Chen et al.

Critical Care (2018) 22:57

https://doi.org/10.1186/s13054-018-1976-2

RESEARCH Open Access

Tight glycemic control in critically ill

pediatric patients: a systematic review
and meta-analysis
Lvlin Chen1, Tiangui Li2, Fang Fang3, Yu Zhang1* and Andrew Faramand4

Abstract
Background: Hyperglycemia is prevalent in patients in the pediatric intensive care unit. The purpose of this study
was to describe the benefits and risks of tight glucose control (TGC) in critically ill children.
Methods: A systemic review and meta-analysis of the literature was carried out on randomized controlled trials
of TGC in critically ill children admitted to the pediatric intensive care unit. The databases searched were Medline,
Embase, and CENTRAL databases until May 1, 2017. Paired reviewers independently screened citations, assessed risk
of bias of included studies, and extracted data. A random-effects model was used to report all outcomes. The Grading
of Recommendations Assessment, Development and Evaluation system was used to quantify absolute effects and
quality of evidence. The primary outcome was hospital mortality. The secondary outcomes were hypoglycemia (any,
severe), sepsis, new need for dialysis, and seizures.
Results: A total of 4030 patients were included from six studies. All six studies were rated as at low risk of bias.
Our meta-analysis showed that TGC did not result in a decrease in risk of hospital mortality (odds ratio (OR), 0.95;
95% confidence interval (CI), 0.62–1.45; I2 = 40%; moderate quality), sepsis (OR, 0.82; 95% CI, 0.63–1.08), or seizures
(OR, 0.98; 95% CI, 0.59–1.63). TGC was associated with a decrease in new need for dialysis (OR, 0.63; 95% CI, 0.45–0.86).
However, TGC was associated with a significant increase in any hypoglycemia (OR, 4.39; 95% CI, 2.39–8.06) and severe
hypoglycemia (OR, 4.11; 95% CI, 2.67–6.32).
Conclusions: Among critically ill children with hyperglycemia, TGC does not result in a decrease in hospital mortality,
but appears to reduce a new need for dialysis. However, TGC is associated with higher incidence of hypoglycemia.
Systematic review registration: PROSPERO registration number CRD42017074039.
Keywords: Hyperglycemia, Randomized controlled trial, Children, Tight glycemic control, Mortality

Background practice of tight glucose control (TGC) with insulin

Hyperglycemia is prevalent in patients in the pediatric treatment in critically ill children has emerged as a
intensive care unit (PICU), with more than 80% having a plausible strategy to improve outcomes. To achieve such
blood glucose concentration greater than 110 mg/dl, ambitious goals in clinical practice, however, there are sig-
more than 60% a concentration greater than 150 mg/dl, nificant challenges in increased risk of hypoglycemia, add-
and more than 30% a concentration exceeding 200 mg/ itional personnel training, efficient utilization of medical
dl [1–4]. The extent of hyperglycemia is associated with resources, and radical revamping of glycemic management
adverse outcomes, including organ failure, length of stay protocols [10]. Furthermore, insulin treatment for critic-
in the PICU, and death [1, 5–9]. Consequently, the ally ill patients only works when the normal healthy fast-
ing ranges for blood glucose concentrations are achieved,
and these are lower in children than in adults [11].
* Correspondence: [email protected]
1
Department of Critical Care Medicine, Affiliated Hospital of Chengdu Several investigations have examined the benefits and
University, No.82, North Section 2, 2nd Ring Road, Jinniu District, Chengdu, risks of using TGC in critically ill children. In 2009,
Sichuan 610081, China Vlasselaers et al. [12] published a single-center,
Full list of author information is available at the end of the article

© The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Chen et al. Critical Care (2018) 22:57 Page 2 of 11

