadequate by the attending clinician in 267 patients, who were prescribed TPN (group 1). EN was given to 231 patients in whom GI function was deemed to be adequate (group 2). There was doubt as to the adequacy of GI function in 64 patients (11.4%); these were therefore randomized, 32 patients to receive TPN (rTPN, group 3) and 32 to receive EN (rEN, group 4) (Fig. 1). The demographic data for each patient group is given in Table I. There were no sex differences between the groups. The median age of patients in group 1 was significantly lower than that in group 2 (64 versus 70 y, P ⬍ 0.01). There was no significant difference in age between the two randomized groups. Group 1 consisted predominantly of postoperative patients (n ⫽ 109, 40.8%), patients with inflammatory bowel disease (n ⫽ 65, 24.3%), and patients with other surgical diagnoses including gastric outlet obstruction (n ⫽ 14, 5.2%) and enterocutaneous fistula (n ⫽ 14, 5.2%). Group 2 contained a much higher proportion of patients with nonsurgical diagnoses, such as stroke (n ⫽ 60, 26.0%) and respiratory failure requiring mechanical ventilation (n ⫽ 25, 10.8%). There were no

4

Woodcock et al.

Nutrition Volume 17, Number 1, 2001 TABLE I. DEMOGRAPHIC DATA

Male:female ratio Median age in years (IQR) Diagnoses Postoperative 199 (35.4%) Acute upper GI Acute lower GI Elective upper GI Elective lower GI Small bowel resection Vascular Other Inflammatory bowel disease 75 (13.3%) Flare-up Top-up Postoperative Fistula Other Pancreatitis 31 (5.5%) Acute Chronic Non-surgical 163 (29.0%) Neurologic Respiratory failure Other Miscellaneous 94 (16.7%) Gastric outlet obstruction Fistula Bowel obstruction Preoperative Intraabdominal sepsis Esophageal perforation Multiple trauma Other

Group 1 (TPN)

Group 2 (EN)

Group 3 (rTPN)

Group 4 (rEN)

156:111 64 (49–74)

137:94 70 (58–76)

18:14 67 (52–75)

20:12 67 (44–75)

109 (40.8%) 14 27 4 13 11 17 23 65 (24.3%) 29 15 14 4 3 20 (7.5%) 13 7 9 (3.4%) 0 2 7 64 (24.0%) 14 14 8 5 5 3 1 14

65 (28.1%) 2 5 32 4 0 14 8 3 (1.3%) 0 1 2 0 0 1 (0.4%) 0 1 139 (60.2%) 79 25 35 23 (10.0%) 0 1 0 1 0 1 7 13

14 (43.8%) 1 3 0 2 1 5 2 4 (12.5%) 4 0 0 0 0 8 (25.0%) 8 0 4 (12.5%) 0 2 2 2 (6.3%) 1 0 0 0 0 0 0 1

11 (34.4%) 2 4 0 2 0 2 1 3 (9.4%) 3 0 0 0 0 2 (6.3%) 2 0 11 (34.4%) 0 7 4 5 (15.6%) 0 0 0 1 1 0 0 3

EN, enteral nutrition; IQR, interquartile range; rEN, randomized EN; TPN, parenteral nutrition; rTPN, randomized TPN.

significant differences between the two randomized groups (groups 3 and 4) with respect to their diagnoses. Nutritional and Physiologic Status Table II contains the details of nutritional parameters and POSSUM physiologic scores at the time of study entry for those patients in whom data are available. Overall, 31.2% of patients had lost 10% or more of their usual body weight, 44.5% were severely malnourished according to the Nutritional Risk Index (score of ⬍83.5), and 24.3% of patients had a BMI of 19 or less. Only 11.2% of subjects were well nourished according to the NRI. The incidences of 10% or more weight loss and severe malnutrition according to the NRI were both significantly higher in group 1 compared with group 2 (P ⬍ 0.001), but there was no significant difference in BMI. There were no significant differences observed between the two randomized groups for any of these parameters. The mean POSSUM physiologic score in group 2 patients was significantly higher than in group 1 (P ⬍ 0.001). There were no significant differences between the two randomized groups in terms of POSSUM scores. The median POSSUM score of the randomized patients (groups 3 and 4) was significantly higher than that of the nonrandomized patients (groups 1 and 2) (P ⬍ 0.01). A total of 210 subjects (37.4%) were in the ICU at the time of

entry into the study. There was a significantly higher proportion of group 2 patients in the ICU (109/231, 47.2%) compared with group 1 (63/267, 23.6%) (P ⬍ 0.001). There was no significant difference between the two randomized groups in terms of the proportion of patients in ICU (rTPN 21/32, 65.6% versus rEN 17/32, 53.1%, P ⬎ 0.05). A significantly higher percentage of ICU patients were randomized compared with non-ICU patients (38/ 210, 18.1% versus 26/352, 7.4%, P ⬍ 0.001).

Feeding Methods The peripheral route was used initially in 100/299 (33.4%) of all patients receiving TPN (groups 1 and 3). Thirty-four (34.0%) of these patients required subsequent conversion to central feeding owing to an inability to maintain adequate peripheral venous access. Of the patients receiving EN (groups 2 and 4), 189/263 (71.9%) were fed by nasogastric tube. Thirty (15.9%) of these patients subsequently required more prolonged enteral nutrition and were converted to PEG feeding, while a further 40 patients were fed using a PEG only. A total of 33 patients were fed using a surgically placed jejunostomy, all but 1 of whom had undergone an elective major gastrooesophageal resection (Fig. 2).

Nutrition Volume 17, Number 1, 2001

Enteral Versus Parenteral Nutrition

5

TABLE II. NUTRITIONAL AND PHYSIOLOGIC STATUS ON COMMENCING NUTRITIONAL SUPPORT

Weight loss Less than 10% UBW 10% or more UBW Mean serum albumin g/L (SD) Nutritional risk index Well nourished Mild/mod malnourished Severely malnourished Body mass index 19 or below 20 or above Mean (SD) Median POSSUM physiologic score (IQR)

Group 1 (TPN)

Group 2 (EN)

n ⫽ 208 121 (58.2%) 87 (41.8%) 30.3 (6.0) n ⫽ 232 21 (9.1%) 87 (37.5%) 124 (53.4%) n ⫽ 164 40 (24.4%) 124 (75.6%) 22.7 (5.3) 23 (19–29)

n ⫽ 180 146 (81.1%) 34 (18.9%) 31.8 (6.1) n ⫽ 198 30 (15.2%) 104 (52.5%) 64 (32.3%) n ⫽ 31 9 (29.0%) 22 (71.0%) 22.5 (5.6) 25 (20–31)

Group 3 (rTPN) n ⫽ 28 21 (75.0%) 7 (25.0%) 29.0 (8.5) n ⫽ 31 3 (9.7%) 13 (41.9%) 15 (48.4%) n ⫽ 15 2 (13.3%) 13 (86.6%) 27.2 (9.6) 26 (21–33)

Group 4 (rEN) n ⫽ 23 14 (60.9%) 9 (39.1%) 28.5 (5.8) n ⫽ 29 1 (3.4%) 13 (44.8%) 15 (51.7%) n⫽8 2 (25.0%) 5 (75.0%) 21.8 (5.6) 31 (23–38)

EN, enteral nutrition; IQR, interquartile range; rEN, randomized EN; TPN, parenteral nutrition; rTPN, randomized TPN.

