Introduction

Patients receiving parenteral nutrition are at risk of septic morbidity following immunosuppression induced by parenteral nutrition. Previous work has shown that parenteral nutrition causes atrophy of the small intestinal gut-associated lymphoid tissue (GALT) (Li et al, 1995), lowers small intestinal immunoglobulin A (IgA) levels (Kudsk et al, 1996), and impairs secretory IgA-mediated mucosal immunity (Kudsk et al, 1996). Apart from its detrimental effect on mucosal defences, parenteral nutrition impairs gut motility (Vantrappen et al, 1977), which is frequently reduced by the underlying disease, and it also impairs biliary flow (Jawaheer et al, 1995). The combination promotes small intestinal bacterial overgrowth. Gut overgrowth, particularly with aerobic Gram-negative bacilli (AGNB) has been identified as a risk factor for impaired systemic immunity (Marshall et al, 1987). Endotoxin released by these bacteria is thought to induce a depression of systemic immunity via liver dysfunction (Billiar et al, 1988). Suppression of both mucosal and systemic immunity is required for the development of sepsis and septicaemia following endotoxin absorption and/or bacterial translocation (Alverdy & Burke, 1992).

A prospective 5 y study (1 February 1992–31 January 1997) in a large homogeneous population of surgical infants less than 6 months of age was undertaken to determine the importance of gut overgrowth with AGNB in the pathogenesis of septic complications related to parenteral nutrition.

Patients and methods

Patients

Two-hundred and eight infants who received parenteral nutrition in Alder Hey Children's Hospital, Liverpool were enrolled in a prospective observational cohort study from 1 February 1992 until 31 January 1997. The neonates and infants, who were less than 6 months of age at the start of parenteral nutrition, all underwent surgical procedures. The indication for parenteral nutrition was gut dysfunction secondary to the following disorders: gastroschisis (n=66), congenital intestinal obstruction (n=58); necrotizing enterocolitis (n=25); Hirschprung's disease (n=20); prolonged postoperative ileus (n=20) and miscellaneous (n=19). All patients were admitted to the neonatal surgical special care unit.

Of the 208 patients 187 received parenteral nutrition through a central venous line, whilst a peripheral line was used in 21 infants. Central venous lines were inserted by percutaneous venepuncture or by surgical cutdown and the catheter was advanced to a central venous position. All central venous lines were made of silicone. The central venous lines inserted via cutdown were tunnelled under the skin leaving the attached Dacron cuff in the subcutaneous tissue to reduce the risk of dislodgement. The same protocol for central venous line care was applied to all patients. Every 4 days, the dressing on the exit site of the intravenous line was removed, the skin was cleansed with an antiseptic solution, and a new sterile dressing applied. A Pall ELD 96 filter (0.22 m; Pall Biomedical, Portsmouth, UK) was placed in-line to remove particulate matter, such as calcium salts or micro-organisms, that may have contaminated the solution. All intravenous tubing and the containers of infusate were changed daily. Chlorhexidine 0.5% in 70% alcohol solution was applied to all joints in the circuit to prevent entry of micro-organisms.

Parenteral nutrition was prepared under strict aseptic conditions which were regularly checked for sterility. Intravenous fluid intake varied from 100 to 180 ml/kg/day according to patient requirement. A standardized protocol for intravenous nutrition was used in all patients. The intravenous diet included amino acids, lipids, carbohydrate, electrolytes, vitamins and trace elements. Intravenous energy intake was progressively increased over a 4 day period. Intravenous amino acid intake was increased from 1.25 g/kg/day on the first day of parenteral nutrition to a maximum of 2.5 g/kg/day on day 2. Vamin 9^® solution (Pharmacia & Upjohn, Milton Keynes, Bucks, UK) or Primene^® 10% (Clintec Nutrition Ltd, Slough, UK) were the sources of intravenous amino acids. Lipid was increased from 1 g/kg/day on the first day of parenteral nutrition to a maximum of 4 g/kg/day by day 4. Intralipid^® 20% (Pharmacia & Upjohn, Milton Keynes, Bucks, UK) and Lipofundin^® MCT/LCT 20% (B Braun Medical Ltd, Aylesbury, UK) were the sources of intravenous lipid. A minimum of 10 g/kg/day and a maximum of 15 g/kg/day of intravenous carbohydrate was prescribed. Non-protein intravenous energy intake, after achieving stabilization, varied from 75 to 90 kcal/kg/day according to weight gain and individual needs. Infants initially received all nutrients intravenously. Enteral feeds were introduced gradually when the gut motility recovered (ie substantial reduction in gastric aspirate and/or passage of faeces). The observational period ended once parenteral nutrition was discontinued.

