Severe Diabetic Ketoacidosis (Diabetic "Coma ) - Diabetes
Severe Diabetic Ketoacidosis (Diabetic "Coma 95 ) 482 Episodes in 257 Patients; Experience of Three Years Paul M. Beigelman, M.D., Los Angeles SUMMARY There were 340 episodes of severe diabetic ketoacidosis in 257 patients, July 1, 1965 to June 30, 1968. Fatality in thirty-two cases was associated with significantly elevated admission serum glucose, urea nitrogen, osmolality and age. Major associated causes of death were infection and myo- cardial infarction. Twenty episodes of acute pancreatitis, with only two deaths, demonstrated severe morbidity and marked elevation of serum glucose and urea nitrogen. Four- teen of the twenty occurred in alcoholics. There were eleven episodes in ten pregnant women, with one maternal and seven fetal deaths. Admission serum potassium, recorded in 336 episodes, was markedly elevated in 22 per cent and decreased in 4 per cent. Including all multiple admissions, there were 482 episodes; forty had hypoglycemic reactions, none with fatality or severe morbidity. Nine cases of non- ketotic hyperosmolar coma, with five deaths, demonstrated very high admission serum glucose, urea nitrogen, and osmo- lality levels. DIABETES 20:490-500, July, 1971. Severe diabetic ketoacidosis is defined by the Diabetes Service of LAC-USC Medical Center* as a state of keto- acidosis with serum bicarbonate decreased to levels of io mEq./L or less. This degree of diabetic ketoacidosis, even without mental obtundation, has been designated by some authorities as "diabetic coma". 1 Admission data on 340 sequential episodes of severe diabetic ketoacidosis in 257 patients admitted between July 1, 1965 and June 30, 1968 are presented. Particu- lar emphasis is given to factors leading to death in thirty-two of these patients. None of these fatalities was From the Departments of Medicine and Pharmacology, School of Medicine, University of Southern California, Los Angeles, California. *Los Angeles County-University of Southern California Med- ical Center. caused by cerebral edema. Also, certain conditions asso- ciated with severe diabetic ketoacidosis are discussed, in- cluding myocardial infarction, alcoholism, pancreatitis, therapy with oral agents, and hypoglycemia. Additional admissions, numbering 142, occurred in certain of the patients who had multiple admissions for severe diabetic ketoacidosis before or after the prin- cipal three-year study period. Inclusion of these yielded a total of 482 episodes in the 257 patients, if time limits are ignored. These data are utilized to ascertain the effect of multiple admissions upon mortality rate. Nine cases of hyperosmolar nonketotic coma, admitted to the Diabetes Service between July 1, 1965 and June 30, 1968, -are also presented for comparison with the ketoacidotic patients. The therapeutic technics utilized by this Diabetes Ser- vice are based on careful monitoring of vital signs and chemical status. Individualized, vigorous therapy is em- phasized with large intravenous doses of insulin ad- ministered in the earlier phases of treatment, followed by smaller subcutaneous doses until ketosis is corrected. Deficits of water, sodium, and potassium may be pro- found, requiring rapid and aggressive parenteral replace- ment therapy. Large doses of parenteral potassium, care- fully monitored, also may be necessary. Shock, infec- tions, and hypoglycemia must be diagnosed immediately and treated vigorously. These therapeutic approaches are fully presented in a separate communication 2 which also summarizes briefly fifty-one episodes of severe diabetic ketoacidosis in forty-six patients admitted between Janu- ary 1 and April 30, 1968. The fifty-one episodes sum- marized in the treatment paper are included in the present report. RESULTS 1. Findings on admission The age range in this series was fourteen to eighty- 490 DIABETES, VOL. 20, NO. 7 Downloaded from http://diabetesjournals.org/diabetes/article-pdf/20/7/490/346430/20-7-490.pdf by guest on 01 January 2022
Severe Diabetic Ketoacidosis (Diabetic "Coma ) - Diabetes
Text of Severe Diabetic Ketoacidosis (Diabetic "Coma ) - Diabetes
482 Episodes in 257 Patients; Experience of Three Years
Paul M. Beigelman, M.D., Los Angeles
There were 340 episodes of severe diabetic ketoacidosis in 257
patients, July 1, 1965 to June 30, 1968. Fatality in thirty-two
cases was associated with significantly elevated admission serum
glucose, urea nitrogen, osmolality and age. Major associated causes
of death were infection and myo- cardial infarction. Twenty
episodes of acute pancreatitis, with only two deaths, demonstrated
severe morbidity and marked elevation of serum glucose and urea
nitrogen. Four- teen of the twenty occurred in alcoholics. There
were eleven episodes in ten pregnant women, with one maternal and
seven fetal deaths. Admission serum potassium, recorded in 336
episodes, was markedly elevated in 22 per cent and decreased in 4
per cent. Including all multiple admissions, there were 482
episodes; forty had hypoglycemic reactions, none with fatality or
severe morbidity. Nine cases of non- ketotic hyperosmolar coma,
with five deaths, demonstrated very high admission serum glucose,
urea nitrogen, and osmo- lality levels. DIABETES 20:490-500, July,
Severe diabetic ketoacidosis is defined by the Diabetes Service of
LAC-USC Medical Center* as a state of keto- acidosis with serum
bicarbonate decreased to levels of io mEq./L or less. This degree
of diabetic ketoacidosis, even without mental obtundation, has been
designated by some authorities as "diabetic coma".1
Admission data on 340 sequential episodes of severe diabetic
ketoacidosis in 257 patients admitted between July 1, 1965 and June
30, 1968 are presented. Particu- lar emphasis is given to factors
leading to death in thirty-two of these patients. None of these
From the Departments of Medicine and Pharmacology, School of
Medicine, University of Southern California, Los Angeles,
*Los Angeles County-University of Southern California Med- ical
caused by cerebral edema. Also, certain conditions asso- ciated
with severe diabetic ketoacidosis are discussed, in- cluding
myocardial infarction, alcoholism, pancreatitis, therapy with oral
agents, and hypoglycemia.
