Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
ARTICLES
Long-Term Neurobehavioral and Quality of Life Outcomes of Critically Ill
Children after Glycemic Control
Katherine V. Biagas, MD1, Veronica J. Hinton, PhD2, Natalie R. Hasbani, MPH3, Peter M. Luckett, MD4, David Wypij, MD3,
Vinay M. Nadkarni, MD5, and Michael S. D. Agus, MD6, on behalf of the HALF-PINT trial study investigators*,
and the PALISI Network
Objectives To investigate adaptive skills, behavior, and quality health-related quality of life in children from 32
centers enrolling in the Heart And Lung Failure-Pediatric INsulin Titration randomized controlled trial.
Study design This prospective longitudinal cohort study compared the effect of 2 tight glycemic control ranges
(lower target, 80-100 mg/dL vs higher target, 150-180 mg/dL) 1-year neurobehavioral and health-related quality of
life outcomes. Subjects had confirmed hyperglycemia and cardiac and/or respiratory failure. Patients aged 2-
16 years old enrolled between April 2012 and September 2016 were studied at 1 year after intensive care
discharge. The primary outcome, adaptive skills, was assessed using the Vineland Adaptive Behavior Scale.
Behavior and health-related quality of life outcomes were assessed as secondary outcomes using the Pediatric
Quality of Life and Child Behavior Checklist at baseline and 1-year follow-up. Group differences were evaluated us-
ing regression models adjusting for age category, baseline overall performance, and risk of mortality.
Results Of 369 eligible children, 358 survived after hospital discharge and 214 (60%) completed follow-up. One-
year Vineland Adaptive Behavior Scale-II composite scores were not different (meanT SD, 79.9T 25.5 vs
79.4 T26.9, lower vs higher target; P = .20). Improvement in Pediatric Quality of Life total health from baseline
was greater in the higher target group (adjusted mean difference, 8.2; 95% CI, 1.1-15.3; P = .02).
Conclusions One-year adaptive behavior in critically ill children with lower vs higher target glycemic control did
not differ. The higher target group demonstrated improvement from baseline in overall health. This study affirms the
lack of benefit of lower glucose targeting. (J Pediatr 2019;-:1-7).
Trial registration ClinicalTrials.gov: NCT01565941.
C
linical management of patients with critical illness-associated hyperglycemia has evolved, provoking questions about
the outcomes of patients managed with glycemic control. After initial reports of reduced mortality and morbidity, sub-
sequent multicenter trials demonstrated no such benefit.1-6 Follow-up analysis of the largest trial and meta-analyses
have called attention to the potential for harm from hypoglycemia. 7-10 Glucose control has the potential for inducing neuro-
logic injury. The mammalian brain is an obligate glucose user requiring a continuous supply. 11 This is especially true for the
developing brain, which undergoes a maturational upregulation of the glucose transporter protein. 12 Tight glycemic control
presents the risk of inducing hypoglycorrhachia and neuronal injury. 12 Because of these risks, assessment of long-term neuro-
logic function is an important outcome of these trials.
We report an a priori aim of the Heart And Lung Failure-Pediatric INsulin
Titration trial (HALF-PINT) of tight glycemic control in critically ill children. 1
From the Department of Pediatrics, Stony Brook
Children’s Hospital and the Renaissance School of
We studied the long-term neurobehavioral (adaptive skills and behavior) and 2
Medicine, Stony Brook, NY; Department of Psychology,
health-related quality of life 13,14
(HRQoL) outcomes 1 year after pediatric intensive Queens College and the Graduate Center of the City
University of New York, New York, NY; Department of 3
care unit (PICU) discharge. Patients were hyperglycemic (2 blood glucose Cardiology, Boston Children’s Hospital and Harvard
4
measurements of ³150 mg/dL [8.3 mmol/L]) infants and children with cardiac
Medical School, Boston, MA; Department of Pediatrics,
University of Texas Southwestern Medical Center and
Children’s Health, Dallas, TX; Department of 5
and/or respiratory failure treated with lower target (80-110 mg/dL [4.4- Anesthesiology and Critical Care Medicine, Children’s
6.1 mmol/L]) vs higher target (150-180 mg/dL [8.3-10.0 mmol/L]) glycemic con- Hospital of Philadelphia and the University of
6
Pennsylvania, Philadelphia, PA; and Division of Medical
trol. PICU-free days and mortality did not differ between lower and higher target Critical Care, Department of Pediatrics, Boston
Children’s Hospital and Harvard Medical School,
groups. Hypoglycemia (<60 mg/dL [<3.3 mmol/L]) occurred more frequently in Boston, MA
*List of the HALF-PINT trial study investigators is
available at www.jpeds.com (Appendix).
