NICE-SUGAR

NICE-SUGAR: Intensive versus Conventional Glucose Control

Finfer et al. N Engl J Med 2009; 360:1283-97.

Clinical Question

  • In critically ill adults that are expected to be in Intensive Care for 3 days or more, does intensively controlled blood glucose compared to conventionally controlled blood glucose reduce mortality at 90 days?

Design

  • Randomised, controlled trial.
  • Stratified randomisation using minimisation algorithm for type of admission (operative vs non-operative) and geographical region.
  • Allocation was appropriately concealed before randomisation.
  • Treating clinicians were not blinded.
  • Aimed to recruit 6,100 patients to provide a power of 90% to detect an absolute difference in mortality of 3.8%, assuming a baseline mortality of 30%, with an accepted two-sided alpha level of 0.05 or less.
  • Intention-to-treat analysis method.

Setting

  • Medical and surgical ICUs across Australia and New Zealand (87.5%), and North America (12.5%).
  • 38 academic tertiary care hospitals and 4 community hospitals
  • December 2004 to November 2008.

Population

  • Inclusion: Patients expected to require 3 or more days of care on an ICU, with an arterial line to allow blood sampling
  • Exclusion: Admitted for DKA or hyperosmolar state; expected to be eating by day 3; high-risk of hypoglycaemia such as insulin-secreting tumour or fulminant liver failure.
  • 40,171 were screened. 6,104 patients were recruited.

Intervention

  • Intensive-control: control of blood glucose to a level between 4.5 and 6.0 mmol/L (81 and 108 mg/dL).
    • Insulin in saline was administered to keep glucose less than 6.0 mmol/L (108 mg/dL)
    • This was titrated to blood glucose, and stopped if it fell below 4.5 mmol/L (81 mg/dL)
    • The intervention was stopped when patients were eating, discharged from ICU, deceased or 90 after randomisation
    • 97.2% required insulin during the trial
    • Mean time-weighted blood glucose level was 6.4 mmol/L (115 mg/dL).

Control

  • Conventional-control: control of blood glucose to a level less than 10.0 mmol/L (180mg/dL)
    • Insulin in saline was administered to keep glucose less than 10.0 mmol/L (180 mg/dL)
    • This was titrated to blood glucose, and stopped if it fell below 8.0 mmol/L (144 mg/dL)
    • The intervention was stopped when patients were eating, discharged from ICU, deceased or 90 after randomisation
    • 69.0% required insulin during the trial
    • Mean time-weighted blood glucose level was 8.0 mmol/L (144 mg/dL).

Outcome

  • Primary outcome: There was a statistically significant difference in 90-day mortality that favoured conventional-control.
    • Intensive-control: 27.5%
    • Conventional-control: 24.9%
    • Absolute difference: 2.6% (95% CI 0.4-4.8)
    • Odds ration for death with intensive control: 1.14 (95% CI 1.02-1.29; p=0.02)
  • Secondary outcome:
    • No difference in
      • the mortality rate at 28 days
      • the median length of stay in ICU or hospital
      • the development of new organ failure
      • the number of days of mechanical ventilation or renal replacement therapy
      • the rates of red-cell tranfusion or positive blood culture
    • a significant difference in
      • the incidence of severe hypoglycaemia (6.8% in intensive-control; 0.5% in conventional-control)

Authors’ Conclusions

  • A blood glucose target of less than 10.0 mmol/L (180 mg/dL) resulted in lower mortality than a target of 4.5 to 6.0 mmol/L (81 to 108 mg/dL), and the authors do not recommend use of the lower target in critically ill adults.

Strengths

  • Well designed, pragmatic study designed to answer a clinically relevant research question.
  • 2.4% withdrew consent and 0.3% were lost-to-follow-up: negligible attrition bias.
  • 99.5% were administered the correct management according to the study algorith: stengthens the internal validity (accuracy) of the results.

Weaknesses

  • The treatment was discontinued prematurely in 10.0% of patients in the intervention group and 7.4% in the conventional-control group. The most common reasons were physician decision or a change to palliative care. This difference may have introduced bias to the results, especially as the clinicians were unblinded. These rates are lower than the primary outcome incidence, and are therefore unlikely to change the direction of the primary outcome, but may have exaggerated the effect size.
  • More patients in the intensive-control group received corticosteroids (34.6% vs 31.7%; p=0.02). This may affect the measured outcome, but predicting how and to what extent is not possible.
  • The median blood glucose level achieved in the intensive-control group was 6.4 mmol/L, which is above the target range of 4.5 to 6.0 mmol/L. This demonstrates the difficulty in achieving this target range, and it may reduce the observed difference between the two groups.
  • The majority of patients received nutrition via enteral feeding routes, perhaps reducing generalisability to patients being fed parenterally.
  • The authors do not attempt to explain why there was a difference in mortality at 90 days, but not at 28 days. Why should intensive-control of blood glucose have a harmful effect on late survival odds?

 

The Bottom Line

  • This trial demonstrates better survival if a conventional blood glucose target of less than 10.0 mmol/L (180 mg/dL) is used in critically ill patients.
    The trial’s strong internal validity and external generalisability suggests this should be used in clinical practice in developed countries for adult patients.

Links

Full text pdf / abstract / doi: 10.1056/NEJMoa0810625

Editorial, Commentaries or Blogs

Metadata

Summary author: Duncan Chambler (@DuncanChambler)
Summary date: 11 June 2014
Peer-review editor: Steve Mathieu (@stevemathieu75)

3 comments

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.