Antibiotic policy

Prophylaxis

Cefotaxime and metronidazole were given for clean and clean-contaminated procedures (such as fundoplication and repair of oesophageal atresia). Three doses were given intravenously, the first at induction of anaesthesia to be followed by two more doses 8 h apart. For contaminated or dirty procedures (such as gastroschisis and necrotising enterocolitis) intravenous gentamicin was added and the antibiotics given for at least 3 days. None of the patients received selective decontamination of the digestive tract.

Therapy

Blind therapy with a combination of teicoplanin and gentamicin was started at the onset of sepsis. The aminoglycoside was stopped if a Gram-positive micro-organism was isolated. In cases of Gram-negative bacillary septicaemia the glycopeptide was discontinued and replaced by a second or third generation cephalosporin. For fungaemia, liposomal amphotericin B was administered. If cultures proved negative, antibiotics were discontinued unless the severity of sepsis warranted a full 5 day course of treatment.

Endpoints

In this homogeneous population three endpoints were evaluated:

1. the incidence of abnormal gut carriage, sepsis and septicaemia;
2. sepsis and septicaemia rates in infants who developed AGNB overgrowth compared with infants who remained free of AGNB; and
3. the temporal relationship between abnormal gut carriage, sepsis and positive blood cultures, in infants developing septicaemia.

Microbiological methods

Surveillance samples comprising throat and rectal swabs were taken at the start of parenteral nutrition, and twice weekly thereafter to detect the carrier state. They were processed using a semi-quantitative method, ie, the four quadrant technique combined with enrichment broth. Both swabs were inoculated onto a solid medium, MacConkey agar, and also placed into 5 ml of brain–heart infusion broth. The target micro-organisms were AGNB. Growth density was classified as very low (equivalent to <10 colony forming units (CFU)/ml), where only the enrichment broth was positive, low (equivalent to <10³ CFU/ml), medium (equivalent to <10⁵ CFU/ml), high (equivalent to <10⁷ CFU/ml), and very high (equivalent to >10⁷ CFU/ml) where the plate had grown completely.

Diagnostic samples of blood for culture were taken from the central venous line and/or from a peripheral vein when sepsis was suspected. Blood was processed using BACTEC 9240 (Becton and Dickinson, Diagnostic Instrument Systems, Sparks, MD, USA). Identification was performed using the ATB system, and sensitivity patterns were determined using the break-point method.

Definitions

1. The abnormal carrier state or carriage existed when the same AGNB strain, in any concentration, was isolated from at least two consecutive surveillance samples over a period of at least one week. Whilst rectal carriage of Escherichia coli was considered as normal, persistence of E. coli in the throat constituted an abnormal carrier state (Sarginson et al, 2001).
2. Sepsis was defined as the clinical signs of generalised inflammation caused by micro-organisms and/or their products. At least three of the following signs had to be present: pyrexia, tachycardia, hypotension, poor peripheral perfusion, lethargy and respiratory distress (Saez-Llorens & McCracken, 1993).
3. Septicaemia was defined as sepsis accompanied by a positive blood culture drawn through the catheter and/or from a peripheral vein (Pierro et al, 1996).
4. Overgrowth was defined as a concentration of AGNB in the throat and/or rectal swabs of 10⁵ bacteria/ml, ie high or very high growth density (Husebye, 1995).