Supported by the National Heart, Lung, and Blood Insti-
tute (U01HL107681 [to M.A. and V.N.]) and
CBCL Child Behavior Checklist (U01HL108028 [to D.W.]) and the National Center for
Advancing Translational Sciences (UL1TR001873 [to
HALF-PINT Heart And Lung Failure-Pediatric INsulin Titration trial K.B.]), and by endowed chairs (to M.A. and V.N.).
HRQoL Health-related quality of life The authors declare no conflicts of interests.
PedsQL Pediatric Quality of Life Version 4.0-Short Form Portions of this study were presented at the Pediatric
PICU Pediatric intensive care unit Academic Societies annual meeting, May 5-8, 2018,
Toronto, Canada.
POPC Pediatric Overall Performance Category
VABS-II Vineland Adaptive Behavior Scales, Second Edition-Survey Interview 0022-3476/$ - see front matter. ª 2019 Elsevier Inc. All rights reserved.
https://doi.org/10.1016/j.jpeds.2019.10.055
1
THE JOURNAL OF PEDIATRICS ● www.jpeds.com Volume - ● - 2019
the lower target group.13 Therefore, an additional aim was to blinded as to treatment group. Parent report forms were
determine the effect of any hypoglycemia on outcomes. separately scored by 2 research assistants; discrepancies
were adjudicated by the training co-investigator. Validated
Spanish-language translation forms were used and Spanish-
Methods speaking research assistants conducted all interviews with
Spanish-speaking parents.
Institutional review and clinical conduct of the study were
coordinated by the Institutional Review Board and Clinical One-Year Assessments
Coordinating Center at Boston Children’s Hospital.13,14 Neurobehavioral outcomes assessing adaptive skills,
Outcome assessments were performed by a Neurobehavioral behavior, and HRQoL were examined. The primary outcome
Research Team from the Gertrude Sergievsky Center at for the study was the adaptive behavior composite score of
Columbia University Medical Center. Data analysis was per- the Vineland Adaptive Behavior Scales, Second Edition-
formed by the Data Coordinating Center at Boston Chil- Survey Interview (VABS-II).19 The VABS-II was chosen as
dren’s Hospital.13 Appropriate reliance and data use and the primary outcome measure because it is a single validated
safe exchange of protected health information agreements instrument (Cronbach alpha of ³0.95 for adaptive behavior
were enacted. composite across all pediatric ages) and is sensitive to detect
HALF-PINT was a randomized trial (ClinicalTrials.gov: changes in a variety of populations.20 The VABS-II assesses
NCT01565941) that enrolled patients with confirmed hyper- adaptive behaviors in 3 primary domains, namely, communi-
glycemia from April 2012 through September 2016.13 The gly- cation, daily living skills, and socialization.19 For children
cemic control methodology of the HALF-PINT trial has been <7 years of age, a motor skills domain is added.
