29/01/19 19.

32

www.medscape.com

Pediatric Diabetic Ketoacidosis With Hyperosmolarity

Clinical Characteristics and Outcomes

Sungeeta Agrawal, MD; Grayson L. Baird, PhD; Jose Bernardo Quintos, MD; Steven E. Reinert, MS; Geetha Gopalakrishnan,
MD; Charlotte M. Boney, MD; Lisa Swartz Topor, MD

Endocr Pract. 2018;24(8):726-732.

Abstract and Introduction

Abstract

Objective: Studies of hyperglycemic emergencies with hyperosmolality, including hyperglycemic hyperosmolar state (HHS) and
"mixed presentation" with features of diabetic ketoacidosis (DKA) and HHS, are lacking in children. Objectives were to
determine the incidence of DKA, HHS, and mixed presentation in a pediatric population, to characterize complications, and to
assess accuracy of associated diagnosis codes.

Methods: Retrospective cohort study of 411 hyperglycemic emergencies in pediatric patients hospitalized between 2009 and
2014. Hyperglycemic emergency type was determined by biochemical criteria and compared to the associated diagnosis code.

Results: Hyperglycemic emergencies included: 333 DKA, 54 mixed presentation, and 3 HHS. Altered mental status occurred
more frequently in hyperosmolar events (P<.0001), and patients with hyperosmolarity had 3.7-fold greater odds of developing
complications compared to those with DKA (P = .0187). Of those with DKA, 98.5% were coded correctly. The majority (81.5%)
of mixed DKA-HHS events were coded incorrectly. Events coded incorrectly had 3.1-fold greater odds of a complication (P =
.02).

Conclusion: A mixed DKA-HHS presentation occurred in 13.8% of characterized hyperglycemic emergencies, whereas HHS
remained a rare diagnosis (0.8%) in pediatrics. Hyperosmolar events had higher rates of complications. As treatment of
hyperosmolarity differs from DKA, its recognition is essential for appropriate management.

Introduction

Diabetic ketoacidosis (DKA) and hyperglycemic hyperosmolar state (HHS) are the main hyperglycemic emergencies seen in
diabetes mellitus, and their management and complications differ. DKA, characterized by hyperglycemia and ketoacidosis, is
the most common hyperglycemic emergency in children with type 1 diabetes mellitus (T1DM).[1–3] HHS is also characterized by
hyperglycemia and dehydration but with hyperosmolarity and without ketoacidosis. While DKA is far more common in children,
HHS is seen in children with T1DM, and its incidence is on the rise.[1–5]

HHS and DKA are distinct diagnoses, though patients with diabetes mellitus can present with a combination of acidosis and
hyperosmolarity. The hyperglycemic emergency with mixed features of HHS and DKA is herein referred to as a "mixed
presentation." There are few reports about the frequency of the mixed presentations in adults: one study of adults with
hyperglycemic emergencies at Rhode Island Hospital found that 33% of subjects had a mixed presentation.[6] A similar study of
adults in Australia found that 30% of subjects had a mixed presentation.[7] Studies of mixed presentation are lacking in children.

Distinguishing HHS from DKA is important, as their treatments and complications differ.[2,3] Untreated DKA can lead to cardiac
arrest and death, and the most common serious complication of DKA is cerebral edema. HHS symptoms can be more insidious
and include significant dehydration on presentation.[3] HHS, like DKA, can also result in significant morbidity and mortality, with
complications including cardiovascular collapse, malignant hyperthermia, and rhabdomyolysis.[2,8] Little is known about the
features and complications of the mixed presentation, given the limited literature on this topic. Characterizing the patients who
present with DKA, HHS, or a mixed presentation and their complications can help physicians recognize and optimize treatment
for these different hyperglycemic emergencies.

Studies of hyperglycemic emergencies in children have utilized diagnosis codes, such as those of the International
Classification of Diseases, Ninth Revision (ICD-9), to identify cases and better understand their management.[5,9,10] While
these studies include large numbers of subjects across multiple hospitals, it is unknown if ICD-9 or other billing codes
appropriately classify all hyperglycemic emergencies and specifically recognize the mixed DKA-HHS presentation.