Statistical analysis

Data were not normally distributed and were therefore expressed as medians. The Mann–Whitney and the Wilcoxon matched-pairs signed-rank non-parametric tests and Fisher's exact test were used for tests of the significance of differences between the abnormal and normal flora groups. A probability of less than 0.05 was considered significant.

Results

Patients

The total population of 208 patients comprised 123 male and 85 female infants. Their median gestational age was 37 weeks (range 25–45 weeks), median birth weight was 2.6 kg (range 0.73–4.28 kg), median age at the start of parenteral nutrition was 5 days (range 1–176 days) and median duration of parenteral nutrition was 13 days (range 2–512 days). Parenteral nutrition was administered for a total of 6271 patient days (a mean of 1254 parenteral nutrition days per year).

Incidence of carriage, sepsis and septicaemia in the total study population

Out of 208 infants abnormal carriage with AGNB occurred in 88 patients (42%). Overgrowth was detected in 68 of these (77%). A total of 198 blood cultures were taken from 52 patients (25%) with clinical evidence of sepsis, but only 79 were positive in 32 patients (15%).

Table 1 shows the AGNB carried by the 88 infants who developed the abnormal carrier state, and the AGNB isolated from the blood cultures of seven patients found to be septicaemic with those target micro-organisms.

Table 1 - Aerobic Gram-negative bacilli (AGNB) isolated from 88 abnormal carriers (42%) and from seven septicaemic children (3%).

Full table (15K)

Abnormal carriage predisposed to blood stream infection with coagulase-negative staphylococci (CNS) in 16 neonates of whom three had septicaemia with dual micro-organisms, CNS and enterococci. Four abnormal carriers developed blood cultures positive for S. pneumoniae (2), S. aureus (1) and Candida albicans (1). The remaining five neonates who developed septicaemia with CNS (4) and Candida albicans (1) were AGNB free.

Comparison between abnormal and normal flora carriers

When the abnormal and normal flora groups were compared there were no significant differences in gender, birth weight, gestational age, age at starting parenteral nutrition or usage of central venous catheters (Table 2). The infants who became carriers of AGNB received a median of 25 days of parenteral nutrition (1–512), whilst the patients who remained free of these abnormal bacteria received only 9 days of parenteral nutrition (1–58; P<0.0001). The total number of days at risk was 4895 in the abnormal flora group and 1376 in the group who carried normal flora. Both the incidence and rate of sepsis and of septicaemia were significantly higher in the group of infants with abnormal flora (Table 3).

Table 2 - Patient demographics in both normal and abnormal carriers.

Full table (12K)

Table 3 - Sepsis and septicaemia during parenteral nutrition.

Full table (17K)

Time relationship between onset of carriage, sepsis and septicaemia

Of the 32 infants with positive blood cultures, 27 (84%) belonged to the group with abnormal carriage. Only five patients (16%) developed septicaemia whilst remaining free of AGNB (P<0.005). In this subset of 32 infants, abnormal carriage with AGNB developed at a median of 14 days (range 1–113 days), signs of clinical sepsis at 19 days (1–104), and bacteriologically proven septicaemia at 23 days (1–118). The intervals between the onset of carriage, sepsis and septicaemia were significantly different; the P-value for the differences between onset of carriage and sepsis and between onset of carriage and septicaemia were <0.05 and <0.0003, respectively. Unlike the 27 abnormal carriers who developed 'late' septicaemia, at a median of 23 days, the five AGNB-free infants developed positive blood cultures 'early', at a median of 7 days.

Discussion

Three findings emerged from this study:

1. the incidence of AGNB carriage was 42%, whilst the sepsis and septicaemia rates were 25 and 15%, respectively. However, only 3% of the patients (7/208) had a septicaemia due to AGNB;
2. sepsis and septicaemia rates were significantly higher in the subset of abnormal carriers, 84% of infants who had septicaemia belonged to the subset of AGNB carriers;
3. carriage of AGNB developed significantly 'earlier' than sepsis and septicaemia.

