OBJECTIVE

The etiology of altered sensorium in diabetic ketoacidosis (DKA) remains unclear. Therefore, we sought to determine the origin of depressed consciousness in DKA.

RESEARCH DESIGN AND METHODS

We analyzed retrospectively clinical and biochemical data of DKA patients admitted in a community teaching hospital.

RESULTS

We recorded 216 cases, 21% of which occurred in subjects with type 2 diabetes. Mean serum osmolality and pH were 304 ± 31.6 mOsm/kg and 7.14 ± 0.15, respectively. Acidosis emerged as the prime determinant of altered sensorium, but hyperosmolarity played a synergistic role in patients with severe acidosis to precipitate depressed sensorium (odds ratio 2.87). Combination of severe acidosis and hyperosmolarity predicted altered consciousness with 61% sensitivity and 87% specificity. Mortality occurred in 0.9% of the cases.

CONCLUSIONS

Acidosis was independently associated with altered sensorium, but hyperosmolarity and serum “ketone” levels were not. Combination of hyperosmolarity and acidosis predicted altered sensorium with good sensitivity and specificity.

Diabetic ketoacidosis (DKA) is frequently associated with altered mental status, which is correlated with the severity of the disease and prognosis (1). However, the etiology of depressed sensorium in DKA remains uncertain and controversial (2,3). Putative factors in the pathogenesis of diabetic ketoacidotic coma include cerebral hypoperfusion due to circulatory collapse and cerebral thrombosis (4), reduced cerebral glucose and oxygen utilization (1,5), acidosis (6,7), hyperosmolarity (8,9), and direct toxic effect of ketone bodies (2). Cerebral edema remains an important precipitant of altered consciousness in DKA, especially in children.

Different studies have yielded conflicting results regarding the role of these etiologic factors in the pathogenesis of altered mentation in patients with DKA. Hence, the origin of clouded sensorium in DKA remains to be fully elucidated. We undertook to study the etiology of depressed consciousness in patients admitted with DKA at the Regional Medical Center in Memphis.

We performed a retrospective analysis of patients admitted with DKA between October 2003 and October 2008. DKA was diagnosed based on American Diabetes Association criteria (10). Demographic and biochemical data were extracted from the medical records of the patients including serum pH, bicarbonate, sodium, potassium, blood urea nitrogen (BUN), serum creatinine, osmolality, “ketones,” arterial pO2 and pCO2, as well as level of consciousness. Serum effective osmolality was calculated as reported before (10).

Patients were categorized by mental status into alert versus altered (semi-comatose and comatose).

Statistical analysis

For comparison of variables among groups, univariate analysis was performed. To identify the independent determinants of level of consciousness, multivariate linear model analysis was applied. The groups were characterized using means ± SD and compared by Student's t test for continuous variables and χ2 test for discrete variables using SAS 9.1.3 software. Using quartiles of pH and serum osmolality as predictors of depressed consciousness, we generated a receiver operating characteristic curve from which we derived the sensitivity and specificity of these independent variables in predicting the presence of depressed sensorium in patients presenting with DKA.

Quartiles of pH were as follows: 6.92–7.03, 7.04–7.15, 7.16–7.25, and 7.26–7.40, while quartiles of serum osmolality were as follows: 231–287, 287.4–301, 301.2–315, and 316–431.

The Institution Review Board of the University of Tennessee Health Science Center approved this study.

We recorded 216 cases of DKA in patients aged 37.7 ± 12.4 years, 57.9% of whom were males, whereas 21 and 15% had type 2 diabetes and new-onset diabetes, respectively. Average blood glucose level was 658.6 ± 276.0 mg/dl, whereas mean serum osmolality and pH were 304 ± 31.6 mOsm/kg and 7.14 ± 0.15, respectively. Serum sodium and potassium were 131.6 ± 6.5 and 5.4 ± 3.5 mEq, respectively. The clinical and biochemical characteristics of the patients according to mental status are shown in Table 1. Variables that demonstrated a significant difference between the two groups were entered into a multiple regression model shown in supplemental Table 1 (available in an online appendix at http://care.diabetesjournals.org/cgi/content/full/dc10-0102/DC1). As can be seen, only pH emerged as an independent determinant of level of consciousness in this cohort of patients with DKA.