Additional admissions, numbering 142, occurred in certain of the
patients who had multiple admissions for severe diabetic
ketoacidosis before or after the prin- cipal three-year study
period. Inclusion of these yielded a total of 482 episodes in the
257 patients, if time limits are ignored. These data are utilized
to ascertain the effect of multiple admissions upon mortality
Nine cases of hyperosmolar nonketotic coma, admitted to the
Diabetes Service between July 1, 1965 and June 30, 1968, -are also
presented for comparison with the ketoacidotic patients.
The therapeutic technics utilized by this Diabetes Ser- vice are
based on careful monitoring of vital signs and chemical status.
Individualized, vigorous therapy is em- phasized with large
intravenous doses of insulin ad- ministered in the earlier phases
of treatment, followed by smaller subcutaneous doses until ketosis
is corrected. Deficits of water, sodium, and potassium may be pro-
found, requiring rapid and aggressive parenteral replace- ment
therapy. Large doses of parenteral potassium, care- fully
monitored, also may be necessary. Shock, infec- tions, and
hypoglycemia must be diagnosed immediately and treated vigorously.
These therapeutic approaches are fully presented in a separate
communication2 which also summarizes briefly fifty-one episodes of
severe diabetic ketoacidosis in forty-six patients admitted between
Janu- ary 1 and April 30, 1968. The fifty-one episodes sum- marized
in the treatment paper are included in the present report.
1. Findings on admission The age range in this series was fourteen
4 9 0 DIABETES, VOL. 20, NO. 7
by guest on 01 January 2022
PAUL M. BEIGELMAN, M.D.
four years. The mean age of thirty-eight was within the range of
mean ages, twenty-nine to forty-two, reported in other series.3'8
The mean admission serum glucose levels reported by others,3'8
371-597 mg./ioo cc, were less than the mean value of 705 mg./ioo
cc. observed in this institution. The mean admission serum urea
nitro- gen in this series, 35 mg./ioo cc, was slightly higher than
mean values of 30-33 mg./ioo cc. in other se- ries.6'8'9' A minor
factor possibly contributing to these discrepancies is the use, in
previous smdies, of whole blood for glucose and urea
determinations, which are lower than equivalent serum values. Mean
admission serum potassium levels reported by others, 4.6-5.7 mEq./
L,3-5'9 were in the same range as the mean potassium of this
series, 5.3 mEq./L. The mean admission serum sodium in this
institution, 131 mEq./L., was within the range reported elsewhere,
131-138 mEq./L. 3-r>-9 In earlier reports, mean admission
leucocyte counts were 16,900-21,900 per cu. mm.,3-15'8 closely
agreeing with the value observed in this series, 18,387 per cu. mm.
The patients of Zieve and Hill6 had a mean admis- sion hemoglobin
of 13.9 gm. per cent, slightly lower than the 14.7 gm. per cent
observed in this series. Mean admission respiratory rate of
30/min., pulse rate of 115/min., and temperature of 98.5 ° F. are
in good agreement with values of other series.3'6
2. Causes Many of the episodes reported in this series of
had multiple causes, and frequently the cause was un- known.
Discontinuance of therapy and infection were the two most common
causes of severe diabetic keto- acidosis in this and other
series.1'3'4'6'8'9 The third most common cause in this series was
diabetes mellitus not being diagnosed before admission. This group
consti- tuted 28 per cent of fatalities, 13 per cent of nonfatal
episodes (15 per cent of total episodes), and nearly 20 per cent of
the 257 subjects admitted during the three-year study period. Other
institutions have reported a high frequency (10 to 35 per cent) of
diabetes mel- litus being diagnosed at time of admission for
3. Treatment (table 1) The mean dose of insulin administered in
(540 units) was within the widely variable twelve to twenty-four
hour range of others, 148-1,280 units.3"5'8'9
The smallest insulin dose (20 units) was given in a puzzling case
with definite hyperglycemia and ketoacid- osis, excellent response
to parenteral bicarbonate therapy and diet, and no evidence of
diabetes mellitus at time of discharge. The highest dose was 10,700
units in a fatal case complicated by pneumonia and
Mean fluid volume infused was 6 L, the range being 0-18. Mean
sodium given was 530 mEq., the range being 0-620. Mean potassium
administered was 165 mEq., the range being 0-620.
The amounts of insulin, fluid and electrolytes given were designed
in each instance to relieve ketosis and correct fluid and
electrolyte deficits. Correction of ketosis (disappearance of
ketonuria) occurred, on the average, in seventeen hours (range
three to forty-four hours). 4. Characteristics of fatal cases
Fatal and nonfatal cases of severe diabetic ketoacidosis admitted
July 1, 1965 to June 30, 1968 are compared in table 2.
During the three-year period, 32 of the 340 episodes (9 per cent)
terminated fatally (death in thirty-one days or less after
admission). The patient mortality rate, 32 of 257, was 13 per cent.
There were more episodes of severe diabetic ketoacidosis in women
than in men, but the male mortality rate was greater. There was no
relation of mortality to presence or absence of a family history of
diabetes. Mortality was higher with increased age and with increase
of serum glucose, urea nitrogen, and osmolality. Similar
associations, particularly with age, have been noted in other
The mean pulse rate and diastolic blood pressure were lower in
fatal cases. The depression of pulse rate was probably due to
greater frequency of bradycardia, less than 50/min., or cardiac
standstill. The lower mean diastolic blood pressure in the fatal
cases probably re- flected the increased frequency of severe
hypotension and shock in such patients.