Recently, scoring systems including the simplified acute physiology score (SAPS), the acute physiology and chronic health evaluation score (APACHE II), and paediatric risk and mortality score (PRISM) have become routine methods for estimating the disease severity in ill patients. Emergence of resistance amongst AGNB has prompted investigators to establish a link between the carriage of these micro-organisms and illness severity as measured by such scoring systems (Garrouste-Orgeas et al, 1996; Lortholary et al, 1995; Toltzis et al, 1997). A study in adult ICU patients reported that one third of patients with a mean SAPS score of 134.6 carried Acinetobacter baumannii and Klebsiella pneumoniae in the oropharynx and/or gut (Garrouste-Orgeas et al, 1996). In another study, high SAPS (12.35.3) and APACHE II (20.69.1) scores were identified as independent risk factors for rectal carriage of multi-resistant A. baumannii (Lortholary et al, 1995). A PRISM score of 6.454.85 predicted carriage of ceftazidime-resistant micro-organisms in a paediatric ICU (Toltzis et al, 1997). Additionally, the second study showed that patients who developed an infection with the multi-resistant A. baumannii were more ill than those who only became carriers (APACHE II 26.79.1 vs 17.79.2). In keeping with these observations, our finding that 84% of infants with a positive blood culture belonged to the subset of patients who carried abnormal flora suggests that the AGNB carrier state is a marker of severity of underlying disease, and hence of infection risk. Indeed, this observation is not new. As early as 1969, Johanson showed that the rate of oropharyngeal carriage of AGNB is proportionate to the severity of illness (Johanson et al, 1969). In critically ill neonates Harris found that systemic infection occurred only in these infants who initially and subsequently carried abnormal flora (Harris et al, 1976). In infants requiring intensive care Sprunt demonstrated an infection rate of only 0.5% with normal flora but 15% with abnormal carriage (Sprunt, 1985). These researchers believed surveillance cultures useful in identifying infants at risk.

Over 40% of our infants were so ill that they were unable to clear the AGNB acquired on the neonatal surgical unit, whilst the less ill did not become carriers despite exposure to the same micro-organisms. The hands of staff caring for neighbouring patients with AGNB are the main vehicle for transmission (Larson, 1981). Some may argue that the duration of exposure to the unit micro-organisms determines acquisition and subsequent carriage, but healthy carers infrequently develop gut carriage of the unit's AGNB despite exposure to them for decades (Chambers et al, 1987). Moreover, in almost all infants who acquired AGNB the carrier state occurred 'early' (median 14 days) denying duration of stay on the unit, per se, as a risk factor (Garrouste-Orgeas et al, 1996).

Our hypothesis is that the sicker the infant, the longer it takes to re-establish normal gut function. Lack of gut peristalsis in an abnormal carrier leads to overgrowth. The subset of patients with AGNB received parenteral nutrition for a median of 25 days, significantly longer than the median of 9 days for non-carriers. The need for and duration of parenteral nutrition is a reflection of a poorer clinical response and suggests that non-use of the gut is an important factor in septic morbidity related to parenteral nutrition. GALT atrophy (Li et al, 1995), impairment of secretory IgA mediated mucosal immunity (Kudsk et al, 1996) and cholestasis (Jawaheer et al, 1995) may be due to long term parenteral nutrition alone. Impaired mucosal defences and cholestasis predispose to the translocation of micro-organisms that are present in high concentration in the small intestine (Hadfield et al, 1995). Nonetheless, our study suggests that these conditions related to parenteral nutrition, on their own, are unlikely to result in septicaemia. Over a period of 5 y, only five patients, or one patient per year, suffered septicaemia whilst being free from AGNB. Gut overgrowth with AGNB is associated with an abnormally high gut endotoxin pool (van Saene et al, 1992), unlikely to be controlled by the GALT impaired by long term parenteral nutrition. AGNB overgrowth alone has been shown to cause liver dysfunction (Marshall et al, 1987; Billiar et al, 1988) and systemic immunity is impaired in proportion to the degree of that liver dysfunction (Alverdy & Burke, 1992). We suggest that it is small intestinal overgrowth with AGNB that is required for systemic immunosuppression (Marshall et al, 1987) that then allows viable translocating bacteria to 'spill over' into the blood.

