The Journal of Emergency Medicine, Vol. -, No. -, pp.

1–7, 2018
Ó 2018 Elsevier Inc. All rights reserved.
0736-4679/$ - see front matter

https://doi.org/10.1016/j.jemermed.2018.01.004

Original
Contributions

THE TWO-BAG METHOD FOR TREATMENT OF DIABETIC KETOACIDOSIS

IN ADULTS

Nathan L. Haas, MD,* Roma Y. Gianchandani, MBBS,†‡ Kyle J. Gunnerson, MD,*†§k Benjamin S. Bassin, MD,*§
Arun Ganti, MD,* Christopher Hapner, DO,* Caryn Boyd, BS,* James A. Cranford, PHD,{ and
Sage P. Whitmore, MD*§#
*Department of Emergency Medicine, Michigan Medicine, Ann Arbor, Michigan, †Department of Internal Medicine, Michigan Medicine, Ann
Arbor, Michigan, ‡Division of Metabolism, Endocrinology and Diabetes, Michigan Medicine, Ann Arbor, Michigan, §Division of Emergency
Critical Care, Michigan Medicine, Ann Arbor, Michigan, kDepartment of Anesthesiology/Critical Care, Michigan Medicine, Ann Arbor,
Michigan, {Department of Psychiatry, Michigan Medicine, Ann Arbor, Michigan, and #Department of Internal Medicine, Division of Pulmonary
and Critical Care Medicine, Michigan Medicine, Ann Arbor, Michigan
Reprint Address: Nathan L. Haas, MD, Department of Emergency Medicine, University of Michigan, 1500 East Medical Center Drive, B1-380
Taubman Center, SPC 5305, Ann Arbor, MI 48109

, Abstract—Background: The ‘‘two-bag method’’ of man-
agement of diabetic ketoacidosis (DKA) allows for titration
of dextrose delivery by adjusting the infusions of two i.v.
fluid bags of varying dextrose concentrations while keeping
fluid, electrolyte, and insulin infusion rates constant. Objec-
tive: We aimed to evaluate the feasibility and potential ben-
efits of this strategy in adult emergency department (ED)
patients with DKA. Methods: This is a before-and-after
comparison of a protocol using the two-bag method opera-
tionalized in our adult ED in 2015. A retrospective electronic
medical record search identified adult ED patients present-
ing with DKA from January 1, 2013 to June 30, 2016. Clin-
ical and laboratory data, timing of medical therapies, and
safety outcomes were collected and analyzed. Results:
Sixty-eight patients managed with the two-bag method
(2B) and 107 patients managed with the one-bag method
(1B) were identified. The 2B and 1B groups were similar
in demographics and baseline metabolic derangements,
though significantly more patients in the 2B group received
care in a hybrid ED and intensive care unit setting (94.1%
vs. 51.4%; p < 0.01). 2B patients experienced a shorter inter-
val to first serum bicarbonate $ 18 mEq/L (13.4 vs. 20.0 h;
p < 0.05), shorter duration of insulin infusion (14.1 vs. 21.8 h;
p < 0.05), and fewer fluid bags were charged to the patient
(5.2 vs. 29.7; p < 0.01). Frequency of any measured hypogly-
cemia or hypokalemia trended in favor of the 2B group
(2.9% vs. 10.3%; p = 0.07; 16.2% vs. 27.1%; p = 0.09;
respectively), though did not reach significance. Conclu-
sions: The 2B method appears feasible for management of
adult ED patients with DKA, and use was associated with
earlier correction of acidosis, earlier discontinuation of insu-
lin infusion, and fewer i.v. fluid bags charged than tradi-
tional 1B methods, while no safety concerns were
observed. Ó 2018 Elsevier Inc. All rights reserved.
, Keywords—diabetic ketoacidosis; DKA; endocrine; crit-
ical care
INTRODUCTION
Diabetic ketoacidosis (DKA) is a common and severe
complication of diabetes mellitus, with the incidence
increasing in the United States over the past 2 decades
from 80,000 reported cases in 1988 to 140,000 reported
cases in 2009 (1). Most cases of DKA occur in adult pa-
tients, with 80% of cases occurring between the ages of
18 and 65 years (2). Management of DKA requires close
laboratory monitoring with frequent titration of fluid,
electrolyte, and dextrose therapies. Outcomes do not