previously described in detail.14 A total of 35 centers were Additional measures included the age-appropriate PedsQL
approved to enroll patients but the trial was stopped early and CBCL checklists.15,16 These were administered at base-
based on recommendations from the data and safety moni- line and follow-up. The PedsQL is a 15-question form that
toring board. Three centers were not enrolling patients at stop- assesses 4 domains (physical, emotional, social, and school
page. The study protocol involved the use of continuous functioning). Four age-appropriate forms of the PedsQL
glucose monitoring, an explicit treatment protocol, and main- were used—ages 2-4, 5-7, 8-11, and 12-18 years (Cronbach
tenance of a minimal glucose infusion rate as safety measures.14 alpha of ³0.88 for all ages for total score). Summary total
Patients eligible for the follow-up study were aged 2-16 at the health, physical health, and psychosocial health scores were
time of enrollment, received the HALF-PINT protocol, and calculated with psychosocial health comprised of the last 3
spoke English or Spanish. Patients were excluded from of these named domains. The primary outcome variable for
follow-up if they did not survive, were under the care of per- the PedsQL was the total health score with higher scores indi-
sons with insufficient knowledge about regular behavior to cating better HRQoL. In defining impairment, we used the
complete assessment, or for whom consent was withdrawn. findings of a previous study of HRQoL in PICU survivors 21
After randomization and during the first week of stay, as PedsQL total, physical, or psychosocial health scores >1
parents or guardians were asked to complete baseline assess- SD below a population mean.22
ments when acclimated to the PICU. Assessments included The CBCL is a comprehensive checklist that assesses total
age-appropriate Pediatric Quality of Life Version 4.0- problems with behavior as well as 2 broad-band scores, inter-
Short Form (PedsQL)15 and Child Behavior Checklist nalizing and externalizing problems, composed of multiple
(CBCL)16 instruments. Parents were asked to represent their age-dependent subscores.16 Two age-appropriate forms of
child’s prehospitalization baseline condition. Parents also the CBCL were used—ages 1.5-5 years and 6-18 years (Cron-
completed demographic and socioeconomic questionnaires. bach alphas of 0.90 and 0.97 for total problems, respectively).
Gross performance was assessed by bedside staff using the Pe- For children aged 6-18 years, 2 questions concerned the exis-
diatric Overall Performance Category (POPC) and Pediatric tence of self-harming behaviors or suicidal ideation/attempts.
Cerebral Performance Category scores.17 Severity of illness If these questions were answered in the affirmative, site inves-
was assessed by the Pediatric Risk of Mortality III-12 score.18 tigators were informed and required to contact primary care
The neurobehavioral research team was notified of eligible physicians for referral for appropriate counseling. CBCL raw
patients bimonthly. Email and postcard reminders were scores were converted into T-scores. At-risk behavior was
sent to parents or guardians at 8 months. At 11 defined as a T-score of >65.16 The primary outcome variable
months after PICU discharge, parents were contacted to was the total problems score.
schedule a telephone interview. Outcome assessments were
conducted up to 13 months after PICU discharge. In- Statistical Analyses
terviews were conducted on weekdays, weekends, and eve- Our original target sample size of 378 (189 per group) pro-
nings at the parent’s convenience. The duration of vided 90% power to detect a one-third SD effect size (5-
interviews was 2-3 hours and, for some, was accomplished point difference) between groups in the VABS-II adaptive
over 2 telephone calls. Parent report forms were used for all behavior composite score using a 2-sided 0.05 level test.
study measures. All members of the neurobehavioral Owing to early stopping of the HALF-PINT trial, these calcu-
research team were trained by a co-investigator and were lations were revised based on our final per-protocol sample
2 Biagas et al
- 2019 ORIGINAL ARTICLES
size of 698. A revised target sample size of 214 patients (107 related differences in carbohydrate requirements, baseline
per group) provided 80% power to detect an effect size of overall functioning, and severity of illness, respectively.