Clinical coma, characterized by the patient's not re- sponding to
pain or to loud voice commands, was more frequent in fatal cases. A
relationship between uncon- sciousness and mortality has been
frequently re- ported.1'4'6-8-9'11'14"17 Severe hypotension,
designated as a diastolic blood pressure of zero, was related to
TABLE 1 Summary of therapy
Therapy Insulin (units) Potassium (mEq.) Fluid (liters) Sodium
(mEq.) Time for correction of
*To nearest five units or 5 mEq. tTo nearest 0.5 L. tTo nearest
JULY, 197 I 491
995±72 7±1 1/16
62±6 5.4±0.2 132±2 320±98 357±6
29±1 100±7 110±10 58±5
291 302 302 302
32±1 5.3±0.1 131±0 242±51 323±1
30±l 117±1 125±1 71 ±1
* Number of episodes. t Standard error of mean. tPer cent of total
fatal or total nonfatal. § Probability of significance of
difference between means for clinical and chemical findings in
fatal and nonfatal "coma"
episodes. Computed by X2 or Student f-test. "P" values from
Fisher's Tables (Fisher and Yates Statistical Tables, Hatner
Publishing Co., New York, 1953).
Serum glucose (mg./100 cc.) Serum nitrogen (mg./lOO cc.) ||2(Na+ +
K+) + + **Not significant (N.S.).
in severe diabetic ketoacidosis, but moderate hypotension,
arbitrarily designated as a systolic pressure of less than ioo
mm./Hg and diastolic blood pressure between o and 50 mm./Hg, was
not. An association between hypotension and mortality has been
reported by others.4-6-8'11-14-16
Among the nonfatal cases in this series, mean serum glucose levels
were significantly lower in individuals with thirteen to
twenty-eight episodes each compared to those admitted only once in
severe diabetic keto- acidosis (table 3). Most deaths occurred
during the first or second admission. The much higher survival rate
of individuals with multiple admissions probably re-
flects their relative youth. Younger, "brittle" diabetics more
frequently endure episodes of diabetic ketoacidosis which may be
severe as measured by degree of acidosis, but are usually
characterized by mild hyperglycemia, little or no azotemia, and few
complications. 5. Survival time and conditions associated with
The time intervals between admission and death in the thirty-two
fatal cases varied from one and one-half hours to thirty-one days.
Six died within six hours of admis- sion; four, seven to twelve
hours after admission; eight, within twelve to forty-eight hours;
seven, within two to seven days; and seven, within seven to
492 DIABETES, VOL. 20, NO. 7
by guest on 01 January 2022
per individual 1 2 3 4 5 6 7 9
10 13 19 28
PAUL M. BEIGELMAN, M.D.
occurring in an individual.
Number of individuals
183 36 13 9 6 1 2 1 2 2 1 1
Nonfatal Number of
6 14 9
Mean±S.E.M 710±22 708±47 568±28 650±46 709 ±44 613±115 654±76
791±167 766±67 537±37 503±32 361±34
glucose levels episodes in 257
<.001 <.001 <.001
*Mg./100 cc. tStandard error of mean. ^Probability of significance
of difference between means of multiple "coma" episodes per
individual compared with mean
of single episode per individual. Computed by Student /-test. "P"
from Fisher's Tables. §Probability of difference between means of
death and survival being significant. Computed by Student /-test.
The two major contributing factors in death were acute myocardial
infarction and infection (table 4 ) . Bacter- emia with sepsis
occurred in three subjects wfth pneu- monia, and in the single case
of pyelonephritis. Four subjects were diagnosed as dying of
uncomplicated severe diabetic ketoacidosis. The uncomplicated
mortality rate of severe diabetic ketoacidosis may be computed as
four of 340 episodes (1 per cent) or four of 257 subjects (1.5 per
The importance of associated complications in the mortality of
severe diabetic ketoacidosis has been fre-
TABLE 4 Major contributing causes of death in thirty-two
fatal cases of severe diabetic ketoacidosis
Coronary artery disease: Myocardial infarction Occlusive coronary
Infection: Pneumonia or pneumonitis Pyelonephritis with
* Questionable as to cause of death.
quently emphasized.1'3'4-6"15'18'19 Also, many of the fatal cases
in this series had extremely complicated courses with multiple
factors contributing to death. A detailed study of these deaths,
including pathological findings, is being prepared. 6. Myocardial
Seven of eleven subjects admitted with acute myocar- dial
infarction and severe diabetic ketoacidosis expired. Autopsies
performed on six of the seven fatal cases confirmed the diagnosis
of myocardial infarction. Mean admission serum osmolality, age,
admission serum glu- cose, urea nitrogen, and sodium values were
higher in fatal cases (table 5). Two of the seven fatal cases dem-
onstrated marked hypotension, with unobtainable dia- stolic blood
pressure. In three of the seven fatal cases, death occurred within
two hours of admission. One pa- tient, with evidence of lactic
acidosis in addition to keto- sis, died twelve days after
admission. It is possible that ketosis was not a major factor
contributing to the acido- sis and death in this person.
Death in seven of eleven subjects with myocardial infarction
emphasizes the particularly ominous prognosis of this complication.
It should be emphasized also that the four subjects who survived
were very ill.
The association of vascular thrombosis and death in diabetic
ketoacidosis has been pointed out in particular by Fitzgerald et
al.20 Myocardial infarction with shock in diabetics, even without
ketoacidosis, is associated with a very poor prognosis.21 Vascular
stasis and hemoconcen- tration secondary to marked hyperglycemia
JULY, 1971 493
TABLE 5 Acute myocardial infarction associated with severe diabetic
Age Serum glucose (mg./lOO ml.) Serum urea nitrogen (mg. per cent)
Serum potassium (mEq./L.) Serum sodium (mEq./L.) Serum
No.* 7 7 7 7 7 7
No. 4 4 4 4 4 4
*Number of episodes. tStandard error of mean.
osmolarity could be important factors in the genesis of coronary
arterial thrombosis associated with severe dia- betic ketoacidosis.
