dme_2102.

fm Page 359 Friday, March 16, 2007 2:41 PM

DIABETICMedicine
DOI: 10.1111/j.1464-5491.2007.02102.x

Does an Integrated Care Pathway enhance the

management of diabetic ketoacidosis?
An
Integrated
Care Pathway
Oxford,
Diabetic
DME
Blackwell
0742-3071
Original
24
Original
UK
article
Medicine
Publishing,
Article
Ltd. for the management of diabetic ketoacidosis S. L.Waller et al.
2007

S. L. Waller, S. Delaney and M. W. J. Strachan

Abstract
Metabolic Unit, Western General Hospital,
Edinburgh, UK
Accepted 21 November 2006

Aims Integrated Care Pathways (ICPs) are management plans that indicate the

sequence and timing of the optimal treatment for individuals with a given disorder.
The treatment of diabetic ketoacidosis (DKA) before and after the implementation
of an ICP in a teaching hospital was examined.
Methods Twenty-seven episodes of DKA were identified during the 13-month

control period and 22 in the 13 months following implementation of the ICP.

Case notes were reviewed and relevant clinical data extracted.
Results The introduction of the ICP was associated with a reduction in the
time taken to initiate intravenous fluid [45.0 (5225) min to 37.5 (0135) min;
P = 0.01]. Time taken to initiate insulin infusion was also reduced [60.0 (5755)
min to 37.5 (0175) min; P = 0.02]. The proportion of patients commenced
on intravenous insulin within 60 min increased from 48 to 77% (P = 0.04). In
addition, there was a reduction in the prescription of antibiotics (4818%;
P = 0.028) and low molecular weight heparin (595%; P < 0.001). Length of
stay was not affected.
Conclusion The ICP significantly improved key areas in the management of

DKA, although there remains room for further improvements.

Diabet. Med. 24, 359363 (2007)
Keywords

diabetic ketoacidosis, Integrated Care Pathway

DKA, diabetic ketoacidosis; ICP, Integrated Care Pathway;

LOS, length of stay

Abbreviations

Introduction
The 1999 British Diabetic Association Cohort Study reported
that metabolic disturbances were the most common cause of
death in people with Type 1 diabetes under the age of 20 years
[1]. Of the deaths due to diabetes, 54% of males and 76% of
females died from diabetic ketoacidosis (DKA). There are multiple
mechanisms by which DKA can cause serious morbidity and
death, including cerebral oedema, cardiovascular insufficiency
secondary to volume depletion and acidosis, renal failure and
aspiration pneumonia [2]. Some of these deaths might be

Correspondence to: Dr Mark W. J. Strachan, Consultant Physician, Metabolic

Unit, Western General Hospital, Crewe Road, Edinburgh, EH4 2XU, UK. E-mail:
mark.strachan@luht.scot.nhs.uk

2007 The Authors.

Journal compilation 2007 Diabetes UK. Diabetic Medicine, 24, 359363

avoided by timely and appropriate interventions such as rapid

initiation of intravenous (i.v.) fluids and insulin and nasogastric
tube insertion in patients with impaired consciousness.
The European Insulin Dependent Diabetes Mellitus (IDDM)
Policy Group [3] emphasized that DKA should be managed
according to specific core protocols by professionals experienced
in their use. However, a previous retrospective survey found
that the presence of clinical guidelines alone were insufficient
to ensure the provision of adequate patient care [4]. In their
study, Singh et al. reported that, in a large UK teaching hospital,
DKA guidelines were followed in less than 22.5% of occasions.
The median time from admission to the initiation of fluids was
80 min and the median time to initiation of i.v. insulin was
greater than 60 min in 69% of cases. In addition, the amounts
of potassium chloride and intravenous fluid administered were
typically inadequate.

359

dme_2102.fm Page 360 Friday, March 16, 2007 2:41 PM

DIABETICMedicine

An Integrated Care Pathway for the management of diabetic ketoacidosis S. L. Waller et al.