RECEIVED: 23 August 2017; FINAL SUBMISSION RECEIVED: 15 December 2017;

ACCEPTED: 6 January 2018

1
2 N. L. Haas et al.
appear to be affected by the managing physician’s spe-
cialty (i.e., family physician vs. internist vs. endocrinolo-
gist), as long as standard written therapeutic guidelines
are followed; therefore, standardized or protocolized
management is appealing in the care of patients with
DKA (3,4).
Insulin is required to clear hyperglycemia and acidosis
and to suppress ketone formation. As hyperglycemia
clears faster than acidosis, insulin is continued past eugly-
cemia, and dextrose is added to prevent hypoglycemia. It
is recommended that DKA patients be given dextrose-
containing fluids as their blood glucose (BG) falls into
high-normal range (2). The two-bag (2B) method of fluid
management maintains constant fluid and insulin infusion
rates, while the concentration of dextrose administered is
titrated by adjusting the infusion rates of two bags of i.v.
fluids, one with dextrose and one without, in response to
changing BG measurements. This variation of the eugly-
cemic clamp technique allows for rapidly titratable and
customizable dextrose infusion rates, while keeping fluid,
insulin, and electrolyte delivery constant (5).
The 2B method has been studied in pediatric popula-
tions, and has been weakly associated with faster
response time in i.v. fluid therapy changes; earlier correc-
tion of serum bicarbonate, ketones, and pH; and more
cost-effective care in small studies (6–8). However, this
method has not been studied previously in adult
patients. We aimed to evaluate the feasibility of the 2B
method in the treatment of adult emergency department
(ED) patients with DKA, and to assess for potential
association with improvement in metabolic and
resource utilization outcomes. We hypothesized that use
of the 2B method would be associated with fewer
episodes of hypoglycemia and shorter duration of
insulin infusion compared to the traditional one-bag
(1B) method with a titratable insulin infusion.
MATERIALS AND METHODS
Study Design and Setting
A protocolized 2B method for management of adult ED
patients with DKA was developed, with the working hy-
pothesis that implementation would be associated with
earlier correction of metabolic acidosis and resource
preservation, and implemented in August 2015. This
was a before-and-after comparison study of the initiation
of this protocol using the 2B method with the hypotheses
mentioned. It was reviewed and approved by the Institu-
tional Review Board at the University of Michigan, and
was granted exception from informed consent. All sub-
jects were treated in the adult ED at a large tertiary
referral center with > 100,000 annual ED visits from
January 1, 2013 to June 30, 2016.
Selection of Participants
Patients were included if they were at least 18 years of age
with an ED clinical impression or admission diagnosis of
DKA, as identified by an electronic medical record
(EMR) search, and either the 1B or 2B order set was
used. Patients were excluded if they did not meet diag-
nostic criteria on manual chart review, which included
hyperglycemia plus initial pH < 7.30, initial bicarbonate
(HCO3) < 18 mEq/L, initial anion gap > 14, if neither or-
der set was used, or if both order sets were used. Before
implementation of the 2B method, 120 patients were
excluded due to no protocolized care plan being used. Af-
ter implementation of the 2B method, 25 patients were
excluded, as both order sets were utilized at some point.
All data were obtained from the EMR. A computerized
search algorithm was used to identify data points of inter-
est. As the search criteria were developed and refined, the
authors reviewed charts from 10-patient sample groups to
ensure accuracy and inter-abstractor reliability in an iter-
ative fashion. Data unable to be obtained by computer-
ized search were abstracted manually.
Interventions (1B and 2B Methodology)
Patients managed with the 1B method were treated with a
hospital-wide DKA protocol, which included a contin-
uous insulin infusion with titration instructions to adjust
the insulin rate depending on the magnitude and rate of
change of BG on hourly checks, and a written prompt
to providers to initiate 5% dextrose-containing i.v. fluids
when glucose dropped below 250 mg/dL (Figure 1). Lab-
