Proefschrift Kerklaan

Evidence for the use of PN in the PICU

Quality assessment revealed low scores for the 2 RCTs by Chaloupecky et al.; namely a Jadad score of 1 and a Black Downs score of 9/31 for both trials, and the RCT by Lekmanov et al.; Jadad score of 2 and Black Downs score of 4/31. The study by Larsen et al. had a higher Jadad score of 3 but the Black and Downs score was only 17/31. The trial of Jordan et al. scored the highest with a Jadad score of 5 and a Black and Downs score of 27/31. As only 6 studies were retained, a funnel plot to assess publication bias could not be created.

DISCUSSION

This systematic review could identify only 6 small RCTs that investigated the impact of a different dose or composition of PN in critically ill infants or children treated in the PICU. Of these 6 studies, 4 investigated infants after cardiac surgery and two included children with sepsis or after other major surgery, or burns respectively. The focus of these few studies was on intermediate or surrogate endpoints, such as nitrogen balances and inflammation markers, which appeared to be beneficially affected by providing more or altered parenteral nutrition early during critical illness. As the studies were small, all were statistically underpowered to detect a clinically relevant effect on patient-centered endpoints. Only the RCT by Lekmanov et al. reported a significant reduction of the duration of mechanical ventilation in children receiving glutamine-supplemented parenteral nutrition. However, with limited information on the used methodology which lacked a statistical analysis plan, the accuracy of these results cannot be determined. Hence, strong clinical conclusions cannot be drawn from these studies. As a result, no recommendations can be made regarding the optimal timing for initiation and composition of parenteral nutrition for use in critically ill infants and children. The lack of large RCTs on the use of parenteral nutrition in critically ill infants and children is striking. However, this is an observation that is not limited to the nutritional field. Indeed, there are only 7 randomized controlled trials of PICU patients that have addressed a clinical question with a large enough sample size to be able to detect a difference in patient-centered, hard clinical outcomes 18-25 , of which 3 are related to metabolic aspects 19,20,23 . This overall lack of large RCTs in PICU patients suggests difficulties in recruiting large numbers of patients, due to the fact that the number of PICU patients and the size of the PICUs worldwide are smaller than for adult intensive care. All the trials retained by the search strategy of this systematic review focused on surrogate endpoints, such as nitrogen balances and inflammation markers. This may hold some risks. Surrogate nutritional outcome measures are often used to describe mechanistic effects of an intervention. However, there is often a weak relationship, if any, between these surrogate endpoints and the important patient-centered clinically relevant outcomes. Sometimes surrogate endpoints can be misleading as they may inadvertently suggest a benefit whereas the clinical outcomes indicate harm. For example, a largewell-designedRCT of critically ill adults

5

95