Monitoring and Evaluating Nutrition Support in the Critically Ill

Abstract

With a growing body of research suggesting the importance of nutrition in critical illness, there is a need for ways to know just how much nutrition to give. For the majority whose ICU stay is short, this might not be of great concern, but it is of crucial importance for longer-stay critically ill patients for whom unnecessary catabolism could be detrimental but who are also at high risk of complications from overfeeding. Guidelines for the provision of energy and protein are not based on a high level of evidence; available measures of energy and protein consumption have significant limitations, and there are no established ways to assess when an individual patient is receiving optimal nutrition during critical illness. Of the variety of available techniques in use for assessing and monitoring nutritional status, most are not easily applicable in the intensive care unit (ICU) setting due to the bodily changes that occur in critical illness or due to the limitations of the ICU setting itself, and many have not been adequately validated for their purpose. Tools for nutritional assessment are usually validated against a gold standard measure of nutritional status, and it is a relatively straightforward process to evaluate the sensitivity and specificity of such a tool against the gold standard tool. In contrast, it is less straightforward to validate a tool for nutritional monitoring. To be meaningful, the validation process would need to assess the tool against patients’ outcomes in a longitudinal study, ideally one that is randomised and controlled to eliminate possible confounders. Such outcomes should be clinically meaningful and patient-focused, that is, measures of how the patient feels, functions or survives. There would be little value in a tool that indicates a significant nutritional improvement without resulting in any better outcomes for the patient. The aim of this thesis is to investigate dietitians’ practices in nutritional management of ICU patients, to evaluate current approaches to nutritional monitoring in critical illness, and to identify methods that could be used to monitor nutritional progress during critical illness, with a view to developing a valid tool for this purpose. Protein was chosen as the focus for this monitoring as it has not been well studied, so this thesis also aims to explore current practices in provision of protein to ICU patients and compare these to the existing evidence on protein requirements in the critically ill and ways of measuring the effects of protein intake. In order to establish current nutritional management practices amongst ICU dietitians, an 18- question survey tool was developed. The survey included validated measures of dietitian involvement in the ICU team, as this parameter was identified a priori as a potential confounder of the survey responses. The survey was then administered by telephone to at least one dietitian from each of the 182 ICUs in Australia and New Zealand. The findings of this project indicate that none of the interviewed dietitians are using methods of monitoring nutritional support that have been validated against patient outcome. Overall 15 indicators were identified that Australasian dietitians report using to assess and monitor nutrition in their critically ill patients. The results of the survey also indicated that there is a wide variation in nutritional practices in ICU; one example in particular is the wide range of target protein intakes for ICU patients which in current practice ranges threefold from 0.75 g/kg (the RDI for healthy Australians and New Zealanders) to more than 2 g/kg body weight. To identify additional methods of monitoring nutritional progress in ICU, a literature search was undertaken, focusing on journal articles that reported nutrition-related studies in a critically ill population. There were 1027 studies found, that mentioned 53 different nutritionrelated parameters, and these included the 15 that dietitians had reported using. All of these nutritional indicators were then assessed for their potential application in a critically ill population, considering both feasibility and availability. The validity of these indicators was then investigated using a systematic review methodology asking the question ‘What nutritional parameters predict clinically significant outcomes in randomised studies conducted in a critically ill population?’ As there are no randomised controlled trials specifically designed to validate the use of any of the common nutritional parameters, the review identified trials of any nutrition-related intervention and then looked at whether these trials also recorded any nutritional monitoring parameters that might be expected to respond to improved nutrition, and whether these parameters were associated with improvements in clinically meaningful outcomes such as mortality, length of hospital or ICU stay, or incidence of infectious complications. Twenty-nine studies were found that did not report any significant difference in outcomes. Twenty studies showed a significant improvement in outcome as well as reporting nutritional monitoring parameters like serum albumin, prealbumin (transthyretin), retinol-binding protein, transferrin and lymphocytes. Of these, the number of studies supporting a statistical relationship between outcome and particular nutritional indicators was equal to, or outnumbered by, the studies not supporting such a relationship. The results of this systematic review indicated that none of the frequently-used parameters consistently predicts clinically significant outcomes in the critically ill. The survey of dietitians had revealed a wide range of target protein intakes used for ICU patients, ranging from the NHMRC’s RDI to more than 2 g/kg body weight. Standard parenteral nutrition products generally provide only the NHMRC amount of amino acids when guideline energy intakes are achieved, and only specialised solutions meet guideline goals for both energy and protein. To establish current practice at the candidate’s hospital, an audit looked at the last 300 patients who received parenteral nutrition. Due to institutionspecific factors, all 300 patients received the same standard three-in-one parenteral nutrition admixture (providing about an RDI level of 0.8 g/kg protein), for median 8 days (interquartile range 6-14 days). The incidence of complications such as line sepsis and metabolic abnormalities was low. The standard solution did not meet the protein requirements of surgical or critically ill patients with a mean deficit of 12 g protein per day. Twenty-seven patients died while receiving parenteral nutrition. Mortality was significantly higher in ICU patients (odds ratio 5.84; 95%CI 1.66-20.57, p=0.006), those requiring any type of dialysis (OR 7.34; 05%CI 1.30-18.49, p<0.001). and in non-surgical patients (OR 2.57; 95%CI 1.07- 6.15, p=0.03). It was not possible to control for severity of illness in this analysis, but it confirmed that standard parenteral nutrition products do not allow ICU patients to reach the level of amino acids recommended in well-accepted ICU guidelines. A systematic review of the literature was then conducted to establish the evidence base for these recommendations for protein intakes in the critically ill. Relevant databases were systematically searched for randomised controlled trials comparing at least two levels of protein intake in a critically ill population, which were then assessed for quality using a previously-published quality ranking scale. Eleven studies were included in this review, reporting on a range of ICU patient groups and levels of protein intake. All but two of the studies made comparisons only within the guideline-recommended range, and the remaining two were both limited by the major confounding factor of differing energy intakes between groups. In most of the studies, nitrogen balance was the only outcome that improved with increased protein. The results of the review indicated that estimation of protein requirements is complex, and recommendations in guidelines are not based on high-level evidence or patient-focused outcomes. A randomised controlled trial was then conducted to compare two different levels of protein intake in 119 critically ill patients who were randomised to receive amino acids at either 1.2 g/kg or the RDI level of 0.8 g/kg body weight from approximately isoenergetic parenteral solutions. In order to identify methods that could possibly be used to monitor nutritional progress during critical illness, a wide range of outcome variables was measured in the first few days of ICU admission, at study days 0, 3 and 7. The higher amount of amino acids was associated with small improvements in handgrip strength (22.09(SD 10.14) kg vs 18.47(11.82) kg, p=0.025 at study day 7), less fatigue (measured using the Chalder fatigue score: 5.35(SD 2.21) vs 6.19(SD2.17), p = 0.045) and greater forearm muscle thickness on ultrasound (3.2(0.4) cm vs 2.8(0.4) cm, p<0.0001), as well as better nitrogen balance. There was no difference between the groups in mortality or length of stay measures. The trial employed a variety of the previously evaluated nutritional indicators, and explored their relationship with intake and patient outcome, identifying four parameters that appear to be both feasible and valid for monitoring nutritional progress as they significantly correlate with both outcome and intake. These were handgrip strength, organ function score (SOFA), serum albumin level, and ultrasound measurement of muscle thickness (sum of three body sites). This analysis was inconclusive in developing a monitoring tool, and future research is recommended. This thesis is unique in comprehensively evaluating the currently-used methods of nutritional monitoring in the critically ill and identifying potential candidate measures for this purpose. Further research is needed to validate these findings in the various groups that make up the heterogeneous ICU population