In order to address these knowledge gaps, we sought to determine the incidence of all hyperglycemic emergencies, as well as
the associated complication rates, among children and adolescents at a single institution. We also sought to assess the
accuracy of diagnosis codes of hyperglycemic emergencies.

https://www.medscape.com/viewarticle/901782_print Page 1 of 9
 29/01/19 19.32

Methods

This was a retrospective cohort study of children and adolescents ages 0 to 21 years with diabetes mellitus who were admitted
to Rhode Island Hospital with a hyperglycemic emergency between January 1, 2009, and December 31, 2014. This study was
approved by the Institutional Review Board of Lifespan Corporation.

To identify all potential hyperglycemic emergencies, we searched the electronic medical record (EMR) database for at least one
of the following laboratory criteria: blood glucose ≥600 mg/dL, pH ≤7.3, bicarbonate ≤15 mEq/L, or anion gap ≥12. To ensure an
underlying diagnosis of diabetes, the search identified subjects with any ICD-9 code for diabetes mellitus (codes 249–251). One
author (S.A.) reviewed all biochemical data, and hyperglycemic emergencies were included if they met criteria for DKA, HHS, or
a mixed presentation. If the event did not meet criteria for a hyperglycemic emergency, or if the subject did not have diabetes,
the event was excluded. Laboratory analyses were performed at Rhode Island Hospital, unless the subject initially presented to
an outside hospital. DKA and HHS were defined in accordance with the International Society of Pediatric and Adolescent
Diabetes (ISPAD) consensus guidelines.[2] DKA criteria included: (1) blood glucose ≥200 mg/dL, (2) pH ≤7.3 or bicarbonate ≤15
mEq/L, and (3) ketonuria and/or elevated serum beta-hydroxybutyrate. Although studies show good concordance between
serum bicarbonate values and venous pH,[11] in the instances where pH and bicarbonate were discordant, pH was used to
make the definitive diagnosis. A pH of ≤7.3 for all blood gases (arterial, venous, capillary) was used to define acidosis.[12] HHS
criteria included: (1) osmolarity ≥320 mOsm/kg (measured when available, and otherwise calculated (2[serum sodium mEq/L] +
[blood urea nitrogen mg/dL]/2.8 + [glucose mg/dL]/18), (2) blood glucose ≥600 mg/dL, and (3) absence of acidosis (pH >7.3
without large ketosis).[2] Events with features of both DKA and HHS were categorized as a "mixed presentation" and included:
(1) osmolarity ≥320 mOsm/kg, (2) initial blood sugar ≥600 mg/dL, (3) acidosis (pH ≤7.3), and (4) ketonuria and/or elevated
serum beta-hydroxybutyrate. Three subjects with unknown pretreatment blood sugar values and posttreatment blood sugar
<200 mg/dL were included, as were 4 subjects without ketone data who otherwise met criteria for DKA and were treated with
intravenous insulin.

DKA treatment at our institution is consistent with American Diabetes Association and ISPAD guidelines.[2,13] Patients are
treated with intravenous insulin starting at 0.1 units/kg/hour and receive intravenous fluids with 0.45 or 0.9% normal saline.
Potassium repletion is added to maintain normokalemia, and the dextrose concentration is titrated to maintain blood sugar in
the target range.

Anthropometric data, vital signs, medical history, physical exam, and laboratory data were collected for all events. Study data
were collected and managed using REDCap (Research Electronic Data Capture) electronic data capture tool,[14] hosted by
Lifespan Corporation. Diabetes type was obtained from the EMR as determined by clinical presentation, pancreatic
autoantibodies (available for 50 of 270 subjects), and subsequent clinical course. Pancreatic autoantibodies were routinely
measured in obese adolescents suspected of having type 2 diabetes mellitus (T2DM), in children receiving chronic steroids or
other medications that can cause diabetes, and in those with suspicion for maturity-onset diabetes of the young.[15–17]

Altered mental status (AMS) included a Glasgow coma scale ≤13, confusion, memory loss, focal neurologic deficits, loss of
consciousness, or other descriptors indicating impaired mental status. Fatigue and somnolence were not considered AMS.
Intellectual disability included developmental delay, learning disability, autism spectrum disorder, cerebral palsy associated with
intellectual impairment, or genetic conditions associated with intellectual disability. History of depression was obtained from the
EMR. Race/ethnicity was recorded from the EMR as Hispanic, non-Hispanic black, non-Hispanic white, Asian, multiracial, or
other.