Thus, when translocation promoted by parenteral nutrition is combined with gut endotoxin- induced systemic immunosuppression (Meyer et al, 1988), micro-organisms present in the gut, particularly in high concentrations, may traverse through the impaired GALT, including the liver, into the circulation. Recently, it has been reported that low level pathogens, including CNS, are able to translocate from the gut to the mesenteric lymph nodes (Ferri et al, 1997), and recent work in infants receiving long-term parenteral nutrition has shown that the bactericidal activity of the blood is initially impaired for CNS, whilst PPM including AGNB are still successfully eliminated (Okada et al, 2000). While systemic immunodysfunction remains moderate the patient develops Gram-positive septicaemias, mainly CNS, and it is only when liver dysfunction and systemic immunosuppression have become more pronounced, that AGNB—apart from their systemic immunosuppressive role—can survive translocation and host defences to make blood cultures positive. The potential pathogens including AGNB caused septicaemia significantly later than CNS, at a time when liver function was also significantly more impaired (Donnell et al, 2002). The fact that AGNB carriage precedes sepsis and septicaemia increases the evidence that gut overgrowth with abnormal AGNB contributes to septic morbidity.

The obvious question is what can be done to bolster the local and systemic defences impaired by parenteral nutrition and gut-derived endotoxin respectively to control septic morbidity? Feeding is the most important single factor in restoring both lines of defence. There is good evidence that enteral feeding is associated with better gut function including return of peristalsis (Hadfield et al, 1995) and gall bladder contractility (Jawaheer et al, 1995). Return of gut motility is required to clear overgrowth of AGNB (Vantrappen et al, 1977). There appears to be good evidence that enteral nutrition significantly reduces the incidence of infection (Minard & Kudsk, 1994) following the reversal of gut leakage (Mainous et al, 1991) and of defective macrophage function (Shou et al, 1994; Okada et al, 1998). Until the patient is able to tolerate enteral feeds, protection can be directed towards both gut and flora. Glutamine which is an essential nutrient for the small intestinal enterocyte has several benefits including reduction of gut permeability and atrophy (van der Hulst et al, 1996). More recently, epidermal growth factor has been shown to have a trophic effect on the small intestine (Fürst & Rombeau, 1996). Cholecystokinin may reverse cholestasis (Rintala et al, 1995), but experience suggests it is only effective once enteral feeding is significant. Our findings suggest that patients who receive parenteral nutrition and who become AGNB carriers might benefit from selective decontamination of the digestive tract (Baxby et al, 1996; Nathens & Marshall, 1999). Oral polymyxin and tobramycin have been shown to be effective in eradicating AGNB overgrowth and in neutralizing gut endotoxin (van Saene et al, 1996) and may reverse the immunosuppression related to parenteral nutrition (Houdijk et al, 1997; Yao et al, 1997). The answer to the question is unlikely to be one single manoeuvre but a combination of interventions (Deitch, 1990).

Finally, the criterion of AGNB carriage increases the homogeneity of a study population and we suggest that it should therefore be a prerequisite for the enrolment of infants into prospective trials aimed at reducing septicaemia related to parenteral nutrition. Patients with normal flora who do not become abnormal carriers are at low risk of infection.