0.38 SD units (a 6-point difference). Prerandomization and postrandomization demographic
For patients with an age-appropriate interview at 1 year af- and clinical characteristics were compared between patients
ter PICU discharge, overall adaptive functioning, HRQoL, with follow-up and those without and between target groups.
and behavioral problems were evaluated. For patients who Differences between groups were evaluated by Wilcoxon
completed both the baseline and 1-year post-PICU discharge rank-sum tests and Fisher exact tests for continuous and cat-
PedsQL and CBCL, the change in score was calculated. Only egorical variables, respectively, except with t tests for baseline
patients with complete datasets for both time points were PedsQL and CBCL T-scores. All P values are 2-tailed and
included in the paired analysis, whereas all data points are have not been adjusted for multiple comparisons. Analyses
shown for unpaired analyses at the baseline and 1-year post- were performed with the use of SAS software, version 9.4
PICU times. Subsequent analyses also compared the (SAS Institute, Cary, North Carolina).
outcomes between patients who experienced hypoglycemia
(any blood glucose <60 mg/dL [3.3 mmol/L]) and those
who did not. Group comparisons used linear and logistic Results
regression adjusted for age group (2-6 years vs ³7 years),
baseline POPC of >1 vs POPC of 1, and Pediatric Risk of A total of 713 patients were randomized by the 32 partici-
Mortality III-12 for continuous and binary outcomes, respec- pating HALF-PINT sites (Figure). Of these, 94 died before
tively. These adjustments were made to account for age- hospital discharge and 15 underwent randomization but
Figure. Eligibility and 1-year follow-up of the study patients. A neurobehavioral or HRQoL assessment was available for 60%
(214/358 eligible patients) of the HALF-PINT study population. Only patients aged 2-16 years at the time of randomization were
eligible for follow-up. Of the 369 eligible for 1-year assessment, 11 patients (3%) did not survive the 1-year follow-up period.
Long-Term Neurobehavioral and Quality of Life Outcomes of Critically Ill Children after Glycemic Control 3
THE JOURNAL OF PEDIATRICS ● www.jpeds.com Volume -
were not treated per protocol. Of the remaining patients, 358 One-Year Neurobehavioral and HRQoL Status
eligible patients survived after hospital discharge and 214 Table V describes the long-term outcomes. The mean 1-year
(60%) families participated in follow-up, 102 from the VABS-II adaptive behavior composite scores were not
lower target and 112 from the higher target group. Reasons significantly different for lower vs higher target patients
that follow-up was not completed are described in Table I (79.9 T 25.5 vs 79.4 T 26.9; P = .20). Moreover, the percent
(available at www.jpeds.com). Families that participated in of patients in each adaptive level did not differ. Total,
follow-up had a lower proportion in the lowest income psychosocial, and physical health scores were similar between
category (45/162 [28%]) compared with families that did groups. CBCL T-scores did not differ between groups for the
not participate (46/106 [43%]; P = .02). Maternal total problems or broad-band scores. The percent of patients
education was higher for families that participated in at risk did not differ (Table V). Between group comparisons
follow-up with 129 of 192 families (67%) reporting more for the PedsQL and CBCL subscale scores did not differ with
than a high school diploma (P = .046). The mean CBCL the exception of one CBCL subscale score in 1 age category
T vs 46.2 12.3;TP = .03) and
total problems (49.2 11.8 (Table VI; available at www.jpeds.com).