randomized controlled trial (RCT) of critically ill chil- Exclusion criteria

dren showing that TGC of 80–110 mg/dl reduced hos- Trials were excluded if the intervention was conducted
pital mortality by half, and reduced the infection rate primarily during the intraoperative period rather than
and length of stay, but also presented extremely high during the PICU stay, or if we were unable to obtain ad-
rates of severe hypoglycemia. However, subsequent mul- equate details of the study methodology or results from
ticenter large RCTs of TGC have failed to replicate this the article or study investigators.
mortality benefit [13, 14]. Furthermore, a recent trial,
the Heart and Lung Failure—Pediatric Insulin Titration
Missing data
(HALF-PINT) trial, was stopped early because the data
indicated a low likelihood of benefit and evidence for We contacted the investigators of all unpublished RCTs as
the possibility of harm [15]. well as any published RCTs in which data were missing to
A meta-analysis [16] has been published on this sub- confirm eligibility and obtain additional study details.
ject. However, results from two RCTs [15, 17] were not
included in the study. Moreover, the meta-analysis failed Duplicate publications
to conducted subgroup analyses on critical variables or a If separate articles from the same RCT were published,
formal evaluation of the quality of evidence (using the the article with the most updated data was selected. In
Grading of Recommendations Assessment, Development the case of duplicate publications, only one publication
and Evaluation (GRADE)). Thus, a comprehensive over- was included.
view of all RCTs involving critically ill children has never
been performed, and the optimal glucose goal remains
largely unknown. Information sources and search strategy
Consequently, the considerable controversy of RCTs Medline, Embase, and the Cochrane Library at the
and the limitations of the prior meta-analysis prompted CENTRAL Register of Controlled Trials were systemat-
us to perform an updated systematic review and meta- ically searched. Gray literature was searched through
analysis examining the risks and benefits of TGC as appropriate databases (British Library Thesis Service,
compared with usual care in critically ill children. More- Database of Abstracts of Reviews of Effects, OpenGrey).
over, we conducted subgroup analyses on three variables We also consulted databases of clinical trial registries
that have been debated in the controversy over TGI: (ClinicalTrials.gov, World Health Organization Inter-
glucose goal (< 110 mg/dl or 110–140 mg/dl), patient national Clinical Trials Registry Platform, European
setting (cardiac surgery or not cardiac surgery), and con- Union Clinical Trials Register, ISRCTN Registry). The
tinuous glucose monitoring. last electronic search was on May 1, 2017. We also hand
searched the references to the retrieved articles and
meta-analyses.
Methods For the search strategy, we used a combination of key-
Protocol and guidance words and MeSH terms for “child” AND “insulin”, using
The study protocol was prepared following PRISMA-P the sensitive search filters for therapeutic interventions
guidelines [18] and was registered at PROSPERO (Additional file 1: Supplemental Digital Content).
(CRD42017074039). The methods of the systematic re-
view and meta-analysis followed PRISMA guidelines
[19]. Reporting of statistical data in the study followed Study selection
SAMPL guidelines [20]. Two reviewers (YZ and LC) independently screened the
titles and abstracts of retrieved reports for potential eligi-
bility. They then screened the full text of potentially rele-
Study selection
vant trials. Disagreements were resolved by discussion and
Inclusion criteria
consensus or by consulting a third reviewer (TL).
We included RCTs that met each of the following cri-
teria: the setting was a PICU, and the patient was child
(age < 16 years); the intervention group received TGC Data collection process
(glucose goal < 140 mg/dl obtained using insulin treat- Following removal of duplicate articles, two reviewers
ment during part or all of the PICU stay); the compari- (YZ and LC) independently extracted data from the in-
son group received usual care (method of insulin cluded RCTs using a standardized electronic form.
administration and glucose goal could vary between tri- Disagreements between the two reviewers were resolved
als); and the primary or secondary outcomes included by discussion and consensus or by consulting a third re-
hospital mortality, hypoglycemia (any, severe), new need viewer (TL). Another reviewer (FF) double-checked the
for dialysis, sepsis, or seizures. extracted data.
Chen et al. Critical Care (2018) 22:57 Page 3 of 11

Fig. 1 Study selection for inclusion in meta-analysis

Outcomes and prioritization 30-day outcomes were reported, the former was used
The primary outcome was hospital mortality because for analysis.
we considered a reduction in hospital mortality to be The secondary outcomes were hypoglycemia (any,
the most important potential benefit of TGC. severe), sepsis, new need for dialysis, and seizures. We
Hospital mortality was defined as death occurring defined severe hypoglycemia as a blood glucose level
during the hospital stay or within 30 days following below 40 mg/dl and any hypoglycemia as a blood glu-
admission. In cases in which both in-hospital and cose level below 60 mg/dl. We defined sepsis to

Table 1 Characteristics of studies comparing tight and usual glucose control

Author Year Size (n) Centers (n) Country Setting Age, median (IQR)
Tight control Usual control
Vlasselaers et al. [12] 2009 700 1 Belgium Mixed with 75% cardiac surgery 1.4 (0.3–5.5) 1.3 (0.3–4.6)
Jeschke et al. [27] 2010 239 1 USA Burns 7.7 (5.2)a 10.8 (5.4)a
Agus et al. [14] 2012 980 2 USA Cardiac surgery 0.4 (0.2–0.8) 0.4 (0.2–0.9)
Alsweiler et al. [17] 2012 88 1 New Zealand Preterm Preterm babies
Macrae et al. [13] 2014 1369 13 UK Mixed with 60% cardiac surgery 0.5 (0.1–2.7)
Agus et al. [15] 2017 713 35 USA Mixed without cardiac surgery 5.5 (1.4–12.5) 6.7 (1.7–12.8)
IQR interquartile range
a
Mean (standard deviation)
Chen et al. Critical Care (2018) 22:57 Page 4 of 11

encompass the terms septicemia, bacteremia, or a de-

scription of positive blood cultures; a general description
of infection did not qualify.