Table 1

Clinical and biochemical characteristics of the patients at presentation

Mental status
P
AlertAltered
n 133 83  
Age (years) 37.1 ± 13.1 38.6 ± 11.5 0.41 
Sex   0.05 
    Male 70 55  
    Female 63 28  
Race   0.39 
    African American 116 71  
    Caucasian 16  
    Hispanic  
    Others  
Type of diabetes   0.12 
    Type 1 94 45  
    Type 2 24 21  
    New-onset 24  
Systolic blood pressure 130 ± 21 122 ± 27 0.022* 
Diastolic blood pressure 73 ± 18 70 ± 18 0.023* 
Leucocytes 13.4 ± 6.7 17.1 ± 8.6 0.001* 
Glucose 603 ± 236 747 ± 311 0.0004* 
Serum ketones   0.06 
    Mild 46 20  
    Moderate 68 43  
    Large 16 19  
Blood pH 7.19 ± 0.10 7.05 ± 0.16 <0.0001* 
Serum bicarbonate 11.8 ± 4.3 8.8 ± 4.7 <0.0001* 
Anion gap 22.8 ± 6.7 29.8 ± 32.4 0.062 
Serum osmolality 300.1 ± 18.0 310.8 ± 45.0 0.041* 
Serum sodium 131.8 ± 6.2 131.2 ± 6.9 0.49 
Serum potassium 5.1 ± 1.2 6.0 ± 5.5 0.14 
Blood urea nitrogen 23.8 ± 16.8 33.8 ± 22.6 0.0007* 
Serum creatinine 2.3 ± 1.1 3.1 ± 2.3 0.003* 
pCO2 50.0 ± 35.4 47.4 ± 38.5 0.61 
pO2 79.1 ± 52.8 88.3 ± 65.3 0.28 
Mental status
P
AlertAltered
n 133 83  
Age (years) 37.1 ± 13.1 38.6 ± 11.5 0.41 
Sex   0.05 
    Male 70 55  
    Female 63 28  
Race   0.39 
    African American 116 71  
    Caucasian 16  
    Hispanic  
    Others  
Type of diabetes   0.12 
    Type 1 94 45  
    Type 2 24 21  
    New-onset 24  
Systolic blood pressure 130 ± 21 122 ± 27 0.022* 
Diastolic blood pressure 73 ± 18 70 ± 18 0.023* 
Leucocytes 13.4 ± 6.7 17.1 ± 8.6 0.001* 
Glucose 603 ± 236 747 ± 311 0.0004* 
Serum ketones   0.06 
    Mild 46 20  
    Moderate 68 43  
    Large 16 19  
Blood pH 7.19 ± 0.10 7.05 ± 0.16 <0.0001* 
Serum bicarbonate 11.8 ± 4.3 8.8 ± 4.7 <0.0001* 
Anion gap 22.8 ± 6.7 29.8 ± 32.4 0.062 
Serum osmolality 300.1 ± 18.0 310.8 ± 45.0 0.041* 
Serum sodium 131.8 ± 6.2 131.2 ± 6.9 0.49 
Serum potassium 5.1 ± 1.2 6.0 ± 5.5 0.14 
Blood urea nitrogen 23.8 ± 16.8 33.8 ± 22.6 0.0007* 
Serum creatinine 2.3 ± 1.1 3.1 ± 2.3 0.003* 
pCO2 50.0 ± 35.4 47.4 ± 38.5 0.61 
pO2 79.1 ± 52.8 88.3 ± 65.3 0.28 

*Statistically significant.

To further elucidate the influence of acidosis measured by arterial pH and effective serum osmolality on the level of consciousness at presentation, we determined the odds ratio of depressed sensorium in all 216 subjects by quartiles of pH and osmolality (see supplemental Fig. 1 in the online appendix). Supplemental Fig. 1 clearly demonstrates that arterial pH is the prime determinant of mental status in DKA, which appeared to act synergistically with hyperosmolarity to produce depressed consciousness. In patients with severe acidosis (lowest pH quartile ≤7.03), the odds ratio for altered sensorium was 1 if serum osmolality was in the lowest quartile (≤287), but increased almost threefold to 2.8 among patients in the highest quartile of osmolality (≥316). However, in the absence of severe acidosis, worsening hyperosmolarity was not significantly associated with altered sensorium (odds ratio <1). Therefore, although acidosis was required for depressed sensorium to occur, the likelihood of altered mentation was higher in the presence of elevated serum osmolality. A total of 45 patients (20.8%) were in the lowest pH quartile, while 52 patients (24.1%) were in the highest osmolality quartile (≥316 mOsm/kg). Fifteen subjects (6.9%) had a combination of low pH and high osmolality. This latter category represented the patients with the highest likelihood of having depressed level of consciousness (odds ratio 2.81). Mortality, which was recorded in 0.9% of the cases, occurred in patients who had presented with severe acidosis and sepsis. The combination of severe acidosis and hyperosmolarity predicted altered sensorium with sensitivity of 61% and specificity of 87% (supplemental Table 2 and supplemental Fig. 2).