oratory draws and electrolyte replacement were at the
discretion of the physician. Patients managed with the
2B method were treated with an ED-specific DKA proto-
col, which included a non-titratable continuous insulin
infusion at 0.1 U/kg/h, combined with a constant fluid
infusion rate of two bags of 0.45% saline, one with
10% dextrose and the other with no dextrose, the ratio
of which was adjusted based on hourly BG checks to
maintain a total fluid rate of 250 mL/h (Figure 2). Labo-
ratory draws were performed every 2 h, alternating be-
tween venous blood gas and basic chemistries.
Electrolyte replacement was guided by a nurse-initiated
electrolyte protocol identical to that used in the inpatient
intensive care units (ICUs). In both groups, long-acting
insulin was routinely given when the patient met the
following parameters: BG < 200 mg/dL, anion gap <
14, HCO3 $ 15 mEq/L, pH > 7.3, and tolerating a diet
by mouth. The insulin infusion was continued for an addi-
tional 2 h once these parameters were met before transi-
tioning to subcutaneous insulin therapy. Patients in both
cohorts with unique circumstances (i.e., renal failure,
fluid overload) were excluded from protocolized
Two-Bag Method of Management of DKA
Figure 1. Insulin and fluid protocol in the control (one-bag
method) group. All blood glucose (BG) units are mg/dL.
management at the discretion of the treating physician.
As neither order set (1B or 2B) was utilized, these pa-
tients were not included in analysis, as management
was not standardized and was instead dictated solely by
physician discretion.
Measurements/Outcomes
End points were related to safety, namely the incidence of
hypoglycemia, defined as any BG < 70 mg/dL on a meta-
bolic panel, venous blood gas, or bedside glucometer, and
hypokalemia, defined as any serum potassium < 3.3 mEq/
L on a metabolic panel or venous blood gas. Additional
safety end points included the lowest recorded BG and
highest and lowest recorded serum potassium from any
metabolic panel or venous blood gas. The time frame
for these measurements was the duration of insulin infu-
sion. If the stop time of insulin infusion was not recorded,
Regular
Insulin
0.45% NS D10-0.45% NS
+ +
20mEq KCl 20mEq KCl
BG >250: 250mL/hr 0 mL/hr
BG 150-250: 125mL/hr 125mL/hr
0.1 u/kg/hr BG <150: 0 ml/hr 250mL/hr
Figure 2. Insulin and fluid protocol in the two-bag method
group. All blood glucose (BG) units are mg/dL. KCl = potas-
sium chloride; NS = normal saline.
3
the time of first administration of long-acting subcutane-
ous insulin was used, or 24 h from ED arrival if neither
time point was documented. Additional end points
included patient disposition, time to correction of
acidosis (first pH $ 7.3, first HCO3 $ 18 mEq/L), dura-
tion of insulin infusion, hospital charges, and the number
of 500-mL and 1000-mL i.v. fluid bags (excluding
smaller medication piggy-backs) charged to the patient.
The number of fluid bags charged to the patient, rather
than total volume of fluid administered, was selected as
an outcome measure to compare episodes of spiking
and wasting partial bags of fluid based on fluctuating
serum glucose values.
Analysis
Between-group differences in proportions were tested
with bivariate c2 analysis, and hypotheses about be-
tween- and within-group differences in means were tested
with independent and paired-samples t-tests, respec-
tively. An a level of 0.05 was used for all analyses. All
statistical analyses were conducted using SAS software,
version 9.4 (SAS Institute, Cary, NC, 2013).
RESULTS
Characteristics of Study Subjects
The EMR search generated 468 initial patient encounters
(Figure 3). Subsequent manual chart review excluded 148
encounters from analysis due to lack of initial diagnostic
criteria. Of the remaining 320 patients, 145 were
Patients with ED Clinical Impression
or Admission Diagnosis of DKA via
EMR search (n=468))
Excluded: initial pH ≥ 7.30,
HCO3 ≥ 18mEq/L, or
H
anion gap ≤ 14 (n=148)
Patients analyzed (n=320)
(
(n=3
E
Excluded: both or neither
protocols utilized (n=145)
p
Patients included in final
fi analysis:
• One-bag: n= 107
• Two-bag: n= 68
Figure 3. CONSORT (Consolidated Standards of Reporting
Trials) flow diagram demonstrating inclusion and exclusion
criteria. DKA = diabetic ketoacidosis; ED = emergency
department; EMR = electronic medical record.
4 N. L. Haas et al.

