1. Blood Sugar Control in ICU
Muhammad Asim Rana
MBBS, MRCP, FCCP, SF-CCM, EDIC
Department of Critical Care Medicine
King Saud Medical City
Riyadh Saudi Arabia
2. Introduction
• Hyperglycemia is common in critically ill patients .
• Current literature shows conflicting results on the
effects of more intensive glucose controls versus
conventional glucose control in critically ill
patients.
3. Case Study 1
• 32 years male without any prior medical history,
admitted to surgical ICU after an assault with head &
chest injuries
• Second day routine labs showed random blood sugar
to be more than 10mmols
• Repeated labs failed to show a fall below 8.5mmols
• Is he a diabetic?
• What should be our plan to control the blood sugar?
Why control of blood sugar is necessary?
4. Case Study 2
• 52 years male with prior medical history of HTN on Rx,
admitted to surgical ICU post laparotomy for bowel
perforation & septic shock
• Labs showed random blood sugar to be more than
8.0mmols
• Repeated attempts failed to control blood sugar on subcut
insulin
• Are we on right track?
• What should be our plan to control the blood sugar?
Why control of blood sugar is necessary?
5. Remember
• Hyperglycemia commonly ensues, even in patients
who do not have pre-existing diabetes mellitus.
• The rationale behind the use of insulin therapy in
critically ill patients is that severe injury or
infection alters carbohydrate metabolism, which
results in insulin resistance.
6. Three Types of Hyperglycemic Patient
• Known history of diabetes
• Existing, but unrecognized, diabetes
• Stress hyperglycemia
Clement et al. Diabetes Care. 2004;27:553-591.
7. Why blood sugar control is important ?
Hyperglycemia Adversely Affects Outcomes
• Hyperglycemia impacts
•
•
•
•
Mortality
Morbidity
Rate of infections
Length of stay in ICU (LOS)
8. Hyperglycemia and Mortality in the MICU
~3x
45
Mortality Rate (%)
40
~4x
~2x
35
30
25
20
15
10
5
0
80-99
100-119
120-139
140-159 160-179 180-199
200-249
Mean Glucose Value (mg/dL)
N=1826 ICU patients.
Krinsley JS. Mayo Clin Proc. 2003;78:1471-1478.
250-299
>300
9. Hyperglycemia:
A Predictor of Mortality Following CABG in Diabetics
10
Postop
Mortality
P<0.0001
BG >200
n=662
5.0% *
1.8%
*P<0.001
Adjusted for 19 clinical and operation variables
First Postop Glucose >200
8.6
8
Postop Mortality (%)
BG <200
n=1369
• 2x LOS
• 3x Vent duration
• 7x mortality !!!
5.8
6
3.8
4
2
1.4
1.7
2.1
0
CABG, coronary artery bypass graft.
Furnary AP et al. Circulation. 1999:100 (Suppl I): I-591.
<150
175200150200
225
175
Blood Glucose (mg/dL)
225250
>250
11. Wake up… here comes…
The milestone study which revolutionized …
… THE HISTORY OF ICM
12. Intensive Insulin Management in Medical-Surgical ICU
P < 0.001
P < 0.002
29.3%
Reduction
Mean BG Levels
(mg/dL)
Baseline group (n = 800)
Krinsley JS. Mayo Clin Proc. 79:992-1000, 2004.
Hospital Mortality
(%)
Glucose management group (n = 800)
13. Benefits of Tight Glycemic Control:
Observational Studies and Early Intervention Trials
Study
Setting
Population
Clinical Outcome
Furnary, 1999
ICU
DM undergoing open
heart surgery
65% infection
Furnary, 2003
ICU
DM undergoing CABG
57% mortality
Krinsley, 2004
Medical/surgical
ICU
Mixed, no Cardiac
29% mortality
Malmberg, 1995
CCU
Mixed
28% mortality
After 1 year
Van den Berghe, 2001*
Surgical ICU
Mixed, with CABG
42% mortality
Lazar, 2004
OR and ICU
CABG and DM
60% A Fib post op
survival 2 yr
*RCT, randomized clinical trial.
Kitabchi & Umpierrez. Metabolism. 2008;57:116-120.
14. Intensive Insulin Therapy in Critically Ill Patients:
The Leuven SICU Study
• Randomized controlled trial:
• 1548 patients admitted to a surgical ICU, receiving mechanical
ventilation. Patients were assigned to receive either:
• Conventional therapy: IV insulin only if BG >215 mg/dL
• Target BG levels: 180-200 mg/dL
• Mean daily BG: 153 mg/dL
• Intensive therapy: IV insulin if BG >110 mg/dL
• Target BG levels : 80-110 mg/dL
• Mean daily BG: 103 mg/dL
Van den Berghe et al. N Engl J Med. 2001;345:1359-1367.
