Analysis of the German Hodgkin Study 21 CarbonePP, Kaplan HS, Musshoff K, following short term megadose

Group (GHSG). EHA Learning Center; SmithersDW, TubianaM. Report of the methylprednisolone treatment. Spinal
2016; S109. committee on Hodgkin’s diseasestaging Cord 1998; 36: 633–6.
19 Hasenclever D, Diehl VA. Prognostic classification. CancerRes 1971; 31: 1860–61. 25 Miyanishi K, Yamamoto T, Irisa T,
score for advanced Hodgkin’s disease. 22 Ficat RP. Idiopathic bone necrosis of the Motomura G, Jingushi S, Sueishi K et al.
International prognostic factors project femoral head. Early diagnosis and Effects of different corticosteroids on the
on advanced Hodgkin’s disease. N Engl J treatment. J Bone Joint Surg Br 1985; development of osteonecrosis in rabbits.
Med 1998; 339: 1506–14. 67: 3–9. Rheumatology 2005; 44: 332–6.
20 Press OW, Li H, Schöder H, Straus DJ, 23 Marsh JCW, Zomas A, Hows JM, 26 Xiao CS, Lin N, Lin SF, Wan R,
Moskowitz CH, LeBlanc M et al. US Chapple M, Gordon-Smith EC. Chen WH. Experimental study on
intergroup trial of response-adapted Avascular necrosis after treatment of avascular necrosis of femoral head in
therapy for stage III to IV Hodgkin aplastic anaemia with antilymphocyte chickens induced by different
lymphoma using early interim globulin and high-dose glucocorticoides. Zhongguo Gu Shang
fluorodeoxyglucose–positron emission methylprednisolone. Br J Haematol 2010; 23: 184–7.
tomography imaging: Southwest 1993; 84: 731–5.
Oncology Group S0816. J Clin Oncol 24 Wing PC, Nance P, Connell DG,
2016; 34: 2020–27. Gagnon F. Risk of avascular necrosis

doi:10.1111/imj.13735

Diabetic ketoacidosis in adult patients: an audit of factors

influencing time to normalisation of metabolic parameters
Melissa H. Lee ,1 Genevieve L. Calder,1 John D. Santamaria2,3 and Richard J. MacIsaac1,3
Departments of 1Endocrinology and Diabetes, and 2Intensive Care, St Vincent’s Hospital Melbourne, and 3Department of Medicine, The University of
Melbourne, Melbourne, Victoria, Australia

Key words Abstract

diabetes, diabetic ketoacidosis, hyperglycaemic
emergencies, metabolic parameters, Background: Diabetic ketoacidosis (DKA) is an acute life-threatening metabolic com-
hyperglycaemia. plication of diabetes that imposes substantial burden on our healthcare system. There is
a paucity of published data in Australia assessing factors influencing time to resolution
Correspondence of DKA and length of stay (LOS).
Melissa H. Lee, Department of Endocrinology Aims: To identify factors that predict a slower time to resolution of DKA in adults with
and Diabetes, St Vincent’s Hospital Melbourne, diabetes.
41 Victoria Parade, Fitzroy, Vic. 3065, Australia. Methods: Retrospective audit of patients admitted to St Vincent’s Hospital Melbourne
Email: melissa.lee@svha.org.au between 2010 to 2014 coded with a diagnosis of ‘Diabetic Ketoacidosis’. The primary out-
come was time to resolution of DKA based on normalisation of biochemical markers.
Received 2 July 2017; accepted Episodes of DKA within the wider Victorian hospital network were also explored.
21 December 2017. Results: Seventy-one patients met biochemical criteria for DKA; median age 31 years
(26–45 years), 59% were male and 23% had newly diagnosed diabetes. Insulin omis-
sion was the most common precipitant (42%). Median time to resolution of DKA was
11 h (6.5–16.5 h). Individual factors associated with slower resolution of DKA were
lower admission pH (P < 0.001) and higher admission serum potassium level (P = 0.03).
Median LOS was 3 days (2–5 days), compared to a Victorian state-wide LOS of 2 days.
Higher comorbidity scores were associated with longer LOS (P < 0.001).
Conclusions: Lower admission pH levels and higher admission serum potassium levels
are independent predictors of slower time to resolution of DKA. This may assist to strat-
ify patients with DKA using markers of severity to determine who may benefit from
closer monitoring and to predict LOS.

