INTRODUCTION
Hyperglycemic crises are associated with substantial morbidity, mortality and healthcare costs. In the U.S., the mean length of stay (LOS) for patients hospitalized with Diabetic Ketoacidosis (DKA) is currently 3-4 days, with costs ranging from $21,215-$36,600 per admission. The inpatient mortality rate during hospital admission for DKA ranges from 0.20%-1.04%, with patients having a 1-year corrected mortality rate that is 13 times higher than the general population after discharge.1
DKA is a serious, acute and life-threatening hyperglycemic emergency characterized by a triad of hyperglycemia (glucose > 200 or prior history of diabetes), increased serum and/or urine ketone concentrations (B-hydroxybutyrate concentration > 3 mmol/L or urine ketone strip > 2+) and metabolic acidosis (pH < 7.3 and/or bicarbonate concentration < 18 mmol/L).1 DKA typically results from severe insulin deficiencies and elevation of counterregulatory hormones, more commonly identified in patients with Type 1 Diabetes (T1D).1 Early diagnosis and management of DKA is essential to improve outcomes. The mainstay of treatment for DKA includes fluid replacement, insulin therapy, electrolyte repletion and treatment of underlying precipitating events, most commonly infection or noncompliance with antidiabetic medications.1
For over a decade, the American Diabetes Association (ADA) has routinely recommended two intravenous (IV) insulin administration techniques for the management of hyperglycemic crises.2 For patients with DKA, a 0.1 unit/kg IV bolus followed by a 0.1 unit/kg/hr insulin infusion or a 0.14 units/kg/hr IV continuous insulin infusion were suggested.2 Utilization of the insulin bolus was rationalized to overcome insulin resistance, promptly initiate reversal of ketogenesis and accelerate metabolic correction. The ADA noted that these recommendations were made with the knowledge of conflicting randomized studies that questioned whether an initial bolus was necessary. In June 2024, the ADA released a consensus report recommending subcutaneous or IV insulin therapy depending on the severity of DKA. Short-acting insulin administered IV by continuous infusion is the preferred choice, especially for moderate-severe DKA. In adults, treatment protocols recommend the initial administration of an insulin bolus (0.1 units/kg IV or subcutaneously) if a delay in obtaining venous access is anticipated, which is then followed by fixed-rate IV insulin infusion (0.1 units/kg/hr).1 Despite the recommendation to administer an initial bolus and a theoretical benefit in DKA, the data to support this practice is limited.3 Additionally, there are concerns regarding the safety of this practice due to potential risks of causing hypoglycemia and hypokalemia. Initial insulin bolus is furthermore recommended against in end-stage renal disease (ESRD) patients based on the potential to cause overly rapid blood glucose (BG) correction leading to hypoglycemia.3
Goyal et al. (2010), a prospective observational cohort study conducted over a 12-month period, found no meaningful differences in outcomes when comparing the initial insulin bolus plus infusion versus infusion alone.4 In 2018, Brown and colleagues similarly assessed whether a bolus accelerates DKA resolution and reported no improvement in recovery time or safety measures.5 Again, Jouini et al. (2022) compared these regimens and likewise observed comparable safety and efficacy between the two approaches.6 In contrast, Fischer and Celmins (2024) recently identified higher complication rates among patients who received a bolus without any evidence of faster DKA resolution, suggesting that bolus administration may introduce additional risk without improving clinical outcomes.3
As presently published literature evaluating insulin bolus plus infusion versus infusion alone in the management of DKA has produced inconsistent and conflicting results, the overall clinical impact of these regimens remains uncertain. Prior studies have reported no meaningful differences between strategies, while others have suggested potential safety concerns without clear evidence of benefit, producing uncertainty regarding the optimal approach for managing these critical patients. Given the lack of consensus within the literature, our study was designed to more clearly determine whether initiating therapy with an insulin bolus plus infusion, compared with infusion alone, leads to meaningful differences in clinical outcomes for patients with DKA.
MATERIALS AND METHODS
Definitions and Data Collection
This study was an IRB-approved, single-center, retrospective cohort evaluation that was conducted at a large tertiary care hospital, high-volume emergency department. All data were obtained from the institution’s electronic health record. Patients were consecutively screened in reverse chronological order until 125 patients were included in each group.