There was no difference observed in the duration of feeding by the designated route between group 1 (median 7 d; IQR 5–13 d) and group 2 (median 8 d, IQR 3–18 d) (Fig. 3); however, the duration of feeding in the randomized patients was significantly longer in those receiving TPN (median 9 d, IQR 7–14 d) compared with those given EN (median 3 d, IQR 1-9 d) (P ⬍ 0.01).

The volume of feed received was expressed as a percentage of target intake for each patient. A mean value was then calculated for each of the four groups. This was significantly higher in group 1 compared with group 2 (93.8% versus 80.1%, P ⬍ 0.001) and in group 3 compared with group 4 (96.7% versus 54.1%, P ⬍ 0.001). There was a significantly higher incidence of patients receiving less than 80% of their target intake associated with EN compared with TPN in both the nonrandomized (EN 32.0% versus TPN

FIG. 2. Methods of delivery of nutritional support. (A) Parenteral nutrition; (B) enteral nutrition.

FIG. 3. Adequacy of nutritional intake. (A) Duration of feeding by designated route; (B) percentage of patients receiving less than 80% of target intake.

Adequacy of Nutritional Intake

6

Woodcock et al.

Nutrition Volume 17, Number 1, 2001 TABLE III.

TABLE IV.

REASONS FOR INADEQUATE NUTRITIONAL INTAKE (RECEIPT OF LESS THAN 80% OF TARGET INTAKE)

INCIDENCE OF SEPTIC MORBIDITY Nonrandomized patients

Parenteral nutrition Group 1 (TPN) Group 3 (rTPN) n ⫽ 22 n ⫽ 2 Failure to tolerate feed Fluid overload Hypernatremia Abnormal liver function Failure of feeding method No available access Confirmed/suspected line sepsis Mechanical problem with delivery system Other reasons Undergoing surgery Acute renal failure

10 (45%) 9 1 0 13 (59%) 4 6 3

2 (100%) 2 0 0 0 (0%) 0 0 0

5 (19%) 3 2

0 (0%) 0 0

Enteral nutrition

Failure to tolerate feed Large nasogastric aspirates Vomiting Diarrhea Aspiration Intestinal leak Failure of feeding method Tube pulled out/no available access Initial build up of feeding rate Infected PEG site External bleeding from PEG site Mechanical problem with delivery system Other reasons Undergoing surgery GI hemorrhage

Group 2 (EN) n ⫽ 74

Group 4 (rEN) n ⫽ 20

26 (35%) 18 5 3 2 1 48 (65%) 34

9 (45%) 8 1 0 0 0 14 (70%) 10

29 3 1 4

4 0 0 0

3 (2%) 2 1

0 (0%) 0 0

EN, enteral nutrition; GI, gastrointestinal; PEG, percutaneous endoscopic gastrostomy; rEN, randomized EN; TPN, parenteral nutrition; rTPN, randomized TPN.

8.2%, P ⬍ 0.001) and randomized patients (EN 62.5% versus TPN 6.3%, P ⬍ 0.001) (Fig. 3). Significantly more randomized EN patients received less than 80% of target intake compared with nonrandomized EN patients (62.5% versus 32.0%, P ⬍ 0.01). The reasons for inadequate nutritional intake in each group are illustrated in Table III. Twenty-nine patients (10.9%) in group 1 were changed from TPN to adjuvant EN. One patient was converted as a result of fluid overload and another because the patient inadvertently removed the central line. The remaining patients were each converted following a clinical decision that they were now able to tolerate enteral feeding. In group 3 five patients (15.6%) were converted from TPN to adjuvant EN. The reasons were deranged liver function tests in one patient and ability to tolerate enteral feeding in the other four. Conversion from EN to TPN was necessary in 15 patients (6.5%) in group 2 and 10 (31.3%) in group 4. Failure to

Incidence of septic complications Mean no. of complications per patient (⫾ SEM) Mean no. of complications per infected patient (⫾ SEM)

Group 1 (TPN)

Group 2 (EN)

P value

84/267 (31.5%) 0.44 ⫾ 0.05

81/231 (35.1%) 0.45 ⫾ 0.05

0.48, NS ⬎0.05, NS

1.40 ⫾ 0.07

1.30 ⫾ 0.07

⬎0.05, NS

Randomized patients

Incidence of septic complications Mean no. of complications per patient (⫾ SEM) Mean no. of complications per infected patient (⫾ SEM)

Group 3 (rTPN)

Group 4 (rEN)

P value

16/32 (50%) 0.75 ⫾ 0.16

10/32 (31.3%) 0.41 ⫾ 0.12

0.13, NS ⬎0.05, NS

1.50 ⫾ 0.16

1.30 ⫾ 0.15

⬎0.05, NS

EN, enteral nutrition; rEN, randomized EN; TPN, parenteral nutrition; rTPN, randomized TPN.

tolerate EN, most commonly signified by large nasogastric aspirates (n ⫽ 13 in group 2 and n ⫽ 7 in group 4), was the most frequent reason for conversion in both groups. Accuracy of Assessment of Gastrointestinal Function In this study, 267 patients were deemed to have inadequate gastrointestinal function and received TPN (group 1). Of these, 35 patients (13.1%) were converted to adjuvant EN or oral diet having received less than 5 d of TPN. These patients represent treatment “failures” in that clinical assessment of gastrointestinal function was incorrect resulting in the employment of an inappropriate mode of delivery. This equates to an accuracy of assessment in this group of 86.9%. A total of 231 patients were considered to have adequate gastrointestinal function and received EN (group 2). Of these, 19 (8.2%) received less than 5 d of EN because of a failure to tolerate the feed, 10 of whom were converted to TPN. The accuracy of assessment in this group was therefore 91.8%. Patients in this study in whom there was clinical uncertainty about gastrointestinal function were randomized to either TPN or EN. No patient in group 3 (randomized TPN) was converted to enteral feeding within 5 d, signifying that assessment was 100% accurate in this group. In contrast, in group 4 (randomized EN) 6 patients (18.8%) received less than 5 d of enteral feeding because of poor tolerance, 3 of whom were converted to TPN. This represents an accuracy of assessment of 81.3% in this group. Septic Morbidity There was no statistically significant difference in the incidence of septic complications between TPN and EN in either the nonrandomized (difference 3.6%, 95% CI ⫺4.7 to 11.9%) or the randomized patients (difference 18.7%, 95% CI ⫺4.9 to 42.3%) (Table