Funding: None.
Conflict of interest: None.

Internal Medicine Journal 48 (2018) 529–534 529

© 2018 Royal Australasian College of Physicians
Lee et al.

Introduction (HHS) (i.e. serum osmolality >320 mOsm/L), mixed

HHS/DKA or incomplete datasets. Only index cases were
Diabetic ketoacidosis (DKA) is an acute, severe and life- included and recurrent episodes of DKA were excluded
threatening metabolic complication of diabetes. It typi- to avoid selection bias.
cally occurs in those with an absolute or relative insulin All clinicians are encouraged to adhere to the DKA
deficient state. The annual incidence of DKA ranges from management protocol used at SVHM (Supporting Infor-
4 to 8 episodes per 1000 patient admissions with diabe- mation Appendix S1). Data were collected retrospec-
tes1 and 30% are newly diagnosed with diabetes.2 Most tively through SVHM medical records. Information
patients admitted with DKA are expected to make a full regarding Victorian DKA presentations was obtained
recovery, and the implementation of timely and effective through the Victorian Admitted Episode Dataset
management strategies have resulted in falling mortality (VAED). This study was approved by St Vincent’s Hospi-
rates in Australia over recent decades to 1.4–3.0%.3,4 tal Ethics Committee.
Should a death occur in the setting of DKA, it is usually The primary outcome was time to resolution of DKA,
due to the underlying precipitating illness rather than calculated as the time from the first blood sample that
the metabolic complications of ketoacidosis.5,6 provided a measure of acid–base state to the normalisa-
DKA imposes substantial burden on the individual, tion of relevant biochemical markers. Resolution of DKA
family and overall healthcare system. To our knowledge, was pre-defined as meeting all three of the following cri-
there are no published studies in Australia that have teria: pH >7.3, HCO3 ≥15mmol/L and blood glucose
assessed factors influencing time to resolution of DKA <11.1 mmol/L, adapted from the American Diabetes
and hospital length of stay (LOS). A retrospective UK Association (ADA) criteria.9 Anion gap and ketones,
study of 50 patient episodes of DKA, confirmed using which were not routinely measured during admission,
biochemical criteria, reported an average time to resolu- were not included in our criteria for the resolution of
tion of DKA of 12 h and 6 min, but did not assess factors DKA. The ADA criteria was used as there are no current
contributing to this.7 ADS consensus guidelines to define resolution of DKA.
The aims of this audit were to assess patient and treat- Secondary outcome measures included precipitating fac-
ment factors that may influence time to resolution of tors, choice of intravenous fluid, adverse events and
DKA, based on normalisation of metabolic parameters. LOS. Over the same 4-year period, we compared LOS
We benchmarked LOS at our tertiary institution to other for admissions with DKA at our institution with other
Victorian hospitals and examined the impact of comor- tertiary and metropolitan Victorian hospitals. Comorbid-
bidity scores on LOS. ity indexes for these episodes of DKA were extracted
from the VAED, using the Charlson Comorbidity Index10
and the Elixhauser measure,11 which includes psychiat-
Methods
ric disorders.
This retrospective audit reviewed the clinical records of Statistical analysis was largely using non-parametric
all patients admitted to St Vincent’s Hospital Melbourne measures. Results are expressed as median (interquartile
(SVHM), between November 2010–2014 inclusive, range, IQR) and number (%). A Kruskal–Wallis rank test
coded with a discharge diagnosis of ‘DKA’. SVHM is a was used for non-parametric data. Multivariate survival
500-bed tertiary public teaching hospital that offers a analysis was performed using a Kaplan–Meier model
variety of medical, surgical, emergency, critical care and and a stepwise backward elimination Cox proportional
mental health services. hazards model. Variables included in the Cox propor-
Patients were eligible for inclusion if aged over 18 years tional hazards model included age, gender, admission
and meeting biochemical criteria for DKA. This was pH, glucose, ketones, potassium, bicarbonate; HbA1c;
defined as a triad of hyperglycaemia, ketosis and acido- presence of concurrent infection, insulin omission as a
sis, based on the Australian Diabetes Society (ADS) precipitant, and use of compound sodium lactate (CSL).
guidelines.8 These parameters included blood glucose A two-tailed P-value <0.05 was considered significant.
>11 mmol/L or known diabetes, bicarbonate (HCO3) Data were analysed using Stata/MP software, version 12
<15 mmol/L or venous pH <7.3, and presence of ketosis (StataCorp, College Station, TX, USA), for Windows XP.
(blood or urine). A history of diabetes was included to
account for cases of euglycaemic DKA, such as in the
Results
context of sodium-glucose co-transporter 2 (SGLT-2)
inhibitor use or pregnancy. Patients were excluded if A total of 239 episodes was identified with a coded dis-
presenting with biochemical evidence of hyperglycaemia charge diagnosis of DKA over a consecutive 4-year
without ketoacidosis, hyperglycaemic hyperosmolar state period. Of these, 168 patients were excluded for the