Subgroup and sensitivity analyses

We performed subgroup analyses based on three vari-
ables prespecified clinically relevant to analysis out-
comes: glucose goal in the tight control group, cardiac
surgery, and continuous glucose monitoring. Subgroup
analyses were performed only if there were at least two
RCTs in each subgroup or a trial’s report permitted a
comparison within the trial.
Differing opinions exist on the optimal level of TGC.
The 2018 recommendations from the American
Diabetes Association recommend targeting blood glucose
levels of 140–180 mg/dl in critically ill patients [21–23].
We stratified studies by glucose goal in the TGC group
into two categories: very tight control (upper limit of glu-
cose goal < 110 mg/dl); and moderately tight control
(upper limit of glucose goal 110–140 mg/dl).
Because of the concern that the pathophysiological ef-
fect of hyperglycemia may differ between patients with
and without cardiac surgery, we stratified trials by PICU
setting into two categories: cardiac surgery and not car-
diac surgery. For trials involving mixed populations but
not presenting separate data for patients with cardiac
surgery, we included the pooled results in the cardiac
surgery subgroup only if ≥ 50% of patients underwent
cardiac surgery. Fig. 2 Risk of bias summary. Agus 2010 [14], Agus 2017 [15], Alsweiler
Continuous glucose monitoring has been shown to be 2012 [17], Jeschke 2010 [27], Macrae 2014 [13], Vlasselaers 2009 [12]
safe and effective in children and adults, and may assist
in the safer provision of tight glycemic control, with less
hypoglycemia [24]. Thus, we stratified trials by whether

Table 2 GRADE evidence profile of outcomes, tight glucose control vs usual glucose control
Outcome Number of Effect Qualitya Importance
children (studies)
Relative effect, OR (95% CI) Absolute risk (95% CI)
Hospital mortality 4021 (6 studies) 0.95 (0.62–1.45) 3 fewer per 1000 ⊕ ⊕ ⊕⊝ moderateb Critical
(from 20 fewer to 23 more)
Severe hypoglycemia 3835 (5 studies) 4.11 (2.67–6.32) 42 more per 1000 ⊕ ⊕ ⊕ ⊕ high Critical
(glucose < 40 ml/dl) (from 23 more to 69 more)b
Any hypoglycemia 3747 (4 studies) 4.57 (2.24–9.33) 157 more per 1000 ⊕ ⊕ ⊕ ⊕ highb Critical
(glucose < 60 mg/dl) (from 61 more to 299 more)
Dialysis 3049 (3 studies) 0.63 (0.45–0.86) 24 fewer per 1000 ⊕ ⊕ ⊕ ⊕ highb Important
(from 9 fewer to 36 fewer)
Seizures 3047 (3 studies) 0.98 (0.59–1.63) 0 fewer per 1000 ⊕ ⊕ ⊝⊝ lowb,c Important
(from 8 fewer to 12 more)
Sepsis 4021 (6 studies) 0.83 (0.63–1.08) 20 fewer per 1000 ⊕ ⊕ ⊕ ⊕ high Important
(from 45 fewer to 9 more)
GRADE Grading of Recommendations Assessment, Development and Evaluation, OR odds ratio, CI confidence interval
a
High quality, further research is very unlikely to change our confidence in the estimate of effect; moderate quality, further research is likely to have an important
impact on our confidence in the estimate of effect and may change the estimate; low quality, further research is very likely to have an important impact on our
confidence in the estimate of effect and is likely to change the estimate; very low quality, we are very uncertain about the estimate
b
Wide confidence
c
High heterogeneity
Chen et al. Critical Care (2018) 22:57 Page 5 of 11

they used continuous glucose monitoring to control components as having a moderate risk of bias, and trials
blood glucose. with more than four high-risk components as having a
We conducted sensitivity analyses to examine the im- high risk of bias. We used the GRADE approach to rate
pact of using alternative effect measures (odds ratio vs the quality of evidence and generate absolute estimates of
relative risk), pooling methods (Peto vs Mantel–Haenszel effect for the outcomes [25].
(M–H) or inverse variance), statistical models (fixed vs
random effects), and removing one study at a time.
Data synthesis
Risk of bias and quality of evidence Computations were performed with RevMan 5.3.3
Two reviewers (YZ and LC) independently assessed risk software (freeware available from The Cochrane Col-
of bias (low risk of bias, high risk of bias, or unclear risk laboration). We used the M–H method as the pri-
of bias) using the Cochrane risk of bias instrument, which mary analysis to estimate the odds ratio (OR) and
deals with random sequence generation and allocation 95% confidence intervals (CIs). Two-tailed P < 0.05
concealment (selection bias), blinding of study partici- was considered a criterion for statistical significance.
pants and personnel (performance bias), blinding of out- We report the results of the random-effects model
come assessment (detection bias), incomplete outcome for all outcomes. We assessed heterogeneity with the
data (attrition bias), selective reporting (reporting bias), Cochran Q test and the I2 test, with I2 values exceed-
and other bias. They resolved any disagreements by dis- ing 25%, 50%, and 75% representing low, moderate,
cussion and consensus or by consulting a third reviewer and high heterogeneity, respectively [26]. If an ana-
(TL). We judged trials with more than two high-risk lysis included 10 or more RCTs, we planned to use a

Fig. 3 Association of tight glucose control vs usual glucose control with hospital mortality, any hypoglycemia and severe hypoglycemia. CI confidence
interval, Agus 2010 [14], Agus 2017 [15], Alsweiler 2012 [17], Jeschke 2010 [27], Macrae 2014 [13], Vlasselaers 2009 [12]
Chen et al. Critical Care (2018) 22:57 Page 6 of 11

funnel plot to explore the possibility of published tight control and usual care groups (Additional file 1:
bias. Supplemental Digital Content Table S1).

