The occurrence of diabetic ketoacidosis (DKA),                                      Berhane Seyoum, MD; Paulos Berhanu, MD
a serious but largely preventable acute com-
plication of diabetes mellitus, has been de-
clining in recent years. However, empiric
observations indicate that DKA continues to
have a major effect on ethnic minority patients                                                    Diabetic ketoacidosis (DKA)
in inner-city settings. In this study, we con-        Diabetic ketoacidosis (DKA) is
ducted a retrospective analysis of five-year      a common and serious acute complica-             is one of the preventable acute
hospital admission data for DKA at a single       tion of diabetes caused by a relative or
inner-city hospital that serves a largely un-
                                                  absolute lack of insulin. Diabetic ketoa-
                                                                                                   complications of diabetes
insured adult African American population. A
computer-assisted search of the International     cidosis (DKA) is one of the preventable          mellitus through appropriate
Classification of Diseases, Ninth Revision,       acute complications of diabetes mellitus
Clinical Modification codes for DKA revealed      through appropriate outpatient diabetes          outpatient diabetes
847 admissions for confirmed DKA in 630           management. Although the manage-
patients. Of these, 592 (94%) were African                                                         management.
Americans, 22 (3.5%) were Whites, and 16
                                                  ment of DKA has been markedly
(2.5%) were Hispanics. The mean age was           improved in recent years in association
43.4 6 .4 years. Five hundred seventy-one         with the general healthcare improve-
(90.6%) of the patients had type 1 diabetes,
                                                                                                   reasons; and 14% stopped insulin
                                                  ments, it is still a public health problem.
and 59 (9.4%) had type 2 diabetes. One                                                             because they did not know what to do
                                                  In 1983, the incidence of DKA was
hundred forty-five patients (23%) were newly                                                       when they became sick.
diagnosed with diabetes. Ninety-four (14.9%)
                                                  reported to be 46 per 10,000 patients
                                                                                                       Empiric observations in our city
of the patients had multiple admissions,          with diabetes.1 During the last two
                                                                                                   indicate that DKA continues to have
ranging from 2 to 23 admissions per patient       decades the trend of DKA admissions
                                                                                                   a considerable effect on ethnic minority
during the five-year period, while the remain-    has been increased.2 Part of this in-
ing 391 (62.1%) patients were single admis-                                                        patients in an inner-city hospital that
                                                  creased frequency of admissions may be
sions. Half of the patients (52%) did not have                                                     serves disproportionately large unin-
                                                  related to the increased prevalence of
health insurance. Major precipitating factors                                                      sured adult African American popula-
for DKA included discontinuation of insulin,      diabetes mellitus. Moreover, multiple
                                                                                                   tion. This study was intended to expand
infection, and other medical illness in 501       episodes of DKA have also contributed
                                                                                                   previous empiric observations by in-
(59.1%), 136 (16.1%), and 30 (3.5%) of the        to the increase of admissions,3 but the
admissions, respectively. In conclusion, these                                                     vestigating the causes of DKA and its
                                                  age-adjusted mortality rate has been
data demonstrate that DKA continues to have                                                        complications during a five-year period
                                                  improving over the last two decades.2
a major effect in urban African American                                                           and to suggest preventive measures to
patients with diabetes. Therefore, multiple           In earlier reports, infection was the        reduce the impact of the problem.
targeted interventions are needed in this         major precipitating factor in 33%–56%
population to improve diabetes care and           of cases, followed by noncompliance
thereby decrease the frequency of DKA. (Ethn
Dis. 2007;17:234–237)
                                                  with therapy, including discontinuation          METHODS
                                                  of insulin and oral medications and
Key Words: Admissions, Diabetes Mellitus,         nonadherence to diet, in 4%–25% of               Patient Population
Diabetic Ketoacidosis (DKA)                       the cases.4–11 Most of these studies were            This study is a retrospective analysis
                                                  done on White populations, the results           of confirmed DKA admission to one
                                                  of which may not apply to African                inner-city hospital in Detroit during the
                                                  American populations. Contrary to                five-year period from January 1, 1999,
    From the Division of Endocrinology,           these reports, a study by Musey et al12          to December 30, 2003. The hospital is
Diabetes, and Metabolism, Wayne State             in a largely African American popula-            a teaching facility associated with a med-
University School of Medicine, Detroit,
                                                  tion showed that the major cause of              ical school. It serves a largely indigent
                                                  DKA was discontinuation of insulin               urban population that is .90% African
    Address correspondence and reprint            therapy in up to 67% of the cases. More          American. Many of the patients are
requests to Berhane Seyoum, MD; Division          than 50% of the patients discontinued            under- or uninsured for health care.
of Endocrinology, Diabetes, and Metabo-           or reduced the insulin dose; 21% did
lism; Wayne State University School of
                                                  not know how to manage their insulin             Diabetic Ketoacidosis Admissions:
Medicine; 4201 St. Antoine, UHC-4H;
Detroit, MI 48201; 313-595-2830; 313-             dosage with change of diet and physical          Criteria and Confirmation
993-0903 (fax); bseyoum@med.wayne.                activity; 14% discontinued insulin be-               A computer assisted search of the
edu                                               cause of behavioral or psychological             International Classification of Diseases,