Patients were included if they were 18 years of age or older, admitted with a diagnosis of DKA per ICD-10 coding and received IV insulin in the ED between October 1, 2024 and December 15, 2025 via standardized order set. Providers may utilize the ED’s DKA order set to select their insulin regimen; the bolus is optional as it is not prechecked; however, if selected, a 5, 7, or 10-unit bolus may be given. The predetermined insulin infusion rate of 5, 7, or 10 units/hr is then selected based on the patient’s weight being between 50-69, 70-99 or > 100 kg respectively. Patients were excluded if pregnant, received subcutaneous insulin therapy for DKA, received multiple IV insulin boluses or were transferred out of the institution within 6 hours of insulin initiation.
The primary outcome was a composite safety outcome of interruption of insulin infusion (documentation of 0 units/hr., indicating a temporary pause in therapy), the incidence of hypoglycemia (serum blood glucose < 70 mg/dL) or hypokalemia (serum potassium < 3.4 mEq/L) within 6 hours of insulin bolus administration or continuous infusion initiation. Secondary outcomes included the time to discontinuation of insulin infusion, need for dextrose 50% administration after insulin administration, ICU and hospital length of stay and all-cause mortality during hospital admission.
Statistical Analysis
Mann-Whitney U was used to analyze continuous data and the chi-square test or Fisher’s exact test was used to analyze categorical data where applicable. Our power analysis included an enrollment goal of 250 patients which would provide the study with a power of 95% to detect a 22% higher incidence of the primary composite outcome occurring in the bolus group. A two-sided alpha level of 0.05 was set and a p-value of < 0.05 was considered to be statistically significant. All continuous data are presented as medians with interquartile ranges (IQR) and all categorical variables are expressed as numbers with percentages unless otherwise stated. All analyses were performed using IBM® SPSS Statistics 31.
RESULTS
Patient Population and Baseline Characteristics
A total of 467 charts were screened and 323 patients were eligible for inclusion in this study. Of those eligible, random selection was performed to obtain 250 patients, 125 who received an insulin bolus plus infusion and 125 who received insulin infusion alone for the final analysis. As there were fewer patients who received the bolus compared to those who only received the infusion, in order to avoid selection bias, data on both groups were collected until 125 patients could be secured within each. A total of 144 patients were excluded from this study (no DKA diagnosis code n = 120, insulin infusion not administered n = 20, pregnant n = 2, received > 1 IV insulin bolus n = 1, ordered IV insulin outside ED DKA order set n = 1).
Baseline characteristics are outlined in Table 1. A statistically significant difference was noted in the age, sex, first episode of DKA, end-stage renal disease status, serum creatinine, initial serum glucose, beta-hydroxybutyrate level and anion gap between the bolus and non-bolus groups.
Primary and Secondary Outcomes
The primary composite outcome occurred in 34 (27.2%) patients who received a bolus and 22 (17.6%) patients who received an infusion only (p = 0.06). Component outcome analysis revealed that there was a 14% difference between groups in the interruption of insulin infusion (22 vs 4). To adjust for observed differences noted in baseline characteristics, a post-hoc binary logistic regression was performed (Table 3). Neither age, sex, first episode of DKA, end-stage renal disease status, serum creatinine, initial serum glucose or beta-hydroxybutyrate level were associated with the composite outcome. There was a non-significant trend toward a shorter insulin infusion time (13:59 vs 15:44, p = 0.43) as well as the ICU length of stay (27:45 vs 30:10, p = 0.06) in the bolus plus infusion group, as compared to those who received an insulin infusion alone. There was no significant difference in the need for dextrose 50% administration within six hours of insulin infusion initiation between the continuous infusion only group versus the bolus group (13 vs 10, p = 0.51). There was no difference in all-cause mortality between the bolus and continuous infusion groups (2 vs 1, p = 1).