530 Internal Medicine Journal 48 (2018) 529–534

© 2018 Royal Australasian College of Physicians
 Factors influencing normalisation of DKA

Using a Kaplan–Meier estimate (Fig. 2), the median

time to resolution of DKA was 11 h (6.5–16.5 h). Indi-
vidual factors that may have been associated with reso-
lution of DKA were identified using a log rank test and
were then included in a multivariate stepwise backward
elimination Cox proportional hazards model (Table 2).
A lower admission pH (P < 0.001) and higher admission
serum potassium level (P = 0.03) were both independent
predictors of a slower time to resolution of DKA. Con-
current infection (P = 0.05) showed a trend towards
slower resolution time. Patients with concurrent infec-
tion had lower admission pH levels compared to those
without infection (median: 7.15 vs 7.22).
Figure 1 Study algorithm. DKA, diabetic ketoacidosis; HHS, hypergly- Hyperglycaemia normalised more rapidly than ketoa-
caemic hyperosmolar state. cidosis (5.5 vs 11 h). Time to transition from intravenous
reasons listed in Figure 1. Seventy-one patients meeting insulin infusion to subcutaneous insulin was 26 h
biochemical criteria for DKA were included in the final (18.8–40.8 h). Two patients, presenting with mild DKA,
analysis (Fig. 1). did not require an insulin infusion as they had rapid res-
Baseline patient characteristics and admission bio- olution of acidosis within 2.5 h following treatment with
chemical parameters are shown in Table 1. Of note, subcutaneous rapid-acting insulin and intravenous
there were no cases of euglycaemic DKA and no use of fluids. Median LOS was 3 days (2–5 days). Twenty-four
SGLT-2 inhibitors. The median HbA1c on presentation patients (34%) became hypokalaemic (K+ < 3.5 mmol/
was 101 mmol/mol (11.4%) indicating poor glycaemic L) during admission. There were no serious adverse
control. The most common precipitating factor for DKA events or deaths.
was insulin omission (42%), which included continuous All patients received intravenous 0.9% sodium chlo-
subcutaneous insulin infusion failure. This was followed ride solution as either initial resuscitation or mainte-
by infection (29%), most commonly urinary and respira- nance fluid replacement. Forty-four per cent also
tory sources. Other precipitants included psychosocial received CSL solution. Plasma-Lyte was used in
stressors (6%), including illicit substance use, recent 10 patients (14%) and hemosol in 8 patients (11%), all
changes in insulin regimen (3%), epileptic seizure (1%), of whom were managed in the intensive care unit (ICU)
radiotherapy treatment (1%) and corticosteroid use on haemofiltration. On univariate analysis, the use of
(1%). No clear cause was identified in 17% of patients. CSL compared to other fluid types was associated with
slower time to resolution of DKA (14.5 vs 10.0 h, P =
Table 1 Baseline characteristics and admission biochemical data in 0.02). However, on multivariate analysis that included
patients admitted with DKA admission pH, this association was no longer signifi-
Characteristics Values cant (P = 0.37).