References

1. Alverdy, JC & Burke, D (1992). Total parenteral nutrition: iatrogenic immuno-suppression. Nutrition, 8, 359–365.
2. Baxby, D, van Saene, HKF & Stoutenbeek, CPet al (1996). Selective decontamination of the digestive tract: 13 years on, what it is and what it is not. Intensive Care Med., 22, 699–706.
3. Billiar, TR, Maddaus, MA & West, MAet al (1988). Intestinal Gram-negative bacterial overgrowth in-vivo augments the in-vitro response of Kupffer cells to endotoxin. Ann. Surg., 208, 532–540.
4. Chambers, ST, Steele, C & Kunin, CM (1987). Enteric colonization with antibiotic resistant bacteria in nurses working in intensive care units. J. Antimicrob. Chemother., 19, 685–693.
5. Deitch, EA (1990). Selective decontamination of the gut. Is it enough?. Crit. Care Med., 18, 1043–1044.
6. Donnell, SC, Taylor, N & van Saene, HKFet al (2002). Infection rates in surgical neonates and infants receiving parenteral nutrition; a five year prospective study. J. Hosp. Infect., 52, 273–280.
7. Ferri, M, Gabriel, S & Gavelli, Aet al (1997). Bacterial translocation during portal clamping for liver resection. Arch. Surg., 132, 162–165.
8. Fürst, P & Rombeau, JL (1996). Growth factors and the intestine in critical illness. In:Gut Dysfunction in Critical Illness, ed. JL Rombeau & J Takala, pp375–393, Berlin: Springer
9. Garrouste-Orgeas, M, Marie, O & Rouveau, Met al (1996). Secondary carriage with multi-resistant Acinetobacter baumannii and Klebsiella pneumoniae in an adult ICU population: relationship with nosocomial infections and mortality. J. Hosp. Infect., 34, 279–289.
10. Hadfield, RJ, Sinclair, DG & Houldsworth, PEet al (1995). Effects of enteral and parenteral nutrition on gut mucosal permeability in the critically ill. Am. J. Respir. Crit. Care Med., 152, 1545–1548. | PubMed | ChemPort |
11. Harris, H, Wirtshafter, D & Cassady, G (1976). Endotracheal intubation and its relationship to bacterial colonization and systemic infection of new born infants. Paediatrics, 56, 816–823.
12. Houdijk, AP, Teerlink, T & Bloemers, FWet al (1997). Gut endotoxin restriction prevents catabolic changes in glutamine metabolism after surgery in the bile duct-ligated rat. Ann. Surg., 225, 391–400. | Article | PubMed | ISI | ChemPort |
13. Husebye, H (1995). Gastrointestinal motility disorders and bacterial overgrowth. J. Intern. Med, 237, 419–427.
14. Jawaheer, G, Pierro, A & Lloyd, DAet al (1995). Gall bladder contractility in neonates: effects of parenteral and enteral feeding. Arch. Dis. Child., 72, F200–F202.
15. Johanson, WG, Pierce, AK & Sanford, JP (1969). Changing bacterial pharyngeal flora of hospitalized patients. Emergence of Gram-negative bacilli. New Engl. J. Med., 281, 1137–1140.
16. Kudsk, KA, Renegar, KB & Li, J (1996). Loss of upper respiratory tract immunity with parenteral feeding. Ann. Surg., 223, 629–638.
17. Larson, EL (1981). Persistent carriage of Gram-negative bacteria on hands. Am. J. Infect. Control, 9, 112–119.
18. Li, J, Gocinski, B & Henken, Bet al (1995). Effects of parenteral nutrition on gut-associated lymphoid tissue. J. Trauma, 39, 44–52.
19. Lortholary, O, Fagon, JY & Hoi, ABet al (1995). Nosocomial acquisition of multiresistant Acinetobacter baumannii: risk factors and prognosis. Clin. Infect. Dis., 20, 790–796.
20. Mainous, M, Xu, D & Lu, Qet al (1991). Oral-TPN-induced bacterial translocation and impaired immune defences are reversed with refeeding. Surgery, 110, 277–284.
21. Marshall, JC, Lee, C & Meakins, JLet al (1987). Kupffer cell modulation of the systemic immune response. Arch. Surg., 122, 191–196.