We collected information on complications through review of hospital records, which included but were not limited to cerebral
edema, pancreatitis, severe hypertriglyceridemia with triglycerides ≥1,000 ± 100 mg/dL, pneumomediastinum, and deep vein
thrombosis. Amylase and lipase measurements were recorded when available. Complications also included additional
diagnoses occurring in relation to the hyperglycemic emergency that were not part of the expected clinical course and resulted
in lengthened stay, additional interventions, increased monitoring, or additional imaging. There were no deaths.

ICD-9 discharge diagnoses were recorded and compared to the biochemical diagnoses. For those with mixed presentation,
coding was considered correct if it included both acidosis and hyperosmolarity.

Statistical Analyses

All analyses were conducted using SAS software 9.4 (SAS Inc, Cary, NC). Comparisons of characteristics for subjects with
new-onset diabetes mellitus (NODM) between DKA and mixed presentation/HHS were examined using generalized linear
modeling, and comparisons of characteristics for multiple events in patients during our observation window between DKA and
mixed presentation/HHS were examined using generalized mixed linear modeling with sandwich estimation assuming either a
binary distribution (for dichotomous outcomes) or normal distribution (for continuous outcomes). For both NODM and all
hyperglycemic emergencies, predictors of hyperosmolarity (mixed and HHS) versus DKA presentation were modeled using
generalized linear modeling and mixed modeling, respectively, using sandwich estimation. Predictors of complications and
miscoding for all events were examined using generalized mixed modeling with sandwich estimation assuming a binary
distribution. All modeling was done using PROC GLIMMIX, and all counts were calculated using PROC FREQ. Statistical
significance was established at the P<.05 level and all interval estimates calculated for 95% confidence.

https://www.medscape.com/viewarticle/901782_print Page 2 of 9
 29/01/19 19.32

Results
Hyperglycemic Emergencies

The EMR search identified 916 potential hyperglycemic emergencies. All records were reviewed, and using the biochemical
criteria for DKA, HHS, or mixed presentation, we identified 411 hyperglycemic emergencies: 333 DKA admissions, 54 mixed
presentation admissions, 3 HHS admissions, and 21 admissions that were unable to be characterized due to insufficient
information in the EMR (Figure 1). Nineteen of the 21 uncharacterized events involved subjects who initially presented to an
outside hospital. Among all characterized emergencies, the frequency of HHS was 0.8%, the frequency of mixed presentation
was 13.8%, and the total frequency of hyperosmolarity (mixed presentation and HHS) was 14.6%.

Figure 1.

Types of hyperglycemic emergencies over 6 years. DKA = diabetic ketoacidosis; HHS = hyperglycemic hyperosmolar state.

The 411 hyperglycemic emergencies occurred in 270 subjects. There were 208 subjects with 1 event, 29 subjects with 2
events, 17 subjects with 3 events, and 16 subjects with 4 or more events. summarizes the characteristics of subjects with DKA
and hyperosmolarity. When comparing all events, there were no statistically significant differences in age, gender, percent
obese, duration of diabetes (for those with known disease), admission hemoglobin A1c, or percent with depression observed.
There was a higher proportion of subjects with NODM in the hyperosmolar group, compared to the DKA group (56.2% vs.
35.8%; P = .028).

Table 1. Characteristics of Hyperglycemic Emergencies

https://www.medscape.com/viewarticle/901782_print Page 3 of 9
 29/01/19 19.32

  DKA Hyperosmolaritya  
All events Value (n) 95% CI Value (n) 95% CI P value
Age, years 13.5 (333) [12.9, 14.1] 13.5 (57) [12.7, 14.3] .887
Male gender, % 57.4 (333) [49.0, 65.3] 69.5 (57) [50.5, 83.5] .225
NODM, % 35.8 (333) [29.2, 43.1] 56.2 (57) [39.2, 71.8] .028b
Duration of DM, years 6.2 (224) [5.5, 6.9] 6.9 (29) [5.6, 8.4] .255
A1c at admission, % 12.2 (213) [11.9, 12.5] 12.1 (37) [11.4, 12.7] .742
BMI, %c 59.3 (282) [54.6, 64.0] 55.3 (42) [47.5, 63.1] .287
Obesity (BMI ≥95%), % 9.6 (282) [6.5, 14.0] 3.8 (42) [0.9, 15.2] .213
Intellectual disability, % 4.7 (326) [2.6, 8.1] 12.6 (56) [5.1, 28.1] .06
AMS, % 10.9 (324) [7.8, 15.2] 42.8 (56) [28.9, 57.9] <.0001b
Depression, % 13.3 (325) [9.4, 18.5] 12.9 (56) [5.8, 26.4] .938
Complications, % 4.3 (306) [2.5, 7.4] 15.0 (56) [7.1, 29.0] .008b