Age (years) 31 (26–45)

Male (%) 59
Known type 1:type 2 diabetes (n) 44:11
Prior episodes of DKA (%) 42
Newly diagnosed (%) 23
Requiring ICU admission (%) 30
Median length of hospital stay, days (IQR) 3 (2–5)

Admission biochemical data Value, median (IQR)

Glucose (mmol/L) 27.6 (21.7–39.2)

pH 7.2 (7.1–7.3)
HCO3 (mmol/L) 12.0 (8.0–15.5)
Ketones (mmol/L) 5.6 (4.4–6.5)
Potassium (mmol/L) 4.7 (4.4–5.2)
HbA1c (mmol/mol, %) 101 (81–119) (11.4
(9.6–13.0))
Figure 2 Time to resolution of diabetic ketoacidosis (DKA) using a
DKA, diabetic ketoacidosis; ICU, intensive care unit; IQR, interquartile Kaplan–Meier estimate. Median time to resolution of DKA was 11 h
range. (blue line).

Internal Medicine Journal 48 (2018) 529–534 531

© 2018 Royal Australasian College of Physicians
Lee et al.

Table 2 Factors associated with slower time to resolution of DKA using insulin. Patients requiring ICU, and those with concur-
a Cox proportional hazards model rent infection, tended to have a greater delay. We also
Individual factors HR (95% CI for HR) P value attribute this delay to other factors such as delay to tran-
sition over meal-times; and ‘after-hours’ factors with
Admission pH 393 (28.7, 5400)† <0.001*
fewer experienced medical personnel to oversee man-
Admission potassium 1.53 (1.06, 2.22) 0.03*
Concurrent infection 0.56 (0.32, 0.99) 0.05 agement. This delay subsequently impacts LOS, which is
notably longer than the time taken for resolution of
*Significance (P<0.05). †Equates to a HR of 1.77 (1.38, 2.28) for every
DKA. Additional factors that may influence delay to dis-
0.1 decrease in pH level. CI, confidence interval; HR, hazard ratio.
charge include treatment of concurrent precipitants; time
taken to involve the diabetes multidisciplinary team and
Using the VAED, there were 2175 adult admissions
other specialty teams; and complex discharge planning
over a 4-year period with a coded diagnosis of ‘DKA’ to
such as those with psychosocial issues. There was no
Victorian hospitals. The median overall LOS was 2 days.
mortality related to DKA in our institution over the 4-
LOS at SVHM was 1 day longer than the state-wide LOS.
year period.
This may be due to a higher percentage at SVHM requir-
Our findings demonstrated that nearly half of the
ing mechanical ventilation (5.1% vs 2.2%, NS). There
admissions were attributable to insulin omission. This
was a significant association between LOS and higher
was followed by infection, which is typically reported as
comorbidity score (P < 0.001) and greater psychiatric
the most common precipitant for DKA in the literature.9
comorbidities (P < 0.001). A total of six (0.3%) deaths
Psychological issues, including eating disorders and sub-
occurred over the 4-year period; the causes of death
stance use, are also known to constitute a significant
were not available.
proportion of DKA presentations.9,14 Our small percent-
age of patients in whom we identified psychosocial
stressors as a precipitant is likely a gross underestimate
Discussion
given the interplay with insulin omission, as well as
This is the first reported study in Australia to assess fac- potential under-reporting due to fear of judgement.15
tors that influence time to resolution of DKA. The major Early identification and appropriate management of pre-
findings are that lower admission pH and higher admis- cipitants are important for the acute recovery of DKA,
sion potassium levels are independent predictors of a and also to minimise LOS and future episodes of DKA.
slower time to resolution of DKA, based on normalisa- The optimal choice of intravenous resuscitation fluid
tion of biochemical markers. This suggests that these two in the management of DKA still remains to be deter-
biochemical markers are most reflective of the severity mined. Crystalloids are favoured over colloids however
of the acidosis. Hyperkalaemia is common initially in evidence is lacking.16 Traditionally, 0.9% sodium chlo-
DKA despite a total body potassium-deplete state.2 This ride is first-line treatment due to its efficacy, safety, cost
is due to the extracellular shift of potassium in an aci- and availability; and its ability to be readily mixed with
dotic state due to insulin insufficiency, and further exac- potassium. However, recent observational data have
erbated in those with DKA by severe dehydration and raised concerns over hyperchloraemic metabolic acidosis
acute renal failure. Concurrent infection also trended as a consequence of excessive sodium chloride use.17,18
towards slower resolution time. We propose that this is Although some studies suggest that hyperchloraemic
due to a more severe metabolic acidosis in patients pre- metabolic acidosis may prolong acidosis and precipitate
senting with concurrent infection, demonstrated by oliguria,19 the clinical significance remains unknown.
lower admission pH levels. Our study did not show that use of sodium chloride slo-
In our study, DKA was treated effectively with a wed the resolution of DKA however rates of hyperchlor-
median time to resolution of 11 h, similar to previous aemic metabolic acidosis were not assessed.
UK and US studies.7,12 Hyperglycaemia was faster to nor- The association between CSL use and slower resolu-
malise than ketoacidosis, also consistent with previous tion of DKA is likely explained by the biochemical char-
findings.9 However, the optimal rate of resolution for the acteristics of patients at time of admission. Those with a
metabolic derangements that characterise DKA remains greater degree of acidosis received CSL more commonly
to be defined. One recent study has suggested that inten- than sodium chloride, presumably because of the per-
sive compared with partial early correction of hypergly- ceived risk of worsening the acid–base status by promot-
caemia is associated with a higher risk of hypoglycaemia, ing hyperchloraemic metabolic acidosis due to the lower
hypo-osmolarity and death.13 chloride content in CSL. Another proposed hypothesis
There was also considerable delay after resolution of for the delay in resolution of DKA is that CSL contains
DKA in implementing transition to subcutaneous lactate which requires additional metabolism in the

