Chapter 87 Enteral and parenteral nutrition
It is standard practice to provide nutritional support to critically ill patients, in order to
However, despite the universality of this practice, the evidence underlying it is often conflicting and of disappointingly poor quality.1 The failings in the evidence seem to extend to some of the resulting debates, in which certainty appears inversely proportional to justification.2 Inevitably, these difficulties have led many to seek clarity in meta-analysis; perhaps equally inevitably,3 they have usually been disappointed. On so basic a question as the relative merits of enteral and parenteral routes of feeding, the two most recent meta-analyses have produced conflicting results.4,5 Instead of choosing on which trials to base patient care, it seems the clinician must now decide which meta-analysis to believe.
The problem persists with the publication of numerous clinical practice guidelines,6–11 which differ in important areas, although one at least has been used in a cluster randomised trial showing a 10% reduction in mortality that just failed to reach statistical significance in those intensive care units (ICUs) randomised to use the guideline – the ACCEPT study.6 While such validation does not mean that each component of the ACCEPT guideline (Figure 87.1) is optimal, it does at least provide some support.
Figure 87.1 Algorithm for nutritional support used in the ACCEPT trial.6 EN, enteral nutrition, PN, parenteral nutrition.
NUTRITIONAL ASSESSMENT
Objective assessment of nutritional status is difficult in ICU, because disease processes confound methods used in the general population. Anthropometric measures such as triceps skin-fold thickness and mid-arm circumference may be obscured by oedema. Voluntary handgrip strength, a test of functional capacity, is impractical in unconscious patients. Laboratory measures, including transferrin, pre-albumin and albumin levels, lymphocyte counts and skin-prick test reactivity, are abnormal in critical illness. Clinical evaluation – the so-called Subjective Global Assessment – is better than objective measurement at predicting morbidity.12 Historical features of malnutrition include weight loss, poor diet, gastrointestinal symptoms, reduced functional capacity and a diagnosis associated with poor intake. Physical signs include loss of subcutaneous fat, muscle wasting, peripheral oedema and ascites.
While laboratory measures are of little value in assessing nutritional status in critically ill patients, intensivists are increasingly involved in preoperative assessment of patients undergoing major surgery. In these patients the serum albumin and operative site were closely associated with the risk of postoperative complications.13 This raises the possibility that outcomes may be improved by treating preoperative malnutrition identified by a simple screening test.
PATIENT SELECTION AND TIMING OF SUPPORT
There are reasonable grounds to believe that it is better to provide nutritional support to critically ill patients than not to do so. This belief is based on the close association between malnutrition, negative nitrogen and calorie balance and poor outcome, and the inevitability of death if starvation continues for long enough. In otherwise healthy humans this takes several weeks to occur. There is also some direct evidence from small studies of parenteral nutrition in patients with severe head injuries14 and of jejunal feeding in those operated on for severe pancreatitis,15 in which the control groups received little or no nutritional support. Both studies showed decreased mortality in the groups receiving adequate nutrition.
Quite good evidence now supports the early institution of nutritional support, and the trend is both to tolerate much shorter periods without nutrition and to begin feeding more rapidly after initial resuscitation.
In 1997, recommendations from a conference sponsored by the US National Institutes of Health, the American Society for Parenteral and Enteral Nutrition and the American Society for Clinical Nutrition suggested that nutritional support be started in any critically ill patient unlikely to regain oral intake within 7–10 days.10 The basis for this was that at a typical nitrogen loss of 20–40 g/day dangerous depletion of lean tissue may occur after 14 days of starvation. Others have suggested a maximum acceptable delay of 7 days. A meta-analysis comparing early (first 48 hours after admission to ICU) with late enteral feeding revealed a reduction in infectious complications.16 Two subsequent meta-analyses comparing early enteral feeding with no artificial nutritional support, and early parenteral feeding with delayed enteral feeding, both found a reduction in mortality with early support.17 Finally, early institution of enteral feeding was an important component of the ACCEPT study guideline.6 Patients in this study received nutritional support if they were thought unlikely to tolerate oral intake in the next 24 hours, and the goal was to start feeding within 24 hours of admission to ICU. The weight of evidence is presently in favour of this more aggressive approach.18
NUTRITIONAL REQUIREMENTS OF THE CRITICALLY ILL
ENERGY