15. Intensive Insulin Therapy in Critically Ill Patients: SICU
34%
*
46%
*
*
*
*
41%
*
*P<0.01
Van den Berghe et al. N Engl J Med. 2001;345:1359-1367.
Relative Risk Reduction (%)
16. Intensive Glucose Management in RCT
Primary
Outcome
Trial
N
Setting
Van den
Berghe
2006
HI-5
2006
1200
MICU
Hospital
mortality
240
CCU AMI
6-mo
mortality
Glucontrol
2007
1101
ICU
ICU
mortality
-1.5%
Ghandi
2007
399
OR
Composite
VISEP
2008
537
ICU
28-d
mortality
De La Rosa
2008
504
SICU
MICU
28-d
mortality
NICE-SUGAR
2009
6104
ICU
3-mo
mortality
*not significant
ARR
RRR
2.7%
7.0%
Odds Ratio
(95% CI)
P-value
0.94*
(0.84-1.06)
N.S.
NR
N.S.
-10%
1.10*
(0.84-1.44)
N.S.
2%
4.3%
1.0*
(0.8-1.2)
N.S.
1.3%
5.0%
0.89*
(0.58-1.38)
NR
N.S.
1.14
(1.02-1.28)
< 0.05
-1.8%* -30%*
-4.2% * -13%*
-2.6%
-10.6
N.S.
18. NICE – SUGAR
March 26, 2009 NEJM Vol 360 (13)
• Open Label RCT, Multinational
• 6104 critically ill patients
• Intensive infusion (81-108 mg/dL) vs
“Conventional” control (144 – 180 mg/dL)
• 90 day survival – primary end point
19. Description
• 3054 patients were assigned to the intensive
control group and 3050 to the conventional control
group.
• 829 patients(27.5%) died in the intensive control
group and 751(24.9%) in the conventional-control
group which is a difference between surgical vs.
medical ICU patients.
• Severe hypoglycemia (<40 mg/dL) was recorded in
6.8% of patients in the intensive control group, vs.
0.5% in the conventional group.
20. Blood Glucose Values, According
to Treatment Group
The NICE-SUGAR Study Investigators.
N Engl J Med 2009;360:1283-1297
21. Probability of Survival
Odds Ratios for Death,
According to Treatment
Group
The NICE-SUGAR Study Investigators. N
Engl J Med 2009;360:1283-1297
22. Probability of Survival and Odds Ratios for Death,
According to Treatment Group
Operative Admission
Diabetes
Severe Sepsis
Trauma
Apache Score
Corticosteroids
All deaths at day 90
Favors Favors
IIT Conventional
Nice Sugar, NEJM 2009;360:1283
23. NICE - SUGAR
•
•
•
•
90 day mortality
Severe hypoglycemia
Glucose control (median)
Insulin infusion
27.5% vs 24.9%
6.8% vs 0.5%
107 vs 141 mg/dL
97% vs 69%
24. No difference
•30 day mortality
•ICU days
•Hospital days
•Days of mechanical ventilation
•Days of renal replacement
•Organ failures
25. Discussion
•
•
•
•
•
•
Positives:
Large multi-center study.
Robust statistical analysis.
Use of a uniform insulin protocol between sites.
The primary outcome in this in unbiased.
Good representation od critically ill patients
Enrolled more patients than trials that preceded it.
26. Discussion
•
•
•
•
•
Limitations:
More patients in the IIT group received corticosteroids
which could effect the variable were studying
10% if the IIT discontinued prematurely.
No significant difference in the primary outcome,
death.
Inclusion criteria ,i.e., length of stay is a subjective
parameter.
The study was not blinded to the treating personnel
27. Why the differences?
When all the available data are considered, an important step is to
establish reasons for the discrepancies in the published literature.
28. Reasons for discrepancies
• Differences in populations of patients (for example,
reasons for admittance to the ICU),
• Insulin-treatment protocols,
• Mortality,
• Glucose goals,
• Glucose concentrations actually achieved
• The use of parenteral nutrition
• The expertise and experience of nursing staff at a
particular institution could also influence the outcome
29. Methods used to measure glucose
• Another critical, but frequently overlooked, factor
is the method used to measure glucose levels.
• Arterial blood gas analyzer
• Capillary blood with point of care meters
• Differences among precesion of different glucometers
30. Patient-specific factors
• Some glucose meters are affected by partial pressure of oxygen
and hematocrit.
• Reduced tissue perfusion in hypotensive patients results in large
differences in glucose concentrations in capillary blood samples,
despite minimal alterations in arterial blood samples.
• Another variable is that the glucose concentrations in arterial,
venous and capillary blood all differ.
• Although these differences are minimal in fasting individuals,
postprandial capillary glucose values are 1.1–1.4 mmol/l higher than
those in venous blood.
• Finally, as a consequence of differences in water content, glucose
concentrations in plasma are ~11% higher than those in whole
blood if the hematocrit is normal.