Risk of bias and quality of evidence

Results
Treating clinicians were not blinded to treatment alloca-
Search results and characteristics of included studies
tion in any of the trials. Most investigators were, however,
The literature search yielded 154 articles, of which 22
blinded to treatment and outcomes. Study quality
were reviewed in full text (Fig. 1). Of these articles, six
appraisal indicated that studies were of variable quality
RCTs [12–15, 17, 27] met the inclusion criteria. The six
(Fig. 2) and that all six trials had a low risk of bias. Table 2
included trials randomized 4030 patients (1980 to tight
presents GRADE summary findings for all outcomes.
glycemic control and 2050 to receiving control) (Table 1).
Trials were conducted in a diverse array of countries, half
of them at a single center. Study sizes ranged widely Primary outcome: hospital mortality
(88–1369 patients), with two trials enrolling fewer than Hospital mortality was reported in six trials. These trials re-
300 patients and four trials enrolling more than 700 pa- ported 223 deaths (TGC, 110/1974 (5.6%); UGC, 113/2047
tients. The study participants encompass a broad distribu- (5.5%)). Our meta-analysis showed no significant difference
tion of critically ill children (Vlasselaers et al. [12], mixed in mortality between tight control vs usual control (OR,
with 75% cardiac surgery; Macrae et al. [13], mixed with 0.95; 95% CI, 0.62–1.45; P = 0.82; I2 = 40%; Fig. 3).
60% cardiac surgery; Jeschke et al. [27], severe burns; Agus Tests for heterogeneity identified the trial by Vlasselaers
et al. [14], cardiac surgery; Agus et al. [15], noncardiac sur- et al. [12] as having outlying results, which appeared to be
gery; Alsweiler et al. [17], preterm). TGI, as well as mean explained by the lowest glucose target in the six trials.
achieved glucose levels, varied between trials in both the Exclusion of the outlying trial resolved this heterogeneity

Fig. 4 Association of tight glucose control vs usual glucose control with sepsis, new need for dialysis, seizures. CI confidence interval, Agus 2010
[14], Agus 2017 [15], Alsweiler 2012 [17], Jeschke 2010 [27], Macrae 2014 [13], Vlasselaers 2009 [12]
Chen et al. Critical Care (2018) 22:57 Page 7 of 11

(I2 = 0%, P = 0.40), but did not significantly change the glucose care, 6.8%). TGC was associated with decrease
findings (OR, 1.13; 95% CI, 0.84–1.52). in dialysis (OR, 0.63; 95% CI, 0.45–0.86; Fig. 4).
Sensitivity analyses using an alternative statistical TGC did not result in a significant decrease in sepsis
method, effect measure, analysis model, and after re- or seizures (Fig. 4).
moving one study at a time showed similar results for
hospital mortality.
Subgroup analyses
We performed three subgroup analyses by the target of
TGC, by whether cardiac surgery or not, and by whether
Secondary outcomes: hypoglycemia, sepsis, new need for
using continuous glucose monitoring for insulin adjust-
dialysis, seizures
ment. However, there were no differences between groups
Hypoglycemia was reported in five published trials.
observed concerning those factors (Tables 3, 4 and 5).
TGC was associated with an increased risk of severe
hypoglycemia (OR, 4.11; 95% CI, 2.67–6.32; I2 = 0%;
42 more per 1000 patients; high quality) and any Discussion
hypoglycemia (OR, 4.39; 95% CI, 2.39–8.06; I2 = 83%; Findings and interpretations
157 more per 1000 patients; high quality) (Table 2 In this meta-analysis of six RCTs of TGC vs usual care in
and Fig. 3). critically ill children, we found no significant difference in
New need for dialysis was reported in three trials, and risk of hospital death, sepsis, or seizures, although TGC
the overall incidence was 5.6% (TGC, 4.5%; usual was associated with a significant reduction in dialysis. On

