Treatment Practices of Diabetic Ketoacidosis at a Large Teaching Hospital
Ruth Ferreri BS, RN, MS, APRN 

Journal of Nursing Care Quality
April/June 2008 
Volume 23 Number 2
Pages 147 - 154

A retrospective chart review of 55 medical records of adults treated for diabetic ketoacidosis (DKA) identified areas of treatment that delayed the improvement of DKA. There was a positive correlation between time to initiation of insulin infusion and resolution of anion gap. Sixty-six percent of cases did not have the recommended overlap between intravenous and subcutaneous insulin. Opportunity exists to educate nurses as experts to partner with their physician colleagues to manage DKA.

DIABETIC KETOACIDOSIS (DKA) is a serious and potentially fatal complication of diabetes. Although traditionally thought to affect only patients with type 1 diabetes, it can also occur with type 2 diabetes.1 In 2004, 1.4 million adults were newly diagnosed with diabetes (type 1 or 2), which was a 54% increase in incidence since 1997.2 In the same year, 14.7 million people in the United States were living with diabetes.3

The 1999 surveillance report by the Centers for Disease Control and Prevention found that mortality from DKA has steadily declined in the adult population since 1980. However, DKA still remains a significant cause of hospital admission.4 In 2003, it was the first-listed diagnosis for 115,000 hospital admissions in the United States, with an average length of stay (LOS) of 3.8 days.5 The estimated annual rate of DKA has been reported in the range of 4.6 to 8 cases per 1,000 patients with diabetes.6 A 21-year British study reported a 3.9% mortality rate, with results suggesting that, in most recent years, mortality was related to concomitant disease.7 Treatment of DKA accounts for 2% to 8% of hospital admissions for diabetes,8 and in 20% to 30% of cases, DKA may be the first indication that a patient has developed diabetes.6,9

Treating DKA can be costly, requiring stay in acute care environments. A recent study found that DKA is responsible for more than 1 of every 4 dollars spent on direct medical care for adults with type 1 diabetes. For patients who experienced multiple episodes of DKA, the cost of DKA was 1 of every 2 dollars spent.10 It has been estimated that the annual cost of treating DKA may be at least $1 billion in the United States.1

Patients treated for DKA require intensive monitoring, initially requiring glucose monitoring every hour. Regular insulin infusions as well as subcutaneous lispro insulin therapies are equally effective, but hourly monitoring is crucial.11 Depending on the hospital facilities, patients are generally admitted to either intensive- or intermediate-care areas where nursing ratios allow for the necessary monitoring. Patients can be treated safely on regular units only if there is adequate nursing availability. Although some regular medical units allow for intravenous insulin infusions, rarely is hourly monitoring possible.12 Because of these constraints, treatment for DKA is costly, with estimates ranging from $6,444 to $14,429 per admission, depending on the use of intensive care facilities.10,11,13 Although some patients may require intensive or intermediate care for coexisting acute morbidities, many will require this level of care only until DKA is resolved, and hemodynamic stability is achieved.

To minimize economic burden on both patients and healthcare facilities, as well as promote the efficient use of resources, it is important that clinicians treat DKA consistently and efficiently to speed safe delivery of care through the healthcare system. It is, therefore, advantageous for both nurses and providers to have a consistent method of care delivery that is familiar to both groups. Various guidelines have been suggested, but the American Diabetes Association's (ADA's) recently published evidence-based position statement 14 can be used as a starting point in the development of a consistent approach to treatment.

One study conducted in a large teaching hospital found that 87% of DKA cases were treated primarily by physicians not specially trained in diabetes or endocrinology.15 A national study of residents in their final year at US academic healthcare centers reported that 95% of internal medicine residents felt “very prepared” to treat DKA.16 However, patients treated by endocrinologists had a significantly shorter LOS and lower rate of readmission than those treated by generalists.15

In a recent study, physicians and nurses in critical care and general medicine areas were instructed in use of a critical pathway on inpatient management of DKA. Results showed a decreased LOS and trend toward decreased cost. Treatment variation was also decreased with the intervention.17 An earlier British study found the management of DKA patients in a large teaching hospital to be suboptimal, even after the introduction and dispersal of clinical guidelines. It might be noted in that study that guidelines were aimed at physicians and were not directed at nursing. A study on the management of DKA in a teaching hospital in Spain found inadequacies in the time to initiation and amount of intravenous fluids, under-replacement of potassium, and excessive use of alkali therapy.18 No literature focusing on specific areas of treatment that delay improvement of DKA or affect LOS was found, with studies in this area focusing more generally on critical pathway implementation.