Integrated Care Pathways (ICPs) or Critical Care Pathways

are management plans that display goals for patients and indicate
the sequence and timing of actions required by staff to fulfil
these goals with optimal efficacy [5]. Some case studies have
shown reduction in costs and improvements in patient care
[6], although this has not been supported by systematic reviews
[7]. Effective pathways have been found to be associated with
uncomplicated conditions requiring a predictable course of
inpatient care [6], such as DKA.
The impact of an ICP for the treatment of DKA was examined in a US study [8] comparing 72 patients treated for DKA
in the year prior to the implementation of the care pathway
with 77 patients treated over the following year. Length of stay
(LOS) and total cost of hospitalization were taken as primary
end points, with treatment outcomes as secondary end points.
The implementation of this pathway was associated with a
significant decrease in mean LOS from 5.2 to 2.4 days, but only
in those patients treated without an endocrine consultation.
A trend for lower hospital costs was found, although this was
not statistically significant. A significant increase in the volume
of intravenous fluid administered over the first 24 h was
demonstrated; however, this was accompanied by significant
increase in the proportion of patients receiving greater than the
recommended volume. The proportion of patients receiving
intravenous insulin within a very undemanding target of 8 h,
remained unchanged at approximately 85%.
In 2004, a Diabetes Specialist Nurse was appointed to the
Lothian National Health Service (NHS) Acute Division with
funding from the Scottish Diabetes Group with the specific
remit of developing an ICP for use in DKA. The pathway was
introduced in November 2004 and the purpose of this study
was to investigate whether the ICP altered the quality of care
of patients subsequently admitted with DKA.

Patients and methods

The ICP was based around a new Scottish protocol for the management of DKA [9]. The local guidelines for the management
of patients admitted with DKA were altered to be in line with
this national protocol and served as the template for the ICP
[10]. The ICP comprised four sections and specified a detailed
management plan for the patient from admission to discharge
[10]. The pathway was designed to be used in conjunction with
a multidisciplinary admissions proforma.
Patients were identified by the Medical Records Department
based on the Scottish Morbidity Record (SMR) coding of discharge summaries. The control group comprised those patients
admitted with DKA in the 13 months prior to the implementation
of the ICP on 1 November 2004. The ICP group were those
patients admitted in the 13 months after the ICP was introduced.
DKA was defined as: plasma glucose > 11 mmol/l; ketonuria or
ketonaemia; arterial hydrogen ion concentration > 50 nmol/l;
plasma bicarbonate 18 mmol/l. The chair of the local Medical
Ethics Committee confirmed that formal ethical approval was
not required as this study constituted a retrospective review of
a clinical service development. The study was approved by the
local Caldicott Guardian.

360

Case notes were reviewed retrospectively for clinical details

including age and sex, previous episodes of DKA, precipitating
factors, biochemical parameters on presentation (glucose,
hydrogen/pH, bicarbonate, potassium, urea and creatinine
concentrations). Urea(mmol/l) : creatinine(mmol/l) ratio was
calculated The primary outcome measures were time from
admission to initiation of i.v. fluid and insulin. Secondary outcome measures included amount of i.v. potassium given in the
first 24 h, volume of i.v. fluid given in the first 24 h, frequency
of biochemical monitoring, adjunctive therapy and length of
inpatient stay. In addition, the proportion of patients meeting
specified targets for the initiation of i.v. fluid (30 min) and
insulin (60 min) and amount of i.v. fluid administered (6.5 l in
the first 24 h) were studied [4]. Adverse consequences, including
hypoglycaemia (plasma glucose < 4.0 mmol/l) and hypokalaemia (plasma potassium < 3.5 mmol/l), were also considered.
Statistical analysis

Normality of the continuous data was established using

ShapiroWilks test. Where both control and ICP data for a given
variable were identified as parametric, the comparison was
made using a non-paired t-test. A MannWhitney U-test was
used for non-parametric data. Chi-square test was used to compare the categorical data. Multivariate logistic regression was
used to adjust for differences in baseline parameters between
the study groups. All statistical analyses were performed using
SPSS (version 13.0) software (SAS Institute, Cary, NC, USA). A
P-value of < 0.05 was defined as the limit of statistical significance. With 27 episodes in the control group and 22 in the ICP
group, the study had over 95% power to detect a large effect
size (1.0 SD), equivalent to a 30-min reduction in time to initiation
of fluids. The study had only 54% power to detect a medium
(0.5 SD) effect size.