532 Internal Medicine Journal 48 (2018) 529–534

© 2018 Royal Australasian College of Physicians
 Factors influencing normalisation of DKA

liver.20,21 Data on lactate were not collected in this study The Victorian dataset is likely an overestimate of the
and would be of interest in future studies. Two large ran- true number of DKA episodes in Victoria. Coded dis-
domised trials comparing sodium chloride with CSL in charge diagnoses were used rather than biochemical
DKA have shown that neither fluid type was superior in parameters, and the dataset also included recurrent DKA
terms of clinical outcomes;20,22 however, glycaemic episodes. The VAED does not have any unique identifiers
recovery was slower in those given CSL.20 or pathology linked to each episode. However, this is the
More recently, Plasma-Lyte, a balanced electrolyte first study to assess comorbidity indices in patients admit-
solution, has been proposed as the fluid of choice in the ted with DKA to Victorian hospitals. The finding that
management of DKA.23 It contains organic acid buffers higher comorbidity scores are associated with longer LOS
and less chloride content than sodium chloride, thus may help to predict patient outcomes and associated mor-
may prevent hyperchloraemic metabolic acidosis.24 Sev- bidity. It is also a timely reminder that diabetes is a multi-
eral studies suggest that Plasma-Lyte may hasten recov- system condition that requires optimal management of
ery of DKA with more rapid increases in mean arterial associated comorbidities and mental health disorders.
pressure and urine output.25 However like CSL, Plasma- There are several limitations to this study. First, our
Lyte contains acetate and gluconate which require addi- sample size was small. Nevertheless, the final patient
tional metabolism and may slow resolution of DKA. cohort was a highly select group, truly reflective of the
These balanced electrolyte solutions may then negate metabolic changes seen in DKA. Furthermore, blood
the proposed benefits in reducing rates of hyperchlorae- ketone measurements are now performed regularly in
mic metabolic acidosis. Moreover, Plasma-Lyte’s safety the management of DKA, and should be incorporated
profile and long-term cost-effectiveness remains largely into the criteria for resolution of DKA. However, bicar-
unknown.24 In our study, Plasma-Lyte was used in only bonate levels are still commonly used as a surrogate
14% of patients, limiting our ability to draw any firm marker. Moreover, as the VAED is purely an administra-
conclusions about the potential benefits of this solution. tive dataset, it only provides preliminary Victorian data
Hemosol, a bicarbonate-buffered solution, was used on DKA episodes and does not allow further clarification
only in patients who were haemofiltered in ICU. The into the clinical and biochemical details of these
role of bicarbonate in the treatment of DKA remains episodes. Future multi-centre prospective studies com-
controversial.26 While there may be some indication for paring the VAED to biochemical criteria for DKA would
its use in severe DKA with pH <6.9,16 there have been provide a more accurate estimate of rates of DKA in Vic-
no prospective randomised studies to support this torian hospitals.
approach. There are also concerns that bicarbonate
replacement may be associated with hypokalaemia, cen-
Conclusion
tral nervous system acidosis and cerebral oedema.27
Hypokalaemia was common following initiation of the Time to resolution of DKA in this Australian study was
insulin infusion. Information was not collected on the comparable to similar studies in the UK and
rates and quantity of potassium supplementation pre- US. Independent predictors of a slower time to resolu-
scribed, and this would be useful for future studies. We tion of DKA are lower admission pH levels and higher
hypothesise that this occurred due to potential lack of admission potassium levels. Implementation of these
adherence to aggressive intravenous potassium replace- findings may assist to stratify patients with DKA using
ment recommended in the hospital DKA protocol. Regu- biochemical markers to determine who may benefit
lar review of patients’ biochemical parameters, and strict from closer surveillance and monitoring, and to predict
implementation of hospital protocols may reduce com- LOS. There still lies debate around the optimal choice of
plications, such as insulin-induced hypokalaemia. Rates fluid resuscitation in DKA. Further prospective rando-
of other adverse events such as hypoglycaemia and other mised trials are required to establish greater evidence for
electrolyte disturbances were not collected in this study the optimal management of DKA. Reassuringly, mortal-
but should be considered in future studies. ity rates for DKA in Victoria remain low.