The current study examined the following research question: What are the specific areas of treatment that delay the improvement of DKA and slow progression of patients to less intensive environments? The goal was to identify areas for improvement in the treatment of DKA at a large teaching hospital to guide nurses, interns, and residents who are responsible, as a group, for planning and implementing care of patients with DKA. Specific outcomes explored included total LOS, time to transfer order to a regular floor, and resolution of anion gap.


A post hoc design was used to review the medical records of patients admitted and treated for DKA at a large New England teaching hospital. No record was excluded on the basis of race, gender, or concomitant disease. Eligibility for the study was based solely on age of more than 18 years and diagnosis of DKA as defined by the following:

* plasma glucose > 250 mg/dL,
* arterial pH < 7.25,
* serum bicarbonate < 15 to 18 mEq/L,
* positive urine ketones,
* positive serum ketones, and
* anion gap >10 mEq/L, calculation: (Na+) – (Cl- + HCO3-) mEq/L.16

The study was approved for human subjects' protection by the institution's internal review board. Medical records for patients admitted with a diagnosis of DKA were identified using the ICD-9 codes for DKA (250.11, 250.12, 250.13; and 250.21, 250.22, and 250.23) and reviewed for inclusion criteria. All charts meeting these criteria from January 1, 2004, through December 31, 2005 (2 full years), were sampled. The researcher transcribed data from included charts onto a collection instrument, including timing of treatments, location within the institution, laboratory values, and LOS.


The sample consisted of 55 cases. One patient died, whereas the other 54 survived to discharge. The mean age was 38 (SD = 15.8) years, with range 18 to 79 years. Thirty (54.5%) were men. Twenty-three charts indicated race, with 9 (16.3%) Hispanic, 7 (12.7%) black, 6 (10.9%) white, and 1 (1.8%) Indian. Diabetes type was indicated in 45 charts. Thirty-one (56.3%) cases were type 1, 8 (14.5%) type 2, and 6 (10.9%) new onset (type not specified). Only 18 records indicated the duration of diabetes, M = 12.1 (SD = 8.15). The most common concomitant disease was pancreatitis, with 6 (10.9%) cases given this diagnosis. Patients were admitted to 3 areas: medical intermediate care (MICA) (n = 27, 49.1%), intensive care unit (ICU) (n = 16, 29.1%), and general medical units (n = 12, (21.8%).

Outliers were found in 2 variables. In one case, the blood glucose level at the start of dextrose infusion was 870 mg/dL, which was 7.07 times the SD of all cases. In another case, the overlap between insulin infusion and subcutaneous insulin was 4.34 SDs away from the mean of all cases with an overlap of at least 60 minutes, and represented an overlap of 4.5 days. In both these cases, the outliers were removed from analysis involving those variables. As this was an exploratory study, the remainder of the variables for these cases was used in analysis.

The ADA recommends an overlap of insulin infusion and subcutaneous insulin of at least 60 minutes.14 Two thirds of the charts reviewed did not indicate an overlap consistent with this recommendation. In the charts that did not indicate an overlap, the mean gap between infusion and subcutaneous insulin was 86 minutes. The mode of these cases was zero (24.2%), indicating that it was common to stop the infusion and start subcutaneous insulin simultaneously. Almost 75% of charts reviewed included an order for a hyperglycemia protocol that does not address the other metabolic derangements of DKA and, therefore, is not for use in patients with this condition. The mean time to initiation of insulin infusion was 206.9 minutes, with mode 150 minutes after presentation (15% of cases). The time to initiation of insulin infusion was longer for patients who were admitted to the less acute areas (Table 1).