Nutrition Volume 17, Number 1, 2001

Enteral Versus Parenteral Nutrition

7

TABLE V. SEPTIC MORBIDITY IN RELATION TO NUTRITIONAL STATUS Nonrandomized patients

Well nourished or mild/ moderately malnourished (NRI ⱖ83.5) Severely malnourished (NRI ⬍83.5)

Group 1 (TPN)

Group 2 (EN)

P value

31/108 (28.7%)

54/134 (40.2%)

0.06, NS

41/124 (33.1%)

20/64 (31.3%)

⬎0.5, NS

Randomized patients

Well nourished or mild/ moderately malnourished (NRI ⱖ83.5) Severely malnourished (NRI ⬍83.5)

Group 3 (rTPN)

Group 4 (rEN)

P value

6/16 (37.5%)

4/14 (28.6%)

⬎0.5, NS

10/15 (66.7%)

5/15 (33.3%)

0.07, NS

EN, enteral nutrition, NRI, nutritional risk index; rEN, randomized EN; TPN, parenteral nutrition; rTPN, randomized TPN.

IV). In addition, there was no significant difference in the number of septic complications per patient or number of complications per infected patient between TPN and EN in either the nonrandomized or randomized group. The incidence of septic morbidity in relation to nutritional status was investigated by analyzing separately the patients who were well nourished or mildly/moderately malnourished according to the NRI and comparing them with those who were severely malnourished. We found no significant differences between TPN and EN in either the nonrandomized or randomized patients (Table V). Overall, there were no statistically significant differences between the incidences of septic morbidity in the well nourished (14/55, 25.5%), mild/moderately malnourished (81/217, 37.3%), and severely malnourished patients (76/218, 34.9%) (P ⬎ 0.4). Septic morbidity was assessed in relation to duration of nutritional support by considering separately those patients who received 10 d or more feeding and those fed for less than 10 d (Fig. 4). The incidence of septic morbidity was significantly higher in those patients fed for 10 d or more in group 1 (50.5% versus 20.2%, P ⬍ 0.001), group 2 (51.4% versus 20.5%, P ⬍ 0.001), and group 3 (73.3% versus 29.4%, P ⫽ 0.01). The difference observed in group 4 (50% versus 26.9%) was not statistically significant.

FIG. 4. Septic morbidity in relation to duration of feeding.

FIG. 5. Nature of septic complications.

Figure 5 illustrates the nature of the septic complications that occurred in each group. There was a higher incidence of respiratory tract infections in the group 2 (EN) patients compared with those in group 1 (TPN), although this was not statistically significant (18.2% versus 13.1%). Seven respiratory tract infections among the enterally fed patients were thought to be due to aspiration, an incidence of 2.7%. The higher proportion of wound infections seen in group 1 compared with group 2 (10.5% versus 5.6%, P ⬍ 0.05) is a reflection of the greater number of postoperative surgical patients in this group (Table I). If the incidence of wound infection is alternatively expressed as a percentage of the number of patients who had surgery, the incidences are almost identical (16.3% versus 15.7%, P ⬎ 0.05). The incidence of line sepsis was 20/205 (9.8%) among those patients fed by the central route in group 1 and 5/28 (17.9%) in group 3. Line infection was the sole septic complication in 14 patients in group 1 (16.7% of all patients who developed septic morbidity in that group) and three patients in group 3 (18.8%). Eight of the line infections occurred in patients who were converted from peripheral to central TPN. There was no line sepsis associated with the use of peripheral lines. Among the enterally fed patients there were almost as many septic complications related to the delivery system, with 14/69 PEGs (20.3%) and 5/33 jejunostomies (15.2%) becoming infected. There were no significant differences between the TPN and EN groups in terms of all other forms of septic morbidity. Other Morbidity There was a significantly higher incidence of nonseptic complications related to the delivery system associated with EN compared with TPN in both the nonrandomized and randomized patients. The most frequent complications in the TPN groups were mechanical problems with the line such as blockage or leakage (n ⫽ 17 in group 1 and n ⫽ 1 in group 3) and dislodgement of the line (n ⫽ 10 in group 1 and n ⫽1 in group 3). The incidence of pneumothorax associated with central line insertion was 4/205 (2.0%) in group 1 and 0/28 (0%) in group 3. Peripheral vein phlebitis was observed in 3/93 (3.2%) group 1 patients and 1/9 (11.1%) group 3 patients fed peripherally. Among the enterally fed patients the most common complication was dislodgement of the feeding tube, which occurred at least once in 66 patients (28.6%) in group 2 and in 14 (43.8%) in group 4. Other complications included blockage of the feeding tube and failure of tube insertion. One patient in group 4 developed peritonitis secondary to a gastric perforation at the site of insertion of a PEG (Fig. 6). The incidence of other feed-related complications was also significantly higher in both EN groups compared with the respective TPN groups. The complications associated with TPN were

8

Woodcock et al.

FIG. 6. Other complications.

fluid overload (21/267, 7.9% in group 1 and 1/32, 3.1% in group 3) and metabolic, such as hyperglycemia, hyperkalemia, and abnormal liver function (10/267, 3.7% in group 1 and 3/32, 9.4% in group 3). In the patients receiving EN the most frequent such complications were diarrhea (30/231, 13.0% in group 2 and 2/32, 6.3% in group 4) and large volume nasogastric aspirates (27/231, 11.7% in group 2 and 7/32, 21.9% in group 4). Other complications included vomiting (n ⫽ 16) and abdominal bloating (n ⫽ 10). Mortality All deaths occurring during the same hospital admission were recorded. The mortality rate in group 2 was significantly greater than in group 1 (36.8% versus 22.1%, P ⬍ 0.001). The mortality rate in the stroke patients alone was 27/55 (49.1%). Although the mortality rate was higher among group 4 patients compared with group 3 (37.5% versus 21.9%), this difference was not statistically significant (difference 15.6%, 95% CI ⫺6.5 to 37.7%, P ⫽ 0.17). There were no deaths directly attributable to TPN during the course of the study; however, one patient randomized to receive EN was fed initially via a nasogastric tube and then subsequently had a PEG inserted. Shortly after this procedure was performed he developed peritonitis secondary to an ischemic perforation in the anterior wall of the stomach adjacent to the entry point of the PEG, from which he died soon after (Fig. 7).

