The role of nutritional support in the physical and functional recovery of critically ill patients: a narrative review

In general, critically ill patients do not meet recommended levels of nutritional intake, particularly when the enteral route is used alone [ 36]. This is true both in routine clinical practice [ 36] and in the setting of RCTs [ 2– 4]. The effect of underfeeding during the period of critical illness on skeletal muscle wasting and physical function is wholly unclear. One-year follow-up from the EDEN trial (Early vs Delayed Enteral Feeding to Treat People with Acute Lung Injury or Acute Respiratory Distress Syndrome) suggested that there was no beneficial effect on physical function from target compared with trophic enteral feeding over the first 6 days of critical illness [ 37], albeit in the context of a number of confounders that would need further consideration. However, more patients in the trophic feeding group were discharged to rehabilitation centres, suggesting that there may be some beneficial effect to improving nutritional intakes [ 37]. Noteworthy in this trial is that patients in the full feeding group only met 70% of the energy targets which may not be sufficient to produce an outcome benefit, at least when predictive equations are used [ 30]. In contrast, a sub-group analysis from the Reducing Deaths due to Oxidative Stress Study (REDOXs) found that increasing nutritional adequacy led to improvements in 3-month Short Form-36 (SF-36) scores relating to the physical domains. However, this effect was diminished by 6 months [ 38]. Other large RCTs have also included physical or quality of life outcomes with varying results (Table  1) [ 2, 3, 31, 38– 42].

Two pre-planned sub-group analyses from the EPaNIC study investigated the impact of the macronutrient dose (in the form of early vs late supplemental PN) on rates of skeletal muscle wasting [ 31, 39]. The first of these [ 31] found that muscle wasting, measured from muscle biopsies, was not different between the two groups. In addition, using the Medical Research Council (MRC) sum-score, weakness was found to recover faster in the group receiving late PN. In the second of these sub-group analyses [ 39], early PN was shown to adversely impact on femoral muscle quality, measured using computed tomography (CT) scans, but did not affect the rates of wasting observed in 15 neurosurgical patients.

It is likely that the timing and dose of energy provision go hand in hand. Indeed, recent thinking suggests that consideration of the endogenous production of energy in early critical illness is essential to the timing and dose of nutritional supplementation [ 1]. However, with no bedside method to measure endogenous energy production, it is impossible to account for this when calculating energy expenditure and devising feeding regimens. It has been postulated that in early critical illness (e.g. within the first 72–96 hours) permissive underfeeding to approximately 15 kcal/kg with full protein nutrition support may be warranted [ 43], but this awaits confirmation of benefit in RCTs. In addition, the use of predictive equations to determine energy targets may heavily influence the results of nutrition trials in the ICU as they are known to produce results which are less accurate than measured energy expenditure (MEE) using indirect calorimetry [ 44]. Indeed, studies feeding to MEE have consistently shown positive benefits and a recent observational study found that feeding to 70% of MEE was optimal in terms of mortality [ 45]. However, limitations preclude the frequent use of indirect calorimetry in clinical practice. These include availability of accurate metabolic monitors, costs, time taken to undertake the measurement and specific exclusions meaning that some of the sickest patients are unsuitable for measurement (e.g. those on continuous renal replacement therapy and those with high oxygen requirements) [ 46]. However, the introduction to the market of a metabolic monitor designed specifically for mechanically ventilated patients, with a reasonable cost, is under development and may bypass some of these limitations for future trials [ 46]. This is particularly pertinent as the effect of this targeted energy feeding on physical and functional recovery remains unknown.

Inadequate protein provision has been considered a contributing factor explaining why RCTs, such as the EDEN trial mentioned earlier [ 3], do not show any beneficial impact of nutrition in the critically ill [ 47]. Early studies investigating protein intake in critically ill patients reported an improvement in whole-body nitrogen balance or whole-body protein turnover, with higher protein intakes [ 48]. Since then, several large observational studies have reported mortality benefits when higher protein delivery is achieved [ 49– 51]. Whilst this may, in part, be because less sick patients may be able to have more protein delivered, this important confounder is accounted for in many of the more recent studies. For this reason, current recommendations range between 1.2 and 2.5 g/kg/day [ 8]. Whilst it seems plausible that higher protein delivery may attenuate skeletal muscle loss, the data supporting enhancement of muscle strength and function are lacking [ 52]. Secondary outcome results relating to the physical function component of the SF-36 score from the Nephro-Protective Trial [ 53], investigating the effect of intravenous amino acid supplementation on development of acute kidney injury, are awaited to contribute to the current evidence base.