References Malone JI et al. Hyperglycemic crises in 4 Venkatesh B, Pilcher D, Prins J,

diabetes. Diabetes Care 2004; 27 (Suppl Bellomo R, Morgan TJ, Bailey M.
1 Johnson DD, Palumbo PJ, Chu CP. 1): S94–102. Incidence and outcome of adults with
Diabetic ketoacidosis in a community- 3 RJ MI, Lee LY, KJ MN, Tsalamandris C, diabetic ketoacidosis admitted to ICUs in
based population. Mayo Clin Proc 1980; Jerums G. Influence of age on the Australia and New Zealand. Crit Care
55: 83–8. presentation and outcome of acidotic 2015; 19: 451.
2 Kitabchi AE, Umpierrez GE, and hyperosmolar diabetic emergencies. 5 Kitabchi AE, Umpierrez GE,
Murphy MB, Barrett EJ, Kreisberg RA, Intern Med J 2002; 32: 379–85. Murphy MB, Barrett EJ, Kreisberg RA,

Internal Medicine Journal 48 (2018) 529–534 533

© 2018 Royal Australasian College of Physicians
Lee et al.

Malone JI et al. Management of et al. Intensity of early correction of 21 Davies PG, Venkatesh B, Morgan TJ,
hyperglycemic crises in patients with hyperglycaemia and outcome of Presneill JJ, Kruger PS, Thomas BJ et al.
diabetes. Diabetes Care 2001; 24: 131–53. critically ill patients with diabetic Plasma acetate, gluconate and
6 Umpierrez GE, Kelly JP, Navarrete JE, ketoacidosis. Crit Care Resusc 2017; 19: interleukin-6 profiles during and after
Casals MM, Kitabchi AE. Hyperglycemic 266–73. cardiopulmonary bypass: a comparison
crises in urban blacks. Arch Intern Med 14 Lee P, Greenfield JR, Campbell LV. of plasma-Lyte 148 with a bicarbonate-
1997; 157: 669–75. “Mind the gap” when managing balanced solution. Crit Care 2011;
7 Crasto W, Htike ZZ, Turner L, Higgins K. ketoacidosis in type 1 diabetes. Diabetes 15: R21.
Management of diabetic ketoacidosis Care 2008; 31: e58. 22 Mahler SA, Conrad SA, Wang H,
following implementation of the JBDS 15 Ng RS, Darko DA, Hillson RM. Street Arnold TC. Resuscitation with balanced
guidelines: where are we and where drug use among young patients with electrolyte solution prevents
should we go? Br J Diab Vasc Dis 2015; type 1 diabetes in the UK. Diabet Med hyperchloremic metabolic acidosis in
15: 11–16. 2004; 21: 295–6. patients with diabetic ketoacidosis.
8 Craig ME, Twigg SM, Donaghue KC, 16 Tran TTT, Pease A, Wood AJ, Zajac JD, Am J Emerg Med 2011; 29: 670–4.
Cheung NW, Cameron FJ, Conn J et al. Mårtensson J, Bellomo R et al. Review 23 Stowe ML. Plasma-Lyte vs. normal
In: Australian Government Department of evidence for adult diabetic saline: preventing hyperchloremic
of Health and Ageing, ed. National ketoacidosis management protocols. acidosis in fluid resuscitation for
