Insulin infused at 0.05 versus 0.

1 units/kg/hr in children admitted

to intensive care with diabetic ketoacidosis*
Said Al Hanshi, MD, MRCPCH; Frank Shann, MD, FRACP, FJFICM

Objective: To compare the effects of infusing insulin at 0.05
units/kg/hr rather than 0.1 units/kg/hr in children admitted to the
intensive care unit with diabetic ketoacidosis.
Design: A retrospective observational study.
Setting: A tertiary pediatric intensive care unit.
Patients: All children with diabetic ketoacidosis admitted dur-
ing the 6-yr period from 2000 to 2005.
Interventions: None.
Measurements and Main Results: The effective plasma osmo-
lality (plasma glucose concentration in mmol/L ⴙ twice the
plasma sodium concentration in mmol/L), plasma glucose,
plasma sodium, fluid intake, and acid-base status 12 hrs after the
commencement of the insulin infusion. Compared to the 34 chil-
dren who received 0.1 units/kg/hr of insulin, the 33 children who
received 0.05 units/kg/hr of insulin were younger (median age, 25
mos vs. 62 mos, p ⴝ .024) and had a more gradual reduction in
the effective plasma osmolality over the first 12 hrs (p < .0005);
this was because plasma glucose decreased more slowly (p ⴝ
.004) and plasma sodium increased faster (p < .0005). Both
groups had a satisfactory improvement in acidosis and ketosis,
and they had a similar length of stay in the intensive care unit.
Conclusions: Further studies are needed to evaluate the role of
using 0.05 units/kg/hr of insulin to treat children with diabetic
ketoacidosis. The smaller dose of insulin may make it easier to lower the
effective plasma osmolality gradually and might, therefore, reduce the
risk of cerebral edema. (Pediatr Crit Care Med 2011; 12:137–140)
KEY WORDS: brain edema; diabetic ketoacidosis; diabetes mel-
litus; insulin

C erebral edema is the major
cause of morbidity and mor-
tality in children with diabetic
ketoacidosis (1– 4). There is
increasing evidence that a rapid de-
crease in the effective plasma osmolality
may increase the risk of cerebral edema
(1, 5– 8). The effective plasma osmolality
is equal to the plasma glucose concentra-
tion in mmol/L plus twice the plasma
sodium concentration in mmol/L (5, 9),
so the rate of change of the effective os-
molality will be determined by the
amounts of glucose, insulin, sodium, and
water administered.
If the effective osmolality is to fall
slowly during the treatment of diabetic
ketoacidosis, the plasma sodium concen-
tration has to rise by nearly 1 mmol/L for
every 2 mmol/L fall in the plasma glucose
*See also p. 217.
From the Pediatric Intensive Care Unit (SAH), The
Royal Hospital, Sultanate of Oman; and Intensive Care
(FS), Royal Children’s Hospital, Melbourne, Victoria,
Australia.
The authors have not disclosed any potential con-
flicts of interest.
For information regarding this article, E-mail:
frank.shann@rch.org.au
Copyright © 2011 by the Society of Critical Care
Medicine and the World Federation of Pediatric Inten-
sive and Critical Care Societies
DOI: 10.1097/PCC.0b013e3181e2a21b
concentration (1). Several factors may
prevent an increase in the plasma sodium
concentration: the use of intravenous flu-
ids with a sodium concentration lower
than the patient’s plasma sodium concen-
tration; the infusion of enough fluid to
cause expansion of the extracellular fluid
volume, which may lead to enhanced re-
nal excretion of sodium and “desalina-
tion”; absorption from the bowel of large
volumes of hypotonic fluid ingested be-
fore admission to hospital; and a high
dose of insulin (1).
Much of the discussion about how to
preserve the effective plasma osmolality
in diabetic ketoacidosis has been about
sodium and water therapy, rather than
the changes in plasma glucose (1, 5– 8).