Abbreviations: A1c = hemoglobin A1c; AMS = altered mental status; BMI = body mass index; CI = confidence interval; DKA =
diabetic ketoacidosis; DM = diabetes mellitus; HHS = hyperglycemic hyperosmolar state; NODM = new-onset diabetes mellitus.
aHyperosmolarity includes both HHS and mixed presentation.
b P<.05.
cBMI percent for age and gender.

In the DKA group, 66.7% were non-Hispanic white, 16.8% were Hispanic, and 13.2% were non-Hispanic black. In the
hyperosmolar group, 50.1% were non-Hispanic white, 28.1% were Hispanic, and 21.1% were non-Hispanic black.

AMS (P≤.0001) and complications (P = .008) both occurred with increased frequency in the hyperosmolar group. As the degree
of acidosis can be associated with AMS and complications,[18] we examined additive and interaction effects of pH with
hyperosmolarity and did not find any significant interaction effect between diagnosis and pH with complications or AMS.
Hyperglycemic emergency type remained a significant predictor of complications and AMS with pH as a covariate.

In subjects with NODM, there was no statistically significant difference in age, gender, percent obese, incidence of intellectual
disability, percent with depression, or rate of complications. There remained a higher proportion of subjects presenting with
AMS in the hyperosmolar group (P = .018).

All 3 HHS admissions occurred in subjects with NODM (2 with T1DM, 1 with steroid-induced diabetes). The average age of
those with HHS was 15.2 years (range, 9.5 to 18.4 years), and none presented with AMS or had intellectual disability.

Predictors of Hyperosmolarity (Mixed Presentation or HHS) Versus DKA

Demographic and biometric variables associated with a hyperosmolar (mixed and HHS presentation) relative to DKA
presentation were examined (). For all presentations, the odds of hyperosmolarity were 3.4 times higher for those who had
intellectual disability relative to those who did not have intellectual disability (P = .0174). In addition, the odds of hyperosmolarity
were 6 times higher for those presenting with AMS relative to those without AMS (P<.0001), while for the subgroup with NODM,
the odds of hyperosmolarity were 3.4 times higher in those with AMS relative to those without AMS (P = .0179).

Table 2. Predictors of Hyperosmolaritya Versus DKA Diagnosis

Predictor % 95% CI OR 95% CI P value

All Presentations
Male 15.6 [10.9, 21.9] 1.6 [0.8, 3.2] .178
Female 10.3 [6.2, 16.7]      
ID present 31.4 [15.3, 53.7] 3.4 [1.2, 9.1] .017b
ID absent 12.0 [8.7, 16.3]      
AMS present 37.7 [24.7, 52.8] 6.1 [3.0, 12.6] <.0001b
AMS absent 9.0 [6.1, 13.0]      
Depression present 19.4 [11.9, 30.2] 1.5 [0.8, 3.0] .21

https://www.medscape.com/viewarticle/901782_print Page 4 of 9
 29/01/19 19.32

Depression absent 13.6 [10.2, 17.9]      

New-Onset Diabetes Mellitus
Male 25.0 (20/80) [16.7, 35.7] 2.3 [0.9, 6.4] .097
Female 12.5 (6/48) [5.7, 25.3]      
ID present 35.7 (5/14) [15.6, 62.6] 2.5 [0.7, 8.2] .141
ID absent 18.4 (21/114) [12.3, 26.7]      
AMS present 39.1 (9/23) [21.6, 60.0] 3.4 [1.2, 9.2] .018b
AMS absent 16.0 (16/100) [10.0, 24.6]      
Depression present 25.0 (1/4) [3.3, 76.6] 1.3 [0.1, 13.4] .821
Depression absent 20.3 (25/123) [14.1, 28.4]      

Abbreviations: AMS = altered mental status; CI = confidence interval; DKA = diabetic ketoacidosis; ID = intellectual disability;
OR = odds ratio.
aHyperosmolarity includes both hyperglycemic hyperosmolar state and mixed
b P<.05.