If vascular stasis is important, it is puzzling, then, that
pulmonary embolism is seen so rarely. There was only one fatal case
in this series, and it is questionable whether pulmonary embolism
was the cause of death in this individual. 7. Alcoholism and
Forty-two alcoholic patients were admitted with fifty- three
episodes of severe diabetic ketoacidosis (table 6 ) . In fourteen
episodes there was presumed acute pancrea- titis, including one
fatality, with typical abdominal find- ings and markedly elevated
serum amylase levels. The diagnosis of acute pancreatitis may be
difficult, the ele-
• vated serum amylase level (above 300 Somogyi units) reflecting a
possible "chemical" pancreatitis with no overt pathological
changes. The diagnosis was provi- sionally accepted or excluded on
the basis of the highest serum amylase level, which was usually the
admission value. These cases were associated with marked eleva-
tions of serum glucose, urea nitrogen, and amylase.
Sixteen of the forty-two alcoholic subjects had no previous
diagnosis of diabetes mellitus, a higher fre- quency than among
nonalcoholics admitted in severe diabetic ketoacidosis. There were
definite histories of hypoglycemia in seven, acute or chronic
fifteen, and pancreatic calcification was noted in roent- genograms
of two patients. Acute delirium tremens was observed during five
episodes. Three of six alcoholic subjects with evidence of hepatic
cirrhosis died shortly after admission. All four alcoholic subjects
with severe infection expired.
There were six additional cases of presumed acute pancreatitis in
nonalcoholic subjects. Mean admission values for all twenty cases
of pancreatitis, including the alcoholics, were: age, forty years;
serum glucose, 909 mg. / ioo cc; urea nitrogen, 50 mg. / ioo cc;
amylase, 1,067 Somogyi units; potassium, 5.1 mEq./L; sodium, 125
mEq./L; and osmolality, 324. Chemical data of al- coholic compared
to nonalcoholic subjects with acute pancreatitis was not markedly
different. There was a 10 per cent mortality rate, two of twenty
expiring. One of the two had an exceedingly complex course with
pneumonia, and severe insulin resistance playing im- portant roles
in his demise. Other institutions have em- phasized the severe
morbidity of acute pancreatitis, which was encountered in this
series. The mortality rate in this series was considerably lower
than that reported by others.22"25
Five patients developed severe ketoacidosis after sur- gery. The
procedures were one embolectomy, two leg
TABLE 6 Severe diabetic ketoacidosis associated with alcoholism (53
Age (years) Serum glucose (mg./lOO ml.) Serum urea nitrogen (mg.
per cent) Serum potassium (mEq./L.) Serum sodium (mEq./L.) Serum
amylase (Somogyi units) Osmolality Insulin therapy (units)
Fatal (6)* .t Mean+S.E.M.t
49+3 870 + 170
57 + 13 4.6+0.6 128+5 339 334 + 14 423 + 110
Pancreatitis (14)* No.
Mean +S.E.M. 42+3
320+7 745 + 113
Nonfatal (34) No. Mean +S.E.M. 34 34 34 34 33 21 33 34
41+2 657+49 33+3 5.5+0.2 130±l 119 + 18 325+4 423+50
*Fatal episode of pancreatitis included in both fatal and
pancreatitis series. tNumber of episodes. tStandard error of
494 DIABETES, VOL. 20, NO. 7
by guest on 01 January 2022
Urea Serum Potas- nitro-
glucose HCO3 sium gen Age (mg./ (mEq./ (mEq./ (mg./ (yrs.) 28
100 cc.) 430 330
100 c 8 10
Ace- Dura- tone Insulin tion of (dilu- Sodium dosage pregnancy
tion) (mEq./L.) (units) (months)
Comment Hyperemesis gravidarum. Fetal demise at six months.
Discharged three weeks after "coma." Fetus probably dead on
admission. Cesarean section two months later. Viable eight-month
old male. Complete abortion. Mediasti- nal emphysema. Survival of
four-pound pre- mature newborn. Cesarean section and hyste-
rectomy. Fetal death. Spontaneous labor. Death of fetus. Stillborn
one day after admis- sion.
Maternal death. Recurrence of coma twenty-four hours after
admission. Fetal death. Cesarean section at eight months. Viable
amputations, a pancreatectomy for pancreatic carcinoma, followed
within four days by onset of newly diagnosed diabetes mellitus with
acidosis, and incision and drain- age of a large (12 cm.) abscess
of the thigh. Two sub- jects died, one fifty hours after admission
due to acute myocardial infarction; the other expired four days
after onset of gram-negative bacteremia.
One of the nonfatal cases, a twenty-six-year-old male requiring
embolectomy shortly after admission, survived acute myocardial
infarction. Gangrene of the left foot developed despite the
embolectomy, and a left below- knee amputation was performed one
month after ad- mission. 9. Pregnancy
Ten pregnant women were admitted to LAC-USC Medical Center eleven
times in severe diabetic keto- acidosis, one having two episodes
during a single preg- nancy (table 7 ) . Severe ketoacidosis
occurred between the first and eighth months of pregnancy.
One of the ten women expired five hours after therapy was begun for
recurrence of severe ketoacidosis success- fully treated
twenty-four hours earlier. Pulmonary edema,
presumably due to severe coronary arteriosclerosis (diag- nosed at
autopsy), was the cause of death. The fetus also died.