This is understandable, because a 1
mmol/L change in plasma sodium con-
centration has double the effect on osmo-
lality of a 1 mmol/L change in plasma
glucose concentration. However, if the
plasma glucose concentration falls very
rapidly, it is difficult to increase the
plasma sodium concentration quickly
enough to maintain the effective plasma
osmolality. It is, therefore, important to
avoid a very rapid fall in the plasma glu-
cose concentration (2).
The International Society for Pediatric
and Adolescent Diabetes 2006 –2007 Clin-
ical Practice Consensus Guidelines on the
management of diabetic ketoacidosis rec-
ommend the infusion of 0.1 units/kg/hr
of insulin from 1 to 2 hrs after the start of
fluid replacement therapy (10). However,
only two papers are cited in support of
this dose of insulin: the first paper stud-
ied only insulin doses of 0.1 units/kg/hr
and 1.0 units/kg/hr in 32 children with
diabetic ketoacidosis (11); and the second
gave 0.01 units/kg/hr, 0.1 units/kg/hr,
and 1.0 units/kg/hr for only 60 mins to
six insulin-dependent adult volunteers
who had been given dexamethasone (12).
Schade and Eaton (12) in the second pa-
per stated explicitly that their results “are
not directly applicable to insulin therapy
for spontaneous diabetic ketoacidosis.”
Neither of these studies provides evidence
that 0.1 units/kg/hr is superior to lower
doses of insulin for the treatment of dia-
betic ketoacidosis in children. In the sec-
ond study (12), an insulin dose of only
0.01 units/kg/hr reduced the total plasma
ketone body concentration by 42% and the
plasma glucose concentration by 8.4% (a
rate of 1.6 mmol/L/hr) in just 1 hr.
For several years, many of the children
admitted to our intensive care unit (ICU)
with diabetic ketoacidosis have been
treated with an infusion of 0.05 units/
kg/hr of insulin, in the hope that this will
make it easier to achieve a gradual reduc-
tion in the effective plasma osmolality.

Pediatr Crit Care Med 2011 Vol. 12, No. 2 137

Approximately half the admissions have Table 1. Biochemistry at the start of treatment and 12 hrs after the commencement of insulin therapy,
been treated with an initial infusion of by dose of insulin: median (interquartile range)
0.05 units/kg/hr of insulin, and half with
Dose of Insulin
0.1 u/kg/hr. This paper describes the ef-
fects of these two regimens in the first 12 0.05 units/kg/hr (n ⫽ 33) 0.1 units/kg/hr (n ⫽ 34)
hrs of treatment.
Effective Osmolality (mOsm/kg)
Start 300 (285,311) 311 (298,327)
12 hrs 295 (284,311) 293 (287,332)
METHODS Glucose (mmol/L)
Start 29 (23,39) 32 (26,68)
A computerized database is used to record 12 hrs 11 (9,17) 14 (8,18)
all admissions to the pediatric ICU at the Royal Sodium (mmol/L)
Start 135 (129,139) 137 (134,140)
Children’s Hospital, Melbourne. In the 6 yrs
12 hrs 140 (138,149) 141 (138,157)
between January 2000 and December 2005, 69 Potassium (mmol/L)
children were admitted with a diagnosis of Start 4.3 (4.0,5.2) 4.0 (3.8,4.7)
diabetic ketoacidosis (plasma glucose of ⬎11 12 hrs 3.6 (3.4,4.0) 3.9 (3.4,4.5)
mmol/L, and an arterial pH of ⬍7.30, or pH
Start 7.03 (6.95,7.16) 7.07 (6.95,7.15)
plasma bicarbonate of ⬍15 mmol/L). The 12 hrs 7.25 (7.21,7.31) 7.28 (7.24,7.32)
medical records of two children were not Base Excess (mmol/L)
available, so 67 children are included in this Start ⫺24 (⫺26,⫺22) ⫺23 (⫺26,⫺19)
review. Children were managed according to 12 hrs ⫺12 (⫺16,⫺11) ⫺12 (⫺14,⫺9)
the section on diabetic ketoacidosis in the Bicarbonate (mmol/L)
Start 5 (4,7) 5 (4,7)
Unit’s book of Pediatric Intensive Care Guide- 12 hrs 13 (10,15) 14 (11,17)
lines (13). After fluid resuscitation using 0.9% PaCO2 (torr)
saline, insulin was given by continuous infu- Start 20 (15,24) 20 (16,25)
sion in 4% albumin with no loading dose, and 12 hrs 27 (22,31) 33 (27,36)
Urine ketones (0 to ⫹⫹⫹)
0.9% saline was used as maintenance fluid
Start 2.9 pluses* 2.8 plusesa
without bicarbonate or citrate. 12 hrs 2.0 pluses* 2.0 plusesa
Each child’s age and weight, and the dose
of insulin and the volume of fluid infused from To convert torr to kPa, multiply by 0.1317.