Table 3 Association of tight glucose control vs usual care with outcomes among critically ill adults, stratified by tight control
glucose goal
Subgroup of trialsa Number Event number/total number (%) OR (95% CI) I2 (%)
of trials
Tight control Usual control
Mortality
Very tight control 2 12/392 (3.1) 19/369 (5.1) 1.11 (0.12–10.60) 73
Moderately tight control 4 98/1582 (6.2) 94/1651 (5.7) 1.10 (0.81–1.49) 0
Overall 6 110/1974 (2.9) 113/2047 (5.5) 0.95 (0.62–1.45) 40
Any hypoglycemia (< 60 mg/dl)
Very tight control 2 112/392 (28.6) 17/396 (4.3) 9.35 (1.49–58.83) 88
Moderately tight control 3 259/1533 (16.9) 99/1514 (6.5) 3.00 (2.07–4.34) 54
Overall 5 371/1925 (19.3) 116/1910 (6.1) 4.39 (2.39–8.06) 83
Severe hypoglycemia (< 40 mg/dl)
Very tight control 2 24/392 (6.1) 5/396 (8.7) 5.23 (1.95–14.00) 0
Moderately tight control 3 85/1533 (5.5) 22/1514 (1.5) 3.88 (2.41–6.26) 0
Overall 5 109/1925 (18.4) 27/1910 (5.8) 4.11 (2.67–6.32) 0
Sepsis
Very tight control 2 121/392 (30.9) 143/396 (36.1) 1.01 (0.43–2.41) 73
Moderately tight control 4 95/1582 (6.0) 132/1651 (8.0) 0.81 (0.59–1.12) 18
Overall 6 216/1947 (11.1) 275/2047 (13.4) 0.83 (0.63–1.08) 32
Dialysis
Very tight control 1 2/349 (0.6) 6/351 (1.7) 0.33 (0.07–1.65) NA
Moderately tight control 2 67/1184 (5.7) 97/1165 (8.3) 0.64 (0.46–0.89) 0
Overall 3 69/1533 (4.5) 103/1516 (6.8) 0.63 (0.45–0.86) 0
Seizures
Very tight control 0 NA NA NA NA
Moderately tight control 3 31/1533 (2.0) 31/1514 (2.0) 0.98 (0.59–1.63) 51
Overall 3 31/1533 (2.0) 31/1514 (2.0) 0.98 (0.59–1.63) 51
OR odds ratio, CI confidence interval, NA not applicable
a
Very tight control, glucose goal < 110 mg/dl; moderately tight control, glucose goal 110–140 mg/dl
Chen et al. Critical Care (2018) 22:57 Page 8 of 11

Table 4 Association of tight glucose control vs usual care with outcomes among critically ill adults, stratified by with cardiac surgery
or without cardiac surgery
Subgroup of trials Number Event number/total number (%) OR (95% CI) I2 (%)
of trials
Tight control Usual control
Mortality
Noncardiac surgery 3 57/441 (12.9) 45/394 (9.2) 1.29 (0.52–3.23) 29
Cardiac surgery 3 53/1533 (3.5) 64/1516 (4.2) 0.79 (0.50–1.26) 26
Overall 6 110/1974 (5.6) 113/2047 (5.0) 0.95 (0.62–1.45) 40
Any hypoglycemia (< 60 mg/dl)
Noncardiac surgery 2 104/342 (30.4) 45/394 (11.4) 2.97 (2.01–4.41) 0
Cardiac surgery 3 267/1533 (17.4) 71/1516 (4.7) 5.72 (1.95–16.73) 91
Overall 5 371/1925 (19.3) 116/1910 (6.1) 4.39 (2.39–8.06) 83
Severe hypoglycemia (< 40 mg/dl)
Noncardiac surgery 2 25/392 (6.4) 9/394 (2.3) 2.95 (1.35–6.42) 0
Cardiac surgery 3 84/1533 (5.5) 18/1516 (1.2) 4.76 (2.84–7.97) 91
Overall 5 109/1925 (18.4) 27/1910 (5.8) 4.11 (2.67–6.32) 0
Sepsis
Noncardiac surgery 3 52/441 (11.8) 79/531 (14.9) 0.82 (0.55–1.22) 67
Cardiac surgery 3 164/1633 (10.0) 196/1516 (19.2) 0.79 (0.62–1.00) 0
Overall 6 216/1947 (11.0) 275/2047 (13.4) 0.83 (0.63–1.08) 32
Dialysis
Noncardiac surgery 3 69/1533 (4.5) 103/1516 (6.8) 0.63 (0.45–0.86) 0
Cardiac surgery 0 NA NA NA NA
Overall 3 69/1533 (4.5) 103/1516 (6.8) 0.63 (0.45–0.86) 0
Seizures
Noncardiac surgery 1 5/349 (1.4) 10/349 (2.9) 0.49 (0.17–1.46) NA
Cardiac surgery 2 26/1533 (1.7) 21/1165 (1.8) 1.22 (0.68–2.18) 50
Overall 3 31/1533 (2.0) 31/1514 (2.0) 0.98 (0.59–1.63) 51
OR odds ratio, CI confidence interval, NA not applicable