DISCUSSION The results of this prospective study demonstrate that EN is associated with a significantly higher incidence of failure to achieve target intakes than TPN. In addition, EN results in more complications related to the method of feed delivery and other feed-related complications. There is no evidence from the present study to confirm an advantage of EN over TPN in terms of septic morbidity. EN was associated with higher mortality rates in this series. Study Design The design of this study is critically important to the interpretation of results. Most clinicians would concur with the view that TPN is indicated only in patients with evidence of permanent or temporary “intestinal failure,” while those with a functioning gastrointestinal tract should receive EN. Notwithstanding this, many previous studies comparing enteral with parenteral nutrition have randomized patients irrespective of gastrointestinal function. This must on occasion result in patients receiving inappropriate nutritional support, which may influence the interpretation of results. We have attempted to overcome the inherent problems involved in comparing the two routes of feeding by adopting the principle of prag-

Nutrition Volume 17, Number 1, 2001

FIG. 7. Overall mortality rates.

matism. Pragmatic trials reflect the natural variations that occur between patients and enable measurement of the effectiveness of a treatment and the benefit it produces in routine clinical practice. Such studies also take into account natural variations in individual clinicians’ preferences. Results are always analyzed on an intention-to-treat basis. This study was pragmatic in that the attending clinician was allowed to instigate nutritional support by whatever route he or she considered appropriate or preferable. This recruitment process facilitated the inclusion of all patients requiring adjuvant nutritional support, ensuring relatively rapid accrual of sufficient patient numbers for data analysis. We further encouraged recruitment into this study by agreeing at the outset on clear and unequivocal indications for nutritional support. It is our view that nutritional support can be justified for all patients who have sustained or who are anticipated to sustain 7 d or more of inadequate oral intake. This is based on the premise that such a period of inadequate intake is associated with deleterious consequences to physiological function.16,17 This principle obviates the need to consider preexisting malnutrition or the underlying disease process in the decision to feed. We are of the opinion that application of the principle of pragmatism and the adoption of clear guidelines for the initiation of nutritional support ensures that the results of this study have widespread applicability. It is important, in this context, to stress that in this study the provision of all nutritional support was overseen by a dedicated nutrition team, with considerable experience in enteral and parenteral feeding techniques. The influence of such teams on feed-related morbidity is well recognized.18 –21 Finally, in designing this study we attempted to incorporate a subgroup of patients in whom randomization was appropriate on the basis that there was genuine uncertainty as to the adequacy of gastrointestinal function. In this way we aimed to create two homogeneous groups in whom valid comparisons of enteral and parenteral nutrition could be made. Demography It might be argued that the differences observed between the groups in this study could be accounted for by demographic differences. The mean age of the patients in group 1 was significantly less than in group 2. This reflects the large number of young patients with inflammatory bowel disease in group 1 who were prescribed TPN either because of malabsorption and nutritional failure or as an adjunct to steroid therapy and bowel rest in an attempt to induce remission of active disease. In contrast, group 2 was composed predominantly of patients with “nonsurgical” conditions such as respiratory failure or cerebrovascular disease. These patients often have dysphagia or reduced levels of consciousness or require endotracheal intubation and are usually considered appropriate for enteral nutrition. It is probable that these variations in age distribution and diagnosis account, to some extent, for the observed differences in outcome with enteral or parenteral nutrition. However, the fact remains that this reflects the

Nutrition Volume 17, Number 1, 2001 real situation in everyday clinical practice. Furthermore, the results obtained in this study from groups 1 and 2 (pragmatic groups) whether relating to intakes received or morbidity and mortality are remarkably similar to those observed in groups 3 and 4 (randomized groups), in whom demography was comparable. Malnutrition, Severity of Illness, and Outcome It is frequently stated that the patients most likely to gain benefit from nutritional support are those with severe preexisting malnutrition.22–24 Almost half of all patients included in this study were severely malnourished according to the NRI, and almost a third had lost 10% or more of their usual body weight. These are higher proportions than in most other studies. For example, in the Veterans’ Affairs (VA) study less than 10% of the randomized patients were severely malnourished according to the NRI.22 The incidence of severe malnutrition was significantly greater in group 1 patients who received TPN compared with group 2 who received EN. It is tempting to speculate that this might explain, at least to some extent, the differences in outcomes of feeding between these two groups. A decrease in the incidence of septic complications in malnourished patients as a consequence of nutritional support, and a concomitant increase in the well-nourished patients, has been reported in previous studies.22,25 Indeed, in 5 of the 14 (36%) well-nourished patients in our study who developed septic morbidity, the only complication that occurred was a direct result of the nutritional support (aspiration pneumonia n ⫽ 3 and PEG site infection n ⫽ 2). It is interesting that all these patients were fed enterally. Nonetheless, we consider it improbable that malnutrition alone can account for the differences observed between groups 1 and 2. There are a number of reasons for this: first, we identified no relationship between nutritional status according to the NRI and the incidence of septic morbidity. The incidence of sepsis was lower in the well-nourished patients (25%) compared with the mild/moderately malnourished (37%) and severely malnourished patients (35%), although this failed to reach statistical significance. We recognize the small possibility that the relatively small number of well-nourished patients in this study (n ⫽ 55) could result in a type II error. Second, in addition to assessing nutritional status we incorporated a physiologic assessment of patients at the time of instigation of nutritional support. The POSSUM scoring system is well validated as a means of predicting morbidity and mortality in surgical patients.13,26 The mean POSSUM score was significantly higher in group 2 compared with group 1 and possibly reflects the fact that a greater number of group 2 patients were in the ICU. These data suggest that while the group 1 patients were more malnourished, the group 2 patients were more severely ill despite being less malnourished. This is almost certainly a reflection of the underlying disease processes. Group 1 comprised more surgical patients who may have sustained acute and rapid weight loss but in whom rapid recovery is often the norm. Indeed, this is reflected in surveys of nutritional support, which indicate that very few patients in the UK receive more than 10 d of TPN. In contrast, group 2 patients, the majority of whom had significant longstanding nonsurgical diseases, had poorer overall prognoses as a consequence of the disease process itself. These results suggest that preexisting disease is a more potent determinant of outcome than malnutrition alone. Third, and most important, the differences in outcome observed between groups 1 and 2 (pragmatic groups) were also found between groups 3 and 4 (randomized groups), in whom there were no differences on entry into the study in either the incidence of malnutrition or POSSUM physiologic scores. Septic Morbidity Many authorities continue to advocate the use of EN in preference to TPN. One of the main reasons for this is concern over TPNrelated septic morbidity.2– 4 It is a commonly held view that TPN