One subject, admitted twice during the course of her pregnancy, had
hyperemesis gravidarum. Fetal demise occurred in the sixth month of
her pregnancy, one month after recovery from the second episode of
Only three of the ten fetuses survived. One four- pound, live,
newborn infant was delivered by induction during the seventh month
of pregnancy, three months after recovery from the episode of
severe ketoacidosis. The other two surviving infants were delivered
by cesarean section, one two months and the other a few days after
maternal recovery from severe ketoacidosis. Deliveries were at the
eighth and ninth months of pregnancy.
The very high fetal fatality rate associated with severe maternal
ketoacidosis is striking. Knowles et al.26 re- ported fetal and
neonatal mortality in six of eleven pregnancies complicated by
severe maternal ketoacidosis characterized by serum bicarbonate
levels of < 15
JULY, 197 I 495
TABLE 8 Change from insulin to oral hypoglycemic agents
three days or less before admission (seven episodes)
Number* Mean ± SEMf
TABLE 9 Admission serum potassium of 336 episodes
Age , Serum glucose (mg./lOO cc.) Serum urea nitrogen (mg./lOO cc.)
Serum sodium (mEq./L.) Serum potassium (mEq./L.) Osmolality Coma
insulin dosage (units)
7 7 6 5 7 5 7
47 ± 6 819 ± 118 56 ± 10
122 ±4 6.4 ± A 331±6
466 ± 105
* Number of episodes. f Standard error of mean.
mEq./L. Kyle27 summarized the literature on the sub- ject and
computed a fetal loss rate of 64.5 per cent associated with
maternal diabetic "coma".
One pregnant woman, after severe ketoacidosis re- quiring 1,350
units of insulin, and complicated by pneumonia, was discharged on
diet without requiring drug therapy. She had delivered an
eight-month still- born fetus one day after admission.
The subject with the complete abortion at one month of gestation
also had mediastinal emphysema. A de- tailed description of this
case has been given previ- ously.28
10. Previous therapy ivith oral hypoglycemic agents Seven episodes
of severe ketoacidosis occurred less
than three days after therapy was changed from insulin to oral
hypoglycemic agents (table 8) . The oral agents prescribed were
chlorpropamide (500-750 mg./day) in three subjects, and tolbutamide
in four. All seven sur- vived and were discharged from the hospital
with in- sulin prescribed in dosages of 15-85 units per day. The
need for caution and constant surveillance when chang- ing from
insulin to oral agents is emphasized.
Long-term oral agent therapy, ranging from two weeks to five years,
preceded twelve nonfatal and three fatal episodes of severe
diabetic ketoacidosis. 11. Potassium
Table 9 summarizes admission serum potassium levels obtained in 336
of the 340 episodes of severe diabetic ketoacidosis. Levels were
elevated ( > 6.0 mEq./L.) in 73 (22 per cent) episodes, low
(<3-5 mEq./L.) in 13 (4 per cent), and within a range of
approximate normality (3.5-6.0 mEq./L.) in 250 (74 per cent).
Greatly elevated levels (above 7.0 mEq./L.) were ob- served in 17
(6 per cent) of all nonfatal episodes, and 5 (16 per cent) of all
fatalities. The association of fatality and greatly elevated serum
potassium level pos- sibly reflected circulatory failure and
diminished renal function. It was previously reported from this
institu- tion that admission serum potassium levels were high
Approximately normal (3.5-6.0 mEq./L.)
Markedly elevated (> 6.0 mEq./L.)
* Number of episodes. t Per cent of nonfatal episodes. t Per cent
of fatal episodes. § Per cent of total episodes.
39 per cent, normal in 43 per cent, and low in 18 per cent of
severely ketoacidotic cases.29 Another study from LAC-USC Medical
Cente,r revealed very low (below 2.9 mEq./L.) admission serum
potassium levels in 2 per cent of the episodes.30
The highest parenteral potassium dose in this series, 620 mEq. in
sixteen hours, was given to a subject whose admission serum
potassium level, 4.5 mEq./L., was some- what lower than the mean of
5.3 mEq./L. Other cases with low, or relatively low, admission
serum potassium levels also required larger amounts than usual of
par- enteral petassium. The experience of this institution,
therefore, would support other studies31*35 emphasizing the need to
greatly increase parenteral potassium ad- ministration in subjects
with low admission serum potas- sium levels. These reports
emphasize that potassium de- pletion may be so profound as to
require considerably greater amounts of potassium than the 100-200
mEq. per twelve to twenty-four hours usually advocated.2-36
The maximum amount administered was 715 mEq. in twenty-four
In an extension of the principal study, acute changes of serum
potassium levels during treatment were ana- lyzed in thirty-nine
episodes of severe diabetic keto- acidosis admitted between
December 22, 1961 and June 15, 1968. The mean admission serum
potassium level, 5.4 mEq./L. (range 3.3-7.3 mEq./L), was higher
than all subsequent mean serum potassium levels. The lowest mean
serum potassium level (3.9 mEq./L.) was observed approximately four
hours after admission. The range of lowest serum potassium levels
was 2.5-4.9 mEq./L. Harwood5 reported similar findings with a mean
serum potassium of 5.7 (range 2.0-8.8) mEq./L before treatment, and
subsequent decline to a mean of 3.4 mEq./L. after treatment.
496 DIABETES, VOL. 2O, NO. 7
by guest on 01 January 2022
12. Hypoglycemia Hypoglycemia, denned as serum glucose levels of
than 50 mg./ioo cc, occurred during treatment in 40 of all 482
episodes (8 per cent) of severe diabetic keto- acidosis. The lowest
serum glucose level observed was 9 mg./ioo cc. There was no direct
association between hypoglycemia and death or severe morbidity.