234                                                  Ethnicity & Disease, Volume 17, Spring 2007
                                                           DIABETIC KETOADICOSIS IN AFRICAN AMERICANS - Seyoum and Berhanu

                                                                                                                  2003, we noted 66,317 medical, surgi-
Table 1. Patterns of diabetic ketoacidosis (DKA) and diabetes-related admissions to
Detroit Regional Hospital from 1999 to 2003                                                                       cal, and other admissions to the hospital
                                                                                                                  under study. Of these, 13,068 were
                                                                   N                          %                   related to diabetes and 847 were for
Diabetes-related admissions                                     13,068                       6.5*                 DKA admissions; DKA made up 1.3%
DKA admissions                                                     847                       1.33                 of total admissions and 6.5% of di-
Known patients admitted once with DKA                              391                      62.14
Known patients admitted multiple times with DKA                     94                      14.94
                                                                                                                  abetes-related admissions. Using the
Newly diagnosed                                                    145                      23.04                 ICD codes for DKA, we originally
 * Out of the total 66,317admission during the study period.
                                                                                                                  found that the total number of DKA
 3 Out of the total 13,068 patients with diabetes.                                                                admissions was 971. However, when the
 4 Out of the total 630 patients admitted to the hospital.
                                                                                                                  medical records were identified and
                                                                                                                  reviewed, 124 cases were excluded
Ninth Revision, Clinical Modification                    Statistical Analysis                                     because they did not meet the criteria
(ICD-9-CM) codes 250.1 (DKA in                               SAS software JMP version 6 (SAS                      for DKA. Data analysis was then
type 1 diabetes) and 250.11 (DKA in                      Institute Inc, Cary, NC) was used for all                performed on the 847 confirmed ad-
type 2 diabetes) were used to identify                   statistical analyses. Data were analyzed                 missions for DKA and included 630
patients from the computer records of                    as means plus or minus standard errors                   patients with diabetes (Table 1).
the hospital. After cases were identified                of mean (SEM) for continuous variables                        Of the 847 admissions for DKA,
in the computer data center, the medical                 and as frequency and percentage of                       145 (17.1%) admissions were for
records of the patients were retrieved,                  study population for categorical vari-                   newly diagnosed patients and the rest
and the diagnosis of DKA was con-                        ables. Nonparametric binomial test was                   were admission for known patients
firmed by detailed review of the records                 used to evaluate the statistical signifi-                with diabetes. Of the 702 admissions
according to clinical criteria that in-                  cance of the different precipitating                     for previously diagnosed patients, 311
dicated DKA along with one or all of                     factors for DKA among the patients                       (36.9%) were multiple admissions
the following laboratory data: 1) serum                  with and without insurance coverage.                     during the study period, while the
bicarbonate #15 mg/L; 2) anion gap                       Chi-square tests were used to explore                    remaining 391 (46.2%) were single
$14; 3) arterial pH #7.3; or 4) positive                 the univariate relationships among cat-                  admissions (Fig. 1). Several character-
serum or urine ketones at any dilution                   egorical variables. Statistical significance             istics are described in Table 2. The
level.                                                   was set at P#.05 for all tests.                          newly diagnosed patients were signifi-
    After confirming the diagnosis, pa-                                                                           cantly younger than the known pa-
tient age, sex, ethnicity, insurance                                                                              tients with diabetes admitted for DKA
coverage, length of hospital stay, type                  RESULTS                                                  (P,.05). Males were seen twice as
and duration of diabetes, medications,                                                                            frequently as females in the known
previous episodes of DKA, physical                          During the five-year period from                      patients but nearly four times as
examination findings, and outcome of                     January 1, 1999, to December 30,                         frequently among the newly diagnosed
DKA were recorded in a database.