Some muscle wasting and nitrogen loss are unavoidable in critical illness, despite adequate energy and protein provision.19 This fact, coupled with the realisation that caloric requirements had previously been overestimated, has led to downward revision of intake, a process which may still be ongoing. In 1997, the American College of Chest Physicians (ACCP) published guidelines recommending a daily energy intake of 25 kcal/kg,9 and this has remained the standard target energy intake for critically ill patients. More recently, concerns have been raised that this may be excessive. An observational study found lower mortality in those patients who received 9–18 kcal/kg/day than in those with higher and lower intakes.20 However, meaningful benefits of hypocaloric feeding have yet to be demonstrated in prospective trials. It is, moreover, extremely important to realise that enterally fed patients frequently fail to achieve their target intake, and that significant underfeeding is certainly associated with worse outcomes.20–22
Indirect calorimetry is the gold standard, and its use is becoming easier with the availability of devices designed for ICU patients. It permits measurement of the resting energy expenditure (REE). This value excludes the energy cost of physical activity, which increases later in the course of an ICU admission.23 Calorimetry reveals deviations from values predicted by equations, such that two thirds of patients in one study were being either under- or overfed.24 On the other hand, it could not be shown that outcomes are improved by the use of calorimetry.25 Moreover, there are no clear data to relate measured REE to total energy expenditure in the individual patient. As a result, many units do not use calorimetry; in those that do, a target energy provision of 1.3 × measured REE is usual.22
There are a large number of equations claiming to predict basal metabolic rate (BMR) on the basis of weight, sex and age. The best known is the Harris–Benedict equation, which dates back more than 80 years. Schofield’s equations were derived anew in the 1980s.26 Correction factors exist to convert predictions of BMR into estimated energy expenditure by adjusting for such variables as diagnosis, pyrexia and activity. In the past these correction factors have been excessive and may have contributed to overfeeding; a more conservative approach is now advocated. The recommendations of the British Association for Parenteral and Enteral Nutrition are:27
Table 87.1 Basal metabolic rate in kcal/day by age and gender26
| Age | Female | Male |
|---|---|---|
| 15–18 | 13.3 W + 690 | 17.6 W + 656 |
| 18–30 | 14.8 W + 485 | 15.0 W + 690 |
| 30–60 | 8.1 W + 842 | 11.4 W + 870 |
| > 60 | 9.0 W + 656 | 11.7 W + 585 |
W = weight in kg.
Table 87.2 Stress adjustment in the calculation of basal metabolic rate27
| Partial starvation (> 10% weight loss) | Subtract 0–15% |
| Mild infection, inflammatory bowel disease, postoperative | Add 0–13% |
| Moderate infection, multiple long bone fractures | Add 10–30% |
| Severe sepsis, multiple trauma (ventilated) | Add 25–50% |
| Burns 10–90% | Add 10–70% |
PROTEIN
Assessment of nitrogen balance by measuring urinary urea nitrogen is too variable to be useful in estimating protein requirements in ICU.28 As there is an upper limit to the amount of dietary protein that can be used for synthesis,29 there is no benefit from replacing nitrogen lost in excess of this. A daily nitrogen provision of 0.15–0.2 g/kg per day is therefore recommended for the ICU population; this is equivalent to 1–1.25 g protein/kg per day. Severely hypercatabolic individuals, such as those with major burns, are given up to 0.3 g nitrogen/kg per day, or nearly 2 g protein/kg per day.27
MICRONUTRIENTS
Critical illness increases the requirements for vitamins A, E, K, thiamine (B1), B3, B6, vitamin C and pantothenic and folic acids.30 Thiamine, folic acid and vitamin K are particularly vulnerable to deficiency during total parenteral nutrition (TPN). Renal replacement therapy can cause loss of water-soluble vitamins and trace elements. Deficiencies of selenium, zinc, manganese and copper have been described in critical illness, in addition to the more familiar iron-deficient state. Subclinical deficiencies in critically ill patients are thought to cause immune deficiency and reduced resistance to oxidative stress. Suggested requirements for micronutrients in critically ill patients vary between authors and depending on route of administration; the most comprehensive guidance30 is reproduced in Tables 87.3 and 87.4. More recent but broadly similar recommendations for some compounds are also available.31
Table 87.3 Vitamin requirements in critical illness30
| Vitamin | Function | Dose |
|---|---|---|
| Vitamin A | Cell growth, night vision | 10000–25000 IU |
| Vitamin D | Calcium metabolism | 400–1000 IU |
| Vitamin E | Membrane antioxidant | 400–1000 IU |
| β-Carotene* | Antioxidant | 50 mg |
| Vitamin K | Activation of clotting factors | 1.5 μg/kg per day |
| Thiamine (vitamin B1) | Oxidative decarboxylation | 10 mg |
| Riboflavin (vitamin B2) | Oxidative phosphorylation | 10 mg |
| Niacin (vitamin B3) | Part of NAD, redox reactions | 200 mg |
| Pantothenic acid | Part of coenzyme A | 100 mg |
| Biotin | Carboxylase activity | 5 mg |
| Pyridoxine (vitamin B6) | Decarboxylase activity | 20 mg |
| Folic acid | Haematopoiesis | 2 mg |
| Vitamin B12 | Haematopoiesis | 20 μg |
| Vitamin C | Antioxidant, collagen synthesis | 2000 mg |
Table 87.4 Trace element requirements in critical illness30
| Element | Function | Dose |
|---|---|---|
| Selenium | Antioxidant, fat metabolism | 100 μg |
| Zinc | Energy metabolism, protein synthesis, epithelial growth | 50 mg |
| Copper | Collagen cross-linking, ceruloplasmin | 2–3 mg |
| Manganese | Neural function, fatty acid synthesis | 25–50 mg |
| Chromium | Insulin activity | 200 mg |
| Cobalt | B12 synthesis | |
| Iodine | Thyroid hormones | |
| Iron | Haematopoiesis, oxidative phosphorylation | 10 mg |
| Molybdenum | Purine and pyridine metabolism | 0.2–0.5 mg |
WATER AND ELECTROLYTES
Water and electrolyte requirements vary widely depending on the patient’s condition; typical basal intakes are shown in Table 87.5.