Enteral Versus Parenteral Nutrition

9

results in alterations in gut barrier function, predisposing the patient to the translocation of enteric bacteria, which then cause septic morbidity at distant sites. There is certainly experimental evidence, predominantly derived from animal studies, that demonstrates that TPN adversely effects villus architecture and intestinal permeability.6,7,27–30 However, studies in humans have shown no association between TPN, changes in barrier function, and bacterial translocation.31–37 Despite this, a number of clinical studies point to an increase in septic morbidity in association with TPN. The majority of these have been in patients following abdominal trauma,15,38,39 but similar findings have also been reported in patients with severe acute pancreatitis.40,41 Moore et al.38 randomized 59 patients after laparotomy for blunt or penetrating abdominal trauma to receive either EN via needle catheter jejunostomy or TPN via central venous catheter. The EN group had a significantly lower incidence of “major” septic morbidity (intraabdominal abscess or pneumonia) compared with the TPN group (3% versus 20%), but if all septic complications were included the difference was not significant (17% versus 37%). No mention was made of feed-related complications. A similar study by Kudsk et al.39 in 51 patients observed a significantly lower incidence of infective complications with EN, but only in the most severely injured patients. Again, nonseptic morbidity was not recorded. A meta-analysis by Moore et al.15 combined the data from eight studies of EN and TPN in high-risk surgical patients, six of which were unpublished, and reported a significantly reduced incidence of septic morbidity in patients given EN compared with those give TPN (18% versus 35%). However, this difference was limited to only the trauma patients, with no difference observed in the nontrauma patients. Also, there were significantly more treatment failures among the EN patients, and there was no difference in noninfective complications. Windsor et al.40 reported a prospective study in patients with severe acute pancreatitis finding significant attenuation of the acute phase response in association with EN compared with TPN together with improvements in disease severity. However, equivalent protein and energy intakes were not received by the two groups, which may have influenced outcome. The results of our study are consistent with many other prospective randomized trials that have compared enteral with parenteral nutrition and found no significant difference in septic complications between the two feeding modalities. These studies have been performed on a wide range of patient groups, including patients following elective surgery for upper gastrointestinal malignancy,42– 44 liver transplantation,45 and severe head injury.46,47 There are several possible explanations for the differences in results obtained from these various studies. First, patients with blunt or penetrating abdominal trauma are quite different in terms of age, physiologic, and nutritional status compared with the majority of patients who require nutritional support. They constitute only a very small minority in the present study. Second, the relative protein and energy intakes in these studies were often not comparable, usually being much higher in the patients fed parenterally. For example, in the Kudsk study the mean energy intake in the TPN group was 19.1 kcal/kg/d compared with only 15.7 kcal/kg/d in the EN group.39 The total energy intake in the TPN patients in the VA study was 3300 kcal/d,22 much higher than the total energy expenditure of around 2000 kcal/d observed in unselected surgical patients.48 This excessive energy intake resulted in an incidence of hyperglycaemia (serum glucose ⬎16.7 mmol/L) of 20% in that study. This has led reviewers to question whether the increased incidence of sepsis in the patients receiving TPN is in reality not related to the actual route of feeding, but is more a reflection of the detrimental effects of overfeeding and consequent hyperglycemia.9,10,49 In the present study the prescribed target intake for both the TPN and EN patients was the same (30 kcal/kg/d and 9 gN/d), aiming to match total energy expenditure. Third, the previous studies involved the randomization of all patients to receive either TPN or EN, with no consideration given to the issue of gut function. Inevitably this will result in some

10

Woodcock et al.

patients who are able to tolerate enteral feeding being given TPN. It is questionable as to whether one should draw conclusions from the results of studies in which patients are fed inappropriately. The design of the present study, whereby the route of feeding was dictated by an assessment of gut function, ensured that TPN was administered only to patients with either inadequate or at least doubtful intestinal function. It has been concluded from recent reviews that there is insufficient evidence to support the hypothesis that EN provides any distinct advantage over TPN in terms of septic morbidity when energy and nitrogen intake are equal.9,50 The results of this study are in accord with this conclusion. The incidence of septic morbidity was significantly higher in patients receiving nutritional support by the designated route for 10 days or more, except in group 4 where only six patients actually received EN for this time. Patients fed for prolonged periods are invariably those who are most severely ill and thus at highest risk of developing septic complications. The spectrum of septic complications was similar in patients receiving TPN and EN. The incidence of respiratory tract infections in the EN patients was slightly higher than in the TPN patients (18.2% versus 13.1%). Based on the findings of studies in rodents, it has been claimed that EN has beneficial effects on IgA-mediated upper respiratory tract immunity, whereas TPN suppresses it.29,51 Aspiration pneumonia is a recognized complication of enteral feeding, with a reported incidence of between 2 and 95%, the wide variation probably reflecting poorly defined criteria for diagnosis.52 It is difficult to explain the relative rates of respiratory tract infection in this study on the basis of aspiration alone, considering that clinically only seven cases were thought to be due to aspiration, an incidence of 2.7%. Four of these patients were fed via nasogastric tube and the other three through a PEG. Postpyloric feeding via a jejunostomy has the potential advantage of preventing aspiration,51 although aspiration of jejunal feeds has been reported.53 There were no cases of aspiration pneumonia among the patients fed through a jejunostomy in the present study. The relatively high incidence of line sepsis is often quoted as a major disadvantage of parenteral feeding, sometimes overshadowing any potential benefits.10 The overall incidence of catheter-related sepsis in our study was 11%, similar to other studies of parenteral nutrition in which the incidence is usually reported to be of the order of 4 –10%.10 A recent retrospective study of central line infections in a surgical ICU, however, reported an incidence as high as 34%.54 It is possible that the risk of line sepsis in parenterally fed patients may be offset by infections encountered with the use of invasive methods of providing EN, namely PEGs and feeding jejunostomies. As mentioned previously, such complications are often not reported in studies comparing EN and TPN. We found that the incidences of PEG and jejunostomy site infections (20% and 15%, respectively) were both higher than that of line sepsis.

Nutritional Intakes Enteral feeding was associated with a significantly higher incidence of inadequate nutritional intake compared with TPN. This was observed in both the nonrandomized and randomized patients. Inadequate intake in the EN patients was most commonly related to high nasogastric aspirates, dislodgement of the feeding tube, or initial build up of the feeding rate. Many other authors have highlighted problems with achieving target intakes with EN.55,56 Stechmiller et al.57 reported a significant difference between nutrients prescribed and nutrients received in the majority of 52 mechanically ventilated critically ill patients fed via nasogastric tube, with interruptions caused by tube displacement in 27% of cases. We found that feeding was interrupted at least once by inadvertent removal of the feeding tube in 42% of patients fed nasogastrically. In the present study, patients receiving EN did not routinely receive prokinetic agents, and invasive methods of access were