a
the start of treatment until the completion of Arithmetic mean.
12 hrs of insulin therapy were obtained from
the medical record. The following biochemical
results were obtained from the medical record RESULTS kg (IQR, 12– 45 kg), and stayed in the
and the Hospital’s computerized database: the intensive care unit for a median of 0.92
urine ketone test results; the plasma pH, PCO2 Of the 67 children, 33 were given 0.05 days (IQR, 0.43–1.45 days). The children
and base excess; and the plasma concentration of units/kg/hr of insulin, and 34 were given in the 0.05-units/kg/hr group were
glucose, bicarbonate, sodium, and potassium— 0.1 units/kg/hr of insulin. Four children younger (p ⫽ .024, Student’s t test of the
which were usually measured every 2– 4 hrs up were started on 0.1 units/kg/hr but had logarithm of age).
to the completion of 12 hrs of insulin therapy. the dose reduced to 0.05 units/kg/hr after The biochemical findings at the start
The effective plasma osmolality in mOsm/kg was 6 –12 hrs; they were classified as having of treatment and at the completion of 12
calculated from the plasma glucose concentra- received 0.1 units/kg/hr. Three children hrs of insulin therapy are shown in Table
tion in mmol/L plus twice the plasma sodium were given doses of 0.075– 0.1 units/kg/ 1. The volume of fluid administered and
concentration in mmol/L (5, 9). hr, 0.08 units/kg/hr, and 0.09 units/kg/hr, the change in the plasma glucose, so-
The effect of the dose of insulin on the respectively; they were also classified as dium, and effective osmolality from the
effective osmolality (or glucose or sodium) at having received 0.1 units/kg/hr. One start of treatment until the completion of
12 hrs (the dependent variable) was assessed, child had the dose reduced from 0.05 to 12 hrs of insulin therapy are shown in
using analysis of covariance, with the indepen- 0.03 units/kg/hr, and another had the Table 2. Children who received 0.05
dent variables being the dose of insulin, the dose increased from 0.05 units/kg/hr to units/kg/hr of insulin had a more gradual
initial value of effective osmolality (or glucose 0.1 units/kg/hr after 9 hrs; both were reduction in effective plasma osmolality
or sodium), and the logarithm of age in classified as having received 0.05 units/ in the first 12 hrs (Table 2 and Fig. 1).
months. The percentage of normal daily fluid kg/hr. Cerebral edema was not detected clini-
intake was calculated from the actual total The 33 children (16 boys) given 0.05 cally in any child, and no child died.
