Early or late parenteral nutrition: ASPEN vs. ESPEN

group had a relative increase, of 6.3%, in the likelihood of being discharged earlier, and alive, from the ICU (hazard ratio 1.06; 95% confi dence interval [CI] 1.00-1,13; P = 0.04). Rates of death in the ICU and survival at 90 days were similar between the two groups. Th e lateinitiation group, as compared to the early-initiation group, had fewer ICU infections (22.8% v. 26.2%; P = 0.008), less days of renal replacement therapy (7 days (interquartile range [IQR] 3-16) v. 10 days (IQR 5-23); P = 0.008) and fewer patients requiring more than 2 days of mechanical ventilation (36.3% v. 40.2%; P = 0.006). Conclusions Late-initiation of PN was associated with faster recovery and fewer complications, when compared with earlyinitiation.


Background
Controversy exists about the timing of the initiation of parenteral nutrition (PN) in critically ill adults in whom caloric targets cannot be met by enteral nutrition (EN) alone.

Methods
Objective: To compare early-initiation of PN (European guidelines) with late-initiation (American and Canadian guidelines) in adults who are receiving insuffi cient enteral nutrition in the intensive care unit (ICU). Design: Prospective, randomized, controlled, parallelgroup, multicenter clinical trial. Setting: Seven multidisciplinary ICUs in Belgium. Subjects: All adults admitted to participating ICUs with a nutritional risk score of 3 or more who did not meet any exclusion criteria. Intervention: After enrollment, 2312 patients were randomized to receive PN 48 hours after ICU admission (early-initiation) and 2328 patients were randomized to receive PN on day 8 (late-initiation group). Both groups received early EN using a standardized protocol. PN was continued until EN met 80% of calorifi c goals, or when oral nutrition was resumed. It was restarted if enteral or oral feeding fell below 50% of calculated calorifi c needs. Outcomes: Primary end point was the duration of dependency on intensive care, defi ned as the number of intensive care days and time to discharge from the ICU.

Results
Th e median stay in the ICU was one day shorter for the late-initiation group (3 v. 4; P = 0.02). Th e late-initiation group had a relative increase, of 6.3%, in the likelihood of being discharged earlier, and alive, from the ICU (hazard ratio 1.06; 95% confi dence interval [CI] 1.00-1,13; P = 0.04). Rates of death in the ICU and survival at 90 days were similar between the two groups. Th e lateinitiation group, as compared to the early-initiation group, had fewer ICU infections (22.8% v. 26.2%; P = 0.008), less days of renal replacement therapy (7 days (interquartile range [IQR] 3-16) v. 10 days (IQR 5-23); P = 0.008) and fewer patients requiring more than 2 days of mechanical ventilation (36.3% v. 40.2%; P = 0.006).

Conclusions
Late-initiation of PN was associated with faster recovery and fewer complications, when compared with earlyinitiation.

Background
Critically ill patients are often unable to feed thems elves, and frequently present in a fasting state. Failure to feed these patients eventually guarantees starvation and, if enough tim e lapses, death. However, it is less certain when fasting becomes starvation. Th is is important because starvation has deleterious eff ects, including immune and organ dysfunction [1]. Further, the critically ill patient is already in a catabolic crisis [2].We therefore expect early feeding to be benefi cial. Indeed, a metaanalysis in 2001 demonstrated that early EN is associated with less infections (relative-risk 0.45; CI 0.30-0.66; P < 0.001) and shorter hospital stays (mean-reduction 2.2 days; CI 0.81-3.3; P = 0.004). [3] Nutrition guidelines on both sides of the Atlantic agree that early EN is important [4][5][6].
Achieving calorifi c goals with EN is challenging in the critically ill. In one study 40% of ICU patients received no nutrition, while the remaining only received, on average, 58% of their calorifi c goals [2]. Th is is because critically ill patients frequently experience gastroparesis and unneces sary delays waiting for post-pyloric feeding tubes and/or bowel sounds, despite contrary evidence [7,8].
Supplementing with PN allows practitioners to reach calorifi c goals earlier, but this maybe harmful. When used alone, PN causes hyperglycemia, hepatocellular injury and immunosuppression [9]. Th e point where the potential hazards of PN outweigh those of starvation, due failure to reach EN goals, is unclear. On this point European (ESPEN) and North American (ASPEN) nutritional societies disagree. ESPEN recommend commencing PN after 2 days, if patients are receiving less than their targeted EN [5]. In contrast, ASPEN recommend waiting until day 8 [4,6].