internalizing problems (50.8 T 10.7 vs 47.7 T 11.2; In the higher target group, the mean PedsQL total health
P = .02) T-scores differed between those who were studied improved from 69.7 T 20.9 at baseline to 76.4 T 20.5 at
and those who were not (Table II; available at www.jpeds. 1 year. The change from baseline was greater in the higher
com). Postrandomization clinical factors were similar as compared with the lower target group (adjusted mean dif-
between groups (Table II). ference, 8.2; 95% CI, 1.1-15.3; P = .02). This improvement in
total health could be ascribed to improvement in mean psy-
Baseline Neurobehavioral and HRQoL Status chosocial health, which changed from 72.8T19.9 at baseline
Prerandomization characteristics were similar between to 81.9 T18.3 at 1 year for the higher target group (adjusted
groups with slight differences in baseline cognitive and mean difference, 8.5; 95% CI, 1.7-5.3; P = .01) between the
overall performance (Table III; available at www.jpeds. higher target and lower target groups. No differences in the
com). The lower target group had fewer patients with
poorer overall performance (POPC, 31% [95% CI, 23%-
41%] vs 46% [95% CI, 36%-55%]; P = .04) and
nonsignificantly poorer cognitive performance (Pediatric Table V. One-year neurobehavioral and HRQoL
Cerebral Performance Category, 28% [95% CI, 20%-38%] outcomes by study group
vs 41% [95% CI, 32%-51%]; P = .06) compared with the Lower Higher P
higher target group. Lower mean baseline PedsQL total Assessments target target value*
health (69.8 T 20.8 vs 77.4 T 19.3; P = .01) and VABS-II† (n = 97) (n = 111)
psychosocial health (73.0 T19.8 vs 79.3 T 17.1; P = .03) Adaptive Behavior Composite
Communication
79.9 T 25.5 79.4 T 26.9 .20
.17
81.5 T 25.3 80.8 T 27.5
were reported for higher vs lower target groups, Daily living skills 81.2 T 27.9 79.7 T 28.2 .35
respectively. Baseline CBCL total problems and subscale T- Socialization 83.3 T 23.9 83.1 T 24.8 .16
scores did not differ. Baseline PedsQL and CBCL subscale Motor skills 76.1 T 28.0 75.7 T 24.9 .93
Adaptive level .59
scores are described in Table IV (available at High (³130) 0 1 (<1)
www.jpeds.com). Postrandomization factors related to Moderately high (115-129) 5 (5) 7 (6)
tight glycemic control differed between the lower and Adequate (86-114) 41 (42) 44 (40)
Moderately low (71-85) 22 (23) 22 (20)
higher target groups. Percent of patients treated with Low (£70) 29 (30) 37 (33)
insulin (P < .001) and number of episodes of any PedsQL‡ (n = 98) (n = 110)
Total health 74.4 T 21.5 75.6 T 20.7 .20
hypoglycemia (P = .005) were greater for the lower target Impaired (£65) 34 (35) 37 (34) .41
group (Table III). Median time-weighted blood glucose Physical health 65.8 T 36.9 64.7 T 37.3 .46
average (110 [IQR, 105-120] vs 124 [IQR, 111-146 mg/ Impaired (£64) 39 (40) 47 (43) .75
Psychosocial health 78.9 T 20.2 81.3 T 18.4 .16
dL], lower vs higher target group) and percent time-in- Impaired (£63) 23 (23) 20 (18) .16
target range (54 [IQR, 42-66] vs 91 [IQR, 82-96], lower CBCL T-scores§ (n = 101) (n = 110)
vs higher target group) were greater for the higher target Total problems 51.5 T 12.0 51.9 T 12.5 .96
At risk 17 (17) 15 (14) .38
group (P < .001 for both comparisons). More patients in Internalizing problems 52.2 T 11.4 51.6 T 11.8 .60
the lower target group (6/102 patients [6%]) experienced At risk 16 (16) 17 (15) .86
an episode of severe hypoglycemia (<40 mg/dL Externalizing problems 49.6 T 11.6 48.9 T 10.9 .74
At risk 11 (11) 11 (10) .86
[<2.2 mmol/L]) as compared with the higher target group
(1/112 patients [<1%]; P = .06). Postrandomization Values are meanTSD or number (%).
*P values for the comparison between target groups were determined using logistic and linear
characteristics not related to tight glycemic control (PICU- regression adjusting for age category, baseline overall performance (POPC >1), and PRISM-III
free days, ventilator-free days, or POPC) did not differ. 12 score for categorical and continuous variables, respectively.
†VABS-II standard scores range from 20 to 160 with mean of 100 and SD of 15. Motor skills
Median time to follow-up interview was scores are only available for patients <7 years old (32 lower target, 26 higher target).
12.3 months (IQR, 11.5-13.3 months) with no difference ‡PedsQL scores range from 0 to 100. Patients were considered impaired if they were >1 SD
below healthy pediatric population mean scores.
in time to follow-up for the lower vs higher target groups §CBCL T-scores range from 20 to 100. A T-score of ³65 indicates a child is at risk for a clin-
(Table III). ically relevant problem for the associated syndrome scale.