DISCUSSION
Management of DKA has undergone meaningful evolution over the past decade, yet clinical practice remains highly variable, particularly regarding the use of an initial IV insulin bolus. Earlier ADA guidelines endorsed either a 0.1 unit/kg bolus followed by infusion or a higher fixed-rate infusion alone, acknowledging limited and conflicting evidence supporting bolus administration.2 Despite this uncertainty, many institutions continue to incorporate bolus dosing into their DKA protocols, driven by theoretical benefits, such as overcoming insulin resistance and accelerating metabolic recovery. However, concerns regarding hypoglycemia and hypokalemia have prompted renewed scrutiny of this practice.
In this large, single-center retrospective cohort study of adults presenting to the emergency department with DKA, administration of an intravenous insulin bolus prior to continuous infusion did not result in a statistically significant increase in the composite safety outcome of hypoglycemia, hypokalemia or interruption of insulin infusion within the first six hours of therapy. Although not significant, our study also found that patients who received a bolus transitioned off the insulin infusion, 1 hr and 45 min sooner and were transferred out of the ICU, 2 hrs and 25 min earlier which may indicate faster time to DKA resolution.
The results of INVOKE align with several earlier studies demonstrating no clear difference in utilizing an initial insulin bolus over infusion alone. Prior prospective and retrospective evaluations consistently reported no significant differences in time to DKA resolution, hypoglycemia, hypokalemia or LOS between bolus and non-bolus regimens. Goyal et al. (2010) found no significant difference in the incidence of hypoglycemia (6% vs. 1%, p = 0.12) or LOS in the emergency department (8 vs. 7 h, p = 0.37) or inpatient (5.6 vs. 5.9 days, p = 0.81).4 However, as that study did not directly measure glucose values and instead relied on administration of dextrose as a surrogate marker for hypoglycemia, their ability to assess true hypoglycemic events was limited. In contrast, our study evaluated hypoglycemia using explicit laboratory measurement of serum glucose, allowing for a more accurate comparison of glycemic outcomes between treatment strategies. Brown and colleagues (2018) found that the time to DKA resolution in the ED did not differ significantly between the insulin bolus and no bolus group (15 vs. 15.9 h, p = 0.24) as well as the incidence of hypoglycemia (2 vs. 7%, p = 0.64), hypokalemia (16 vs. 29%, p = 0.65), or length of hospital stay (3.2 vs. 2.7 days, p = 0.27).5 Similarly in 2022, Jouini found no differences when it came to time to recovery of DKA (17 vs. 16 hours, p = 0.76), hypoglycemia (11.5% vs. 15.9%, p = 0.57) and hypokalemia (56.1% vs. 46.9%, p = 0.30).6 These studies unfortunately did not collect data on renal function, including serum creatinine or end-stage renal disease status, despite their relevance to insulin clearance and electrolyte disturbances. Baseline serum glucose levels also significantly differed between groups in the 2018 study, yet this imbalance was not adjusted for using multivariable regression. Relative to these findings, our study incorporated collection of serum creatinine, ESRD status and controlled for all baseline characteristics through logistic regression, allowing for a more rigorous comparison between strategies.
In our study, the incidence of hypoglycemia and hypokalemia remained low and nearly identical between groups, reinforcing the notion that choice of either regimen does not significantly increase these specific safety risks, although the difference of about 10% may be of clinical significance. Despite the fact that an analysis cannot be performed on components of the primary outcome, interruption of insulin infusion was seen at a much higher frequency in those who received a bolus, which was the primary driver of the composite outcome. Additionally, time to insulin discontinuation and hospital LOS were similar between groups, suggesting that the impact of bolus dosing on overall recovery may be limited.