One recent RCT investigated the effect of different protein intakes on muscle strength, wasting and fatigue in critically ill patients receiving PN [ 42]. In this study, 119 patients were randomised to receive 0.8 or 1.2 g/kg protein. There was no difference in the primary outcome of handgrip strength at ICU discharge. However, despite a smaller than planned difference in the delivery of protein (0.9 g/kg vs 1.1 g/kg), the study found that a higher protein intake resulted in differences in secondary outcomes including greater handgrip strength at day 7, improved measures of forearm muscle thickness and rectus femoris cross-sectional area and reduced fatigue scores. These results support the concept that a higher protein intake, at least when supplied via the parenteral route, leads to a reduction in muscle wasting during the first week of critical illness. However, such preliminary findings await confirmation in larger studies that would, in particular, need to correct for baseline heterogeneity, as these results are in contrast to observational data from the EPaNIC Study [ 31, 39] and from the MUSCLE-UK group where higher protein delivery was observed to be associated with greater skeletal muscle wasting [ 16].

Taken together, these data have led to the hypothesis that it may not be the amount of protein delivered, but the way in which we deliver the feed in a continuous manner that drives skeletal muscle wasting [ 32]. In healthy subjects, muscle protein synthesis increases from 45 to 90 minutes after provision of amino acids, either oral or intravenous, but then decreases after 90 minutes [ 54, 55]. This effect is observed despite the continued availability of amino acids in both the plasma and muscle, and has been termed the ‘muscle full effect’. It is not unreasonable to consider that this effect is also relevant in critically ill patients, and this hypothesis underpins the rationale for the current multi-centre RCT comparing intermittent and continuous feeding to investigate the effect on skeletal muscle wasting [ 35].

One of the major challenges in demonstrating benefit from nutrition in critically ill patients is the heterogeneous nature of the clinical population. In this regard it is unlikely that all patients will benefit from the same treatment at the same time point. Whilst feeding protocols are recommended as a means of encouraging early enteral feeding, a ‘one size fits all’ approach to the treatment of critically ill patients is no longer considered appropriate. However, defining patients who are likely to benefit from a nutritional intervention will be challenging and will require rigorous investigation. One method currently suggested to distinguish those patients who may benefit from a targeted nutrition intervention from those who may not is the determination of nutrition risk. The most popular nutritional risk scoring systems for critically ill patients are the Nutrition Risk Score 2002 (NRS 2002) [ 56] and the Nutrition Risk in the Critically Ill (NUTRIC) score [ 57, 58]. Although recent clinical trials have used the NRS 2002 in their inclusion criteria, this score may lack the specificity to determine the true nutrition risk of critically ill patients as an APACHE II score > 10 automatically gives the highest risk score. Although the NUTRIC score was developed specifically for critically ill patients, it is yet to be validated prospectively. However, post-hoc analyses from the PermiT (The Permissive Underfeeding versus target Enteral Feeding in Adult Critically Ill Patients) trial did not show a mortality improvement when stratifying by this score [ 59]. In addition, these scoring systems do not include variables relating to muscle mass or baseline physical function therefore more work is required in the area of nutrition risk before it can be used as a criterion in future nutrition studies.

Potential tools for patient selection in clinical trials include those to determine both baseline skeletal muscle quantity and quality and change over time. Such tools include muscle ultrasound and CT. However, there are limitations with the use of these tools outside the research setting. First, there is currently no standardised, universal technique to perform muscle ultrasound [ 60]; and second, only patients who have required a CT scan for clinical purposes have been investigated, meaning that a selection bias may be present [ 61]. Nonetheless, with further work to standardise techniques, these measures may prove useful in the future either on their own or as an adjunct to existing nutrition risk tools. Lastly, if physical or functional outcomes are to be included as important clinical outcomes, then the limitations surrounding obtaining baseline measurements for non-elective ICU admissions will be equally as important and further work needs to be undertaken to understand this in detail.

Ensuring adequate delivery of the intervention is an essential factor in being able to interpret the outcome of such studies. Several studies in critical care nutrition have not been successful in reaching either target energy or protein delivery (Table  1) [ 2– 4, 7, 40– 42]. Notwithstanding that there may be physiological differences in the utilisation of substrates depending on the route and timing of nutrition delivery, strategies to enhance nutrition delivery [ 62, 63] should be factored into clinical trials where the outcome is dependent upon meeting a target.

Biological plausibility is fundamental to any nutrition research. This includes the likelihood that an outcome will be observed at a particular time point as a result of the specified duration of an intervention. Effects from nutrition dosing are unlikely to be observed immediately and any effect will be reliant on the consistency of delivery of the intervention. For example, the EDEN trial reported no difference in physical outcomes at 1 year when trophic or full feeding was given for 6 days [ 37]. Given that current data suggest that oral intake is inadequate post extubation [ 64, 65] and in the post-ICU recovery phase [ 66], it may be unlikely that an effect would be observed 1 year after such a short intervention. Observational studies tend to include patients fed over a longer period of time and exclude those patients fed over shorter durations. Because of the risk of bias associated with the short duration of trials, we propose that critical care nutrition studies should continue for longer than the first week of critical illness and should consider post-ICU nutritional intake.