the other hand, we found clear evidence for the main harm could be explained by the small sample size of their study.
of TGC: hypoglycemia increased roughly 4-fold. However, Further, we have also provided absolute as well as relative
the rate of hypoglycemia varied greatly across RCTs. We risks and a formal rating of the quality of the evidence.
performed three prespecified subgroup analyses, stratified We quantified a new finding, a decreased risk of
by cardiac surgery, by continuous glucose monitoring, and dialysis with TGC. The previous meta-analysis study-
by glucose goal in tight control group, to explore potential ing the effect of TGC in critically ill children did not
areas of bias, but subanalyses did not differ from the overall report the outcome of dialysis, whereas the meta-
analysis. In short, our meta-analysis does not support the analysis of adults did not show this renoprotective
benefits of TGC reported in the initial trial by Vlasselaers effect [16, 28]. How to explain the conflicting results
et al. [12], yet it suggests a high risk of hypoglycemia. between our study and the other meta-analyses? One
of the reasons may be the inclusion of different types
Compared with other studies of patients. The evidence for a renoprotective effect
A previous meta-analysis of four RCTs examined the of TGC appears most pronounced in cardiac surgery
benefits and risks of TGC in critically ill children [16]. patients [29]. In our study, dialysis was reported in
Similar to our findings, the meta-analysis found no signifi- three trials [12–14], and more than half of children in
cant differences in mortality but an increased risk of those trials underwent cardiac surgery.
hypoglycemia between TGC and usual care in critically ill
children. They reported, however, that TGC appeared to Strengths and limitations
reduce acquired sepsis in critically ill children (OR 0.76; Strengths of this review include a comprehensive search
95% CI 0.59–0.99). This discrepancy with our findings for evidence; duplicate assessment of eligibility, risk of
Chen et al. Critical Care (2018) 22:57 Page 9 of 11

Table 5 Association of tight glucose control vs usual care with outcomes among critically ill adults, stratified by whether using continuous
glucose monitoring (CGM)
Subgroup of trials Number Event number/total number (%) OR (95% CI) I2 (%)
of trials
Tight control Usual control
Mortality
Using CGM 3 97/1553 (12.9) 86/1514 (9.2) 1.13 (0.83–1.53) 0
Not using CGM 3 13/441 (3.5) 27/533 (4.2) 0.70 (0.18–2.69) 49
Overall 6 110/1974 (5.6) 113/2047 (5.0) 0.95 (0.62–1.45) 40
Any hypoglycemia (< 60 mg/dl)
Using CGM 3 259/1533 (30.4) 99/1514 (11.4) 3.00 (2.07–4.36) 54
Not using CGM 2 24/392 (6.1) 5/396 (4.7) 5.23 (1.95–4.00) 0
Overall 5 371/1925 (19.3) 116/1910 (6.1) 4.39 (2.39–8.06) 83
Severe hypoglycemia (< 40 mg/dl)
Using CGM 2 85/1533 (30.4) 22/1514 (11.4) 3.88 (2.41–6.26) 0
Not using CGM 3 112/392 (28.6) 17/396 (4.7) 9.35 (1.49–58.83) 88
Overall 5 109/1925 (18.4) 27/1910 (5.8) 4.11 (2.67–6.32) 0
Sepsis
Using CGM 3 91/1553 (12.9) 101/1514 (9.2) 0.88 (0.66–1.18) 0
Not using CGM 3 125/441 (3.5) 174/533 (4.2) 0.76 (0.18–2.69) 69
Overall 6 216/1947 (11.0) 275/2047 (13.4) 0.83 (0.63–1.08) 32
Dialysis
Using CGM 2 67/1184 (4.5) 97/1165 (6.8) 0.64 (0.46–0.89) 0
Not using CGM 1 2/349 (4.5) 6/351 (6.8) 0.33 (0.07–1.65) NA
Overall 3 69/1533 (4.5) 103/1516 (6.8) 0.63 (0.45–0.86) 0
Seizures
Using CGM 3 31/1533 (2.0) 31/1514 (2.0) 0.98 (0.59–1.63) 51
Not using CGM 0 NA NA NA NA
Overall 3 31/1533 (2.0) 31/1514 (2.0) 0.98 (0.59–1.63) 51
OR odds ratio, CI confidence interval, NA not applicable

bias, and data abstraction; and assessments of risk of Second, since we have pooled results from individual
bias. We included a rigorous assessment of the quality RCTs, our analysis is limited by any flaws in the method-
of evidence and of the credibility of subgroup analyses. ology of these underlying trials. For example, all trials not
Moreover, we have presented absolute and relative risks, using a standard care group led to variable control groups.
which are crucial for making decisions regarding use of Third, all included studies were conducted in devel-
TGC in critically ill children. oped countries. Thus, our findings are applicable only to
Our study also has limitations. First, although there developed countries. Further research in other countries
were many similarities to the methodology of the in- would add to the generalizability.
cluded RCTs, there was also some variability, including Fourth, the small numbers of studies and those in
nutritional supplementation, target of tight glycemic individual subgroup analyses limited power in our con-
control, definition of hypoglycemia, blood glucose moni- clusions. Moreover, the limited number of included trials
toring, quality of glucose control, and duration and afforded modest ability to detect the presence of publi-
route used for the insulin therapy protocols. These cation bias [18]. However, publication bias is unlikely as
diversities may have influenced the pathophysiology and most of included RCTs had negative results.
implications of hyperglycemia. We present the findings
stratified by some widely debated variables—glucose goal Conclusions
in the tight control group, blood glucose monitoring, In summary, we believe the six trials included in our
and whether cardiac surgery. However, we were unable meta-analysis allow us to conclude about the benefits
to assess the effect of other important variables for lack and risks of TGC critically ill children. We found that
of adequate data. TGC was not associated with a significant reduction in
Chen et al. Critical Care (2018) 22:57 Page 10 of 11