Table 1. Patient Demographics and Baseline to 1B (5.2 vs. 29.7; p < 0.001), while overall hospital
Characteristics
charges were similar between groups.
One-Bag, Two-Bag, The percentage of patients that experienced an episode
Mean (95% CI) Mean (95% CI) p of hypoglycemia (2.9% vs. 10.3%; p = 0.07) or hypoka-
Characteristic (n = 107) (n = 68) Value lemia (16.2% vs. 27.1%; p = 0.09) was not significantly
Age, y 34.8 (31.5–38.1) 32.4 (29.0–35.7) 0.33 different between the 2B and 1B groups, respectively
Male, % 45.8 (36.7–55.2) 41.2 (30.3–53.0) 0.55 (Table 3). The lowest BG, highest potassium, and lowest
HR, triage, 114 (110–118) 116 (111–122) 0.53 potassium values were similar between groups. Signifi-
beats/min
SBP, triage, 133 (127–138) 127 (121–132) 0.14 cantly fewer patients in the 2B group were admitted to
mm Hg the hospital ward (55.9% vs. 69.2%; p < 0.01) or to the
Lactate, mmol/L 3.3 (2.9–3.8) 2.7 (2.4–3.1) 0.08 ICU (0% vs. 15.9%; p < 0.001). More patients in the
BG, mg/dL 546 (502–590) 539 (478–601) 0.89
pH (venous) 7.11 (7.09–7.13) 7.12 (7.10–7.15) 0.49 2B cohort were discharged from the ED (44.1% vs.
HCO3, mEq/L 9.6 (8.9–10.4) 10.3 (9.3–11.2) 0.31 15.0%; p < 0.001), and among patients discharged from
Anion gap 27.6 (26.4–28.8) 26.9 (25.3–28.5) 0.51 the ED, there was no increase in ED return visits within
Care transferred 51.4 (42.0–60.6) 94.1 (85.3–98.0) <0.01
to EC3, % 72 h (7.4% vs. 3.7%; p = 0.29).

BG = blood glucose; CI = confidence interval; EC3 = emergency
critical care center; HR = heart rate; SBP = systolic blood pres-
sure.
excluded due to treatment with multiple insulin protocols
or no identifiable insulin protocol. Ultimately 175 DKA
patients were included for analysis, 107 managed with
the 1B method and 68 managed with the 2B method.
The groups were similar in demographic and baseline
data, including age, sex, vital signs, and initial laboratory
data (Table 1).
Main Results
The duration of metabolic acidosis was shorter for 2B
compared to 1B (time to first HCO3 $18 mEq/L 13.4
vs. 20.0 h; p < 0.05) (Table 2). Using the Kaplan-Meier
method, curves for the 1B and 2B groups were signifi-
cantly different (c2[1] = 14.5; p < 0.05) (Figure 4). The
2B patients spent significantly less time receiving an in-
sulin infusion (14.1 vs. 21.8 h; p < 0.05) (Figure 5).
The interval to first administration of subcutaneous
long-acting insulin was shorter in the 2B cohort (14.7 h
vs. 21.6 h; p < 0.05). The number of 5001000-mL i.v.
fluid bags charged to patients was lower in 2B compared
Emergency Critical Care Center
The Emergency Critical Care Center (EC3), a hybrid ED-
ICU setting, opened in February 2015, and the 2B proto-
col was operationalized in August 2015. Significantly
more patients in the 2B group received care at some point
in EC3 (64 of 68 [94.1%] vs. 52 of 107 [51.4%]; p < 0.01).
Of patients managed in EC3 (2B, n = 64 and 1B, n = 52),
baseline demographics did not differ aside from initial
lactate (2B, 2.7 mmol/L vs. 1B, 3.5 mmol/L; p = 0.02).
For 2B and 1B cohorts, respectively, time to first HCO3
$18 mEq/L was 13.2 h vs. 17.3 h (p = 0.15), time on in-
sulin infusion was 14.3 h vs. 16.5 h (p = 0.34), and total
i.v. fluid bags charged was 5.4 vs. 27.3 (p < 0.001). There
were no observed differences in safety outcomes (hypo-
glycemia, hypokalemia, hyperkalemia) between groups.
More patients in the 2B cohort were discharged from
the ED (45.3% vs. 26.9%; p = 0.04) and fewer were
admitted to the ICU (0% vs. 11.5%; p = 0.01).
DISCUSSION
This study demonstrates the feasibility of the 2B method
and potential benefits over the traditional 1B method in
treating adult ED patients with DKA. Our study