22. Meyer, J, Yurt, RW & Duhaney, Ret al (1988). Differential neutrophil activation before and after endotoxin infusion in enterally vs parenterally fed volunteers. Surg. Gynecol. Obstet., 167, 507–509.
23. Minard, G & Kudsk, KA (1994). Effect of route of feeding on the incidence of septic complications in critically ill patients. Semin. Respir. Infect., 9, 228–231.
24. Nathens, AB & Marshall, JC (1999). Selective decontaminaton of the digestive tract in surgical patients. Arch. Surg., 134, 170–176.
25. Okada, Y, Klein, N, van Saene, HKF & Pierro, A (1998). Small volumes of enteral feedings normalise immune function in infants receiving parenteral nutrition. J. Paediatr. Surg., 33, 16–19.
26. Okada, Y, Klein, NJ & van Saene, HKFet al (2000). Bactericidal activity against coagulase-negative staphylococci is impaired in infants receiving long-term parenteral nutrition. Ann. Surg., 231, 276–281.
27. Pierro, A, van Saene, HKF & Donnell, SCet al (1996). Microbial translocation in neonates and infants receiving long-term parenteral nutrition. Arch. Surg., 131, 176–179.
28. Rintala, RJ, Lindahl, H & Pohjavuori, M (1995). Total parenteral nutrition-associated cholestasis in surgical neonates may be reversed by intravenous cholecystokinin: a preliminary report. J. Pediatr. Surg., 30, 827–830.
29. Saez-Llorens, X & McCracken, GH (1993). Sepsis syndrome and septic shock in pediatrics: current concepts of terminology, pathophysiology, and management. J. Pediatr., 123, 497–508.
30. Sarginson, RE, Taylor, N & van Saene, HKF (2001). Glossary of terms and definitions. Curr. Anaesth. Crit. Care, 12, 2–5.
31. Shou, J, Lappin, J & Minnard, EAet al (1994). Total parenteral nutrition, bacterial translocation, and host immune function. Am. J. Surg., 167, 145–150.
32. Sprunt, K (1985). Practical use of surveillance cultures for prevention of nosocomial infection. Semin. Perinatol., 9, 47–50.
33. Toltzis, P, Yamashita, Y & Vilt, Let al (1997). Colonization with antibiotic-resistant Gram-negative organisms in a pediatric intensive care unit. Crit. Care. Med., 25, 538–544.
34. van der Hulst, RRWJ, von Meyenfeldt, MF & Soeters, PB (1996). Glutamine: an essential amino acid for the gut. Nutrition, 12, 578–581.
35. van Saene, HKF, Stoutenbeek, CP & Faber-Nijholt, Ret al (1992). Selective decontamination of the digestive tract contributed to the control of disseminated intravascular coagulation in severe liver impairment. J. Pediatr. Gastroenterol. Nutr., 14, 436–442.
36. van Saene, JJM, Stoutenbeek, CP & van Saene, HKFet al (1996). Reduction of the intestinal endotoxin pool by three different SDD regimens in human volunteers. J. Endotoxin Res., 3, 337–343.
37. Vantrappen, G, Janssens, J & Hellemans, Jet al (1977). The interdigestive motor complex of normal subjects and patients with overgrowth of the small intestine. J. Clin. Invest., 59, 1155–1166.
38. Yao, YM, Yu, LR & Yu, Yet al (1997). Influence of selective decontamination of the digestive tract on cell-mediated immune function and bacteria/endotoxin translocation in thermally injured rats. J. Trauma, 42, 1073–1079.

Acknowledgements

We are indebted to the nursing and junior medical staff for taking samples, to the laboratory and infection control staff of the Departments of Microbiology and Biochemistry for their co-operation, and to Dr Paul Baines and Mr Pradeep Morar for their help with the statistical analyses. We are very grateful to Lisa Rent for her skillful typing of the manuscript. The review of our manuscript by Professor CA Hart is much appreciated.