Our findings diverge from Fischer & Celmins (2024), who observed a significantly higher rate of early hypoglycemia, hypokalemia and infusion interruptions in bolus-treated patients (47.4% vs 25%, p = 0.003). This was driven primarily by hypokalemia which occurred in nearly twice as many patients who received an insulin bolus as the group who did not (22% vs 12%, p = 0.088).3 While INVOKE similarly identified more frequent infusion interruptions in the bolus group, the absence of increased metabolic complications suggests that the mechanism driving interruptions may be multifactorial. One explanation may be that nursing staff in the ICU closely obtained and monitored trends in serum glucose and potassium levels hourly for patients with DKA, therefore, assessing the need to interrupt the insulin infusion if a downtrend in these values was noted to prevent hypoglycemia and hypokalemia from occurring. Our study also observed that patients with baseline characteristics exhibiting a higher degree of laboratory abnormalities were more likely to receive an insulin bolus plus infusion. More patients in the bolus cohort presented with a higher initial serum glucose and potassium level, which may have influenced provider decisions on utilization of an insulin bolus. Conversely, our study discovered that significantly more patients who presented with ESRD status received an insulin bolus plus infusion despite the potential of hypoglycemia. This discrepancy suggests that bedside decision-making may have been influenced more by perceived illness severity and metabolic derangements rather than physiologic considerations unique to ESRD and insulin clearance.
This study has several strengths including its relatively large sample size; the largest to date evaluating insulin bolus versus infusion only strategies in DKA with a focus on early safety outcomes within 6 hours, a clinically relevant window during which insulin-related complications are likely to occur in the ED. Our study collected key baseline demographic information on serum creatinine and end-stage renal disease status, which may impact insulin management. We additionally adjusted for statistically significant differences in baseline characteristics through logistic regression. The composite primary endpoint was chosen in order to capture both definitive and potential complications that may result from the insulin therapy technique chosen. While hypoglycemia and hypokalemia are objective measures of adverse drug effects from insulin administration, collecting data on interruption of insulin infusion identified situations where intervention may have been required to prevent harm, including impending hypoglycemia or hypokalemia. Selecting only one of these components as a single primary outcome would miss key safety events that can occur when administering insulin to these critical patients. The use of a standardized ED DKA order set at our institution also reduced variability in insulin dosing and monitoring practices.
Several limitations must be noted. The retrospective single-center design limits generalizability of our results and introduces the possibility of selection and documentation biases. Residual confounding by indication may still be present despite multivariable logistic regression as clinicians may have preferentially administered insulin bolus therapy to patients perceived to have more severe illness, pertaining to characteristics potentially not assessed in our study at baseline. Use of a primary composite outcome may preclude this study to inflation bias, whereas having any one of three components could facilitate easier achievement of significance. DKA resolution was assessed through discontinuation of the insulin infusion as a surrogate outcome, as our institution has not yet transitioned from use of anion gap closure to serial plasma ketone collection to objectively declare resolution in accordance with the ADA 2024 consensus report. Only data on dextrose 50% administration were collected; therefore, alternative dextrose-containing fluids administered for hypoglycemia were not captured. According to our study, the median insulin dose administered for DKA patients in the ED was 0.08 units/kg, which is less than the ADA recommended dose. This may explain the overall lower incidence of adverse effects observed as compared to previous studies.
Prospective multicenter randomized controlled trials are needed to improve generalizability, evaluate both safety outcomes and efficacy of DKA resolution and identify patient subgroups who may benefit from insulin bolus administration.
CONCLUSION
The use of an insulin bolus plus continuous infusion in the emergency department for the management of DKA was not associated with an increase in the composite of hypoglycemia, hypokalemia or interruption of insulin infusion. This is the largest study to date comparing these two insulin regimens and the results of this study add to existing literature suggesting no meaningful difference between them. Emergency medicine provider awareness of ADA’s 2024 shift toward insulin infusion-only therapy and literature that supports this shift is essential to minimize injudicious use of insulin bolus therapy, long considered to be standard practice.
Funding
No external funding was received for this study.
Conflicts of Interest
The authors declare no financial relationships or other relevant conflicts related to this work.
INSTITUTIONAL REVIEW BOARD STATEMENT
The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of Lakeland Regional Health (IRB#FY2026-2; date of approval: December 30, 2025).
INFORMED CONSENT STATEMENT
Patient consent was waived due to the retrospective nature of this study which does not allow access to research subjects in real-time to obtain informed consent.
DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors on request.
ACKNOWLEDGEMENTS
The authors thank the Pharmacy Data Team and Department of Emergency Medicine for their invaluable support in statistical analysis software guidance and review of results.