Table 2. Duration of Treatment and Resource Utilization

One-Bag, Two-Bag,
Mean (95% CI) Mean (95% CI)
Resource Utilization (n = 107) (n = 68) p Value

Time to pH $ 7.3, h 10.8 (9.7–11.9) 12.4 (11.3–13.6) 0.12

Time to HCO3 $ 18 mEq/L, h 20.0 (16.1–24.0) 13.4 (11.3–15.6) 0.03
Time on insulin infusion, h 21.8 (16.8–26.7) 14.1 (11.5–16.7) 0.03
Time to long-acting insulin 21.6 (17.2–26.0) 14.7 (13.5–15.8) 0.02
administration, h
Total hospital charges, US$ 43,364 (34,621–52,107) 44,324 (38,44350,206) 0.88
No. of i.v. fluid bags charged to patient 29.7 (26.2–33.2) 5.2 (4.3–6.2) <0.001
Total length of stay, h 82.0 (61.4–102.6) 74.6 (50.6–98.7) 0.73

CI = confidence interval; US = United States.

Two-Bag Method of Management of DKA
100
90
80
One-Bag
70
Two-Bag
Remaining
patients with 60
acidosis* (%)
50
40
30
20
10
0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Time (hours)
Figure 4. Kaplan-Meier curve comparing the percentage of
patients with continued metabolic acidosis (*HCO3 < 18
mEq/L) between groups over time.
demonstrated that a protocol utilizing a constant insulin
infusion rate, titratable dextrose containing fluid infusion,
and a nurse-driven electrolyte protocol was associated
with earlier correction of acidosis; decreased duration
of insulin infusion; fewer i.v. fluid bags charged; and no
increase in rates of hypoglycemia, hypokalemia, or hy-
perkalemia. We also observed decreased rates of hospital-
ization and ICU admission, an increased rate of ED
discharge, and no increase in ED return visits with use
of the 2B method. Our results observed in adult patients
are in agreement with published pediatric data, which
demonstrated signals toward faster rates of pH, bicarbon-
ate, and ketone correction and improved resource utiliza-
tion with use of the 2B method (6–8).
In adult patients, DKA carries a mortality of < 1%, and
most deaths in the setting of DKA are related to underly-
ing medical illness rather than acute electrolyte derange-
ments (9,10). Nevertheless, one of the most important
considerations in using a protocol with a constant
Figure 5. Kaplan-Meier curve comparing the percentage of
patients remaining on insulin infusion between groups over
time.
5
insulin infusion rate is whether it would lead to unsafe
nadirs in glucose or potassium levels. We did not
observe concerning safety issues with our studied
protocol, and rates of hypokalemia or hypoglycemia did
not significantly differ between cohorts. In-hospital mor-
tality was 0% in each cohort. Importantly, patients with
unique characteristics (e.g., renal failure, volume over-
load) were excluded from protocolized care in each
cohort at the physician’s discretion.
In addition to its safety profile, the 2B method may
also be associated with decreased resource utilization.
Most notable was a decreased number of 500-mL and
1000-mL i.v. fluid bags charged to the patient (5.2
vs. 29.7; p < 0.001), which we suspect is due to less
