Diabetic ketoacidosis (DKA) is a common endocrine emergency encountered in the emergency department.  DKA associated mortality is relatively low in adults, but in children with type 1 diabetes, the elderly, and adults with concomitant illnesses have a mortality rate is > 5% [cite source=”pubmed”]19564476[/cite].  Guidelines for the management of hyperglycemic crisis in adults provide recommendations for intravenous fluid administration, correction of electrolyte abnormalities, insulin and bicarbonate therapy. While the recommendations made in the American Diabetes Association (ADA) consensus statement are intended to be evidence based, there are two recommendations which have less than optimal supporting evidence which results in controversy in the emergency department: 1. Use of regular insulin boluses of 0.1 units/kg and 2. patients with a pH < 6.9 should receive sodium bicarbonate therapy. Today we will attempt to answer the question, is there any benefit to an initial insulin bolus in DKA?
Is there a benefit to an initial insulin bolus in diabetic ketoacidosis?
Many prospective randomized trials have laid bare the use of low-dose insulin infusion leading to the successful recovery of patients with DKA. However, the data supporting an initial insulin bolus prior to the initiation of insulin infusions is not nearly as robust. The rationale for such a bolus is to overcome the relative insulin deficiency seen in DKA in order to suppress lypolysis and hepatic gluconeogenesis and limit further acidosis (more on that next time). However, insulin boluses may lead to harm including hypoglycemia, hypokalemia, and if glucose levels are too rapidly corrected, cerebral edema [1]. Since the publication of the ADA consensus statement, two investigations have attempted to answer the question of what affect insulin bolus has on patients with DKA [1] [2].
Take Home Points
- Insulin boluses at the start of an insulin infusion DO NOT:
-
- Decrease time to normalization of glucose, pH, or bicarbonate levels
- Affect the rate of change of glucose or anion gap
- Reduce ED or hospital length of stay
- Insulin boluses are associated with numerically higher, but statistically insignificant incidence of hypoglycemia requiring treatment with dextrose
- Clinical Bottom Line:Â There is no benefit to an insulin bolus before the start of an insulin infusion in DKA and may cause worsening hypoglycemia and hypokalemia.
References:
- Goyal N et al. Utility of Initial Bolus Insulin in the Treatment of Diabetic Ketoacidosis. J Emerg Med 2010. PMID: 18514472
- Kitabchi AE et al. Is a Priming Dose of Insulin Necessary in a Low-Dose Insulin Protocol for the Treatment of Diabetic Ketoacidosis? Diabetes Care 2008. PMID: 18694978
Darrel Hughes
Latest posts by Darrel Hughes (see all)
- Benzodiazepine-Refractory Alcohol Withdrawal - April 28, 2016
- Why You Should More Than Consider a Vasopressin, Steroid, and Epinephrine (VSE) Cocktail - October 29, 2015
- Topical Tranexamic Acid for Epistaxis or Oral Bleeds - July 14, 2014
English
Chinese (Simplified)
English
French
German
Italian
Portuguese
Russian
Spanish
Turkish
19 thoughts on “Is There Any Benefit to an Initial Insulin Bolus in Diabetic Ketoacidosis?”
Great article Darrel.. I love practical info I can use. It validates my practice this time.
Hey Larry,
Thanks for reading. Be expecting more “practical info” for clinical practice….
Salim
Great stuff Darrel. Looking forward to you taking down the Bicarb crowd in your next post.
Hey Darrel! Are these non-inferiority studies powered to detect the secondary outcome differences, since these are the major arguments in favor of using bolus?
Hello Bin,
Great point, the studies were not large enough to be statistically significant for secondary outcome differences, but there was a trend toward bolus causing more harm (i.e. hypokalemia & hypoglycemia), and absolutely no benefit (i.e primary outcomes).
Salim
Hi Bin,
Neither study is powered for the secondary outcomes. In fact, the Goyal paper is not powered for its primary outcome of hypoglycemia requiring dextrose (post hoc power analysis of ~ 40%). In the absence of powered research to address these questions, we must cautiously interpret the data we have available and understand the limitations of them. Great question, and thanks for reading.
darrel
Sorry to join this discussion (very) late. While the study was not designed as a non-inferiority analysis, our study established clinically relevant ranges for our secondary endpoints, and examined whether the 95% CI for the data fell within the relevant ranges. We did not merely presume equivalence based on failure to find a statistically significant difference. This is described in detail in the manuscript Table 2.
This approach to establishing equivalence based on confidence intervals is described here: http://www.graphpad.com/guides/prism/6/statistics/index.htm?stat_testing_for_equivalence_with_c.htm
Hello Nikhil,
First of all TY for taking the time to come on to the site and leave a comment. We truly appreciate when PIs and authors do that. It is important to hear all sides of the discussion before coming to a conclusion. Your point is very valid that statistical significance does not always equate to clinical significance. Bolus insulin however does still cause a 6% vs 1% hypoglycemia event rate based on the study and that is both clinically significant as well as statistically significant. I think the point you are trying to make is that there were clinical trends toward longer ED length of stay and longer hospital stays with bolus insulin that did not necessarily meet statistical significance in this study . Please correct me if you were trying to suggest differently and again TY so much.
Salim
In our study, the 6% vs 1% hypoglycemia rate (defined as need for D50) was not statistically significant. Therefore we cannot conclude that the difference observed was purely due to chance, vs. an actual difference in hypoglycemia rates when insulin is bolused or not.
In the comment above, I was trying to clarify our conclusion, that “initial bolus dose of insulin … demonstrated equivalent changes in clinically relevant (secondary) endpoints when compared to patients not administered the bolus.” We judged “equivalence” based on a confidence interval comparison standard as I linked above, not from non-inferiority analyses or by failing to find a statistically significant difference.
For example, per our study we can be 95% certain that the rate of change of anion gap for the bolus group was in the range of 0.5 mEq/L/h slower to 0.32 mEq/L/h faster than the non-bolus group. Even if the “true” value were at either end of this range, we proposed that most clinicians would consider this difference not large enough to justify a management decision, hence functionally “equivalent.”
http://dx.doi.org/10.1016/j.jemermed.2007.11.033
Great article Darrel. Now to update the DKA order set.
Nice work Darrel! Thanks for the quick side-by-side comparison of the two trials!
Awesomeness. Welcome to the blogosphere, Darrel! Looking forward to reading more.
Good food for thought Darrel and smart profile picture too. At the end of the day I think there is still not great data on the topic. The take home point for me is we need better and larger studies. Seems like the secondary outcomes are the bait, but the catch? Not so interesting primary outcomes. The other take home point is the ADA should probably utilize better methodology when disseminating guidelines. Too bad they do not use the GRADE. I look forward to more of your posts on REBELEM.
Our current hospital DKA protocol uses an insulin bolus, The hospital cites the 2013 ADA DKA recommendations as reason to do so. I can’t find any guidelines from 2013. Anyone have that link?
The 2009 ADA guidelines state there is evidence that shows a bolus is NOT necessary. But, unfortunately they give both options…. 0.1 bolus followed by gtt of 0.1 U per hour OR 0.14 U per hour-no bolus
other thoughts?
Hey Scott,
This is the 2009 ADA Guidelines Link: http://care.diabetesjournals.org/content/32/7/1335 . I am not sure I have seen a 2013 guidelines update….as far as I know 2009 was the last iteration. There is a thorough review article from 2014 which what I think you are talking about and the link is here: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4085289/pdf/dmso-7-255.pdf . Hope this helps.
Salim