Diabetes Type

The majority (261 subjects, 96.7%) of subjects had T1DM. Seven subjects (2.6%) had T2DM, 1 subject had steroid-induced
diabetes (0.4%), and 1 subject had tacrolimus-induced diabetes (0.4%). Six of the 7 subjects with T2DM had pancreatic
autoantibody testing, which confirmed the T2DM diagnosis. Both subjects with drug-induced diabetes had negative pancreatic
autoantibodies.

T1DM accounted for 98.2% (327/333) of DKA events, 96.3% (52/54) of mixed presentation events, and 66.7% (2/3) of HHS
events. Six of the subjects with T2DM presented with DKA, and one had a mixed presentation. The subject with steroid-induced
diabetes had HHS, while the subject with tacrolimus-induced diabetes had a mixed presentation.

Complications

Nineteen subjects developed complications during 24 hospitalizations. The odds of developing complications were 3.7 times
greater for those who presented with hyperosmolarity relative to DKA (P = .0187). The types of complications differed between
DKA and mixed presentation group (). None of the 3 subjects with HHS developed complications. Three subjects had recurrent
complications on subsequent admissions: one with hyper-triglyceridemia during three different admissions, one with
hypertriglyceridemia during one admission and re-admission the following day with hypertriglyceridemia and pancreatitis, and
one with endoscopy-confirmed esophagitis twice in the setting of DKA ().

Table 3. Complications in Subjects With Hyperglycemic Emergencies

Subject Complication type

Mixed presentation
1 Pancreatitis
2 Pulmonary edema, acute tubular necrosis
3 Transaminitis with peak ALT 1,144 IU/L, AST 459 IU/L
4 Transaminitis with peak ALT 843 IU/L, AST 195 IU/L, hepatomegaly
5 Pancreatitis
11a Hypertriglyceridemia, peak TG 925 mg/dL
17 Acute respiratory failure, hypertriglyceridemia>2,000 mg/dL
18 Pneumomediastinum
19 Stroke, PEA, DVT, ARDS, sacral ulcer woundrequiring colostomy
DKA
6 Hypertriglyceridemia >1,000 mg/dL
  Admission 1 peak TG 1,480 mg/dL

https://www.medscape.com/viewarticle/901782_print Page 5 of 9
 29/01/19 19.32

  Admission 2 peak TG 4,602 mg/dL

  Admission 3 peak TG 2,819 mg/dL
7 Pneumomediastinum
8 Hypertriglyceridemia, peak TG 988 mg/dL
9 Esophagitis
10 Pneumomediastinum
11a Hypertriglyceridemia to peak TG 23,942 mg/dL
  Returned day after discharge with DKA, TG 2,409 mg/dL, and pancreatitis
12 DVT
13 Severe sepsis
14 DVT, cerebral edema
16 Renal calculi
Uncharacterized emergency
15 Cerebral edema

Abbreviations: ALT = alanine transaminase; ARDS = acuterespiratory distress syndrome; AST = aspartate transaminase;DKA =
diabetic ketoacidosis; DVT = deep vein thrombosis;PEA = pulseless electrical activity; TG = triglyceride.
aSubject 11 had complications during mixed presentation and DKA.

Table 3. Complications in Subjects With Hyperglycemic Emergencies

Subject Complication type

Mixed presentation
1 Pancreatitis
2 Pulmonary edema, acute tubular necrosis
3 Transaminitis with peak ALT 1,144 IU/L, AST 459 IU/L
4 Transaminitis with peak ALT 843 IU/L, AST 195 IU/L, hepatomegaly
5 Pancreatitis
11a Hypertriglyceridemia, peak TG 925 mg/dL
17 Acute respiratory failure, hypertriglyceridemia>2,000 mg/dL
18 Pneumomediastinum
19 Stroke, PEA, DVT, ARDS, sacral ulcer woundrequiring colostomy
DKA
6 Hypertriglyceridemia >1,000 mg/dL
  Admission 1 peak TG 1,480 mg/dL
  Admission 2 peak TG 4,602 mg/dL
  Admission 3 peak TG 2,819 mg/dL
7 Pneumomediastinum
8 Hypertriglyceridemia, peak TG 988 mg/dL
9 Esophagitis
10 Pneumomediastinum
11a Hypertriglyceridemia to peak TG 23,942 mg/dL
  Returned day after discharge with DKA, TG 2,409 mg/dL, and pancreatitis
12 DVT