waste of partially used bags in response to fluctuations
in serum glucose. In the event of iatrogenic hypoglyce-
mia in patients managed with a 1B method, insulin and
fluid delivery is paused, presenting a potential time
delay in therapy, and the fluid bag is wasted and re-
placed with a new bag with a higher dextrose concen-
tration. Once hypoglycemia is corrected, this new
higher dextrose concentration bag may be wasted in
favor of yet another new, lower dextrose concentration
fluid. This sequence of events may occur multiple
times during the initial management of DKA, resulting
in significant delays in treatment and inefficient use of
time and resources. In comparison, iatrogenic hypogly-
cemia in patients managed with a 2B method requires
only a titration of currently administered bags of fluid,
and no wasting or spiking of new bags. The 2B method
was also associated with earlier correction of acidosis
and shorter duration of insulin infusion, which may
have translated into our observation of decreased ICU
and hospital admissions. On a large health systems
scale, these results are promising for overall cost sav-
ings.
Limitations
Several important limitations of this study are recog-
nized. The retrospective nature of this study suggests
association but cannot directly demonstrate causation
of the outcome measures mentioned. The before-and-
after nature of this study is inherently prone to favoring
the ‘‘after’’ group. In addition, the sample size is rela-
tively small, and all patients were enrolled in a single
center. Furthermore, there are several important con-
founders impacting patient care in the 2B group related
to their increased rate of transfer from the main ED to
the EC3, a nine-bed ED-based ICU contained within
the adult ED, which opened midway through the study
period. Ninety-four percent of patients in the 2B cohort
were managed in EC3 at some point, whereas 51% of
patients in the 1B cohort were managed in EC3 at
6 N. L. Haas et al.

Table 3. Safety and Disposition Outcomes

One-Bag Two-Bag
Outcomes (n = 107) (n = 68) p Value

Lowest recorded BG, mg/dL, mean 134 (123–146) 142 (132–151) 0.34
Hypoglycemia (BG < 70 mg/dL), % 10.3 (5.7–16.7) 2.9 (0.3–10.8) 0.07
Lowest recorded K, mEq/L, mean 3.5 (3.4–3.6) 3.7 (3.6–3.8) 0.10
Hypokalemia (K < 3.3 mEq), % 27.1 (19.6–36.3) 16.2 (9.2–26.9) 0.09
Highest recorded K, mEq/L, mean 5.5 (5.2–5.8) 5.5 (5.3–5.7) 0.88
Hyperkalemia (K > 5.2 mEq), % 51.4 (42.1–60.6) 57.4 (45.5–68.4) 0.44
Severe hyperkalemia (K > 6.0 mEq), % 20.6 (14.0–29.3) 23.6 (15.0–35.0) 0.64
Discharged from the ED, % 15.0 (9.4–23.1) 44.1 (33.0–56.0) <0.001
ED return visit within 72 h, % 3.7 (1.2–9.6) 7.4 (2.9–16.6) 0.29
Admitted to the ward, % 69.2 (59.8–77.1) 55.9 (44.1–67.0) 0.006
Admitted to the ICU, % 15.9 (10.1–24.1) 0 (0–6.6) <0.001

BG = blood glucose; CI = confidence interval; ED = emergency department; ICU = intensive care unit; K = potassium.
Values in parentheses are 95% confidence intervals.