Table 87.5 Water and electrolyte requirements/kg per day
| Water | 30 ml |
| Sodium | 1–2 mmol |
| Potassium | 0.7–1 mmol |
| Magnesium | 0.1 mmol |
| Calcium | 0.1 mmol |
| Phosphorus | 0.4 mmol |
ROUTE OF NUTRITION
When possible patients should be fed enterally. The advantages over the parenteral route are:
These appear to be the only advantages of the enteral route. Despite the fervour with which some pursue the debate,2 there is little basis for the widespread belief that the enteral route provides a clear benefit in terms of outcome, and that the advantages of enteral feeding are unassailable.
Two hypotheses are commonly advanced in support of the putative superiority of enteral feeding. First, it appears that the lipid contained within TPN is immunosuppressive. Intravenous lipid is known to suppress neutrophil and reticuloendothelial system function, and a comparison of TPN with and without lipid in critically ill trauma patients showed a lower complication rate in those not receiving lipid.32 Second, enteral feeding may protect against infective complications. Absence of complex nutrients from the intestinal lumen is followed in rats by villus atrophy and reduced cell mass of the gut-associated lymphoid tissue (GALT). Starved humans show these changes to a much lesser extent. Lymphocytes produced in the GALT are redistributed to the respiratory tract, and contribute heavily to mucosal immunity. In mice, this contribution is lost during TPN.
The possibility that multiple organ failure may be driven by translocation of bacteria or endotoxin across an impaired mucosal barrier has been extensively investigated in animals. While it is known that TPN is associated with increased gut permeability to macromolecules in humans, this does not seem to result in translocation.33 Although translocation does occur following surgery, and seems to be associated with sepsis;34 a causal relation with multiple organ failure is unproven. In fact, a reduction in septic morbidity has only been found in certain groups, primarily abdominal trauma victims,35,36 in whom parenteral nutrition was associated with a higher incidence of abdominal abscess and pneumonia. A third study found no difference.37 In head-injured patients there is one trial showing no effect and one each supporting either route; however in the study favouring TPN the enteral nutrition group were grossly underfed.14,38,39
None of these studies is less than 15 years old, and the techniques of both enteral and parenteral feeding have changed a great deal in that time. Reductions in septic complications have also been found using enteral feeding in pancreatitis.40 In contrast, no benefit was found in sepsis, though enteral feeding was instituted late.41 A review of 31 clinical trials comparing enteral with parenteral feeding found no consistent difference.42
More recent systematic analyses have, as mentioned earlier, produced conflicting results. One found a reduction in infectious complications with enteral feeding, but no difference in mortality.5 The most recent and most robust meta-analysis considered only high-quality trials using an intention-to-treat principle. It showed a clear reduction in mortality in patients fed parenterally.4
The same authors performed a prospectively defined subgroup analysis comparing early enteral with parenteral feeding, which showed no difference in mortality. Another recent meta-analysis of this question reached the same conclusion.43 On this basis, as in the ACCEPT study, the authors recommended early use of the enteral route, with rapid recourse to parenteral nutrition if this was not possible.6,17 At present it appears that timing of nutritional support matters as much as the route used.
A well-powered randomised study comparing early parenteral with enteral feeding will soon start.18 However, in view of the practical and financial advantages of enteral feeding, it will probably need to find a significant mortality difference in favour of the parenteral route if it is to change the present preference for enteral feeding.
