Differential DNA methylation by early versus late parenteral nutrition in the PICU: a biological basis for its impact on emotional and behavioral problems documented 4 years later

Table 3 Added explanatory power by early-PN-induced altered DNA methylation in predicting more emotional/behavioral problems with early-PN

Outcome assessed at 4-years follow-up	Non-linear regression models
	Without randomized intervention		With randomized intervention
	R^2a	Fold-increase in R^2b	R^2c	Fold-increase in R^2d
Internalizing problems as reported by parents/caregivers	0.618	2.026	0.619	1.851
Externalizing problems as reported by parents/caregivers	0.603	1.813	0.603	1.710
Total problems as reported by parents/caregivers	0.613	1.868	0.614	1.771

All models were evaluated in 100 bootstrapped replicates. Baseline risk factors included age, center, race [7], gender, geographic origin [7], language, history of malignancy, diabetes, a predefined “syndrome” (Additional file 2), diagnosis and severity of illness (PIM3 and PeLOD) [9, 10], and risk of malnutrition (STRONGkids) [8]. R² for non-linear models computed according to Rubin’s rules
PN parenteral nutrition
^aOptimism-corrected R² of multivariable non-linear random forest regression models for outcomes assessed at 4-years follow-up including 37 differentially methylated CpG-sites and baseline risk factors
^bOptimism-corrected R² fold-increase with respect to multivariable non-linear regression models including baseline risk factors only
^cOptimism-corrected R² of multivariable non-linear random forest regression models for outcomes assessed at 4-years follow-up including 37 differentially methylated CpG-sites, baseline risk factors and the randomized intervention
^dOptimism-corrected R² fold-increase with respect to multivariable non-linear random forest regression models only including baseline risk factors and the randomized intervention

ISSN: 1868-7083