Hypo- glycemia occurred in individuals aged twelve to eighty- three
years (mean, thirty-four). Admission serum glu- cose levels ranged
from 72 to 1,520 mg./ioo cc. (mean, 604 mg./ioo cc). The interval
between commencement of therapy and occurrence of hypoglycemia
varied from four to twenty-six hours (mean, thirteen hours). Cumu-
lative insulin dose at the time of hypoglycemia ranged from 40 to
1,400 units (mean, 400 units). None of these variables was related
to the occurrence of hypo- glycemia. Bortz and Spoont9 reported a
lower incidence, four of 213 episodes (2 per cent), of
"significant" hypo- glycemia. 13. Subjects requiring no drug
therapy after treatment of the acute episodes of severe diabetic
Eight patients were aglycosuric, and required no drug therapy at
time of discharge or within two months after the acute episode.
Seven of the eight subjects- had had no personal history, and none
had a family history, of diabetes mellitus. Three subjects were
alcoholic, two with acute pancreatitis.
The mean age, forty-seven years (range thirty-five to sixty-nine),
was greater than that of subjects with non- fatal, severe
ketoacidosis who continued to require long- term therapy for
diabetes mellitus. Admission mean values of serum glucose, 584
mg./ioo cc. (range 290- 1,020); urea nitrogen, 30 mg./ioo cc.
(range 18-51); potassium, 4.7 mEq./L. (range 3.9-7.4); sodium, 133
mEq./L. (range 111-151); bicarbonate, 6 mEq./L. (range <5- io )
; ketone, 1/16 dilution (range 1/6- 1/32); and insulin dosage, 485
units (range 20-1,480) did not differ markedly from those of
subjects requiring long-term therapy.
Recently, Genuth37 reported two cases with prolonged clinical
remission of diabetes in adults following suc- cessful therapy of
severe diabetic ketoacidosis. These re- missions were characterized
by normal fasting blood sugar levels, and marked rise of plasma
immunoreactive insulin after administration of glucose,
tolbutamide, or glucagon. It is not known to what extent glucose
toler- ance may have improved in the present cases. 14. Nonketotic
Nine cases of nonketotic hyperosmolar coma, defined on the basis of
a serum osmolality of 380 or more and absence of ketoacidosis, were
admitted to the Diabetes
Serum urea nitro- gen (mg./lOOcc.)
Serum potassium (mEq./L.)
Serum sodium (mEq./L.)
Serum bicarbonate (mEq./L.)
66 ± 8 402 ± 7
N.S. >.O2, <.05
* Standard error of mean. t Probability of significance of
difference between means of
fatal and nonfatal cases.- N.S. (Not significant).
Service between July 1, 1965 and June 30, 1968 (table 10).
Probably, numerous other cases were admitted to LAC-USC Medical
Center during this period, but were either not recognized or
reported. Five of the nine cases died, confirming the high
mortality rate reported by others.38-39 In contrast to severe
diabetic ketoacidosis, there was no association in these cases
between mortality and elevated serum glucose (range, 670-2075
mg./ioo cc) , urea nitrogen (range, 43-308 mg./ioo cc) , or os-
molality levels (range, 383-594). Also, there was no marked
difference in age (range, 46-79), or insulin dos- age (range,
70-1200) between fatal and nonfatal cases. Considerably larger
amounts of parenteral fluid were ad- ministered in the surviving
(mean, 7.3 L. and range, 4.5-8.5 L.) as compared with the fatal
nonketotic "coma" cases (mean, 4.8 L. and range, 3.5-5.9 L). One of
the four nonfatal and three of the five fatal cases had a past
history of known diabetes mellitus. All four surviving subjects
were discharged on drug therapy for diabetes mellitus.
This series of patients in severe diabetic ketoacidosis is
characterized by the frequency of severe hypergly- cemia, azotemia,
and elevated serum osmolality. Acidosis was marked in that all
cases selected had serum bicarb- onate values of 10 mEq./L. or
less, with arterial blood pH determinations as low as 6.81,
performed on many individuals.
The rationale of therapy in this series is presented in a separate
communication.2 Insulin requirements were usu-
JULY, 1971 497
SEVERE DIABETIC KETOACIDOSIS (DIABETIC "COMA")
ally very high (mean, 540). Previous studies40-41 have indicated
that smaller insulin doses are usually as effec- tive as larger
amounts, but recent experience at this in- stitution and elsewhere
indicates the value of higher doses.1-3-5-7-9'12-13
The relatively low mortality rates of 9 per cent (340 episodes) and
13 per cent (257 patients) may be at- tributed in part to the
intensive therapy. Careful mon- itoring probably accounted for the
infrequency of severe hypoglycemia, pulmonary edema, or
There have been reports of fatal cerebral edema de- veloping in
severe diabetic ketoacidosis following insti- tution of
therapy.42'46 In certain of these reports, fatal cerebral edema may
have been associated with persistent hypokalemia, acidosis, or
other metabolic derangement. No instance of fatality due to
cerebral edema could be
confirmed in the present series, either clinically or by necropsy.
This observation, associated with the emphasis given rapid
correction of metabolic and electrolyte ab- normalities in this
institution, would support Bradley and Young's denial47 of
allegations that aggressive ther- apy is the sole or principal
cause of this fatal compli- cation.
Four cases of a usually benign clinical entity (media- stinal and
subcutaneous emphysema) occurring in this series, have been
reported separately.27 An additional case has been recently
The larger series reported during the last two decades from Los
Angeles, London, Cincinnati, Philadelphia, Paris, and Sydney
indicate a fairly constant mortality rate in severe diabetic
ketoacidosis ranging from 9 per cent to 16 per cent (table n ) .