• Some, or perhaps all, of these factors might have contributed to
the results reported by the NICE-SUGAR investigators.
31. Questions
1. What is the optimal target for the glucose therapy.
2. Does a particular sub-set of patients benefits from
tight glucose control
3. What about hypoglycemia?
4. Strategies for the future management of blood
glucose in the ICU
32. AACE - Consensus Conference
Blood Glucose Targets
• Upper Limit Inpatient Glycemic Targets:
ICU: 110 mg/dl (6.1 mmol/L)
Non-critical care (limited data)
• Pre-prandial: 110 mg/dl (6.1 mM)
• Maximum: 180 mg/dL (10 mM)
The current ADA guideline for pre-prandial
plasma glucose is now < 126 mg/dL
AACE- Endocrine Practice 10 (1): 77-82, 2004
ADA- Diabetes Care 27: 553-591, 2004
Diabetes Care 31:S12-S54, 2008 - The language around glycemic targets
has softened in the 2008 version of the ADA Standards.
35. Favors IIT
All Mixed ICU
0.99 (0.87-1.12)
All Medical ICU
1.00 (0.78-1.28)
All Surgical ICU
0.63 (0.44-0.91)
ALL ICU
0.93 (0.83-1.04)
Favors Control
36. Points to ponder
• The NICE-SUGAR trial adds to the accumulating data on the
use of tight glucose control protocols in patients in the ICU;
however, this study does not 'close the case' on these
protocols.
• Further large trials are necessary if the question of whether
intensive insulin therapy improves the outcomes of selected
ICU patients is to be unequivocally resolved.
• As the dose of insulin used in critically ill patients is
determined exclusively by their blood glucose value, accurate
measurement of glucose concentration is essential to achieve
the desired targets and to avoid hypoglycemia.
• In multicenter trials, a particularly important issue is that
glucose measurements among institutions are standardized
to avoid variability among patients.
• Highly accurate measurements of glucose concentration will,
therefore, be necessary in future research
37. Practice Points
• Intensive insulin therapy to maintain blood glucose
concentrations <6.0 mmol/l in patients hospitalized in
the intensive-care unit might increase mortality
• Patients enrolled in NICE-SUGAR who received
intensive insulin therapy had increased hypoglycemia
and cardiovascular mortality compared with patients
who received conventional insulin therapy
• Accurate measurement of blood glucose concentration
is necessary to achieve the desired target and avoid
hypoglycemia
The slide shows results of trials of glucose management in critical care patients.1-10 Some early randomized trials suggested that intensive glucose lowering can improve outcomes.1,2 However, more recent studies in the critical care population were unable to replicate earlier studies, and identified severe hypoglycemia as a significant risk of intensive glucose control.3-10 In the study by Ghandi et al,7 intensive insulin therapy during cardiac surgery did not reduce perioperative death or morbidity. In the NICE-SUGAR study,10 critically ill patients treated in the intensive glucose control group (81-108 mg/dL) were 14% more likely to die (27.5% vs 24.9%) than were those in the conventional glucose control group (144-180 mg/dL). Severe hypoglycemia (blood glucose ≤40 mg/dL) occurred in 6.8% of the intensive-control group versus 0.5% of the conventional-control group (P<.001).
Another critical, but frequently overlooked, factor is the method used to measure glucoselevels. In their 2001 study, Van den Berge et al.[1] measured glucose in arterial blood withan accurate arterial blood gas analyzer. Many of the subsequent studies used capillaryblood and measured glucose with point-of-care meters; however, in numerous publicationsthe sample type and/or method of analysis are not actually specified. Variability in glucoseresults could be a consequence of differences in the types of specimen analyzed, themethod used and/or patient-specific variables (for example, physiology or presence ofinterfering substances).[8] In NICE-SUGAR, glucose measurements were performed onarterial blood "whenever possible" using "point-of-care or arterial blood gas analyzers orlaboratory analyzers in routine use at each center." The different glucose values producedby these diverse methods and samples will lead to different insulin doses and potentiallywide variations in the true glucose concentrations among patients. Glucose meters areconsiderably less precise than blood gas analyzers or central laboratory analyzers. Datafrom 19,597 sites in College of American Pathologists proficiency tests show largevariation.[9] The coefficients of variation among 17 glucose meter types were 12–14%, withbias between two types as high as 41%. Bias of 41% at a glucose concentration of 8.0mmol/l is equivalent to 3.2 mmol/l, which is twice the difference (1.6 mmol/l) in mean bloodglucose concentrations between the intensive and conventional groups observed in NICESUGAR.If a glucose meter has high bias (that is, consistently reports higher values thanthe patient's actual glucose concentration), the patient will receive too much insulin andmight develop hypoglycemia (which might not be identified as many of the patients in theICU are unconscious).