hospital mortality, seizures, or sepsis, but appears to be Received: 20 October 2017 Accepted: 5 February 2018
associated with a reduction in new need for dialysis.
However, TGC was associated with a markedly increased
risk of hypoglycemia. These findings were consistent References
with recent guidelines [22, 23]. Thus, adoption of TGC 1. Preissig CM, Rigby MR. Pediatric critical illness hyperglycemia: risk factors
associated with development and severity of hyperglycemia in critically ill
in critically ill children cannot be recommended for rou-
children. J Pediatr. 2009;155(5):734–9.
tine use unless further high-quality and well-powered 2. Allen HF, Rake A, Roy M, Brenner D, McKiernan CA. Prospective detection of
evidence shows benefit. hyperglycemia in critically ill children using continuous glucose monitoring.
Pediatr Crit Care Med. 2008;9(2):153–8.
3. Bowlby D, Rapaport R, Hojsak J. Hyperglycemia in critically ill children. J
Key messages Pediatr. 2006;148(6):847. author reply 847–8
 Tight glycemic control does not appear to improve 4. Srinivasan V, Spinella PC, Drott HR, Roth CL, Helfaer MA, Nadkarni V.
Association of timing, duration, and intensity of hyperglycemia with
mortality in critically ill children. intensive care unit mortality in critically ill children. Pediatr Crit Care Med.
 Tight glycemic control reduces a new need for dialysis 2004;5(4):329–36.
in critically ill children. 5. Gordillo R, Ahluwalia T, Woroniecki R. Hyperglycemia and acute kidney
injury in critically ill children. Int J Nephrol Renovasc Dis. 2016;9:201–4.
 Tight glycemic control greatly increases the risk of
6. Marsillio LE, Ginsburg SL, Rosenbaum CH, Coffin SE, Naim MY, Priestley MA,
hypoglycemia in critically ill children. Srinivasan V. Hyperglycemia at the time of acquiring central catheter-
associated bloodstream infections is associated with mortality in critically ill
children. Pediatr Crit Care Med. 2015;16(7):621–8.
Additional file 7. Kyle UG, Coss Bu JA, Kennedy CE, Jefferson LS. Organ dysfunction is
associated with hyperglycemia in critically ill children. Intensive Care Med.
Additional file 1: is Supplemental Digital Content: Table S1. 2010;36(2):312–20.
(DOCX 17 kb) 8. Otto-Buczkowska E, Dworzecki T, Mazur U. Alterations of blood glucose
homeostasis in critically ill children—hyperglycemia. Pediatr Endocrinol
Diabetes Metab. 2007;13(1):43–6.
Abbreviations 9. Faustino EV, Apkon M. Persistent hyperglycemia in critically ill children. J
CI: Confidence interval; OR: Odds ratio; PICU: Pediatric intensive care unit; Pediatr. 2005;146(1):30–4.
RCT: Randomized controlled trial; TGC: Tight glucose control 10. Kovalaske MA, Gandhi GY. Glycemic control in the medical intensive care
unit. J Diabetes Sci Technol. 2009;3(6):1330–41.
Acknowledgements 11. Brown TC, Connelly JF, Dunlop ME, McDougall PN, Tibballs J. Fasting plasma
None. glucose in children. Aust Paediatr J. 1980;16(1):28–9.
12. Vlasselaers D, Milants I, Desmet L, Wouters PJ, Vanhorebeek I, van den Heuvel I,
Funding Mesotten D, Casaer MP, Meyfroidt G, Ingels C, et al. Intensive insulin therapy
This work is supported by projects of the National Natural Science Foundation for patients in paediatric intensive care: a prospective, randomised controlled
of China (contract/grant number: 81100925, 81472361). study. Lancet (London, England). 2009;373(9663):547–56.
13. Macrae D, Grieve R, Allen E, Sadique Z, Morris K, Pappachan J, Parslow R,
Availability of data and materials Tasker RC, Elbourne D. A randomized trial of hyperglycemic control in
The datasets used and/or analyzed during the current study are available from pediatric intensive care. N Engl J Med. 2014;370(2):107–18.
the corresponding author on reasonable request. 14. Agus MS, Steil GM, Wypij D, Costello JM, Laussen PC, Langer M, Alexander
JL, Scoppettuolo LA, Pigula FA, Charpie JR, et al. Tight glycemic control
Authors’ contributions versus standard care after pediatric cardiac surgery. N Engl J Med. 2012;
YZ and LC carried out study conception, systematic search, manual search, data 367(13):1208–19.
retrieval, assessment of risk of bias, analysis, draft of manuscript and revision 15. Agus MS, Wypij D, Hirshberg EL, Srinivasan V, Faustino EV, Luckett PM,
according to other authors’ suggestions, and submission. TL and FF carried out Alexander JL, Asaro LA, Curley MA, Steil GM, et al. Tight glycemic control in
analysis and critically revising the manuscript. AF carried out drafting and critically critically ill children. N Engl J Med. 2017;376(8):729–41.
revising the manuscript. All authors read and approved the final manuscript. 16. Srinivasan V, Agus MS. Tight glucose control in critically ill children—a
systematic review and meta-analysis. Pediatr Diabetes. 2014;15(2):75–83.
Ethics approval and consent to participate 17. Alsweiler JM, Harding JE, Bloomfield FH. Tight glycemic control with insulin
Not applicable. in hyperglycemic preterm babies: a randomized controlled trial. Pediatrics.
2012;129(4):639–47.
Consent for publication 18. Shamseer L, Moher D, Clarke M, Ghersi D, Liberati A, Petticrew M, Shekelle P,
Not applicable. Stewart LA, Group P-P. Preferred reporting items for systematic review and
meta-analysis protocols (PRISMA-P) 2015: elaboration and explanation. BMJ
Competing interests (Clinical research ed). 2015;349:g7647.
The authors declare that they have no competing interests. 19. Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gotzsche PC, Ioannidis JP, Clarke M,
Devereaux PJ, Kleijnen J, Moher D. The PRISMA statement for reporting
systematic reviews and meta-analyses of studies that evaluate health care
Publisher’s Note interventions: explanation and elaboration. Ann Intern Med. 2009;151(4):W65–94.
Springer Nature remains neutral with regard to jurisdictional claims in 20. Lang TA, Altman DG. Basic statistical reporting for articles published in
published maps and institutional affiliations. biomedical journals: the “Statistical Analyses and Methods in the Published
Literature” or the SAMPL Guidelines. Int J Nurs Stud. 2015;52(1):5–9.
Author details 21. Rhodes A, Evans LE, Alhazzani W, Levy MM, Antonelli M, Ferrer R, Kumar A,
1
Department of Critical Care Medicine, Affiliated Hospital of Chengdu Sevransky JE, Sprung CL, Nunnally ME, et al. Surviving Sepsis Campaign:
University, No.82, North Section 2, 2nd Ring Road, Jinniu District, Chengdu, International Guidelines for Management of Sepsis and Septic Shock: 2016.
Sichuan 610081, China. 2Longquanyi Hospital, Chengdu, Sichuan, China. Crit Care Med. 2017;45(3):486–552.
3
West China Hospital, Sichuan University, Chengdu, Sichuan, China. 22. American Diabetes Association. 14. Diabetes Care in the Hospital: Standards of
4
University of Pittsburgh Medical Center, Pittsburgh, PA, USA. Medical Care in Diabetes—2018. Diabetes Care. 2018;41(Suppl 1):S144–51.
Chen et al. Critical Care (2018) 22:57 Page 11 of 11