Evidence-Based Clinical Care Guidelines for Front Endocrinol 2017; 8: 106. diabetic ketoacidosis. School of Physician
Type 1 Diabetes in Children, Adolescents 17 Kitabchi AE, Umpierrez GE, Assistant Studies Paper 296. Pacific
and Adults. Canberra; 2011. Murphy MB, Kreisberg RA. University, USA; 2012.
9 Kitabchi AE, Umpierrez GE, Miles JM, Hyperglycemic crises in adult patients 24 Rizoli S. PlasmaLyte. J Trauma 2011; 70:
Fisher JN. Hyperglycemic crises in adult with diabetes: a consensus statement S17–8.
patients with diabetes. Diabetes Care from the American Diabetes 25 Chua HR, Venkatesh B, Stachowski E,
2009; 32: 1335–43. Association. Diabetes Care 2006; 29: Schneider AG, Perkins K, Ladanyi S
10 Charlson ME, Pompei P, Ales KL, 2739–48. et al. Plasma-Lyte 148 vs 0.9% saline for
CR MK. A new method of classifying 18 Funk GC, Zauner C, Bauer E, fluid resuscitation in diabetic
prognostic comorbidity in longitudinal Oschatz E, Schneeweiss B. ketoacidosis. J Crit Care 2012; 27:
studies: development and validation. Compensatory hypochloraemic alkalosis 138–45.
J Chronic Dis 1987; 40: 373–83. in diabetic ketoacidosis. Diabetologia 26 Viallon A, Zeni F, Lafond P, Venet C,
11 Elixhauser A, Steiner C, Harris DR, 2003; 46: 871–3. Tardy B, Page Y et al. Does bicarbonate
Coffey RM. Comorbidity measures for 19 Savage MW, Dhatariya KK, Kilvert A, therapy improve the management of
use with administrative data. Med Care Rayman G, Rees JA, Courtney CH et al. severe diabetic ketoacidosis? Crit Care
1998; 36: 8–27. Joint British diabetes societies guideline Med 1999; 27: 2690–3.
12 Umpierrez GE, Jones S, Smiley D, for the management of diabetic 27 Cardoso L, Vicente N, Rodrigues D,
Mulligan P, Keyler T, Temponi A et al. ketoacidosis. Diabet Med 2011; 28: Gomes L, Carrilho F. Controversies in
Insulin analogs versus human insulin in 508–15. the management of hyperglycaemic
the treatment of patients with diabetic 20 Van Zyl DG, Rheeder P, Delport E. Fluid emergencies in adults with diabetes.
ketoacidosis: a randomized controlled management in diabetic-acidosis – Metab Clin Exp 2017; 68: 43–54.
trial. Diabetes Care 2009; 32: 1164–9. Ringer’s lactate versus normal saline: a
13 Martensson J, Bailey M, Venkatesh B, randomized controlled trial. QJM 2012;
Pilcher D, Deane A, Abdelhamid YA 105: 337–43.

Supporting Information
Additional supporting information may be found in the online version of this article at the publisher’s web-site:
Appendix S1. Diabetic ketoacidosis (DKA) policy.

534 Internal Medicine Journal 48 (2018) 529–534

© 2018 Royal Australasian College of Physicians