The present systematic review and meta-analysis of eight RCTs (911 patients) showed that IC-guided energy delivery significantly reduced short-term mortality in critically ill adults compared with predictive equations. The IC-guided nutrition therapy achieved higher mean energy and protein intake per day and percent delivered energy over measured REE. Additionally, using IC-guided energy delivery did not prolong the MV duration, length of stay in ICU or hospital, and increase AEs and long-term mortality.
In this updated meta-analysis, we found a better prognosis in the IC-guided group patients. This finding contrasts with a most recent meta-analysis by Tatucu-Babet and colleagues [11]. They reported an IC-guided energy delivery regimen showed no between-groups difference regarding mortality (both hospital and ICU mortality) and ICU LOS but a longer duration of MV. However, the authors reviewed only four trials with a total of 398 patients [7, 20-22]. By contrast, our meta-analysis had a larger sample size than the previous meta-analysis as it added four other trials published recently [12-15, 23], with more power to assess this effect. In the 2018 ESPEN guideline on ICU nutrition [10], Singer et al. performed a meta-analysis of RCTs that focused only on using IC as a calorie target and found a trend (P=0.07) to improve the short-term mortality. Four RCTs were identified, of which two were included in the present meta-analysis [7, 20]. The two other trials were not included because patients recruited in the control group also received calories guided by IC measurements [9, 24].
Our findings support using IC rather than predictive equations as the gold standard to assess REE in ICU patients. Previous studies have demonstrated the low accuracy of various REE predictive equations based on weight, height, age, gender, etc., for critical illness [2, 25]. Firstly, compared with healthy individuals, REE predictive equations are frequently affected by disease characteristics such as respiratory failure, body temperature, severe trauma, burns, especially in obese or low-weight critically ill populations [8, 25]. Though adjusted by patient populations and modification factors, the estimated energy deviations still exist, and even high up to 60% [26]. Of note, researches have demonstrated that too high or too low calorie supply is associated with a worse prognosis [1].
On the other hand, caloric requirements in ICU patients may change during hospitalization. These changes may be irregular and unpredicted due to the phase of critical illness, nutritional support, analgesia, neuromuscular blockade, sedation, early rehabilitation, and other unknown factors [1]. As shown in the TICACOS study [7], the authors reported a significant day to day variation in measured REE by IC, though the mean EE was comparable between the IC-guided and the control groups. Similar findings were also mentioned in other included studies [7, 13-15]. Therefore, an IC-guided nutrition therapy allows for capturing the daily variations of REE and matching calorie supply and demand, thus avoiding the known adverse effects of under- and overfeeding due to the low precision of calculation-based REE strategy.
Our results suggested that besides accurate REE measurement, the implementation process of calorie supply is also important to achieve successful nutrition therapy. In the current study, the IC group received calorie intake closer to the measured targets than the control group (89-106% vs. 56-79%, respectively). This resulted in a higher cumulative caloric difference between the two groups (MD=622 kcal/day, P<0.00001). The cumulative calorie deficits may harm clinical outcomes, as shown in the study by Villet et al.[27], which reported a positive correlation between the number of infection complications and cumulative calorie deficits over four weeks of ICU stay. Another large retrospective study found that in patients receiving less than 40% of REE, the increase in energy debt harmed clinical outcomes. However, it is not the higher calorie delivery, the better. Zusman et al. investigated the relationship between calorie delivery and REE by IC [1]. Their results indicated a U-shaped relationship between caloric supply and mortality, with the best survival when around 70% REE was provided. This may help to explain the negative results of some included trials [7, 20]. In the TICACOS study, despite the large difference of cumulative energy balance between groups, the IC-guided group tend to overfeed, while the control group was underfed. Thus, a negative result was observed [7]. In contrast, in the EAT-ICU study [20], the median cumulative energy balance was closely related and close to and slightly below their target values (-249 and -747 kcal, respectively). Caloric supply in both groups was in the optimal range, and therefore the clinical results were indistinguishable between the two groups.
On the other hand, the calorie target supplied should vary according to different phases of critical illness. A previous large RCT (SPN study) measured REE by IC on day 3 to guide PN supplement and found the delivery of 100% of the energy target from days 4 with EN plus supplemental PN significantly reduced nosocomial infections [9]. Hypocaloric nutrition (less than 70% of EE) should be administered in the early phase of the illness [10]. Early full energy target feeding might lead to overfeeding and increase endogenous energy production. Moreover, the early phase of the illness is usually a period for resuscitation with frequent treatment adjustment for critically ill patients, which may hinder the accuracy of REE by IC. In the current study, some included RCTs reached 100% of the IC targets in the first 24 hours after randomization [7, 21]. Again, the overfeeding during the early period of critical illness may contribute to the increased MV duration, length of ICU or hospital stay among these trials, and even infection rate, thus diluting the benefit of REE guided by IC.
Additionally, we found no differences between the groups regarding the other clinical outcomes, such as duration of MV, ICU, and hospital stay, long-term mortality and AEs. The main explanation is that these are not the primary outcome among the most included studies. In fact, only a few or around half of the included trials have reported these outcomes. Second, for critically ill patients, ICU discharge was not always determined by the patients' condition. Third, only two trials [14, 20] had reported the long-term mortality of patients. The limited evidence suggested that IC guided strategy did not reduce long-term mortality. Therefore, more RCTs are required to explore the effect of ET on long-term prognosis.
The current meta-analysis provides evidence to support and expands the weak suggestion in the 2018 ESPEN guidelines, i.e., using IC-guided strategy during nutrition therapy in critically ill patients. However, our meta-analysis has some limitations. First, only eight studies were included in the current review. This may be explained by the infrequent application of IC in daily clinical practice. As shown in a large prospective study, only 0.8% of the more than 8,000 cases collected underwent IC measurement [28]. Moreover, we included only RCTs in the current meta-analysis to avoid the select bias of observational studies. Second, most of the included RCTs are unblinded and single-center in design and maybe likely underpowered to demonstrate actual differences in clinical and functional recovery outcomes. Third, there were differences among included trials with regards to the adopted IC measured devices, non-nutritional calories such as propofol and dextrose 5% used during the study period, enteral and parenteral formula, and patient intolerance of EN, which might result from the observed heterogeneity and thus compromise the robustness of our findings. Forth, the uneven distribution of different underlying diseases among included studies might also exert a prognostic value. For instance, one previous study showed that the higher the REE in severe sepsis, the higher the mortality. Finally, our results showed that the IC-guided group received more protein than the control group. However, we could not investigate the effect of protein delivery on ICU outcomes since it was not the study target in the current study. The timing and dose of protein delivery during critical illness remain unclear.