4 Biagas et al
- 2019 ORIGINAL ARTICLES
change from baseline performance in the CBCL total prob- tight glycemic control and underscore the potential for harm
lems T-score or the 2 broad-band scores were observed be- with hypoglycemia.
tween target groups (Table VII). To date, there have been >20 observational and randomized
trials of tight glycemic control in children and adults. Meta-
Status by Hypoglycemia Experience analyses of these differ slightly in that some suggest no
Table VIII describe the outcomes for patients who mortality benefit for adults or children, whereas another
experienced any hypoglycemia in the trial and those who demonstrates higher mortality with lower targets in adult
did not. The VABS-II adaptive behavior composite, VABS- subpopulations.8-10,25 These studies are in agreement,
II subscale scores, or CBCL total problems or broad-band however, that targeted glycemic control confers a risk of
T-scores were similar between groups. However, the hypoglycemia.8,10,25 Outcomes of critically ill infants and chil-
PedsQL physical health scores were lower for those with dren managed with tight glycemic control have been studied in
any hypoglycemia (adjusted mean difference,—13.9; 95% limited fashion. One-year follow-up of children in a multi-
— to 0.1; P = .05). The proportion with
CI, 27.9 center UK study demonstrated no difference in mortality be-
impairment on the PedsQL Psychosocial subscale was tween lower and higher target groups.23 In the single-center
greater in those who experienced any hypoglycemic event Leuven study of largely postoperative cardiac surgical patients,
(P = .04). Only 7 patients in the follow-up cohort intelligence and visual-motor skills were lower than normal
experienced severe hypoglycemia (<40 mg/dL [<2.2 mmol/ children but did not differ between those with lower target
L]). These numbers are insufficient to make comparisons glucose control vs usual therapy.3,26 We previously reported
concerning severity of hypoglycemia. neurodevelopmental outcomes at 1 year of age in a neonatal
cardiac surgical population and noted compromised intelli-
Discussion gence and visual-motor skill levels similar to the Leuven study
only among those who had an episode of moderate or severe
In this study of lower vs higher target tight glycemic control hypoglycemia. Those without moderate or severe hypoglyce-
in children ³2 years of age, the 1-year neurobehavioral (adap- mia had substantially higher scores and were, in fact, indistin-
tive skills and behavior) and HRQoL outcomes were not guishable from normal children.27
significantly different. These results support the safety of gly- Our current results are in agreement with the mortality
cemic control, especially with the use of explicit protocols, rates of the UK study and with our prior results in pediatric
close glucose monitoring, and provision of minimal carbohy- cardiac surgery patients. In the current study we found no
drate supply.14 An improvement in psychosocial perfor- decrement in behavior or adaptive skills. Glycemic control
mance was noted at 1 year in the higher target group, can be performed safely. Moreover, these results add to the
suggesting that higher targeting may be the preferred range. growing body of knowledge on the deleterious effects of hy-
However, these results should be interpreted cautiously given poglycemia during critical illness. HRQoL may be negatively
the differences noted in baseline functioning. Finally, delete- impacted by such events. Yet, it is not reasonable to conclude
rious effects in physical health and incidence of psychosocial that glycemic control should be abandoned altogether, rather
impairment were associated with any episode of hypoglyce- only that the low target range (80-110 mg/dL) should be
mia, underscoring the possibility of long-term dysfunction avoided. Hyperglycemia has direct neurotoxic effects, exag-
with glucose deprivation of the brain during critical illness. gerating brain injury as demonstrated in animal models
The HALF-PINT trial had a lower rate of hypoglycemia and children.28,29 Our results support intensive glucose
than found in other trials or than occurs spontaneously in monitoring and provision of minimal carbohydrate supply
a similar PICU population.1,2,4,6,23,24 Viewed from a wider during insulin therapy and suggest a safer range of glucose
context, these results affirm the lack of benefit of lower target targeting in the 150-180 mg/dL range.
