CAT reviews

The Intensive Care Society 2009

Intensive versus conventional glucose

control in critically ill patients
Intensive glucose control with a target blood glucose 4.5-6.0 mmol/L increases the risk of death at 90 days (number
needed to harm (NNH)=38) and severe hypoglycaemia (NNH=16).
Level of evidence: 1+ (RCT with a low risk of bias)

Appraised by: O Tanner

Citation: The NICE-SUGAR Study Investigators. Intensive
versus conventional glucose control in critically ill patients.
New Engl J Med 2009;360:1283-97.
Lead Author: Simon Finfer, sfinfer@med.usyd.edu.au
Three part clinical question:
Patients: Adult ICU patients (6,104) in Australia, New Zealand
and Canada.
Intervention: Intensive vs conventional blood glucose control.
Outcomes: Primary 90-day all-cause mortality. Secondary
survival time during the first 90 days, cause-specific death,
duration of mechanical ventilation and renal replacement
therapy, and duration of ICU and hospital stay. Tertiary death
from any cause with 28 days, place of death, incidence of new
organ failure, positive blood cultures, blood transfusion,
volume of blood transfusion. Serious adverse events
incidence and severity of hypoglycaemia (blood glucose
2.2 mmol/L).
Search terms: Insulin, blood glucose, ICU, RCT, outcome,
mortality.
The study: Multi-centre, open label, randomised controlled
trial with intention-to-treat analysis.
The study patients: Adult medical or surgical ICU patients
who, within 24 hours of admission, were predicted to require
ICU treatment for three or more consecutive days. An arterial
line was in situ or inserted within one hour.
Exclusion criteria: Age <18 years, imminent death confirmed by
a documented treatment limitation order, diabetic ketoacidosis
or hyperosmolar state, previous hypoglycaemia without
documented full neurological recovery, high risk of
hypoglycaemia, previous enrolment in the study, unable to gain
consent, over 24 hours post admission to ICU.
Patients were stratified into the following sub-groups for the
purpose of primary outcome group analysis:

operative/non-operative (medical)
diabetic/non-diabetic
trauma/non-trauma
severe sepsis
systemic steroid treatment
APACHE II 25 or <25.
Blood glucose management in both control and
experimental groups was guided by a specific web-based
treatment algorithm. Blood glucose samples were obtained
from arterial or central venous lines. Blood glucose
measurements were performed initially hourly, then reduced to
two- and then four-hourly once insulin infusion, blood glucose
concentrations and caloric intake were stable.
The intervention continued until the patient no longer
required supplementary enteral or parental nutrition and had
started eating, or until ICU discharge or death or completion of
the 90-day trial period.

Control group: (n=3,014; 3,012 analysed) Conventional

glucose control (target blood glucose of 10.0 mmol/L or
less); insulin administered if blood glucose >10.0 mmol/L,
adjusted when required to maintain blood glucose between
8-10 mmol/L and discontinued when blood sugar was
<8.0 mmol/L.
Experimental group: (n=3,016; 3,010 analysed) Intensive
glucose control to maintain blood glucose between
4.5-6.0 mmol/L.
During the first 14 days, 70% of the intensive control group
and 71.4% of the conventional control group of patients
received enteral nutrition.
In the sub-group analysis, there was no significant
mortality difference demonstrated between therapies.
When combined, all-cause cardiovascular death was more
common in the intensive glucose control group (p=0.02).
Despite the increased mortality in the intensive control

The evidence: (all patients)

Outcome
Death

Time to outcome

CER

EER

RRR

ARR

NNT

90 days

0.249

0.275

-10%

-0.026

-38

-19% to -2%

-0.048 to -0.004

-262 to -21

95% confidence intervals:

Key CER: control event rate EER: experimental event rate RRR: relative risk reduction ARR: absolute risk reduction NNT: number needed to treat

216

Volume 10, Number 3, July 2009 JICS

Downloaded from inc.sagepub.com by guest on August 2, 2015

CAT reviews

The evidence: (hypoglycaemia)

Outcome
Severe
Death hypoglycaemia

Time to outcome

CER

EER

RRR

ARR

NNT

90 days

0.005

0.068

-1260%

-0.063

-16

-1447% to -1073%

-0.072 to -0.054

-19 to -14

95% confidence intervals:

Key CER: control event rate EER: experimental event rate RRR: relative risk reduction ARR: absolute risk reduction NNT: number needed to treat

group, no difference was demonstrated in either the length

of ICU and hospital stay or in the rates of organ
dysfunction.
EBM questions:
1. Do the methods allow accurate testing of the hypothesis? Yes.
This was a well-designed, large multicentre randomised
controlled trial.
2. Do the statistical tests correctly test the results to allow
differentiation of statistically significant results? Yes. The
correct statistical analysis was used and the study was
sufficiently powered.
3. Are conclusions valid in light of the results? Yes. The
intervention in 10% of patients randomly designated to the
intensive group was discontinued prematurely. However,
these patients were all included in the intention-to-treat
analysis.
4. Did results get omitted, and why? No.
5. Did they suggest areas of further research? Yes. The exact
cause of the increased mortality in the intensive control
group is unclear. Subgroup analysis demonstrated a trend
towards reduced mortality with tight control in trauma
patients and patients receiving corticosteroids at baseline.
This may merit further study.
6. Did they make any recommendations based on the results and
were they appropriate? Yes. A target blood glucose range of 810 mmol/L is more appropriate in the ICU setting than 4.56.0 mmol/L.
7. Is the study relevant to my clinical practice? Yes. This is the
first randomised, controlled study to look at a large,
heterogeneous group of critically ill patients in the intensive

care setting with a long follow-up period. Insulin

administration is commonplace. Many units in the UK have
adopted a tight glycaemic control policy following previous
studies.
8. What level of evidence does this study represent? 1+. The
authors recognised that the bedside treatment strategies
could not be blinded from the treating physicians.
Precautions taken to minimise bias included the use of a
centralised computerised algorithm to guide glucose control
and a feeding regimen protocol with supplementary
collection of details pertaining to nutritional support for all
patients.
9. What grade of recommendation can I make on this result alone?
B.
10.What grade of recommendation can I make when this study is
considered along with other available evidence? A.
11.Should I change my practice because of these results? Yes. The
findings from this study suggest that in a heterogeneous
critical care population, there is no additional benefit to be
gained through intensive blood glucose control in those who
have been on ICU for more than three days and that the
strategy is associated with more severe hypoglycaemic
events. A target blood sugar of between 8-10 mmol/L may be
more appropriate. In this study, 69% of patients required
insulin to achieve this.
12.Should I audit my current practice because of these results? Yes.
Appraised by:
Oona Tanner, ST2 in Anaesthesia, Department of Anaesthetics,
Stirling Royal Infirmary
omtanner@doctors.org.uk

217

JICS Volume 10, Number 3, July 2009

Downloaded from inc.sagepub.com by guest on August 2, 2015