Study
Casaer and colleagues designed a multicenter trial to challenge this disparity head on. Th ey compared earlyinitiation of PN (day 3) with late-initiation (day 8) [10]. Th eir primary outcome measure was ICU length of stay (LOS). Both groups received standardized EN, and they were largely successful in achieving uniform EN. To keep fl uid intake even, the late-initiation group received 5% dextrose at the same rate as they would have received PN, were they in the early-initiation group. Recruitment was excellent, with just 5.3% refusing consent. Exclusion criteria were entirely reasonable, most patients being excluded because they were too young, (age <18), too well (no central-access, taking oral-nutrition) or not expected to survive (moribund). Patients with a body mass index (BMI) <17 were also excluded, withholding PN in these patients may be considered unethical. In the end, 4640 patients were randomized.
Overall, this study found that delayed PN was superior to early PN. Th e late-initiation group remained in the ICU one day less (median ICU-LOS; p=0.02) which refl ected a 6.3% relative increase in the likelihood of earlier discharge alive from the ICU (hazard ratio, 1.06; CI 1.00-1.13; P = 0.04). Secondary outcomes were either the same for both groups, or favored late-initiation. In particular, late-initiation had fewer infections, (22.8% v. 26.2%; P = 0.008) fewer patients requiring >2 days of mechanical ventilation (36.3% v. 40.2%; P = 0.006) and less renal replacement therapy (7-days (IQR 3-16) v. 10-days (IQR 5-23); P = 0.008). Th is translated into an average saving of over €1100 (P = 0.04) per patient. Th e authors concluded that there is no added benefi t to initiating PN earlier and that delayed PN, if needed, was superior.

Critique
Th is study has many strengths. It's a large clinical trial with excellent recruitment where both groups are well matched, with balanced fl uid intake. Early PN volumes were matched by 5% dextrose in the late-initiation group. Further, enteral feeding was similar in both groups and both groups received equal micronutrients. Th is is refl ective of a good trial comparing two treatment alterna tives, with robust results that allow practice patterns to change.
However, there are some weaknesses that deserve consideration. First, forced early EN and early prokinetic use in all subjects does not match generalized practice patterns, though it might be considered a reasonable baseline for us all to adopt. Second, early placement of post-pyloric tubes is not achievable in many institutions, making the generalizability of these dependent on the ability to place post-pyloric tubes quickly. Th ird, the study was necessarily unblinded, although physicians making discharge decisions and outcome data-collectors were unaware of treatment allocation. Further, intro duction of unblinded bias would be expected to favor earlyinitiation, as the study authors' were directly involved in development of the ESPEN feeding guidelines, which were shown to be inferior.
Th e results also raise an intriguing question, what about giving these patients no parenteral nutrition at all? Th is is important because more than 70% of the lateinitiation group did not receive PN at all, on account of being discharged before day 8. It is therefore conceivable that the benefi ts of late-initiation were seen because the majority of patients received no PN. Future trials should investigate delaying parentera l nutrition beyond 8 days, or indeed never giving parenteral nutrition at all, in patients already receiving enteral nutrition.

Recommendation
Casaer and colleagues are to be commended for conducting a large randomized controlled trail to tackle the current trans-Atlantic guideline disparity for timing of PN. Th ey showed that late-initiation of PN is superior to early-initiation, in patients not meeting targeted EN goals. In the absence of a "no PN" arm we recommend that practitioners wait until at least day 8 before considering PN in patients with a BMI >17. In patients who are close to 80% of their calorifi c goal by day 8, practitioners should consider that there might be no benefi t to adding PN.