Table VII. Baseline and 1-year neurobehavioral and HRQoL outcomes by study group*
Lower target Higher target
Average change Average change
Assessments Baseline One-year follow-up in score Baseline One-year follow-up in score P value*
PedsQL† (n = 82) (n = 87)
Total health 78.0 T 19.0 75.9 T 20.5 —2.0 T 23.5 69.7 T 20.9 76.4 T 20.5 6.7 T 23.9 .02
Physical health 73.3 T 32.2 67.8 T 35.6 —4.7 T 39.1 64.3 T 36.1 65.9 T 37.7 1.6 T 38.9 .31
Psychosocial health 80.2 T 16.5 80.0 T 18.5 —0.2 T 22.3 72.8 T 19.9 81.9 T 18.3 9.1 T 23.0 .01
CBCLT-scores (n = 83) (n = 91)
Total problems 48.3 T 12.3 52.0 T 11.4 3.6 T 11.6 50.8 T 11.4 51.7 T 13.1 0.9 T 11.0 .07
Internalizing problems 50.3 T 10.6 52.8 T 11.1 2.6 T 12.5 51.6 T 11.2 52.1 T 12.5 0.5 T 11.8 .18
Externalizing problems 46.9 T 10.5 49.7 T 10.9 2.8 T 11.0 47.3 T 11.2 48.5 T 11.1 1.2 T 9.1 .19
Long-Term Neurobehavioral and Quality of Life Outcomes of Critically Ill Children after Glycemic Control 5
THE JOURNAL OF PEDIATRICS ● www.jpeds.com Volume -
17.
Long-Term Neurobehavioral and Quality of Life Outcomes of Critically Ill Children after Glycemic Control 7
THE JOURNAL OF PEDIATRICS ● www.jpeds.com Volume -
The following investigators contributed subjects to the analyses presented in this manuscript.
7.e1 Biagas et al
- 2019 ORIGINAL ARTICLES
Study Organization
Long-Term Neurobehavioral and Quality of Life Outcomes of Critically Ill Children after Glycemic Control 7.e2
THE JOURNAL OF PEDIATRICS ● www.jpeds.com Volume -
7.e3 Biagas et al
- 2019 ORIGINAL ARTICLES
CPR, cardiopulmonary resuscitation; PCPC, Pediatric Cerebral Performance Category; PRISM-III 12, PRISM Score III from the first 12 hours of PICU admission.
Values are median (IQR), number (%),number/total (%), orTmean SD.
*Patients in the lower target group had their blood glucose level controlled to a target range of 80-110 mg/dL (4.4-6.1 mmol/L), and those in the higher target group to a target range of 150-180 mg/
dL (8.3-10.0 mmol/L) during their critical care stay. To convert values for glucose to millimoles per liter, divide by 18.
†Race and ethnic group were as reported in the medical record.
‡The scales for the PCPC and POPC range from 1 to 6, with lower scores indicating less disability.
§Denotes a significant difference between target groups at the 0.05 significance level. For PedsQL and CBCL T-scores, P values for comparison between target groups were determined using t tests.
For other variables, P values for comparison between target groups were determined using Fisher’s exact tests and Wilcoxon rank-sum tests for categorical and continuous variables, respectively.
{Other includes oncologic, renal, metabolic, and hematologic reasons.
**The scale for the PRISM III score from the first 12 hours in the PICU (the PRISM III-12 score) ranges from 0 to 74, with higher scores indicating a higher risk of death.
††PedsQL scores range from 0 to 100, with higher scores indicating higher quality of life. Age-appropriate PedsQL scores were available for 85 lower target patients and 88 higher target patients.
‡‡CBCL T-scores range from 20 to 100, with lower scores indicating the absence of behavioral problems. Age-appropriate CBCL T-scores were available for 89 lower target patients and 99 higher
target patients.