[Email Jumpstart To Image]
Table 1. Best practice parameters for diabetic ketoacidosis

Using a Mann-Whitney U test, there was no significant difference in LOS between those patients with or without an overlap of at least 60 minutes. Similarly, an independent t test did not show a significant difference in time to transfer order to floor between patients with or without an overlap of at least 60 minutes.

MICA was the only unit that averaged an overlap between insulin infusion and subcutaneous insulin. The ICU and other units averaged no overlap, with the infusion stopping before the subcutaneous insulin was started. The overlap in MICA was 54.4 minutes (SD = 170.3), ICU -25.5 minutes (SD = 137.3), and other units -88.8 minutes (SD = 203.1). Positive numbers indicate an overlap, whereas negative number indicated a gap between the stopping of infusion and starting of subcutaneous insulin. Results indicated no significant difference in length of overlap between any of the units.

A cross-tabulation analysis with [chi]2 test was used to compare admission unit with the presence or absence of an overlap of at least 60 minutes. The relationship was not significant, indicating that admission unit was not significantly predictive of overlap 60 minutes or more. In MICA, 48.1% of patients had an overlap of the recommended length, whereas in ICU, only 20% and in other units 15.4% met this recommendation.

A cross-tabulation analysis with [chi]2 test was used to examine the relationship between presence/absence of protocol and presence/absence of overlap 60 minutes or more. However, with n = 50, only 2 cases (4%) had an overlap 60 minutes or more and did not have the protocol ordered, whereas 48% were ordered the protocol and did not have an overlap 60 minutes or more. No statistically significant relationship was found between admission unit and whether or not the insulin infusion protocol was ordered.

There was a positive correlation between the length of time before insulin infusion started (206.9 ± 165.9 minutes) and the length of time before the anion gap resolved (13.2 ± 6.7 hours), r = 0.43, P = .001. There was a nonsignificant positive correlation between the length of time before insulin infusion was started and LOS (76.7 ± 44.7 hours), r = 0.13, P = .34.

The total time that a patient was on an insulin infusion was a mean of 19.6 (SD = 21.4) hours. There were only 3 cases in which the infusion was stopped and restarted before resolution of the anion gap.

A regression analysis was conducted to explore the effect that overlap, hours in emergency department, presenting pH, presenting anion gap, and minutes to start of insulin infusion had on the clinical outcomes of LOS, time to transfer order to floor, and resolution on anion gap. To achieve normal distribution, the variables LOS and time to transfer order to floor were transformed by taking their square root. The variable resolution of anion gap could not be transformed to achieve normal distribution. This variable was dichotomized at the 50th percentile, and a logistic regression was performed. Results of these tests did not show that any variable was significantly predictive of the clinical outcomes.

Total LOS in the hospital was 76.6 (SD = 44.7) hours. The longest mean LOS was in the ICU, 96.9 (SD = 62.3) hours, and was similar between MICA and other units, 70.7 (SD = 34.0) and 71.4 (SD = 44.0), respectively. Hours spent in the emergency department before transfer to admission unit increased as the admission unit level of acuity decreased. This difference was significant between the ICU and MICA, and the ICU and regular units (P < .001). The number of hours spent before a transfer order to move a patient from the ICU or MICA to a regular floor came was longer in MICA, 34.8 (SD = 15.64) hours than in the ICU, 29.2 (SD = 11.52) hours (Table 2).


[Email Jumpstart To Image]
Table 2. Setting and length of stay

The mean blood glucose levels at the start of dextrose infusion are noted in Table 2. The difference between ICU and regular units was significant (P = .016). There were no significant correlations between the blood glucose level at which dextrose infusion was started and the LOS, resolution of anion gap, or time to transfer order to a regular unit.


The aim of this study was to identify the specific areas of treatment—from intensive and intermediate care areas to less intensive environments in a large teaching hospital—that delay the improvement of DKA and slow progression of patients. While pursuing this aim, the goal was to suggest areas for improvement in the care of patients with DKA.