some point (p < 0.01). EC3 management entails a tight REFERENCES

patient-to-nurse and patient-to-physician ratio (2:1 and
9:1, respectively); detailed disease-specific order sets,
including prescribed frequency of laboratory checks
and a nursing-driven ICU electrolyte protocol; rapid
availability of stat laboratories and blood gases; and
close monitoring. It is likely that the resource-
intensive care received in EC3 is more akin to an
ICU than to the typical ED or ward setting. Specif-
ically, more frequent/prompt laboratory draws may
have influenced the recorded time to resolution of
acidosis.
CONCLUSIONS
The 2B method appears to be a feasible management
strategy for adult ED patients with DKA. Compared to
management with a titratable insulin infusion and a single
dextrose containing i.v. fluid bag, use of the 2B method of
fluid and dextrose delivery was associated with earlier
correction of acidosis, shorter duration of insulin infu-
sion, fewer bags of i.v. fluid charged, more ED dis-
10. Umpierrez GE, Kelly JP, Navarrete JE, Casals MM, Kitabchi AE.
charges, and fewer ICU admissions, while no safety
concerns were observed.
1. Number (in Thousands) of Hospital Discharges with Diabetic Ke-
toacidosis (DKA) as First-Listed Diagnosis. National Center for
Health Statistics, Centers for Disease Control and Prevention.
Available at: 1988–2009. http://www.cdc.gov/diabetes/statistics/
dkafirst/fig1.htm. Accessed March 6, 2016.
2. Kitabchi AE, Umpierrez GE, Miles JM, Fisher JN. Hyperglycemic
crises in adult patients with diabetes. Diabetes Care 2009;32:1335–43.
3. Kitabchi AE, Sacks HS, Fisher JN. Clinical trials in diabetic ketoa-
cidosis. In: Larner J, ed. Methods in Diabetes Research. New York:
John Wiley; 1986:315–32.
4. Kitabchi AE, Materi R, Murphy MB. Optimal insulin delivery in
diabetic ketoacidosis (DKA) and hyperglycemic hyperosmolar non-
ketotic coma (HHNC). Diabetes Care 1982;5:78–87.
5. DeFronzo RA, Tobin JD, Andres R. Glucose clamp technique: a
method for quantifying insulin secretion and resistance. Am J Phys-
iol 1979;237:E214–23.
6. Poirier MP. A prospective study of the ‘‘two-bag system’’ in diabetic
ketoacidosis management. Clin Pediatr 2004;43:809–13.
7. So TY, Grunewalder E. Evaluation of the two-bag system for fluid
management in pediatric patients with diabetic ketoacidosis. J Pe-
diatr Pharmacol Ther 2009;14:100–5.
8. Grimberg A, Cerri RW, Satin-Smith M, Cohen P. The ’two bag sys-
tem’ for variable intravenous dextrose and fluid administration: ben-
efits in diabetic ketoacidosis management. J Pediatr 1999;134:376–8.
9. National Center for Health Statistics. National hospital discharge
and ambulatory surgery data. Available at: http://www.cdc.gov/
nchs/ about/major/hdasd/nhds.htm. Accessed March 6, 2016.
Hyperglycemic crises in urban blacks. Arch Intern Med 1997;
157:669–75.
Two-Bag Method of Management of DKA 7

ARTICLE SUMMARY
1. Why is this topic important?
Use of a two-bag method for management of diabetic
ketoacidosis (DKA) has demonstrated faster response
time in i.v. fluid therapy changes, earlier correction of
serum bicarbonate, ketones, and pH, and more cost-
effective care in pediatric patients. However, this method
has not been studied previously in adults.
2. What does this study attempt to show?
This study compares resource utilization and safety
outcomes between adult emergency department (ED) pa-
tients in DKA managed with the traditional one-bag and
novel two-bag methods.
3. What are the key findings?
Patients managed with the two-bag method experi-
enced a faster resolution of metabolic acidosis, shorter
duration of insulin infusion, and fewer i.v. fluid bags
were charged to the patient. Rates of hypoglycemia and
hypokalemia did not significantly differ.
4. How is patient care impacted?
The two-bag method is an effective management strat-
egy for adult ED patients with DKA, and may have
resource utilization and patient safety benefits over the
traditional one-bag method.