https://www.medscape.com/viewarticle/901782_print Page 6 of 9
 29/01/19 19.32

13 Severe sepsis
14 DVT, cerebral edema
16 Renal calculi
Uncharacterized emergency
15 Cerebral edema

Abbreviations: ALT = alanine transaminase; ARDS = acuterespiratory distress syndrome; AST = aspartate transaminase;DKA =
diabetic ketoacidosis; DVT = deep vein thrombosis;PEA = pulseless electrical activity; TG = triglyceride.
aSubject 11 had complications during mixed presentation and DKA.

ICD-9 Codes

The ICD-9 discharge diagnosis codes for all events were compared to the biochemical diagnoses. Nearly all (98.5%) DKA
events were coded correctly. Only 18.5% (10/54) of the mixed presentation events and 66.7% (2/3) of the HHS events were
coded correctly. Of the 370 admissions coded as DKA, 88.4% were biochemically confirmed as DKA, and the remainder were
mixed presentation events. The odds of being incorrectly coded were 238.9 times greater for those with a mixed presentation
relative to DKA (P<.0001). In general, events coded incorrectly (mixed and DKA) had 3.1-fold greater odds of having
complications compared to those coded correctly (P = .02). Of the 5 DKA admissions coded incorrectly, all were coded as
uncomplicated diabetes, and none had complications. Of the 45 mixed presentation admissions coded incorrectly, 95.6% were
coded as DKA.

Discussion

At our institution, while DKA remains the most common hyperglycemic emergency among pediatric and adolescent subjects
and HHS was rare, mixed presentation accounted for 13.8% of characterized emergencies. The frequency of HHS was lower
than the 3% rate reported by Canarie et al[1] in a study of 10- to 30-year-old subjects. However, as several of the 20 subjects in
their HHS cohort had acidosis and ketonuria, mixed presentation may have been characterized as HHS in their analysis. To our
knowledge, our study is the first to report incidence of a mixed DKA/HHS presentation in pediatric diabetes.

The odds of complications were higher for those with a mixed presentation compared to those with DKA. The increased rate of
complications in those with mixed presentation may be related to the underlying dehydration and initial fluid deficit. ISPAD
guidelines recommend more-aggressive fluid resuscitation and a lower initial insulin infusion rate for children with HHS
compared to those with DKA,[2,3] and insufficient fluid administration during the first 6 hours of treatment for HHS is associated
with increased morbidity and mortality.[1] Treatment recommendations for children with mixed presentation are similar to those
of DKA but with greater intravenous fluid administration and close monitoring of circulation to ensure adequate hydration.[2,3]
Given the higher odds ratio of complications, identifying hyperosmolality early in assessment of a diabetes emergency is
essential for providers to recognize patients who may benefit from closer monitoring and more-aggressive treatment of
dehydration. The prevalence of AMS was greater among those with mixed presentation compared to those DKA, and the
increased frequency of AMS in those with mixed presentation was maintained even when controlling for degree of acidosis.
Similarly, an earlier study of 70 adult subjects with DKA showed that hyperosmolarity, rather than acidosis, best correlated with
AMS,[19] though other reports have identified acidosis as the primary factor in AMS.[18] Recognition of hyperosmolarity is
essential, as the patients with AMS require aggressive treatment that should be tailored to the underlying diagnosis.