In this small retrospective study of medical records, no variable was significantly predictive of any of the 3 clinical outcomes—total LOS, time to transfer order to a regular unit, or resolution of anion gap. The positive correlation between length of time before insulin infusion was started and time to resolution of anion gap, as well as the nonsignificant correlation between length of time before insulin infusion was started and total LOS, suggest an area for improvement. Sharing the correlation data with practitioners would inform them and might promote shorter times to initiate the insulin infusion.

Data gathered about the overlap between the insulin infusion and the first dose of subcutaneous insulin are important to clinical practice. Sixty-six percent of patients did not receive the ADA's recommended overlap of at least 60 minutes. Similar suboptimal results were described by Ilag et al in a 2003 study of DKA-treatment practices, where 35% to 48% of patients did not receive depot (subcutaneous) insulin before the discontinuation of insulin infusion. Subcutaneous insulin must be given time to start acting before the intravenous insulin is stopped to keep plasma insulin levels therapeutic during the change. If this overlap does not occur, plasma insulin levels may drop, leading to reduced control. Both physicians and nurses need to understand this concept so that orders can be both placed and carried out correctly. Good communication is essential between the nurses who are closely monitoring hourly blood glucose levels and the physicians who must place orders for specific doses, routes, and times of insulin administration. No literature on the impact of communication in the care of DKA patients was found, although the Joint Commission lists “Improve the effectiveness of communication among caregivers” as a 2008 national patient safety goal.19

MICA had more than twice as many patients receiving an overlap, compared with remaining units. These data indicate that clinicians in all areas, but especially those in intensive care and regular units, may benefit from education regarding appropriate overlap. It is interesting that MICA, which receives about half of DKA cases, had the highest number of cases with appropriate overlap, perhaps because the clinicians there are more familiar with treating patients in DKA. The regular units (other) had the lowest number of cases with appropriate overlap. This also may be an indicator of the lesser capacity for close monitoring and care for DKA patients on these units, possibly due to factors such as lower staffing ratios.

In DKA, it is essential for the physician to follow closely both the blood glucose as well as the other laboratory tests so that measures such as adding dextrose to fluids, laboratory testing, and electrolyte replacements can be ordered in an appropriate and timely manner. In almost 75% of medical records, an order was found to place the patient on the insulin drip protocol for hyperglycemia. A hyperglycemia protocol is not appropriate for patients in DKA. The protocol allows the nurse to regulate the blood glucose level with a combination of regular insulin infusion and intravenous boluses of regular insulin, without taking into account the anion gap or other metabolic derangements that are present in DKA. This often results in correcting hyperglycemia without treating DKA. In all of these cases, the order was present before the anion gap was resolved, that is, while the patient was still in DKA. The insulin infusion protocol was ordered frequently, irrespective of the unit in which the patient was treated.

It is important to point out that although the protocol was ordered, it cannot be assumed that the nurse in charge of the patient's care used it to regulate the blood glucose level. In some cases, however, it is possible that although an order for the protocol existed in the chart, the nurse notified the physician hourly with glucose results, so that proper monitoring was employed. Because of this limitation, presence of an order for the insulin infusion protocol cannot be correlated with clinical outcomes.

The use of evidence-based protocols to reduce unnecessary variation in practice is generally accepted as a beneficial tool when used along with clinical judgment.20 However, a protocol can become a hindrance to care when it is ordered in an inappropriate clinical situation. It is important for everyone involved in the patient's care to be aware of criteria for the use of a specific protocol.

The hospital studied used computer-order entry; as such, physicians were given a limited number or orders from which to choose. From the researcher's clinical experience, it can be difficult to enter properly an order for an insulin infusion without also ordering titration of the infusion using the “insulin drip protocol.” For this reason, it may be that orders are placed inadvertently or with the false assumption that all patients on an insulin infusion must also be on the protocol. It seems prudent to suggest that if an institution uses computer-order entry, order choices for insulin be labeled in a way that distinguishes hyperglycemia protocols from insulin infusions used in the treatment of DKA. This would decrease chance for error in both placing and following orders.

