The criteria included patients who were critically ill children, admitted to the PICU aged 1 month to 16 years and unable to receive their entire caloric requirements via the enteral route.

Children expected to stay in the PICU<3 days (e.g., diabetic ketoacidosis patients), declined to participate in the study and aged less than 1 month or more than 16 years.

Patients were assessed upon admission via a thorough medical history, clinical examination, and blood and serum analyses. The pediatric sequential organ failure assessment (pSOFA) score was used to measure the severity of the clinical condition at admission and to assess organ dysfunction and depending on the patient's baseline risk level [9]. The pediatric risk of mortality (PRISM) [10] was assessed at the end of the first 24 hours for each patient, using 14 clinical and laboratory variables where data were entered on the website: http://www.sfar.org/scores2/prism2.php that automatically calculates the expected death rate. Early or late PN was given to the patients at random.

Patients were randomized to early PN and late PN groups using computer-generated random numbers. The patients and the treating physicians were blinded to the type of parenteral fluids administered. The early PN group, which included well-nourished and malnourished children, was given PN on the first day of PICU admissions. Children randomized to the late PN group, children who identified as malnourished, were started on PN on the fourth day and well-nourished were initiated on the seventh day.

The PN solution contained glucose, protein, electrolytes (sodium, magnesium phosphorus, calcium, and potassium), water-soluble vitamins, trace elements, and amino acids. The PN dose and composition conformed to the local protocol (shown in Supplementary Tables 1–3) [11]. In the early PN group, we included 10%–12.5% glucose (as D-glucose) starting on PN administration day 1. We increased the glucose proportion by 2.5% per day (2.5g/kg/day) until glucose was ~50% to 60% of total kcal consumed at a glucose infusion rate (GIR) of 4–8mg/ kg/min with a maximum GIR of 12mg/kg/min. The glucose level was maintained at 80–150 mg/dL. If hyperglycemia occurred, the GIR was decreased. If hyperglycemia persisted after GIR decrease, insulin was given until the blood glucose level becomes lower than 150 mg/dL was achieved. If hypoglycemia (<80 mg/dL) developed, we increased GIR or the concentration of infused glucose. Protein (Aminoven infant 10%) was administered from day 1 at a dose of 1g/kg and increased daily by 0.5g/kg to reach 3.5g/kg. Electrolytes were also administered from day 1, in the following doses: sodium at 3–5 mEq/kg, potassium at 2–4 mEq/kg, calcium (calcium gluconate) at 0.5–2 mEq/kg, and magnesium (magnesium sulfate) at 2–10 mEq (0.25–1.25 g) daily. Phosphorus (Glycophos) was administered at a dose of 1–1.5 mmol/kg/day. Trace elements were provided via Peditrace administration. The dose of Peditrace in infants and children weighing up to 15 kg was 1mL/kg/day, and in children weighing more than 15 kg, 15mL/day. Water-soluble vitamins were provided by Soluvit N administration at a dose of 1mL/kg in children weighing less than 10 kg and 1 vial/day in children weighing more than 10 kg. Fat-soluble vitamins were provided by administration of Vitalipid Novum Infant at a dose of 1mL/kg/day. Lipids were administered; Smoflipid was started on day 2 at a dose of 0.5g/kg and increased daily by 0.5g/kg to reach 3g/kg/day.

Well-nourished children in the late PN group received a mixture of 5% glucose and saline until the seventh day after admission, and malnourished children received that mixture until the fourth day. To prevent the refeeding syndrome, late PN patients received intravenous trace elements, minerals, and vitamins starting on day 1 and continued until full enteral nutrition [8,12]. PN was discontinued when more than 80% of the nutritional requirement was tolerated enterally. In both groups, enteral nutrition was initiated as soon as possible and gradually advanced up to the target as tolerated.

The Academy of Nutrition and Dietetics and the American Society for Parenteral and Enteral Nutrition (A.S.P.E.N.) defined pediatric malnutrition in the clinical setting as “an imbalance between nutrient requirements and intake that results in cumulative deficits of energy, protein, or micronutrients that may negatively affect growth, development, and other relevant outcomes.” [13] For each patient admitted, the admission weight and length/height were measured. If the weight-for-length z score was ≤-2 standard deviation, the patient was considered malnourished using World Health Organization scale [14].

Enteral nutrition was initiated and progressed gradually in parallel with a gradual reduction of PN. The decision to start enteral feeding was initiated when the patient showed (1) hemodynamic stability, (2) lack of significant respiratory distress, and (3) audible intestinal sounds, no abdominal distention, and normal gastric residual.

The enteral feeding route used was in the form of nutrition administered through the mouth (oral feeding) or through a tube directly connected to the stomach, called nasogastric tube feeding. We typically use a standard polymeric formula feed, advancing to full calories at 10–20 mL/hr [15] and administering continuously or intermittently.