                                                         Table 2. Patient characteristics of newly diagnosed and previously diagnosed
                                                         patients with diabetes who were admitted with diabetic ketoacidosis (DKA) during
                                                         the five-year period

                                                                                               Newly Diagnosed Patients              Patients with Previously
                                                                                                    with Diabetes                     Diagnosed Diabetes
                                                                                                   Mellitus n5145                        Mellitus n5702*
                                                         Mean age (years)                               36.2 6 .1                              43.4 6 .53
                                                         Sex: M/F                                        114/31                                 468/234
                                                         Hospital stay (days)                            5.1 6 .2                               6.8 6 .2
                                                           African Americans                               142                                    658
                                                           Hispanics                                         2                                     27
                                                           Whites                                            1                                     17
                                                           * Based on the total number of admissions. Some patients had multiple admissions.
Fig 1. Frequency of admissions among                       3 P,.05.
702 patients with known diabetes

                                                         Ethnicity & Disease, Volume 17, Spring 2007                                                            235

                                                                                              acute complications of diabetes mellitus.
                                                                                              Diabetic ketoacidosis (DKA) is a pre-
                                                                                              ventable complication in educated pa-
                                                                                              tients who are compliant with their
                                                                                              treatment regimens, including diet, in-
                                                                                              sulin, or other diabetic medications.
                                                                                              The incidence of DKA is reported to
                                                                                              be 30–46 per 10,000 patients with
                                                                                                   However, in spite of the preventable
                                                                                              nature of the complication and the
                                                                                              growing availability of health education
                                                                                              for patients with diabetes, DKA still
                                                                                              occurs among the underprivileged, un-
                                                                                              derinsured, or uninsured segments of
                                                                                              the population. Our study hospital is
                                                                                              a major tertiary inner-city hospital that
Fig 2. Precipitating factors for DKA in 847 admissions during the 5-year period               serves primarily poor and uninsured
                                                                                              African Americans throughout Detroit.
patients. Most of the patients were              The overall mean hospital stay was
                                                                                              During the study period, DKA made up
African Americans.                           6.7 6 .2 days. No difference was
                                                                                              1.3% of the total admissions and 6.5%
    The precipitating factors for DKA        observed in hospital stay among the
                                                                                              of diabetes-related admissions. Of the
among all the patients are shown in          new and the known patients. The major
                                                                                              total DKA patients, 23% were newly
Figure 2. In 59% of the admissions,          complications observed were anoxic
                                                                                              diagnosed individuals admitted for the
patients discontinued insulin either be-     brain damage in four patients, and one
                                                                                              first time. The occurrence of DKA
cause of lack of insulin or for reasons      patient developed respiratory distress
                                                                                              admissions is consistent with several
related to noncompliance. Patients who       that eventually required permanent
                                                                                              other reports. 12,14 Most patients
stopped taking insulin for various reasons   tracheostomy. Two patients (.3%) died,
                                                                                              (90.6%) in this study were African
while they have it were labeled as           one secondary to myocardial infarction
                                                                                              Americans, of whom .50% had no
noncompliant. The second common              and the other secondary to sepsis.
                                                                                              insurance. This pattern reflects the
precipitating factor was infection. Fig-
                                                                                              population served by our urban inner-
ure 3 shows the precipitating factors
                                                                                              city hospital. A similar pattern was seen
according to the insurance coverage. Lack    DISCUSSION                                       in other inner-city hospitals that provide
of insulin and infection were the main
                                                                                              medical care primarily to uninsured
causes of precipitating factors for DKA in      Diabetic ketoacidosis (DKA) re-
patients without insurance (P,.001).         mains one of the most commonly seen
                                                                                                   As described in Figure 2, among
                                                                                              most of the admissions (59%) for DKA,
                                                                                              the main precipitating factor was dis-
                                                                                              continuation of insulin. Of those who
                                                                                              discontinued insulin, 51.7% were non-
                                                                                              compliant—they discontinued the in-
                                                                                              sulin for no apparent reason—whereas
                                                                                              in 48.3%, the patients discontinued
                                                                                              insulin treatment because they were
                                                                                              not able to afford or obtain the
                                                                                              medication. Again, similar data were
                                                                                              reported from other centers that serve
                                                                                              largely indigenous patients with diabetes
                                                                                              who had no insurance. Musey et al12
                                                                                              reported that among their largely Afri-
Fig 3. Precipitating factors for DKA according to insurance status of patients with           can America patients, 67% of them
known diabetes                                                                                developed DKA after discontinuation of