Only 2 cases in the sample had an overlap of 60 minutes or more as recommended by the ADA and did not have an order for the insulin infusion protocol. In contrast, about half of the cases had both an ordered insulin infusion protocol and an insufficient overlap between the insulin infusion and subcutaneous insulin. Singh et al 21 found inadequacies in the DKA treatment by nonspecialist medical staff at a large teaching hospital in the United Kingdom. There was moderate improvement with implementation of clinical guidelines. Ilag et al 17 also found that educating nurses and physicians on a critical pathway for DKA reduced practice variation and decreased LOS in a US teaching hospital.

Bull et al 22 implemented a mandatory protocol for DKA in an ICU in a US teaching hospital. They reported a 23% decrease in LOS in intensive care and 30% decrease in hospital LOS. There was also a significant decrease in length of time to closure of anion gap and ketone clearance.22

Another interesting finding was the level of blood glucose at which dextrose was added to intravenous fluids. The ADA recommends the addition of dextrose to fluids when the blood glucose level reaches 250 mg/dL. The ICU had an average closest to this, at 229 mg/dL, which was significantly different from then regular units at 141 mg/dL. The ADA recommends keeping blood glucose level between 150 and 200 mg/dL until resolution of DKA. Again, this suggests a need for education, especially of clinicians on regular floors. This finding may reflect the decreased capacity for close monitoring on those units.

In a large teaching hospital, it seems sensible to suggest specific education for clinicians, including nurses, interns, and residents who care for patients with DKA, regarding the important tenets of treatment. It also seems prudent to suggest that unless intensive care is necessary, triaging be aimed at directing patients with DKA consistently to one area of the hospital, such as intermediate care. Nurses in this area can be educated to be experts in and become familiar with treating DKA. In large teaching hospitals, it is common for the medical house staff, who direct much of the care, to rotate over a period (eg, months), creating the possibility of inconsistent patterns of care. If the nurses who work in the designated area are experts in DKA, they may act as partners with the rotating medical staff to guide treatment in this group of patients.


This study is limited because it was a retrospective study. Accordingly, no cause-and-effect relationships can be drawn from the results. The small convenience sample was limited by the data available from the medical records. Some charts were more thoroughly documented than others. Retrospective chart reviews are subject to interpretive bias by the investigator. This was controlled for as much as possible by using carefully defined criteria on the data collection instrument.

Of the 143 charts reviewed, only 55 met criteria for the study. Although some charts did not clearly meet the ADA criteria for DKA, others were not included in the study because of missing data. It is probable that some patients met the ADA criteria but were not included in the study because initial laboratory results were not gathered to prove their eligibility.

Further research would benefit from larger sample sizes, with prospective designs to test the usefulness of specific educational interventions and reveal specific areas that are predictive of LOS. It would be interesting to test the concept of deliberately creating groups of RN experts who can manage the care of these patients.


There is an opportunity to educate clinicians who care for patients with DKA, specifically regarding the need for overlap between insulin infusion and subcutaneous insulin, level of blood glucose at which to start dextrose infusion, use of hyperglycemia protocols, and importance of timely initiation of insulin infusion upon presentation. Communication between nurses and physicians at the point of care is vital for positive health outcomes for patients with DKA.


1. Kitabchi AE, Umpierrez GE, Murphy MB, et al. Management of hyperglycemic crises in patients with diabetes. Diab Care. 2001;24(1):131–153. [Context Link]

2. Incidence of diabetes. National Diabetes Surveillance System. National Center for Chronic Disease Prevention and Health Promotion. . Access-ed February 13, 2006. [Context Link]

3. Prevalence of diabetes. National Diabetes Surveillance System. National Center for Chronic Disease Prevention and Health Promotion. . Accessed February 13, 2006. [Context Link]

4. Diabetes Surveillance System. Chapter 1: the public health burden of diabetes mellitus in the United States; Diabetic Ketoacidosis. National Center for Chronic Disease Prevention and Health Promotion. . Accessed February 13, 2006. [Context Link]