fluid intake up to 12 hrs, divided by the nor- units/kg/hr of insulin had a median age of
mal daily intake for a child of that weight 25 mos (interquartile range [IQR], 14 – 87 DISCUSSION
(0 –10 kg 100 mL/kg, 11–20 kg 1000 mL ⫹ 50 mos, 20 aged ⬍36 mos), a median weight
mL/kg, ⬎20 kg 1500 mL ⫹ 20 mL/kg) (14). of 12 kg (IQR, 9 –25 kg), and stayed in the This study is a retrospective survey of
Stata (StataCorp, College Station, TX) was ICU for a median of 0.91 days (IQR, 0.74 – children admitted to the ICU with dia-
used for statistical analysis. 1.14 days). The 34 children (n ⫽ 15 boys) betic ketoacidosis. By chance, approxi-
The study was approved by the Royal Chil- given 0.1 units/kg/hr of insulin had a mately half the children were treated
dren’s Hospital Ethics and Human Research median age of 62 mos (IQR, 20 –97 mos; with 0.05 units/kg/hr of insulin and half
Committee. 13 aged ⬍36 mos), a median weight of 20 received 0.1 units/kg/hr. The dose was

138 Pediatr Crit Care Med 2011 Vol. 12, No. 2

Table 2. Volume of fluid administered and the change in plasma sodium,glucose and effective
osmolality from the start of treatment until the completion of 12 hrs of insulin therapy, by dose of
insulin: median (interquartile range)
Insulin 0.05 Insulin 0.1 Adjusted
units/kg/hr units/kg/hr p R2
Fluid before intensive care (mL/kg) 20 (19,29) 20 (10,31) .612a
Total fluid intake to 12 hr (mL/kg) 75 (68,91) 69 (57,86) .075a
Percentage of normal daily fluid 90 (79,117) 101 (82,131) .273a
Change in sodium 0–12 hrs (mmol/L) 8 (5,11) 5 (1,17) ⬍.0005b .63b
Change in glucose 0–12 hrs (mmol/L) ⫺17 (⫺26,⫺12) ⫺21 (⫺52,⫺15) .004b .38b
Change in osmolality 0–12 hrs (mOsm/kg) ⫺4 (⫺12,⫹5) ⫺15 (⫺24,⫺6) ⬍.0005b .80b
a
Kruskal-Wallis test; bEffect of insulin dose on the value at 12 hours adjusted for the initial value
and age using analysis of covariance (see Methods).
Figure 1. Change in the effective plasma osmolality in the first 12 hrs of insulin therapy, by age and
dose of insulin in units/kg/hr. Boxes show the interquartile range; whiskers indicate 1.5 times beyond
the first or third quartile.
chosen by the doctors who happened to insulin receiving more fluid during re-
be looking after the child, and allocation suscitation and the first 12 hrs in the
was not formally randomized; intensive pediatric ICU when this was expressed in
care specialists usually prescribed 0.05 mL per kg, but the trend was reversed
units/kg/hr of insulin and endocrinolo- when the amount of fluid was expressed
gists 0.1 units/kg/hr. The children who as a percentage of the normal daily fluid
received the lower dose were younger intake for the age of the child (Table 2);
(median age, 25 mos vs. 62 mos, p ⫽ neither of these differences was statisti-
.024), so the analysis was adjusted for cally significant.
age. However, there may well be other There was a smaller reduction in ef-
undetected differences, so only very ten-
fective plasma osmolality among the chil-
tative comparisons can be made between
dren who received only 0.05 units/kg/hr
the outcomes in the two groups. Our ICU
of insulin (p ⬍ .0005) (Table 2), partly
(13) has published a book of treatment
because of a smaller reduction in plasma
guidelines that includes a protocol for the
management of diabetic ketoacidosis; pa- glucose (p ⫽ .004) and partly because of
tients in the unit are usually treated in a larger increase in plasma sodium (p ⬍
accordance with the guidelines, so the .0005). Because the comparison was not
management of these children is likely to randomized, we cannot be sure that the
have been fairly consistent. smaller reduction in effective osmolality
There was a trend toward children was due to the lower dose of insulin.
who were treated with 0.05 units/kg/hr of However, the study does show that chil-
Pediatr Crit Care Med 2011 Vol. 12, No. 2
dren who received the lower dose of in-
sulin did not have persistent acidosis or
persistent ketosis (Table 1).
It is very uncommon for a child with
diabetic ketoacidosis to present with cir-
culatory shock caused by hypovolemia
(14), but when this does occur, intravas-
cular volume needs to be restored quickly
with 0.9% saline. However, it is rarely
necessary to infuse ⬎7.5–10 mL/kg of
0.9% saline in the first hour of treatment
(15). Most children presenting with dia-
betic ketoacidosis have adequate tissue
perfusion, and cerebral edema is by far
the greatest risk—and it is, therefore,
prudent to correct the hyperglycemia, ac-
idosis, and dehydration slowly over
48 –72 hrs (1, 5– 8).