Results
In the control group, 48 episodes of DKA were identified. Of
these, only 27 episodes met the strict criteria for DKA. These
episodes occurred in 22 individuals, as one person presented
on four occasions in DKA, and two presented twice. Of those
excluded, 13 had a degree of hyperglycaemia, but the hydrogen
ion concentration and/or bicarbonate concentration did not
meet the necessary criteria. Three others had a metabolic acidosis
secondary to other causes. One was incorrectly coded (did not
have diabetes) and four sets of notes could not be retrieved.
In the ICP group, 50 episodes were identified. Of these,
24 episodes met the biochemical criteria for entry into the
study, these episodes occurred in 21 individuals, as one patient
presented on four occasions. Two patients were excluded as
the ICP was not used. One of these was human immunodeficiency virus (HIV) positive and was therefore immediately
admitted to the Infectious Diseases Unit, bypassing acute medicine; there was no obvious reason why the other patient was
not treated according to the ICP. A further 21 episodes had a
degree of hyperglycaemia, but did not meet the necessary
hydrogen and or bicarbonate criteria. Five sets of notes could
not be retrieved.

2007 The Authors.

Journal compilation 2007 Diabetes UK. Diabetic Medicine, 24, 359363

dme_2102.fm Page 361 Friday, March 16, 2007 2:41 PM

Original article

DIABETICMedicine

Table 1 Background characteristics

Age (years)
Sex (% male)
Duration of Type 1 diabetes (years)
Known previous episode(s) of DKA (%)
Plasma glucose (mmol/l)
Arterial H+ (nmol/l)
Arterial pH
Standard bicarbonate (mmol/l)
Plasma potassium (mmol/l)
Plasma creatinine (mol/l)
Plasma urea (mmol/l)
Urea : creatinine ratio

Control group

ICP group

P-value

37.0 (1460)
48.2
9.1 6.3
51.9
38.8 15.4
65.0 (51115)
7.1 0.1
9.6 3.6
5.6 0.9
126 (73690)
8.2 (3.418.6)
70.0 24.5

27.0 (1658)
54.6
8.1 6.5
36.4
33.9 14.3
61.0 (5096)
7.2 0.1
10.6 3.5
5.3 0.8
92 (4993)
6.3 (2.517.8)
70.3 23.0

0.42
0.66
0.59
0.28
0.17
0.05*
0.02*
0.06
0.29
0.18
0.23
0.96

*P < 0.05.
Parametric data are mean SD; non-parametric data are median (range).
DKA, diabetic ketoacidosis; ICP, Integrated Care Pathway.

Table 2 Outcome measures

Time from admission to initiation of i.v. fluid (min)

Time from admission to initiation insulin infusion (min)
Volume of fluid in first 24 h (ml)
Amount of potassium in first 24 h (mmol)
Frequency biochemical parameters measured in first 24 h
Thirty-minute target for i.v. fluid initiation met (%)
Sixty-minute target for initiation insulin infusion met (%)
Target met for 6.5 l i.v. fluid over 24 h (%)
Heparin given (%)
Antibiotics given (%)
Length of stay (days)
Episodes of hypoglycaemia (%)
Episodes of hypokalaemia (%)

Control group

ICP group

P-value

45.0 (5225)
60.0 (5755)
7762 1965
100.0 (20340)
6.0 (49)
37.5
48.0
69.2
59.3
48.2
3.0 (3.87.2)
3.7
22.2

37.5 (0135)
37.5 (0175)
7044 1449
80.0 (20140)
5.0 (38)
63.6
77.3
80.1
4.6
18.2
2.0 (16)
0.0
13.6

0.01*
0.01*
0.28
0.06
0.80
0.08
0.04*
0.36
< 0.001**
0.03*
0.68
0.36
0.44

*P < 0.05; **P < 0.001.

Parametric data are mean SD; non-parametric data are median (range).
ICP, Integrated Care Pathway; i.v., intravenous.

Background characteristics

Outcome measures

Table 1 shows the background characteristics of the DKA

patients studied in the control and ICP periods. There was no
significant difference in demographic data, duration of Type 1
diabetes and previous known episodes (P > 0.05). Most baseline
biochemistry including plasma glucose, potassium, urea and
creatinine concentrations, urea(mmol/l) : creatinine(mmol/l)
ratio and standard bicarbonate concentrations did not differ
between groups (P > 0.05). However, arterial hydrogen ion
concentration was significantly higher in the control group
(P = 0.05).