Long-Term Neurobehavioral and Quality of Life Outcomes of Critically Ill Children after Glycemic Control 7.e4
THE JOURNAL OF PEDIATRICS ● www.jpeds.com Volume -
Table IV. Baseline subscale scores according to study Table VI. One-year subscale scores according to study
group group
Assessments Lower target Higher target P value* Assessments Lower target Higher target P value*
PedsQL (n = 85) (n = 88) PedsQL (n = 98) (n = 110)
Emotional functioning 82.1 T 18.8 74.6 T 23.5 .02 Emotional functioning 79.6 T 21.8 82.8 T 20.7 .16
Social functioning 83.7 T 18.6 80.6 T 23.6 .33 Social functioning 82.3 T 24.8 86.1 T 22.5 .12
School functioning† 70.1 T 28.1 61.4 T 32.2 .07 School functioning† 73.1 T 28.8 73.9 T 28.5 .57
CBCL T-scores CBCL T-scores
Ages 2-5 years (n = 31) (n = 25) Ages 2-5 years (n = 29) (n = 17)
Emotionally reactive 53.4 T 5.4 52.6 T 4.7 .57 Emotionally reactive 56.9 T 8.6 51.7 T 3.2 .01
Anxious/depressed 53.1 T 4.8 52.2 T 3.7 .46 Anxious/depressed 56.6 T 9.2 53.4 T 4.3 .14
Withdrawn 55.0 T 6.9 55.8 T 8.7 .68 Withdrawn 58.0 T 9.8 56.4 T 9.6 .53
Somatic complaints 54.1 T 5.6 54.7 T 6.2 .69 Somatic complaints 56.0 T 7.1 55.8 T 6.8 .76
Attention problems 54.6 T 6.9 52.9 T 6.0 .32 Attention problems 57.8 T 8.2 54.1 T 6.8 .16
Aggressive behavior 52.2 T 4.3 52.1 T 3.9 .92 Aggressive behavior 55.6 T 9.4 52.4 T 3.9 .16
Sleep problems 54.2 T 5.5 54.6 T 5.7 .82 Sleep problems 54.5 T 7.4 53.8 T 5.3 .69
Ages 6-18 years (n = 58) (n = 74) Ages 6-18 years (n = 72) (n = 93)
Anxious/depressed 52.3 T 4.6 54.3 T 8.0 .08 Anxious/depressed 54.1 T 6.8 54.8 T 7.6 .35
Withdrawn/depressed 55.1 T 6.8 55.9 T 8.0 .50 Withdrawn/depressed 57.2 T 8.8 57.5 T 9.3 .84
Somatic complaints 57.8 T 7.9 58.0 T 8.8 .90 Somatic complaints 54.9 T 5.9 56.3 T 6.3 .15
Rule-breaking behavior 53.2 T 4.9 53.3 T 5.0 .88 Rule-breaking behavior 54.0 T 5.9 53.5 T 5.1 .64
Aggressive behavior 52.9 T 5.2 54.2 T 7.2 .23 Aggressive behavior 54.3 T 6.6 55.0 T 6.9 .49
Social problems 54.5 T 5.8 56.4 T 8.3 .11 Social problems 57.3 T 7.5 58.3 T 8.2 .95
Thought problems 54.3 T 6.5 56.5 T 7.4 .07 Thought problems 55.3 T 7.4 57.3 T 8.2 .26
Attention problems 54.9 T 6.1 57.2 T 9.9 .11 Attention problems 56.3 T 7.7 58.3 T 9.0 .23
Data are meanTSD unless otherwise noted. Data are meanTSD unless otherwise noted.
*P values for comparison between target groups were determined using t tests. *P values for the comparison between target groups were determined using linear regression
†School functioning scores were not available for 6 (7%) lower target and 10 (11%) higher adjusting for age category, baseline overall performance (POPC > 1), and PRISM-III 12 score.
target patients at baseline. †School functioning scores were not available for 8 (8%) lower target and 5 (5%) higher target
patients at the 1-year assessment.
7.e5 Biagas et al