Nutrition Volume 17, Number 1, 2001 only employed once successful nasogastric feeding had been established or in a minority of patients in whom a jejunostomy was fashioned at the time of surgery. We recognize that the routine adoption of these techniques has been shown to be associated with an increased tolerance of EN. Nevertheless, the results of this study suggest that, at least in patients who are considered to have adequate gastrointestinal function on clinical assessment, only a minority would benefit from ancillary means to enhance tolerance. Assessment of Gastrointestinal Function Assessment of gastrointestinal function was considered to have been incorrect if patients were converted to an alternative route of feeding within 5 d. The accuracy of assessment was high in both the patients considered to have inadequate gut function (group 1, 87%) and those thought to have adequate function (group 2, 92%) at the time of entry into the study. These results suggest that clinical assessment is a reliable indicator of gastrointestinal function. In this study 64 patients were identified in whom there was uncertainty as to the adequacy of gastrointestinal function. Of the patients randomized to receive TPN, none were converted to EN or oral diet within 5 d of commencing feeding, whereas only 14 (44%) of those randomized to EN were fed “successfully” for 5 d or more. A total of 10 patients (31%) in this randomized EN group were converted to TPN, 6 within 5 d of study entry. The most common reason was poor tolerance of the feed as indicated by large-volume nasogastric aspirates. The incidence of inadequate intake was also significantly higher in the randomized EN patients (group 4) compared with the nonrandomized EN patients (group 2). These findings suggest that when there is reasonable doubt as to the adequacy of gastrointestinal function patients should be fed by the parenteral route, as it is improbable that EN alone will be successful in providing an adequate nutritional intake. Complications Related to Delivery Systems We found a much higher incidence of complications associated with the delivery system in patients given EN, predominantly relating to dislodgement or blockage of the nasogastric tube. Such complications are common reasons for interruption of feeding and inadequate intakes among enterally fed patients. This, together with the fact that small bowel motility is reestablished before that of the stomach in the postoperative period,58 has produced interest in more invasive methods of gaining access to both the stomach and small bowel, such as PEGs, surgically placed jejunostomies, and needle catheter jejunostomies (NCJs). It is increasingly being recognized that these techniques are associated with significant morbidity. PEG complications include a significant procedurerelated mortality, together with entry site infection, intestinal necrosis and perforation, and abdominal wall migration.59 – 63 In the present study, one patient died as a direct consequence of peritonitis secondary to a gastric perforation following the insertion of a PEG. The Kudsk study of postoperative abdominal trauma patients reported that one patient with a feeding jejunostomy required a second laparotomy because of a small bowel obstruction at the site of the jejunostomy.39 NCJs are associated with several potential complications, including leakage of feed, peritonitis, bowel obstruction, and occlusion.59,64 We observed a high incidence of infection associated with both PEGs and feeding jejunostomies (20% and 15%, respectively). These findings indicate that EN is not without significant procedure-related morbidity and even mortality. A third of the patients receiving TPN in the present study were fed peripherally (PPN). Only a third of these required subsequent conversion to central TPN as a result of inadequate peripheral venous access. Cyclical PPN over a period of 12 hours through a short cannula with daily rotation of venous access sites has been shown to be associated with a significant reduction in thrombo-

Nutrition Volume 17, Number 1, 2001 phlebitis compared with continuous feeding via both a short cannula65 and an indwelling ultrafine cannula.66 Cyclical as opposed to continuous TPN may also facilitate improvements in substrate use, with the additional benefits of increased mobility and psychologic well-being.66 In the present study, the incidence of thrombophlebitis in the patients receiving peripheral TPN was only 4%, and other feed-related morbidity was minimal. The use of PPN wherever possible also eliminates the risk of sepsis associated with central venous access devices. These results confirm our view that PPN employing cyclical infusion with rotation of venous access sites should be the method of choice for the administration of TPN.

Enteral Versus Parenteral Nutrition

11

revealed a perforation in the anterior wall of the stomach, immediately adjacent to the site of the PEG tube. Despite active management his condition deteriorated and he died. Many other studies have reported mortality in association with invasive methods employed for the provision of EN, emphasizing the importance of taking into account any procedure-related complications that may occur when comparing enteral with parenteral nutrition. No deaths directly attributable to TPN occurred during the course of the present study.

CONCLUSIONS Feed-Related Complications In this study the incidence of feed-related complications was higher in the enterally fed patients compared with those receiving TPN. In the EN patients these complications were predominantly related to symptoms of gastrointestinal intolerance, such as largevolume nasogastric aspirates, vomiting, abdominal bloating, and diarrhea. We have already described how these complications are common reasons for reducing the rate or stopping feeds, and they are also quite labor intensive in terms of nursing care.67 The prospective study by Jones et al.63 observed abdominal discomfort in 83% of EN patients, and the Moore meta-analysis reported incidences of abdominal distension and diarrhea of 46% and 34%, respectively.14 One study reported an 80% incidence of diarrhea in critically ill patients receiving EN.68 The frequency of diarrhea has been shown to be inversely related to serum albumin,69 which is often low in critically ill patients as a consequence of hepatic reprioritization of protein synthesis and increased endothelial permeability. This suggests that it may be the oncotic pressure between plasma and the bowel lumen69 or intestinal mucosal edema70 causing the diarrhea. The incidence of feed-related morbidity in the TPN patients in this study was very low, the most common complication being fluid overload. In marked contrast to many other comparative studies of EN and TPN, hyperglycemia necessitating suspension of feeding occurred in only one patient in this series, emphasising the importance of avoiding excessively high protein and energy intakes. Mortality There was a higher mortality in patients receiving EN compared with TPN in this series. This difference was statistically significant in the nonrandomized patients (groups 1 and 2) and is almost certainly a consequence of the different patient populations in these two groups. Group 2 comprised large numbers of both critically ill ICU patients and elderly patients with cerebrovascular disease, in whom mortality would be expected to be higher. In contrast, the two randomized groups (groups 3 and 4) were composed of homogeneous patient populations in whom there were no differences in terms of age, diagnosis, or nutritional or physiologic status. Among these patients there was also a higher mortality associated with EN compared with TPN (38% versus 22%), although this failed to reach statistical significance. This increase in mortality, though not significant, remains a matter of concern. Previous authors have commented on the possibility that enteral nutrition, despite the widely held belief that it is safer than TPN, may actually have potentially harmful effects particularly in the critically ill. It has been suggested, for example, that increased splanchnic blood flow that occurs with enteral feeding rather than TPN may result in increased delivery of tumour necrosis factor to the intestinal mucosa with deleterious consequences.71,72 This mechanism has been proposed as a reason for poor tolerance of EN in the critically ill. Another factor in the increased mortality associated with EN is procedure related. One patient randomized to receive EN developed generalized peritonitis soon after an apparently straightforward PEG insertion. Subsequent laparotomy