We have demonstrated that acidosis measured by arterial pH at presentation is the prime determinant of mental status in DKA. Severe acidosis appeared to act in concert with hyperosmolarity in a synergistic manner to produce depressed sensorium in these patients. Therefore, a combination of acidosis and hyperosmolarity at presentation may identify a subset of patients with severe DKA (7% in this study) who may benefit from more aggressive treatment and monitoring. Identifying this class of patients, who are at a higher risk for poorer prognosis, may be helpful in triaging them, thus further improving the outcome.

The dominant effect of acidosis on mental status obtained in this study is consistent with the findings of other clinical studies (6,7).

Some studies have reported that depressed sensorium in DKA is attributable to elevated serum osmolality (8,9). However, despite some biochemical similarities between these two hyperglycemic crises, there are notable differences that may suggest different pathogenic mechanisms. We observed that hyperosmolarity, which was associated with depressed sensorium in a univariate model, did not prove to be an independent predictor of altered sensorium in a multivariate analysis. However, in the presence of hypertonicity, the likelihood of altered mental status was increased in patients with severe acidosis. Studies that reported hyperosmolarity as the sole etiology of altered consciousness DKA did not perform logistic regression (8) as we have done. This could be a confounding factor, since up to 30% of patients with DKA may have high serum osmolality (9,11).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

No potential conflicts of interest relevant to this article were reported.

E.A.N.: research design, data collection, statistical analysis, manuscript writing, and review of manuscript. L.N.R.: data extraction and collation, statistical analysis, and manuscript writing. A.E.K.: research design, manuscript writing, and critical review of manuscript for intellectual content. A.N.K.: data extraction and collation and manuscript writing. J.Y.W.: statistical analysis and manuscript writing.

We acknowledge the efforts of the following residents in internal medicine and fellows in endocrinology who assisted with data collection: Jevaria Ahmed, MD, Swaroopa Bartakke, MD, Malini Gupta-Ganguli, MD, Omid Rad Pour, MD, Amanda Long, DO, and Sadiya Zaidi, MD.

1.
Kety
SS
,
David Polis
B
,
Dadler
CS
:
The blood flow and oxygen consumption of the human brain in diabetic acidosis and coma
.
J Clin Invest
1948
;
27
:
500
510
2.
Fisher
P
:
The role of ketone bodies in the etiology of diabetic coma
.
Am J Med Sci
1951
;
221
:
384
397
3.
Alberti
KG
,
Nattrass
M
:
Severe diabetic ketoacidosis
.
Med Clin North Am
1978
;
62
:
799
814
4.
Guisado
R
,
Arieff
AI
:
Neurologic manifestations of diabetic comas: correlation with biochemical alterations in the brain
.
Metabolism
1975
;
24
:
665
679
5.
Kuschinsky
W
,
Suda
S
,
Sokoloff
L
:
Local cerebral glucose utilization and blood flow during metabolic acidosis
.
Am J Physiol
1981
;
241
:
H772
H777
6.
Rosival
V
:
The influence of blood hydrogen ion concentration on the level of consciousness in diabetic ketoacidosis
.
Ann Clin Res
1987
;
19
:
23
25
7.
Edge
JA
,
Roy
Y
,
Bergomi
A
,
Murphy
NP
,
Ford-Adams
ME
,
Ong
KK
,
Dunger
DB
:
Conscious level in children with diabetic ketoacidosis is related to severity of acidosis and not to blood glucose concentration
.
Pediatric Diabetes
2006
;
7
:
11
15
8.
Fulop
M
,
Rosenblatt
A
,
Kreitzer
SM
,
Gerstenhaber
B
:
Hyperosmolar nature of diabetic coma
.
Diabetes
1975
;
24
:
594
599
9.
Kitabchi
AE
,
Fisher
JN
:
Insulin therapy of diabetic ketoacidosis: physiologic versus pharmacologic doses of insulin and their routes of administration
. In
Handbook of Diabetes Mellitus
.
Brownlee
M
: Ed.
New York
,
Garland ATPM Press
,
1981
, p.
95
149
10.
Kitabchi
AE
,
Umpierrez
GE
,
Miles
JM
,
Fisher
JN
:
Hyperglycemic crises in adult patients with diabetes
.
Diabetes Care
2009
;
32
:
1335
1343
11.
Nyenwe
E
,
Loganathan
R
,
Blum
S
,
Ezuteh
D
,
Erani
D
,
Palace
M
,
Ogugua
C
:
Admissions for diabetic ketoacidosis in ethnic minority groups in a city hospital
.
Metabolism
2007
;
56
:
172
178
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. See http://creativecommons.org/licenses/by-nc-nd/3.0/ for details.

Supplementary data