Web site designed and maintained by Chris Cairns  © SICS EBM Group 2004                                  

Up

 

Tight blood glucose control improves ICU survival

 

 

For every 29 patients given intensive insulin therapy, to keep glucose 4.4-6.1 mmol.l-1, compared to standard therapy, one less patient dies in ICU. (95% CI 17 to 101)

Increased risk of biochemical, but not symptomatic, hypoglycaemia.
Level 1+ evidence

 

Citation/s:                Intensive Insulin Therapy in Critically Ill Patients NEJM 2001; 345: 1359 - 67.
 

Lead author's name and fax: Dr Van Den Berghe, Department of Intensive Care, Hospital Gasthuisberg, University of Leuven. greta.vandenberghe@med.kuleuven.ac.be

 

Three-part Clinical Question: In ICU patients, does the use of intensive insulin therapy to maintain tight blood glucose control, compared to standard therapy, lead to improvements in ICU outcome?
 

Search Terms: 1. exp sepsis/ or severe sep$.tw or sept$.tw or sepsi$.tw (50301), 2. exp critical care/ or critical ca$.tw or intensive ca$.tw (22553), 3. exp insulin or insuli$.tw (50202), 4. 1 and 2 and 3 (25), 5. therapy filter (652119), 6. 4 and 5 (17)

 

The Study: Single-blinded randomised controlled trial with intention-to-treat.
 

The Study Patients: All patients admitted to a surgical ICU in Belgium (62% had cardiac surgery). Median APACHE 9 (interquartile range 7-13). Median TISS 43. 13% had diabetes. Randomised at time of ICU admission. All patients given iv glucose on admission, following day started on parenteral / enteral nutrition or enteral nutrition alone. Matched for blood glucose at admission.
 

Control group (N = 783; 783 analysed): Insulin infusion (1 unit/ml) started if glucose > 12 mmol.l-1, and titrated to keep glucose in range 10.0 - 11.1 mmol.l-1. Blood glucose checked 1 - 4 hourly, algorithm used and discussion with study clinician not involved in patient care.
 

Experimental group (N = 765; 765 analysed): Insulin infusion (1 unit/ml) started if glucose > 6.1 mmol.l-1, and titrated to keep glucose in range 4.4 - 6.1 mmol.l-1. Blood glucose checked 1 - 4 hourly, algorithm used and discussion with study clinician not involved in patient care.

 

The Evidence:

Outcome

Time to Outcome

CER

EER

RRR

ARR

NNT

Death

In ICU

0.080

0.046

43%

0.034

29

95% Confidence Intervals:

12% to 73%

0.010 to 0.058

17 to 101

Hypoglycaemia

In ICU

0.0077

0.0509

-561%

-0.043

-23

95% Confidence Intervals:

-778% to -344%

-0.060 to -0.026

-38 to -17

 

Comments:

 

1.    Predominantly cardiac surgery patients (59% had CABG) could this group be more like the DIAGMI group of patients than "average" ICU patients? 

2.    No, main effect was reduction in deaths due to multiple organ failure due a proven septic focus.

3.    No details provided of algorithm ion article – aimed for normoglycaemia.   Now available via NEJM website.

4.    Reductions in sepsis and critical illness neuropathy, but EMG recordings are a surrogate end-point.

5.    Insulin is an inexpensive drug, especially compared to activated protein C, and may be more widely applicable.   Only single episodes of hypoglycaemia reported with no physical complications.

6.    We have a higher mortality rate, therefore as death (and death due to sepsis) is a more common event per 100 patients, we need to treat fewer patients to save a life = NNT /  f = 29 / 3 = 10.   Note this is a rough estimate.

 

Appraised by: Malcolm Daniel, Department of Anaesthesia, Glasgow Royal Infirmary; Sunday, November 18, 2001     Email:  md23s@udcf.gla.ac.uk

 

Printer-friendly format