236                                             Ethnicity & Disease, Volume 17, Spring 2007
                                               DIABETIC KETOADICOSIS IN AFRICAN AMERICANS - Seyoum and Berhanu

insulin therapy. More than 50% of                                                                   2. Center for Disease Control, Division of
                                                                                                       Diabetes Translations. National diabetes sur-
those who discontinued their insulin         In our study, infection was the                           veillance system. Available at: http://
did so because they could not afford to
purchase insulin.                            second most common cause of                     
    Despite all the progress in diabetes     DKA.                                                   3. Cefalu WT. Diabetic ketoacidosis. Crit Care
care and continuing increase in health                                                                 Clin. 1991;7(1):89–108.
care budget, some patients who depend                                                               4. Lebovitz HE. Diabetic ketoacidosis. Lancet.
on insulin for survival are unable to                                                               5. Taft P, Stockigt JR, Harrison JW, Cameron
obtain insulin until they develop DKA        infection rather than discontinuation of                  DP. Diabetic ketoacidosis: its causes and their
and get admitted to the hospital. In-        insulin.4–11 In our study, infection was                  prevention. Med J Aust. 1968;2(19):825–829.
sulin, which is one of the essential life    the second most common cause of                        6. Umpierrez GE, Kitabchi AE. Diabetic ketoa-
saving drugs, has to be accessible to the    DKA. Those studies that showed in-                        cidosis: risk factors and management strategies.
                                             fection to be the main precipitating                      Treat Endocrinol. 2003;2(2):95–108.
people who depend on it, in order to
                                                                                                    7. Gale EA, Dornan TL, Tattersall RB. Severely
avoid serious and potentially life-threat-   cause were probably done in an eco-
                                                                                                       uncontrolled diabetes in the over fifties.
ening complications such as DKA.             nomically more privileged population                      Diabetologia. 1981;21(1):25–28.
Patients who are non-compliant and           where financial resources and diabetes                 8. Hockaday TD, Alberti KG. Diabetic coma.
discontinue insulin until they develop       education are not a major problem, thus                   Clinical Endocrinol Metab. 1972;1(3):750–788.
                                             the social environment did not adversely               9. Davoren PM, Bowen KM. Precipitating
DKA need to go through intensive
                                                                                                       factors in diabetic ketoacidosis. Med J Aust.
diabetes education programs. They have       contribute to DKA. In the inner-city
to be educated by a health professional      setting, where there are larger number of             10. Johnson DD, Palumbo PJ, Chu CP. Diabetic
and must understand the consequences         indigent population and the health                        ketoacidosis in a community based population.
of discontinuing insulin. It is econom-      delivery status is meager, the main                       Mayo Clin Proc. 1980;55(2):83–88.
                                             precipitating factor of DKA continues                 11. Wachtel TJ, Tetu-Mouradjian LM, Goldman
ically much cheaper for the hospital or
                                             to be discontinuation of insulin, and                     DL, Ellis SE, O’Sullivan PS. Hyperosmolarity
any healthcare provider to identify these                                                              and acidosis in diabetes mellitus: a three year
patients and provide them with the           infection comes in second.12 The dif-                     experience in Rhode island. J Gen Intern Med.
necessary diabetes education. In addi-       ferent trends seen among the econom-                      1991;6(6):495–502.
tion, healthcare providers need to           ically privileged and indigent popula-                12. Musey VC, Lee JK, Crawford R, Klatka MA,
