Glycemic Control in the Critically Ill
Critically ill patients admitted to the ICU with severe injury
or infection, such as burns, trauma, or sepsis, commonly
enter into a hypermetabolic state (see also Chapter 39).
This state is associated with enhanced peripheral glucose
uptake and use,47 hyperlactatemia,48 increased glucose
production,49 depressed glycogenesis,50 and insulin resistance.
49 Glucose intolerance develops after uptake of
glucose in skeletal muscle, adipose tissue, and liver, and
the heart becomes saturated,51 and hyperglycemia occurs
because of defective suppression of gluconeogenesis and
a resistance to the peripheral action of insulin. These
mechanisms all work to generate a hyperglycemic state to
satisfy an obligatory requirement for glucose as an energy
substrate. The intensity of the metabolic response peaks
several days after the initial insult and then diminishes as
the patient recovers.48 However, a prolonged hyperglycemic
response may occur in patients who continue to have
tissue hypoperfusion or persistent infection, which then
predisposes them to progressive metabolic derangements
and multisystem organ failure.
Traditionally, hyperglycemia, secondary to sepsis, was
viewed as a beneficial response because it promoted cellular
glucose uptake when cells were energy deprived.
A glucose concentration of 160 to 200 mg/dL was commonly
recommended and believed to maximize cellular
glucose uptake without causing hyperosmolarity.52 However,
neutrophil function is impaired in patients with
hyperglycemia because of decreased bacterial phagocytosis,
53 and many studies report the negative effects of
high blood sugar. Hyperglycemia in diabetic patients is
associated with an increased rate of postoperative infections,
54 and decreased long-term outcomes after myocardial
infarction.55 Hyperglycemia is also associated with
a poorer prognosis after stroke or head injury56 (see also
Chapter 70).
Van den Berghe and coworkers57 hypothesized that
even mild hyperglycemia (i.e., blood glucose levels
between 110 and 200 mg/dL) could be harmful by predisposing
critically ill patients to increased morbidity
and mortality. They performed a prospective, controlled
study involving 1548 patients in the surgical ICU who
were randomized to receive intensive insulin therapy (i.e.,
blood glucose maintained between 80 and 110 mg/dL) or
conventional treatment (i.e., blood glucose maintained
between 180 and 200 mg/dL). In patients who remained
in the ICU for longer than 5 days, intensive insulin therapy
reduced the mortality rate from 20.2% with conventional
therapy to 10% with intensive therapy (P = 0.005).
The group receiving intensive insulin therapy also had a
lower incidence of bloodstream infections (4.2% versus
7.8%, P = 0.003), renal failure requiring dialysis (4.8% versus
8.2%, P = 0.007) and critical illness polyneuropathy
(28.7% versus 51.9%, P = 0.001). Patients in the intensive
insulin group were also less likely to require prolonged
mechanical ventilation and intensive care. The results of
this trial made a persuasive argument for tighter glucose
control, at least in patients in the surgical ICU.
Opponents of the use of strict glycemic control in critically
ill patients argued that the risks of hypoglycemia
should be seriously considered and that the therapeutic
effect of insulin rather than glycemic control leads to the
beneficial outcomes. Insulin has multiple effects, including
the inhibition of tumor necrosis factor alpha (TNF-
α),58 which triggers procoagulant activity and fibrin
deposition and inhibits macrophage inhibitory factor,
thereby contributing to endotoxemia and toxic shock.59
To determine whether it was insulin effect or glycemic
control, van den Berghe and colleagues60 used multivariate
analysis to reanalyze their previous data. It appeared
that decreasing blood glucose levels rather than the
actual amount of insulin given was more closely correlated
with the beneficial reductions in mortality, polyneuropathy,
and bloodstream infections. Instead of the
glucose level, the dose of insulin correlates with the
incidence of renal failure. Investigators thought that
this difference might be the result of the direct effect
of insulin on the kidney or the need for less exogenous
insulin in patients with renal failure because insulin is
cleared through the kidney. Finney and associates61 in
a prospective, observational study provided additional
evidence that glycemic control, rather than insulin
administration, provided the benefit. They examined the
effects of glucose control in 523 patients admitted to a
single surgical ICU. In this trial the primary determinant
of a bad outcome was hyperglycemia rather than hypoinsulinemia,
and a lower mortality rate was associated
with glycemic control rather than a protective effect of
insulin administration. Increased insulin dosing resulted
in an increased mortality rate across all ranges of glycemia.
With regression analysis, their data also suggest
that keeping blood glucose below 145 mg/dL may provide
a survival benefit similar to that achieved with the
tighter range of 80 to 110 mg/dL.
A major criticism of the original van den Berghe study
was that it was performed on relatively homogeneous
surgical populations. The same group then published a
follow-up study examining tight glucose control in 1200
patients in medical ICUs.62 The results showed reduced
morbidity defined as a reduction in newly acquired renal
injury, earlier weaning from mechanical ventilation, and
earlier discharge from the ICU and the hospital but no difference
in mortality. With subgroup analysis, they were
able to determine a mortality benefit from tight glucose
control if the patient was admitted to the ICU for 3 days
or longer (43% versus 52.5%, P = 0.009). From the study
design, it is unclear whether intensive insulin therapy for
less than 3 days causes harm or perhaps the benefit from
intensive insulin therapy requires time to be realized.
Since then, several multicentered randomized controlled
studies have examined the risk-benefit ratio of tight glucose
control.63-65 Two studies (Volume Substitution and Insulin
Therapy in Severe Sepsis [VISEP] and Glucontrol) were
stopped early because of a high rate of hypoglycemia (17%
versus 4.1%, P < 0.001, and 8.7% versus 2.7%, P < 0.001,
respectively). The Normoglycemia in Intensive Care Evaluation
and Surviving Using Glucose Algorithm Regulation
(NICE-SUGAR) trial, the most recent and largest of the studies,
involved 42 ICUs and enrolled 6000 patients. The investigators
reported no difference between the two groups in
terms of days in the ICU, days on mechanical ventilation,
and days requiring renal replacement therapy. Disturbingly,
they found an increased incidence of hypoglycemia