This study demonstrates that enteral feeding fails to provide a sufficient nutrient intake in a significant number of patients, predominantly because of gastrointestinal intolerance and complications related to the mode of delivery. The incidence of nonseptic morbidity is higher in patients receiving EN compared with TPN, and invasive techniques for access to the gastrointestinal tract are associated with appreciable morbidity and mortality. TPN-related morbidity may be further reduced by the use of cyclical peripheral feeding where appropriate. There is no evidence from this study to substantiate the widely held view that TPN predisposes to septic morbidity and that EN is protective. We propose that the EN versus TPN debate has run its course. The choice of feeding route should be determined by clinical assessment of gastrointestinal function. Medical and nursing attendants, particularly of the critically ill, should be encouraged to administer EN to the limit of patient tolerance. If adequate volumes are tolerated then TPN is clearly not required and the absence of intestinal failure is probably a favorable prognostic indicator. Close monitoring of intakes is essential, and invasive access to the gastrointestinal tract should only be undertaken when absolutely necessary. Any patient who cannot tolerate adequate volumes of EN, irrespective of diagnosis, should receive TPN, preferably by the peripheral route. Those patients in whom gastrointestinal function is in doubt should be fed parenterally if adequate nutritional intakes are to be achieved. The two modalities of nutritional support are not mutually exclusive. Administering small volumes of an enteral formula supplemented by TPN in order to fulfill protein and energy requirements may best provide optimal nutritional support, particularly in critically ill patients. As enteral intakes increase, then parenteral supply should diminish. Future research seeking to establish the role of new substrates or the effects of immunostimulatory products should investigate these against a background of optimal nutritional support.

ACKNOWLEDGMENTS The authors acknowledge the cooperation of all the surgeons and physicians in Scarborough Hospital who, without exception, kindly permitted us to include their patients in this study. In particular we are grateful to Mr. J. D. Harrison, Mr. E. P. Perry, and Mr. N. El-Barghouti.

REFERENCES 1. Satyanarana R, Klein S. Clinical efficacy of perioperative nutrition support. Curr Opin Clin Nutr Metab Care 1998;1:51 2. Silk DBA, Green CJ. Perioperative nutrition: parenteral versus enteral. Curr Opin Clin Nutr Metab Care 1998;1:21 3. Payne-James J. Nutritional support. In: AlJafri AM, Kingsnorth A, eds. Fundamentals of surgical practice. London: Greenwich Medical Media, 1998:59 4. Sigurdsson G. Enteral or parenteral nutrition? Pro-enteral. Acta Anaes Scand 1997;41:143 5. Buchman AL, Moukarzel AA, Bhuta S, et al. Parenteral nutrition is associated with intestinal morphologic and functional changes in humans. JPEN 1995;19:453 6. Deitch EA, Xu D, Naruhn MB, et al. Elemental diet and IV TPN-induced bacterial translocation is associated with loss of intestinal mucosal barrier function against bacteria. Ann Surg 1995;221:299

12

Woodcock et al.

Nutrition Volume 17, Number 1, 2001

7. Alverdy JC, Aoys E, Moss GS. Total parenteral nutrition promotes bacterial translocation from the gut. Surgery 1988;104:185 8. Chang RWS, Jacob S, Lee B. Gastrointestinal dysfunction among intensive care unit patients. Crit Care Med 1987;15: 909 9. Keith ME, Jeejeebhoy KN. Enteral nutrition. Curr Opin Gastroenterol 1998;14:151 10. Nordenstrom J, Thorne A. Benefits and complications of parenteral nutritional support. Eur J Clin Nutr 1994;48:531 11. Roland M, Torgerson DJ. What are pragmatic trials? BMJ 1998;316:285 12. Baker JP, Detsky AS, Wesson DE, et al. Nutritional assessment: a comparison of clinical judgement and objective measurements. N Engl J Med 1982;306:969 13. Copeland GP, Jones D, Walters M. POSSUM: a scoring system for surgical patients. Br J Surg 1991;78:355 14. MacFie J. Infusion phlebitis and peripheral parenteral nutrition. Nutrition 1998; 14:233 15. Moore FA, Feliciano DV, Andrassy RJ, et al. Early enteral feeding, compared with parenteral, reduces postoperative septic complications: the results of a meta-analysis. Ann Surg 1992;216:172 16. Allison S. Uses and limitations of nutritional support. Clin Nutr 1992;11:319 17. ASPEN Board of Directors. Guidelines for use of parenteral and enteral nutrition in adult and paediatric patients. JPEN 1993;17(suppl):1SA 18. Faubion WC, Wesley JR, Kharlidi N, Silva J. Total parenteral nutrition catheter sepsis: impact of the team approach. JPEN 1986;10:641 19. Traeger S, Williams GB, Milliren G, et al. Total parenteral nutrition by a nutrition support team: improved quality of care. JPEN 1986;10:408 20. Elia M. Artificial nutritional support in clinical practice in Britain. J R Coll Phys Lond 1993;27:8 21. Reynolds N, McWhirter JP, Pennington CR. Nutrition support teams: an integral part of developing a gastroenterology service. Gut 1995;37:740 22. The Veterans’ Affairs Total Parenteral Nutrition Cooperative Study group. Perioperative total parenteral nutrition in surgical patients. N Engl J Med 1991;325:525 23. von Meyenfeldt MF, Meijerink WJHJ, Rouflart MMJ, et al. Perioperative nutritional support: a randomized clinical trial. Clin Nutr 1993;11:180 24. Heyland DK, MacDonald S, Keefe L, Drover JW. Total parenteral nutrition in the critically ill patient: a meta-analysis. JAMA 1998;280:2013 25. Sandstrom R, Drott C, Hyltander A, et al. The effect of postoperative intravenous feeding (TPN) on outcome following major surgery evaluated in a randomized study. Ann Surg 1993;217:185 26. Sagar P, Copeland G, Hartley M, Taylor B, MacFie J. Comparison of individual surgeon’s performance: Risk adjusted analysis with POSSUM scoring system. Dis Col Rect 1995;38:640 27. McCauley RD, Heel KA, Christiansen KJ, Hall JC. The effect of minimum luminal nutrition on bacterial translocation and atrophy of the jejunum during parenteral nutrition. J Gastroenterol Hepatol 1996;11:65 28. Lin M-T, Saito H, Fukushima R, et al. Route of nutritional supply influences local, systemic and remote organ responses to intraperitoneal bacterial challenge. Ann Surg 1996;223:84 29. Janu P, Li J, Renegar KB, Kudsk KA. Recovery of gut-associated lymphoid tissue and upper respiratory tract immunity after parenteral nutrition. Ann Surg 1997;225:707 30. Shou J, Lappin J, Minnard EA, Daly JM. Total parenteral nutrition, bacterial translocation and host immune function. Am J Surg 1994;167:145 31. Illig KA, Ryan CK, Hardy DJ, et al. Total parenteral nutrition-induced changes in gut mucosal function: atrophy alone is not the issue. Surgery 1992;112:631 32. Guedon C, Schmitz J, Lerebours E, et al. Decreased brush border hydrolase activities without gross morphologic changes in human intestinal mucosa after prolonged total parenteral nutrition of adults. Gastroenterology 1986;90:373 33. O’Boyle CJ, MacFie J, Dave K, et al. Alterations in intestinal barrier function do not predispose to translocation of enteric bacteria in gastroenterologic patients. Nutrition 1998;14:358 34. Sedman PC, MacFie J, Sagar P, et al. The prevalence of gut translocation in humans. Gastroenterology 1994;107:643 35. O’Boyle CJ, MacFie J, Mitchell CJ, et al. Microbiology of bacterial translocation in humans. Gut 1998;42:29 36. Deitch EA, Xu D, Lu Q, Berg RD. Bacterial translocation from the gut impairs systemic immunity. Surgery 1991;109:269 37. Sedman PC, MacFie J, Palmer MD, et al. Preoperative TPN is not associated with mucosal atrophy or bacterial translocation in humans. Br J Surg 1995;82:1663 38. Moore FA, Moore MM, Jones TN, et al. TEN versus TPN following major abdominal trauma-reduced septic morbidity. J Trauma 1989;29:916 39. Kudsk KA, Croce MA, Fabian TC, et al. Enteral versus parenteral feeding: effects on septic morbidity after blunt and penetrating abdominal trauma. Ann Surg 1992;215:503 40. Windsor ACJ, Kanwar S, Li AGK, et al. Compared with parenteral nutrition,