In addition to osmotic effects, vaso-
genic, metabolic, and inflammatory fac-
tors may be involved in the pathogenesis
of cerebral edema in children with dia-
betic ketoacidosis (1, 3, 16 –18). However,
the evidence (1, 5– 8) suggests that it is
prudent to avoid a rapid fall in the effec-
tive plasma osmolality by ensuring that
the plasma glucose concentration does
not fall too rapidly, and that the reduc-
tion that does occur is offset by a rise in
the plasma sodium. It is interesting that
this recommendation remains controver-
sial (5, 19), when there is general agree-
ment that, when cerebral edema does oc-
cur, it should be treated with measures
that result in a higher effective osmolal-
ity: the administration of mannitol or hy-
pertonic saline, and a reduction in the
rate of administration of fluid and insulin
(5, 10, 15).
Two large case-control studies (2, 20)
of children with diabetic ketoacidosis
concluded that osmolality had no effect
on cerebral edema, but they did not study
the effective osmolality because urea was
included in the calculation. Urea passes
quickly between extracellular and intra-
cellular fluid, and so does not influence
water shifts in diabetic ketoacidosis (5, 9)
The U.K. case-control study (2) found
that children given insulin in the first
hour had an increased risk of cerebral
edema; these children are likely to have
received a higher total dose of insulin in
the first few hours of treatment.
Although the standard dose of 0.1
units/kg/hr of insulin is widely referred to
as “low dose,” it actually achieves high
physiological or low pharmacological
concentrations of insulin (100 –200 ␮U/
mL) with complete inhibition of lipolysis
and ketogenesis, complete suppression of
139
hepatic glucose production, and near-
maximal stimulation of tissue glucose
uptake - except in an occasional patient
with severe insulin resistance that per-
sists despite the correction of hypovolae-
mia (11, 21).
A recent observational study sug-
gested that 0.05 units/kg/hr of insulin
was as effective as 0.1 units/kg/hr in the
treatment of 93 episodes of diabetic keto-
acidosis in children (22), and an earlier
study found that a median dose of 0.045
(range 0.02– 0.1) units/kg/hr of insulin
normalized blood hydroxybutyrate levels
in 35 episodes of diabetic ketoacidosis in
children (23). These papers, and the re-
sults of our study, suggest that it may be
safe to treat most children who have di-
abetic ketoacidosis with an infusion of
0.05 units/kg/hr of insulin, and that even
lower doses may be optimal in some cir-
cumstances. The use of doses of insu-
lin ⬍0.1 units/kg/hr may make it easier
to achieve a very gradual reduction in the
effective plasma osmolality in children
with diabetic ketoacidosis, and this might
reduce the risk of cerebral oedema. Fur-
ther studies are needed to evaluate this
possibility.
REFERENCES
1. Hoorn EJ, Carlotti APCP, Costa LAA, et al:
Preventing a drop in effective plasma osmo-
lality to minimize the likelihood of cerebral
edema during treatment of children with di-
abetic ketoacidosis. J Pediatr 2007; 150:
467– 473
2. Edge JA, Jakes RW, Roy Y, et al: The UK
case-control study of cerebral oedema com-
plicating diabetic ketoacidosis in children.