Table 2 displays the outcome measures of the patients in the

control and ICP groups. Both time from triage to initiation
of intravenous fluid and insulin infusion were significantly
shorter in the ICP group (P = 0.01 and 0.02, respectively). In
addition, there was an increase in the proportion of patients
receiving i.v. fluids (37.5 to 63.6%; P = 0.08) and insulin (48.0
to 77.3%; P = 0.04) within 30 and 60 min, respectively. The
proportion of patients receiving the target amount of intravenous fluid (6.5 l over 24 h) increased from 69.2 to 80.1%,
however, this was not statistically different (P > 0.05). Mean

2007 The Authors.

Journal compilation 2007 Diabetes UK. Diabetic Medicine, 24, 359363

361

dme_2102.fm Page 362 Friday, March 16, 2007 2:41 PM

DIABETICMedicine

An Integrated Care Pathway for the management of diabetic ketoacidosis S. L. Waller et al.

volume of fluid and amount of potassium administered in the

first 24 h, and the frequency that biochemical parameters were
measured in the first 24 h, showed no significant difference
(P > 0.05). Length of stay and frequency of episodes of hypoglycaemia and hypokalaemia also did not differ significantly.
Statistical adjustment for baseline hydrogen concentration did
not materially alter these results (data not shown).
Antibiotics were used less frequently in the ICP group
(P = 0.03). Of the control group, 13 episodes were treated with
antibiotics; of these only three were thought to have an infective
precipitant for DKA. By contrast, only 18% (four) episodes in
the ICP group were treated with antibiotics. Three of these
were thought to have an infective precipitant and the other was
a new presentation of Type 1 diabetes. The use of low molecular
weight heparin was also significantly reduced (P < 0.001),
with patients in the control group treated with heparin on 16
occasions as opposed to only once in the ICP group.
Central venous catheters were inserted on three occasions in
the control group and once in the ICP group. There were three
High Dependency Unit (HDU) admissions in the control group
and two in the ICP group. There were no fatalities. No patients in
either group were treated with intravenous sodium bicarbonate.

Discussion
The present study demonstrated a significant reduction in the
time taken to initiate both i.v. fluids and insulin following the
implementation of the ICP. There is a paucity of randomized
trial evidence in virtually every aspect of the management of
DKA, so it is not possible to cite robust evidence that early
recourse to institution of i.v. fluids and insulin materially
affects patient outcomes. However, good clinical practice
would dictate that delay in the initiation of such therapy in
individuals who are dehydrated and acidotic should be minimized. It would seem logical, therefore, that timely initiation
of treatment would lead to a reduction in complications and
reduce recovery time in a larger group of patients.
There was also a signification reduction in both the use of
antibiotics and low molecular weight heparin between the
control and ICP periods. With regard to antibiotic treatment,
our hospital guidelines for the management of DKA stated that
patients should only be treated with antibiotics if infection is
proven or strongly suspected; this was highlighted in the ICP.
The majority of patients prescribed antibiotics in the control
group had no signs of infection documented. Local guidelines
for the prescription of low molecular weight heparin for prophylaxis of thromboembolic disease recommended prescription
in cases of diabetic coma and where prolonged bed rest was
likely. It appears that, by compelling doctors to consider
actively the indication for antibiotics and heparin, the ICP
significantly reduced overmedication of patients in these areas.
The cost savings for our cohort were not substantial (we would
estimate in the order of 134), but clearly if these findings were
replicated in a larger cohort of patients, substantial financial
savings could accrue.