TABLE 11 Summary of mortality rates in severe diabetic
ketoacidosis: Survey of literature
18 19 20 21 22
23 24 25 26
140 172 61 69
42 67 32 30
44 25 5
31 15 8
New York New York Seattle Seattle Boston Cincinnati Cincinnati
London, England Sydney, Australia
Boston Melbourne, Australia Melbourne, Australia Baltimore Paris,
Paris, France Philadelphia San Francisco Los Angeles
Years 1923-66 1923-34 1923-29
1952-55 1952-55 1956-67 1958-59 1959-62
1963-67 1960-64 1964-68 1965-68
Hudson et al.18
Hudson et al.18
Shaw et al.41
Derot and Tchobroutsky12
Assan et al.18
Bortz et al.9
Kiraly et al.54
*Mortality rate 4.6 per cent between 1946 and 1966 (Joslin Clinic).
tCases with serum bicarbonate levels of <15 mEq./L. included.
Remainder of series based on bicarbonate levels of
approximately 10 mEq./L. or less.
498 DIABETES, VOL. 2O, NO. 7
by guest on 01 January 2022
PAUL M. BEIGELMAN, M.D.
has been reported from certain institutions as more intensive
therapeutic regimens have been instituted. It has been suggested
that a higher mortality rate is to be expected in the larger county
or city hospitals as com- pared with the more selective private or
The helpful advice of Helen E. Martin, M.D., Chief of the Diabetes
Service until December 31, 1968, is gratefully acknowledged.
This study was assisted by grants from the Diabetes Association of
Southern California, U.S. Public Health Service Grant AM06510-07, a
nd University of South- ern California Medical School Grant
REFERENCES 1 Root, H. F.: Diabetic acidosis and coma. In Joslin, E.
Root, H. F., White, P., and Marble, A.: The Treatment of Diabetes
Mellitus. Philadelphia, Lea and Febiger, 1959, pp. 348-94.
2 Beigelman, P. M., Martin, H. E., Miller, L. V., and Grant, W. J.:
Severe diabetic ketoacidosis. J.A.M.A. 210:1082, 1969.
3 Cohen, A. S., Vance, V. K., Runion, W., and Hurwitz, D.: Diabetic
acidosis and evaluation of the cause and course of therapy in 73
cases. Ann. Intern. Med. 52:55, i960.
4 Baker, T. W.: A clinical survey of 108 consecutive cases of
diabetic coma. Arch. Intern. Med. 58:373, 1936.
5 Harwood, R.: Diabetic acidosis: Results of treatment of 67
consecutive cases. New Eng. J. Med. 245:1, 1951.
6 Zieve, L., and Hill, E.: Descriptive characteristics of a group
of patients with moderate or severe diabetic ketoacidosis: Relation
to recovery or death. Arch. Intern. Med. 92:51, 1953.
7 Skillman, T. G., Wilson, R., and Knowles, H. C.: Mortal- ity of
patients with diabetic acidosis in a large city hospital. Diabetes
8 Dillon, E. S., and Dyer, W. W.: Factors influencing the prognosis
in diabetic coma. Ann. Intern. Med. 11:602, 1937.
9 Bortz, C. H., and Spoont, S.: Diabetic acidosis and transi- tion:
A report of 213 admissions to Philadelphia General Hos- pital in
comparison with a similar study done 30 years earlier. Penn. Med.
10 Barnett, D. M., Wilcox, S., and Marble, A.: Diabetic coma in
persons over 60. Geriatrics 17:327, 1962.
11 Rabinowitch, I. M., Fowler, A. F., Bensley, E. H.: Dia- betic
coma. An investigation of mortalities and report of a severity
index for comparative studies. Ann. Intern. Med. 12: 1403,
12 Derot, M., and Tchobroutsky, G.: L'acidose diabetique. Facteurs
decichants; evolution et mortalite dans 133 observa- tions
consecutives. Presse Med. 71:2017, 1963.
13 Assan, R., Aubert, P., Souchal, B., Tchobroutsky, G., and Derot,
M.: Analyse de 154 cas d'acidocetoses graves chez des diabetiques
(1963-1967). Experience d'un centre urbain de praitement d'urgence
du coma diabetique. Presse Med. 77: 423, 1969.
14 Collen, M. F.: Mortality in diabetic coma. Arch. Intern.
Med. 70:347, 1942. 15Beardwood, J. T., and Rouse, G. T.: Diabetic
J.A.M.A. 17:1701, 1941. i6Danowski, T. S., Winkler, A. W., and
Peters, J. P.: Salt
depletion: Peripheral vascular collapse, and treatment of dia-
betic ketoacidosis. Yale J. Biol. Med. 18:405, 1945.
17 Owens, L. B., and Rockwern, S. S.: Prognosis in diabetic coma:
Basic importance of mental state. Amer. J. Med. Sci. 198:252,
18 Hudson, B., Bick, M., and Martin, F. I. R.: Observations on the
treatment of severe diabetic ketosis. Aust. Ann. Med. 9:34,
1 9 Greenaway, J. M., and Read, J . : Diabetic coma: A re-
view of 69 cases. Aust. Ann. Med. 7:151-58, 1958. 2 0 Fitzgerald,
M. G., O'Sullivan, D . J., and Malins, J. M. :
Fatal diabetic ketosis. Brit. Med. J. 1:247, 1961. 2 1 Partamian,
J. O., and Bradley, R. F.: Acute myocardial
infarction in 258 cases of diabetes: Immediate mortality and
five year survival. New Eng. J. Med. 273 :455 , 1965. 2 2 Root, H.
F.: Diabetic coma and acute pancreatitis with
fatty livers. J.A.M.A. 108:777, 1937. 2 3 Bossak, E. T., and
Joelson, R. H . : Acute pancreatitis com-
plicating diabetes mellitus. Arch. Intern. Med. 9 7 : 2 0 1 , 1956.