allocate appropriate resources and pro-      tion can be improved if appropriate                       McAdams D, Phillips L. Diabetes in urban
                                             measures are undertaken to improve the                    African Americans. I. Cessation of insulin
visions for these patients in order to                                                                 therapy is the major precipitating cause of
ensure reliable access to insulin. In this   services of diabetic management to these
                                                                                                       diabetic ketoacidosis. Diabetes Care.
study, more than 50% of the patients         patients. It is intriguing that the main                  1995;18(4):483–489.
indicated that they stopped the insulin      precipitating factors for DKA did not                 13. Fishbein HA. Diabetic ketoacidosis, hyperos-
because they could not get insulin and       change in the last two decades in spite of                molar nonketotic coma, lactic acidoisis, and
                                             major advances in the healthcare de-                      hypoglycemia. In Harris MI, Hamman RF,
they were not able to afford it, even                                                                  eds. Diabetes in America (National Diabetes
though, they knew the consequences of        livery system. While medical advances
                                                                                                       Data Group). Washington, DC: US Dept
discontinuing insulin. The issue of lack     and modernization are important,                          Health and Human Sciences, 1985;XII-1–XII-
of an insulin prescription to purchase       health education and access to appro-                     16.
insulin was not a factor since the           priate diabetes education remain the                  14. Ellemann K, Soerrensen JV, Pedersen L,
                                             core issue in managing diabetes mellitus                  Edsberg B, Sndersen OO. Epidemiology and
commonly used insulins are readily
                                             and preventing DKA. Consistent with                       treatment of diabetic acidosis in a community
available in pharmacies without pre-                                                                   population. Diabetes Care. 1984;7(6):528–532.
scription. For the healthcare system, it     this view, some large employers are                   15. University of Michigan will cut drug costs in
would be cost-effective to provide insulin   beginning to provide diabetes medica-                     2 year test. Detroit Free Press. April 26, 2006.
to patients who cannot afford it, rather     tions and supplies to their employees
                                             with diabetes by waiving the patient’s                AUTHOR CONTRIBUTIONS
than face the cost of managing DKA in
                                             components of the cost.15 Such ap-                    Design concept of study: Seyoum, Berhanu
the inpatient setting. Providing patients                                                          Acquisition of data: Seyoum, Berhanu
with insulin as outpatients and prevent-     proaches should, hopefully become
                                                                                                   Data analysis and interpretation: Seyoum,
ing the occurrence of DKA would reduce       more common.                                              Berhanu
hospital admissions and limit the direct                                                           Manuscript draft: Seyoum, Berhanu
and indirect costs from DKA-related          REFERENCES                                            Statistical expertise: Seyoum, Berhanu
                                              1. Faich GA, Fishbein HA, Ellis SE. The              Acquisition of funding: Berhanu
morbidities and mortalities.                     epidemiology of diabetic acidosis: a population   Administrative, technical, or material assis-
    The major precipitating factor for           based study. Am J Epidemiol. 1983;117(5):             tance: Berhanu
DKA in a number of earlier reports was           551–558.                                          Supervision: Berhanu

                                             Ethnicity & Disease, Volume 17, Spring 2007                                                          237

To top