5. Diabetic ketoacidosis as first-line diagnosis. Average length of stay (LOS) in days of hospital discharges with diabetic ketoacidosis as first-listed diagnosis, United States, 1980–2003. National Center for Chronic Disease Prevention and Health Promotion. . Accessed February 13, 2006. [Context Link]

6. Fishbein H, Palumbo P. Acute Metabolic Complications in Diabetes. Bethesda, MD: National Diabetes Data Group, National Institutes of Health; 1995. NIH Publication No. 95–1468. [Context Link]

7. Basu A, Close CF, Jenkins D, Krentz AJ, Nattrass M, Wright AD. Persisting mortality in diabetic ketoacidosis. Diab Med. 1993;10(3):282–284. [Context Link]

8. Delaney MF, Zisman A, Kettyle WM. Diabetic ketoacidosis and hyperglycemic hyperosmolar nonketotic syndrome. Endocrinol Metab Clin N Am. 2000;29(4):683–705, V. [Context Link]

9. Umpierrez GE, Kitabchi AE. Diabetic ketoacidosis: risk factors and management strategies. Treat Endocrinol. 2003;2(2):95–108. [Context Link]

10. Javor KA, Kotsanos JG, McDonald RC, Baron AD, Kesterson JG, Tierney WM. Diabetic ketoacidosis charges relative to medical charges of adult patients with type I diabetes. Diab Care. 1997;20(3):349–354. [Context Link]

11. Umpierrez GE, Latif K, Stoever J, et al. Efficacy of subcutaneous insulin lispro versus continuous intravenous regular insulin for the treatment of patients with diabetic ketoacidosis. Am J Med. 2004;117(5):291–296. Bibliographic Links [Context Link]

12. Haas RM, Hoffman AR. Treatment of diabetic ketoacidosis: should mode of insulin administration dictate use of intensive care facilities? Am J Med. 2004;117(5):357–358. [Context Link]

13. Maldonado MR, Chong ER, Oehl MA, Balasubramanyam A. Economic impact of diabetic ketoacidosis in a multiethnic indigent population: analysis of costs based on the precipitating cause. Diab Care. 2003;26(4):1265–1269. [Context Link]

14. Kitabchi AE, Umpierrez GE, Murphy MB, et al. Hyperglycemic crises in diabetes. Diab Care. 2004;27(suppl 1):S94–S102. [Context Link]

15. Levetan CS, Passaro MD, Jablonski KA, Ratner RE. Effect of physician specialty on outcomes in diabetic ketoacidosis. Diab Care. 1999;22(11):1790–1795. [Context Link]

16. Wiest FC, Ferris TG, Gokhale M, Campbell EG, Weissman JS, Blumenthal D. Preparedness of internal medicine and family practice residents for treating common conditions. JAMA. 2002;288(20):2609–2614. Bibliographic Links [Context Link]

17. Ilag LL, Kronick S, Ernst RD, et al. Impact of a critical pathway on inpatient management of diabetic ketoacidosis. Diab Res Clin Pract. 2003;62(1):23–32. Bibliographic Links [Context Link]

18. Sola E, Garzon S, Garcia-Torres S, Cubells P, Morillas C, Hernandez-Mijares A. Management of diabetic ketoacidosis in a teaching hospital. Acta diabetol. 2006;43(4):127–130. Bibliographic Links [Context Link]

19. The Joint Commission. 2008 National Patient Safety Goals Hospital Program. . Accessed December 3, 2007. [Context Link]

20. Meade MO, Ely EW. Protocols to improve the care of critically ill pediatric and adult patients. JAMA. 2002;288(20):2601–2603. Bibliographic Links [Context Link]

21. Singh RK, Perros P, Frier BM. Hospital management of diabetic ketoacidosis: are clinical guidelines implemented effectively? Diab Med. 1997;14(6):482–486. [Context Link]

22. Bull SV, Douglas IS, Foster M, Albert RK. Mandatory protocol for treating adult patients with diabetic ketoacidosis decreases intensive care unit and hospital lengths of stay: results of a nonrandomized trial. Crit Care Med. 2007;35(1):41–46. [Context Link]

Key words: diabetic ketoacidosis; hyperglycemia; insulin; protocol