362

The data from the study by Singh et al. [4] demonstrated

that management of DKA was suboptimal and that clinical
guidelines were not followed. Ilag et al. [8] supported the use
of ICPs in the treatment of DKA. However, they found fewer
and different improvements than demonstrated in the present
study. This may be related to differences in the quality of care
prior to the implementation of the ICP. For example, Ilag et al.
demonstrated that the ICP was associated with significant
decrease in mean LOS in those treated without an endocrine
consultation from 5.2 to 2.4 days. In the present study, no
reduction in LOS was observed, but as the initial LOS was only
2.9 days there was less scope for improvement. Similarly Ilag
et al. demonstrated a significant increase in the amount of i.v.
fluids given over the first 24 h, whereas in the present study,
despite an increase in the proportion of patients meeting the
target from 69.2 to 80.1%, no significant difference was
found. This may be because of the high proportion of patients
meeting the target during the control period.
The main limitation of this study was the small number of
patients in each arm. While the study was powered to detect
large effect sizes (and indeed did demonstrate such effects), it
is conceivable that smaller differences might have been overlooked. Moreover, a study involving a substantially greater
number of patients would be required before any potential
impact on morbidity and mortality could be reliably examined. As this was a retrospective, historical control study, as
opposed to a randomized control trial, it is also possible that
other unrecognized factors may have confounded the apparent
impact of the ICP on the outcome measures. The variation in
baseline hydrogen ion concentration between the control and
ICP groups suggested that the groups were not entirely equivalent. However, as the control group was more acidotic, it
could be assumed that this would have favoured even earlier
intervention in this group. Moreover, adjustment for baseline
hydrogen ion concentration did not affect the main outcome
measures, reinforcing the evidence that the ICP hastened the
initiation of i.v. fluid and insulin infusions.
In conclusion, our findings support the use of an Integrated Care Pathway in the management of DKA. Both time
from triage to initiation of i.v. fluids and insulin infusion
showed significant improvement and significantly more
patients were commenced on i.v. insulin within the recommended time frame. However, there remained patients who
did not meet the treatment targets so there remains room for
improvement.

Competing interests
None declared.
Acknowledgements

We wish to acknowledge the financial support from the Scottish

Diabetes Group and Sanofi Aventis in the development of the
Integrated Care Pathway.

2007 The Authors.

Journal compilation 2007 Diabetes UK. Diabetic Medicine, 24, 359363

dme_2102.fm Page 363 Friday, March 16, 2007 2:41 PM

Original article

DIABETICMedicine

References
1 Laing SP, Swerdlow AJ, Slater SD, Botha JL, Burden AC, Waugh NR
et al. The British Diabetic Association cohort study, II: causespecific mortality in patients with insulin-treated diabetes mellitus.
Diabet Med 1999; 16: 466471.
2 Strachan MWJ, Nimmo GR, Noyes K, Simpson D, Kelnar CJ.
Management of cerebral oedema in diabetes. Diabetes Metab
Res Rev 2003; 19: 241247.
3 European IDDM Policy Group. Consensus guidelines for the management of insulin-dependent (Type 1) diabetes. Diabet Med 1993;
10: 9901005.
4 Singh RK, Perros P, Frier BM. Hospital management of diabetic
ketoacidosis: are clinical guidelines implemented effectively? Diabet
Med 1997; 14: 482486.
5 Pearson SD, Goulart-Fisher D, Lee TH. Critical pathways as a strategy

2007 The Authors.

Journal compilation 2007 Diabetes UK. Diabetic Medicine, 24, 359363

6
7

10

for improving care: problems and potential. Ann Intern Med 1995;
123: 941948.
Every NR, Hochman J, Becker R, Kopecky S, Cannon CP . Critical
pathways: a review. Circulation 2000; 101: 461465.
Dy SM, Garg P, Nyberg D, Dawson PB, Pronovost PJ, Morlock L
et al. Critical pathway effectiveness: assessing the impact of the
patient, hospital and pathway characteristics using qualitative comparative analysis. Health Serv Res 2005; 40: 499516.
Ilag LL, Kronick S, Ernst RD, Grondin L, Alaniz C, Liu L et al.
Impact of a critical pathway on inpatient management of diabetic
ketoacidosis. Diabetes Res Clinl Prac 2003; 62: 2332.
McGeoch SC, Philip S, Park C, MacRury S, Pearson DWM. Introducing a new protocol for the management of diabetic ketoacidosis
in a medical admissions unit. Diabet Med 2006; 23: 81.
Lothian Diabetes Network. Local guidelines. 2006 Available from:
http://www.lotdiabetnet.scot.nhs.uk (accessed 29 January 2007).

363