2 4 Tully, G. T., and Lowenthal, J. J . : The diabetic coma
acute pancreatitis. Ann. Intern. Med. 48 :310 , 1958. 2 5 Hayduk,
K., Durr , F., and Schollmeyer, P . : Diabetic coma
and pancreatitis. German Med. Monthly 13:432, 1968. 2 6 Knowles, H.
C , Guest, G. M., Lampe, J., Kessler, M., and
Skillman, T. G.: The course of juvenile diabetes treated with
unmeasured diet. Diabetes 14:239, 1965. 2 7 Kyle, G. C.: Diabetes
and pregnancy. Ann. Intern. Med.
59:Suppl. 3, 1963. 2 8 Beigelman, P. M., Miller, L. V., and Martin,
Mediastinal and subcutaneous emphysema in diabetic coma.
J.A.M.A. 208 :2315 , 1969. 2 9 Martin, H. E., Smith, K., and
Wilson, M. L.: The fluid
and electrolyte therapy of severe diabetic acidosis and
Amer. J. Med. 24 :376 , 1958. 3 0 Martin, H. E., Hillier, P., and
Wer tman , M.: Potassium
deficits in diabetic acidosis with particular reference to
lems in therapy. Proc. Amer. Diabetes Ass. 10 :161 , 1950. 3 1
Abramson, E., and Arky, R.: Diabetic acidosis with ini-
tial hypokalemia. J.A.M.A. 196 :401 , 1966. 3 2 Pullen, H., Doig,
A., and Lambie, A. T. : Intensive intra-
venous potassium or replacement therapy. Lancet 2 :809 , 1967. 3 3
Seftel, H . C , and Kew, M. C.: Early and intensive potas-
sium replacement in diabetic acidosis. Diabetes 15:694, 1966. 3 4
Clementsen, H. J . : Potassium therapy. A break with tra-
dition. Lancet 2 :175 , 1962. 3 5 Bergen, S.: Extreme hypokalemia
during treatment of
diabetic acidosis. New York J. Med. 67 :2267 , 1967. 3 6 Wil l
iams, R. E.: The pancreas. In Will iams, R. E.:
Textbook of Endocrinology. Philadelphia, W . B. Saunders Co.,
1968, pp . 613-802. 3 7 Genuth, S. M. : Clinical remission in
Studies of insulin secretion. Diabetes 19:116, 1970. 3 8 Danowski,
T. S., and Nabarro, J. D . : Hyperosmolar and
other types of non-ketoacidotic coma in diabetes. Diabetes
162, 1965. 3 9 Assan, R., Souchal, B., Aubert, P. H., Tchobroutsky,
and Derot, M. : Comas metaboliques non acido-cetosiques chez
JULY, 1971 499
SEVERE DIABETIC KETOACIDOSIS (DIABETIC "COMA")
des diabetiques. Presse Med. 77:787, 1969. 4 0 Smith, K., and
Martin, H . : Response of diabetic coma
to various insulin dosages. Diabetes 5 :287 , 1954. 4 1 Shaw, C.
E., Jr., Hurwitz, G. E., Schmukler, M., Brager,
F. H., and Bessman, S. P. : A clinical laboratory study of
lin dosage in diabetic acidosis. Comparison with small and
large doses. Diabetes 11 :23 , 1962. 4 2 Dillon, E. S., Riggs, H .
E., and Dyer, W . W . : Cerebral
lesions in uncomplicated fatal diabetic acidosis. Amer. J.
Sci. / 9 2 : 3 6 c , 1936. 4 3 Young, E., and Bradley, R.: Cerebral
edema with irre-
versible coma in severe diabetic ketoacidosis. New Eng. J.
276 :665 , 1967. 4 4 Taubin, H., and Matz, R.: Cerebral edema,
sipidus, and sudden death during the treatment of diabetic
ketoacidosis. Diabetes 17:108, 1968. 4 5 Hayes, T. M., and Woods,
C. J. : Unexpected death dur-
ing treatment of uncomplicated diabetic ketoacidosis. Brit.
J. 4 :32 , 1968. 4 6 Man ignan , D. , Auzepy, P. H. , Leguillant,
and Depar i s :
Le m o r t par oedeme cerebral au cours de l'acidocetose dia-
betiques grave. Presse Med. 7 7 : 7 9 0 , 1969 .
4 7 Bradley, R. F., and Young, E.: Cerebral edema in severe
diabetic ketoacidosis (Letters to the Editor). New Eng. J. Med.
48Tashima, C. K., Reyes, C. U., and Kerlow, A.: Mediastinal
emphysema in diabetic coma (Letter to the Editor). J.A.M.A.
4 9 Marble , Alexander . Personal communicat ion. 5 0 Bowen, B. D .
, and Hek imian , I.: Diabetic coma: Report
of 81 cases. Ann . In tern . Med. 3 : 1 1 0 4 , 1930. 5 1 Pines, K.
L.: Discussion of article by Wi l l iam H. Daugh-
aday, M.D. : T h e N a t u r e and Correction of Diabetic
acidosis. Diabetes 7 : 2 3 5 , 1958. 5 2 Crampton , J. H. , Mel l
inger , G. W . , and Palmer, L. J.:
P o t a s s i u m i n t h e t rea tment of d i a b e t i c c o m a
. D i a b e t e s 2:1,
1953- 5 3 Naba r ro , J. D . N . : Diabet ic acidosis: Clinical
Leibel, B. S., and Wrensha l l , G . A. (Eds . ) : O n the Nature
and Treatment of Diabetes. Amsterdam, Excerpta Medica Founda- tion,
1965, pp. 545-57-
54Kiraly, J. F., Becker, C. E., and Williams, H. E.: Dia-
betic ketoacidosis. A review of cases at a university medical
center. Calif. Med. 112:1, 1970.
5 0 0 DIABETES, VOL. 2O, NO. 7
by guest on 01 January 2022