Diabetologia 2006; 49:2002–2009
3. Glaser NS, Wootton-Gorges SL, Marcin JP, et
140
al: Mechanism of cerebral edema in children
with diabetic ketoacidosis. J Pediatr 2004;
145:164 –171
4. Lawrence SE, Cummings EA, Gaboury I, et
al: Population-based study of incidence and
risk factors for cerebral edema in pediatric
diabetic ketoacidosis. J Pediatr 2005; 146:
688 – 692
5. Rosenbloom AL: Comment on: Treatment of
diabetic ketoacidosis and the risk of cerebral
edema. J Pediatr 2007; 150:467– 473. J Pedi-
atr 2008; 152:146 –147; author reply,
147–149
6. Carlotti APCP, Bohn D, Halperin ML: Impor-
tance of timing of risk factors for cerebral
oedema during therapy for diabetic ketoaci-
dosis. Arch Dis Child 2003; 88:170 –173
7. Harris GD, Fiordalisi I, Yu C: Maintaining
normal intracranial pressure in a rabbit
model during treatment of severe diabetic
ketoacidemia. Life Sci 1996; 59:1695–1702
8. Harris GD, Fiordalisi I, Harris WL, et al:
Minimizing the risk of brain herniation dur-
ing treatment of diabetic ketoacidemia: A
retrospective and prospective study. J Pediatr
1990; 117:22–31
9. Oster JR, Singer I: Hyponatremia, hyposmo-
lality, and hypotonicity: tables and fables.
Arch Intern Med 1999; 159:333–336
10. Wolfsdorf J, Craig ME, Daneman D, et al:
Diabetic ketoacidosis. Pediatr Diabetes 2007;
8:28 – 43
11. Kitabchi AE: Low-dose insulin therapy in di-
abetic ketoacidosis: Fact or fiction? Diabetes
Metab Rev 1989; 5:337–363
12. Schade DS, Eaton RP: Dose response to in-
sulin in man: differential effects on glucose
and ketone body regulation. J Clin Endocri-
nol Metab 1977; 44:1038 –1053
13. Shann F, Henning R, Shekerdemian L, et al:
Paediatric Intensive Care Guidelines. Third
Edition. Melbourne, Collective; 2008. Avail-
able at www.drugdoses.net/iPhone/. Accessed
May 8, 2010
14. Holliday MA, Segar WE: The maintenance
need for water in parenteral fluid therapy.
Pediatrics 1957; 19:823– 832
15. Bohn D, Daneman D: Diabetic ketoacidosis
and cerebral edema. Curr Opin Pediatr 2002;
14:287–291
16. Cameron FJ, Kean MJ, Wellard RM, et al:
Insights into the acute cerebral metabolic
changes associated with childhood diabetes.
Diabet Med 2005; 22:648 – 653
17. Toledo JD, Modesto V, Peinador M, et al:
Sodium concentration in rehydration fluids
for children with ketoacidotic diabetes: Ef-
fect on serum sodium concentration. J Pedi-
atr 2009; 154:895–900
18. Hoffman WH, Stamatovic SM, Andjelkovic
AV: Inflammatory mediators and blood brain
barrier disruption in fatal brain edema of
diabetic ketoacidosis. Brain Res 2009; 1254:
138 –148
19. Rosenbloom AL: Treatment of diabetic keto-
acidosis and the risk of cerebral edema. J Pe-
diatr 2008; 152:146 –147; author reply,
147–149
20. Glaser N, Barnett P, McCaslin I, et al: Risk
factors for cerebral edema in children with
diabetic ketoacidosis. The Pediatric Emer-
gency Medicine Collaborative Research Com-
mittee of the American Academy of Pediat-
rics. N Engl J Med 2001; 344:264 –269
21. DeFronzo R, Matsuda M, Barrett E: Diabetic
ketoacidosis: A combined metabolic-nephro-
logic approach to therapy. Diabetes Rev
1994; 2:209 –238
22. Puttha R, Cooke D, Subbarayan A, et al: Low
dose (0.05 units/kg/h) is comparable with
standard dose (0.1 units/kg/h) intravenous
insulin infusion for the initial treatment of
diabetic ketoacidosis in children with type 1
diabetes—an observational study. Pediatr Di-
abetes 2010; 11:12–17
23. Noyes KJ, Crofton P, Bath LE, et al: Hydroxy-
butyrate near-patient testing to evaluate a
new end-point for intravenous insulin ther-
apy in the treatment of diabetic ketoacidosis
in children. Pediatr Diabetes 2007;
8:150 –156
Pediatr Crit Care Med 2011 Vol. 12, No. 2