Patient weight was also monitored twice weekly throughout their PICU stay. Blood glucose was measured 4 hours after starting or changing the glucose concentration of infusate, then daily for 2 days or until the patient was stable. Serum electrolytes, bicarbonate, and blood urea nitrogen (BUN) were measured twice daily after starting or changing the infusion rate or composition, then twice weekly. Liver function tests, albumin, phosphate, magnesium were measured initially, then weekly unless the patient was unstable or warranted closer monitoring (e.g., refeeding syndrome). Finally, serum triglycerides (if fat emulsion in use) were measured daily after starting or when changing the quantity of fat administration, then weekly. Patients were followed up until discharge from PICU or death.

The primary outcome was the need for mechanical ventilation (MV). The secondary outcomes included (1) PICU stay (among survivors); (2) vasoactive-infusion days; (3) incidence of new infections acquired throughout the course of receiving health care that was not present at the time of admission are known as healthcare-associated infections; 2 examples of which include ventilator-associated pneumonia, occurring 48 hours after endotracheal intubation [16], and central line-associated bloodstream infection (CLABSI) of a central venous catheter (CVC) [17]; and (4) PICU mortality.

Based on the primary outcome of the study to detect need for MV and based on review of past literature [18], who found that, there was significant relation between need for MV and nutrition feeding; need for MV was lower in children receive nutrition feeding than those do not receive 31.7% versus 68.3 respectively. Sample was calculated by the following equation {n=P₁ (1-P₁)+P₂ (1-P₂) / (P₁–P₂)²* C} where n=desired sample size, P1, 2=the proportion in each group and C=standard values of α (with β equal to 7.85 at 80% power) at a confidence level of 95%. The minimum calculated total sample size required after adding a dropout rate of 10% was 82 participants. Due to availability of cases during our practical part of the study sample was increased to 140 participants which were allocated by randomization into 2 group early PN (n=71) and late PN (n=69).

All procedures performed in the study were in accordance with the ethical standards of the Menoufia University and Faculty of Medicine (approval number: 3/2021 PEDI 3-2). Informed consent was obtained from the parents (or legal representative).

The IBM SPSS Statistics ver. 23.0 (IBM Co., Armonk, NY, USA) was used to perform all statistical calculations. Numbers and percentages were used to represent categorical variables. Quantitative variables were presented as the median (range) with a skewed distribution and compared using the MannWhitney U test (for 2 groups) or the Kruskal-Wallis test (for 3 or more groups). The association between categorical variables was assessed using chi-square, yet Fisher exact test was employed when more than 25% of the cells had anticipated counts of less than 5. Multivariate regression analysis was used to detect independent predictors for MV need. The threshold for statistical significance was defined as P<0.05.

Two hundred critically ill children were assessed for eligibility, of which 60 children were excluded. The remaining 140 children were randomized to receive either early or late PN. There were 71 patients in the early PN group and 69 in the late PN group; the CONSORT (consolidated standards for reporting of trials) 2010 flow diagram of the current trial is shown in Fig. 1.

The baseline characteristics of both groups are shown in Table 1. No significant difference was found in illness severity between the 2 groups on admission as assessed by pSOFA and PRISM scores.

Daily caloric and macronutrients intake from PICU admission until day 14 (total caloric, carbohydrate, protein, and lipid intake) for enteral and parenteral routes in both groups are shown in Fig. 2.

Table 2 shows the nutritional data of patients. Patients given early PN were able to start enteral feeding earlier than those provided late PN (median, 6 days; interquartile range [IQR], 2–20 days vs. median, 12days; IQR, 3–30 days; P<0.001), had a lower risk of feeding intolerance (5.6% vs.18.8%, P=0.035) and required less time to reach full calories enterally (median, 8 days; IQR, 7–10 days vs. median, 12.5 days; IQR, 11–20 days; P=0.004). There was no significant difference between the 2 groups in the total duration of PN (P=0.78).

Patients with early PN had lower need and shorter duration of MV with more MV-free days (P=0.018, P<0.001, P=0.001, respectively). There was no significant difference in mortality between the early and late PN groups (P=0.21). Patients with early PN were more likely to have shorter PICU stays (median, 8 days; IQR, 6–55 days vs. median, 16.5 days; IQR, 7–75 days; P<0.001) and has shorter free PICU stays (median, 9 days; IQR, 8–29 days vs. median, 15 days; IQR, 10–37 days; P=0.033). They were also more likely to experience fewer vasoactive-infusion days (P=0.013). Patients receiving early PN had fewer CVC days and lower nosocomial infection rates than late PN patients (P=0.001) (Table 3).

Accordingly, the clinical outcome of patients with and without malnutrition was not significantly different regarding illness severity, as assessed by PRISM and pSOFA. No significant difference was found between the 2 groups regarding disease outcome, including mortality, PICU stay, and the incidence of acute respiratory distress syndrome, MV duration, vasoactive-infusion days or incidence of new infection (Table 4).

Multivariate regression analysis revealed that randomization to late PN compared with early PN was an independent predictor for MV need. Also, pSOFA score, presence of new infection, increased duration of vasoactive drugs, and CVC are independent predictors for MV need (Table 5).