Subjects with intellectual disability had increased odds of a hyperosmolar event, compared to those without intellectual
disability. An increased incidence of intellectual disability has been reported previously in pediatric patients who develop HHS.
[4,20] The association between intellectual disability and hyperosmolarity may be related to lack of an adequate thirst
mechanism, decreased access to fluids due to immobility, or inability to request fluids,[21] and inadequate fluid intake leads to
worsening dehydration and subsequent hyperosmolarity. Similarly, nursing home residents, a population also at risk of limited
access to fluids, limited mobility, and/or inadequate thirst mechanism, also have an increased prevalence of mixed presentation
or HHS compared to DKA.[6]

Multicenter studies of hyperglycemic emergencies in pediatrics frequently rely upon diagnosis codes,[5,9,10] and we sought to
assess the accuracy of these codes in our institution. The majority (79.6%) of mixed presentation events were incorrectly coded
as DKA, and therefore, they were likely treated as DKA. As patients with incorrectly coded events were 3.1-fold more likely to
develop complications, it is possible that suboptimal fluid administration contributed to the increased frequency of
complications. Unlike mixed presentation, DKA was almost always coded correctly. However, of those coded as DKA, only
88.4% had DKA, and the remainder had a mixed presentation. Thus, studies that rely solely upon diagnosis codes may miss
and underreport hyperosmolar events and include mixed presentation events in the DKA cohort. Further studies are needed to
validate our findings in other institutions.

Though this study has several limitations, including those inherent to retrospective chart reviews, we followed best-practice
guidelines for review of medical records to minimize errors.[22] To maximize identification of hyperglycemic emergencies, we

https://www.medscape.com/viewarticle/901782_print Page 7 of 9
 29/01/19 19.32

used two methods, an EMR search for diabetes-related and specific laboratory values and detailed review of medical records,
though we may have missed some events. Diabetes autoimmune panel information was available in a minority of patients, and
clinical assessments in the medical record were used to determine the diagnosis the remainder of cases.[23] Finally, our
hospital has the only pediatric intensive care unit in the state, so our results may not be generalizable to other institutions.
Additional studies across multiple institutions are needed to determine the incidence and optimal treatment of the mixed DKA-
HHS presentation in pediatric patients.

This study is the first to estimate the incidence of mixed presentation among pediatric hyperglycemic emergencies at a single
institution and demonstrated that mixed presentation is associated with a higher odds of complications compared to DKA.
Mixed presentation was often coded as DKA using ICD-9 codes, and if this reflects lack of recognition of hyperosmolarity, then
those with mixed presentation may have higher risk of complications due to inadequate fluid resuscitation. Recognition of
hyperosmolarity and mixed presentation is the first step to ensure appropriate treatment and monitoring for complications.
Prospective studies of hyperglycemic emergencies are needed to better understand the presentation, optimal treatment, and
outcomes in mixed HHS-DKA events.

Conclusion

A mixed presentation with features of DKA and HHS is a frequent yet underrecognized and underreported emergency in
pediatric diabetes. Identifying hyperosmolarity is important, as its treatment requires more-aggressive fluid administration
compared to treatment of DKA, and improper management may increase the risk of complications. Further studies on diagnosis
and management of mixed presentation across multiple institutions will help physicians optimally recognize and treat this
hyperglycemic emergency.

References

1. Canarie MF, Bogue CW, Banasiak KJ, Weinzimer SA, Tamborlane WV. Decompensated hyperglycemic hyperosmolarity
without significant ketoacidosis in the adolescent and young adult population. J Pediatr Endocrinol Metab.
2007;20:1115–1124.

2. Wolfsdorf JI, Allgrove J, Craig ME, et al. ISPAD Clinical Practice Consensus Guidelines 2014. Diabetic ketoacidosis and
hyperglycemic hyperosmolar state. Pediatr Diabetes. 2014;(15 suppl 20):154–179.

3. Zeitler P, Haqq A, Rosenbloom A, Glaser N. Hyperglycemic hyperosmolar syndrome in children: pathophysiological

considerations and suggested guidelines for treatment. J Pediatr. 2011;158:9–14, 14e11–14e12.

4. Rosenbloom AL. Hyperglycemic hyperosmolar state: an emerging pediatric problem. J Pediatr. 2010;156:180–184.

5. Bagdure D, Rewers A, Campagna E, Sills MR. Epidemiology of hyperglycemic hyperosmolar syndrome in children
hospitalized in USA. Pediatr Diabetes. 2013;14:18–24.

6. Wachtel TJ, Tetu-Mouradjian LM, Goldman DL, Ellis SE, O'Sullivan PS. Hyperosmolarity and acidosis in diabetes
mellitus: a three-year experience in Rhode Island. J Gen Intern Med. 1991;6:495–502.