41. 42.

43.

44. 45. 46. 47. 48. 49.

50. 51. 52.

53. 54. 55.

56. 57. 58. 59. 60. 61. 62. 63. 64.

65.

66. 67. 68. 69. 70. 71.

72.

enteral feeding attenuates the acute phase response and improves disease severity in acute pancreatitis. Gut 1998;42:431 Kalfarentzos F, Kehagias J, Mead N, et al. Enteral nutrition is superior to parenteral nutrition in severe acute pancreatitis. Br J Surg 1997;84:1665 Reynolds JV, Kanwar S, Welsh FKS, et al. Does the route of feeding modify gut barrier function and clinical outcome in patients after major upper gastrointestinal surgery? JPEN 1997;21:196 Braga M, Gianotti L, Vignali A, et al. Artificial nutrition after major abdominal surgery: impact of route of administration and composition of the diet. Crit Care Med 1998;26:24 Sand J, Luostarinen M, Matikainen M. Enteral or parenteral feeding after total gastrectomy: prospective randomized pilot study. Eur J Surg 1997;163:761 Wicks C, Somasundaram S, Bjarnason I, et al. Comparison of enteral feeding and total parenteral nutrition after liver transplantation. Lancet 1994;344:837 Borzotta AP, Pennings J, Papasadero B, et al. Enteral versus parenteral nutrition after severe closed head injury. J Trauma 1994;37:459 Young B, Ott L, Twyman D, et al. The effect of nutritional support on outcome from severe head injury. J Neurosurg 1987;67:668 MacFie J. Active metabolic expenditure of gastroenterological surgical patients receiving intravenous nutrition. JPEN 1984;8:371 Klein S, Kinney J, Jeejeebhoy K, et al. Nutrition support in clinical practice: review of published data and recommendations for future research directions. JPEN 1995;21:133 Lipman TO. Grains or veins: is enteral nutrition really better than parenteral nutrition? A look at the evidence. JPEN 1998; 22:167 Kudsk KA, Li J, Renegar KB. Loss of upper respiratory tract immunity with parenteral feeding. Ann Surg 1996;223:629 American Gastroenterological Association. American Gastroenterological Association technical review on tube feeding for enteral nutrition. Gastroenterology 1995;108:1282 Coates NE, MacFadyen BV. Endoscopic jejunal access for enteral feeding. Am J Surg 1995;169:627 Charalambous C, Swoboda SM, Dick J, et al. Risk factors and clinical impact of central line infections in the surgical intensive care unit. Arch Surg 1998;133:1241 Payne-James JJ, de Gara CJ, Grimble GK, et al. Artificial nutrition support in hospitals in the United Kingdom, 1994: third hospital survey. Clin Nutr 1995; 14:329 Grimble GK. The adequacy of enteral nutrition. Curr Opin Gastroenterol 1996; 12:174 Stechmiller J, Treloar DM, Derrico D, et al. Interruption of enteral feedings in head injured patients. J Neurosci Nurs 1994;26:224 Sagar PM, Kruegener G, MacFie J. Nasogastric intubation and elective abdominal surgery. Br J Surg 1992;79:1127 Heyland DK, Cook DJ, Guyatt GH. Enteral nutrition in the critically ill: a critical review of the evidence. Int Care Med 1993;19:435 Botterill I, Miller G, Dexter S, Martin I. Deaths after delayed recognition of percutaneous endoscopic gastrostomy tube migration. BMJ 1998;317:524 Light VL, Slezak FA, Porter JL, et al. Predictive factors for early mortality after percutaneous endoscopic gastrostomy. Gastrointest Endosc 1995;42:330 Hull MA, Rawlings J, Murray FE, et al. Audit of outcome of long term enteral nutrition by percutaneous endoscopic gastrostomy. Lancet 1993;341:869 Schunn CDG, Daly JM. Small bowel necrosis associated with postoperative jejunal tube feeding. J Am Coll Surg 1985;180:410 Jones TN, Moore FA, Moore EE, McCroskey BL. Gastrointestinal symptoms attributed to jejunostomy feeding after major abdominal trauma: a critical analysis. Crit Care Med 1989;17:1146 Kerin MJ, Pickford IR, Jaeger H, et al. A prospective and randomized study comparing the incidence of infusion phlebitis during continuous and cyclic peripheral parenteral nutrition. Clin Nutr 1991;10:315 MacFie J. Cyclic parenteral nutrition. Nutrition 1997;13:46 Anderson ID, Fearon KCH. Paralytic ileus and enteral feeding. Br J Int Care 1995;5:80 Cerra FB, McPherson JP, Konstantinides FN, et al. Enteral nutrition does not prevent multiple organ failure syndrome (MOFS) after sepsis. Surgery 1988;104:727 Woods MS, Kelley H. Oncotic pressure, albumin and ileus: the effect of albumin replacement on postoperative ileus. Am Surg 1993;59:758 Koretz L. Intravenous albumin and nutrition support: going for the quick fix. JPEN 1995;19:166 Raina N, Cameron RG, Jeejeebhoy KN. Gastrointestinal, hepatic and metabolic effects of enteral and parenteral nutrition in rats infused with tumor necrosis factor. JPEN 1997;21:7 Bistrian B. Effects of enteral and parenteral nutrition in rats infused with tumour necrosis factor. JPEN 1997;21:305

(For an additional perspective, see Editorial Opinions)