Proefschrift Kerklaan

Chapter 2

because only 20 children received glucose infusion exclusively during the first 24 hours of admission. Target blood glucose levels varied between tight control 20 and a target glucose less than 10 mmol/L or 180 mg/dL. This discrepancy in definitions and implementation in glucose management has been highlighted before 39,40 . The discrepancy in definitions and implementation stems from the fact that uncertainties about risks and benefits remain 20,41 . A recent U.K. trial showed no benefit 42 and another trial in North America is underway. The strength of our study is the fact that we surveyed the local nutritional strategies as well as their implementation in clinical practice. Furthermore, to our knowledge, it is the first study to describe the practices in relation to income characteristics of countries in six continents. However, our survey may not provide accurate representations of these geographic regions. No response rate can be calculated, since the exact number of PICUs represented byWFPICCS is unknown. The total number of PICUs in all countries joined in theWFPICCS, as identified in the literature, is at least 969, so our 156 PICUs represent a small proportion of all PICUs worldwide. Our point prevalence data represent a small fraction of children in the PICUs per center as well as in the cohort invited to participate. The smaller number of PICUs in the point prevalence study may have caused an aggravation of the selection bias, since it is possible that we mainly received point prevalence data from PICUs with a strict protocol adherence. Hence, observations may not depict actual practices in these centers. However, characteristics of responding PICUs for the point prevalence were similar compared with the overall survey respondents (Table 1). Furthermore, many physicians have limited knowledge of nutritional practices in their centers. Our study may also be limited by the possibility that nonrespondents of this survey were less interested in nutritional practices leading to a selection bias and possible distorted reflection. On the other hand, this selection bias may strengthen our conclusion, if even in the nutrition- minded respondents, adherence to available guidelines is limited. Finally, the heterogeneity of the PICU population may have caused some difficulties; many of the questions required an unambiguous answer, so only most applicable answers were provided. And, as feeding practices differ between populations, answers from combined PICUs (with neonates or adults; respectively, 20 and 6% of the responding PICUs in this study) may falsely increase the perception of variability. Nevertheless, our survey clearly demonstrates the international variation in nutritional practice in critically ill children and the differences due to the limited available guidelines, especially on macronutrient administration and calculation of energy targets. Evidence-based guidelines are needed, but are challenging to develop due to a heterogeneous PICU population. Guidelines can be either very specific in respect to disease and settings, leading to wide variation of practice, or be generally applicable with risk of being unfocused and therefore irrelevant in specific situations.

44