7. MacIsaac RJ, Lee LY, McNeil KJ, Tsalamandris C, Jerums G. Influence of age on the presentation and outcome of
acidotic and hyperosmolar diabetic emergencies. Intern Med J. 2002;32:379–385.

8. Price A, Losek J, Jackson B. Hyperglycaemic hyperosmolar syndrome in children: Patient characteristics, diagnostic
delays and associated complications. J Paediatr Child Health. 2016;52:80–84.

9. Crossen SS, Wilson DM, Saynina O, Sanders LM. Outpatient care preceding hospitalization for diabetic ketoacidosis.
Pediatrics. 2016;137.

10. Tieder JS, McLeod L, Keren R, et al. Variation in resource use and readmission for diabetic ketoacidosis in children's
hospitals. Pediatrics. 2013;132:229–236.

11. von Oettingen J, Wolfsdorf J, Feldman HA, Rhodes ET. Use of serum bicarbonate to substitute for venous ph in new-
onset diabetes. Pediatrics. 2015;136:e371–377.

12. Malatesha G, Singh NK, Bharija A, Rehani B, Goel A. Comparison of arterial and venous pH, bicarbonate, PCO2 and
PO2 in initial emergency department assessment. Emerg Med J. 2007;24:569–571.

13. Kitabchi AE, Umpierrez GE, Murphy MB, et al. Hyperglycemic crises in diabetes. Diabetes Care. 2004;27(suppl 1):S94–
S102.

14. Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research electronic data capture (REDCap)--a
metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed
Inform. 2009;42:377–381.

https://www.medscape.com/viewarticle/901782_print Page 8 of 9
 29/01/19 19.32

15. American Diabetes Association. 2. Classification and diagnosis of diabetes. Diabetes Care. 2016;39(suppl 1):S13–S22.

16. Zeitler P, Fu J, Tandon N, et al. ISPAD Clinical Practice Consensus Guidelines 2014. Type 2 diabetes in the child and
adolescent. Pediatr Diabetes. 2014;(15 suppl 20):26–46.

17. Craig ME, Jefferies C, Dabelea D, et al. ISPAD Clinical Practice Consensus Guidelines 2014. Definition, epidemiology,
and classification of diabetes in children and adolescents. Pediatr Diabetes. 2014;(15 suppl 20):4–17.

18. Nyenwe EA, Razavi LN, Kitabchi AE, Khan AN, Wan JY. Acidosis: the prime determinant of depressed sensorium in
diabetic ketoacidosis. Diabetes Care. 2010;33:1837–1839.

19. Fulop M, Tannenbaum H, Dreyer N. Ketotic hyperosmolar coma. Lancet. 1973;2:635–639.

20. Fourtner SH, Weinzimer SA, Levitt Katz LE. Hyperglycemic hyperosmolar non-ketotic syndrome in children with type 2
diabetes*. Pediatr Diabetes. 2005;6:129–135.

21. Rubin HM, Kramer R, Drash A. Hyperosmolality complicating diabetes mellitus in childhood. J Pediatr. 1969;74:177–
186.

22. Gilbert EH, Lowenstein SR, Koziol-McLain J, Barta DC, Steiner J. Chart reviews in emergency medicine research: where
are the methods? Ann Emerg Med. 1996;27:305–308.

23. von Oettingen JE, Wolfsdorf JI, Feldman HA, Rhodes ET. Utility of diabetes-associated autoantibodies for classification
of new onset diabetes in children and adolescents. Pediatr Diabetes. 2016;17:417–425.

Abbreviations
AMS = altered mental status; DKA = diabetic ketoacidosis; EMR = electronic medical record; HHS = hyperglycemic
hyperosmolar state; ICD-9 = International Classification of Diseases, Ninth Revision; ISPAD = International Society of Pediatric
and Adolescent Diabetes; NODM = new-onset diabetes mellitus; T1DM = type 1 diabetes mellitus; T2DM = type 2 diabetes
mellitus

Endocr Pract. 2018;24(8):726-732. © 2018 American Association of Clinical Endocrinologists

https://www.medscape.com/viewarticle/901782_print Page 9 of 9