23. American Diabetes Association. 12. Children and Adolescents: Standards of

Medical Care in Diabetes—2018. Diabetes Care. 2018;41(Suppl 1):S126–36.
24. Yeh HC, Brown TT, Maruthur N, Ranasinghe P, Berger Z, Suh YD, Wilson LM,
Haberl EB, Brick J, Bass EB, et al. Comparative effectiveness and safety of
methods of insulin delivery and glucose monitoring for diabetes mellitus: a
systematic review and meta-analysis. Ann Intern Med. 2012;157(5):336–47.
25. Guyatt GH, Oxman AD, Vist GE, Kunz R, Falck-Ytter Y, Alonso-Coello P,
Schunemann HJ, Group GW. GRADE: an emerging consensus on rating
quality of evidence and strength of recommendations. BMJ (Clinical
research ed). 2008;336(7650):924–6.
26. Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis.
Stat Med. 2002;21(11):1539–58.
27. Jeschke MG, Kulp GA, Kraft R, Finnerty CC, Mlcak R, Lee JO, Herndon DN.
Intensive insulin therapy in severely burned pediatric patients: a prospective
randomized trial. Am J Respir Crit Care Med. 2010;182(3):351–9.
28. Wiener RS, Wiener DC, Larson RJ. Benefits and risks of tight glucose control
in critically ill adults: a meta-analysis. JAMA. 2008;300(8):933–44.
29. Gunst J, Schetz M. Clinical benefits of tight glycaemic control: effect on the
kidney. Best Pract Res Clin Anaesthesiol. 2009;23(4):431–9.

Submit your next manuscript to BioMed Central

and we will help you at every step:
• We accept pre-submission inquiries
• Our selector tool helps you to find the most relevant journal
• We provide round the clock customer support
• Convenient online submission
• Thorough peer review
• Inclusion in PubMed and all major indexing services
• Maximum visibility for your research

Submit your manuscript at

www.biomedcentral.com/submit