The timing of basal insulin administration in DKA has been controversial for years. I posted the following poll on twitter recently to get a sense of current opinion on this:
So, it seems that the majority of people are using traditional management (wait for the anion gap to close, give basal insulin, then shut off the insulin infusion two hours later). Many are using early basal insulin. And a few people chose the last option, which is incorrect.
It’s not surprising that differences in practice exist, given that guidelines disagree on this topic. The British guidelines (which are generally forward-thinking) have been recommending early basal insulin for many years. Meanwhile, the American guidelines make no mention of this (however, currently 2009 USA guidelines are outdated and moderately awful).1
This post will explore the rationale for early basal insulin, available evidence in adult DKA, and conclude with some technical pointers.
rationale for early basal insulin
(1) It may take glargine or detemir longer than two hours to reach peak effect.
The traditional practice of stopping insulin two hours after initiation of basal insulin was originated when NPH insulin was more commonly used as long-acting insulin. Newer types of basal insulin (especially detemir) may take more than two hours to reach peak effect. This could theoretically lead to a period of time when the insulin level is sub-therapeutic (leading to hyperglycemia and perhaps even recrudescence of DKA).
(2) Initiation of basal insulin immediately protects against premature discontinuation of the insulin infusion.
One common mistake in the management of DKA is premature discontinuation of the insulin infusion (either accidentally or in a misguided attempt to manage hypoglycemia). This may cause rapid widening of the anion gap.
Immediate initiation of basal insulin doesn’t protect entirely against this error. If the insulin infusion is discontinued while the patient is still acidotic, the patient may be highly insulin-resistant and thus still at-risk for widening of the anion gap. However, the presence of basal insulin will tend to mitigate the harm caused by premature discontinuation of the insulin infusion.
(3) Streamlined discontinuation of the insulin infusion
Immediate initiation of basal insulin may eliminate the ~3 hour waiting period which is otherwise required prior to discontinuation of the insulin infusion. Traditionally, the provider waits until the patient is ready to discontinue the insulin infusion. Then the provider will first order long-acting insulin from pharmacy. Subsequently, after the long-acting insulin arrives and is administered, the insulin infusion can be stopped two hours later. The process of ordering long-acting insulin, receiving it from pharmacy, and then waiting two hours takes about three hours.
If basal insulin is provided up-front, there is no need to delay discontinuation of the insulin infusion as above. Once the patient meets criteria for discontinuation of the insulin infusion, it is simply turned off.
brief literature review
Hsia et al. 2012
This is a single-center prospective RCT involving 61 diabetic patients with hyperglycemia.2 Patients were randomized to receive treatment with an insulin infusion with or without weight-based glargine (0.25 U/kg). Early administration of glargine reduced the incidence of rebound hyperglycemia when the insulin infusion was discontinued, without increasing the risk of hypoglycemia. Unfortunately, only 25 patients in this study had DKA. However, the beneficial effect of early glargine was observed in the subset of patients with DKA.
Doshi et al. 2015
This is a two-center pilot RCT comparing early basal glargine versus glargine initiated after the anion gap closed.3 Glargine was dosed in a weight-based fashion (0.3 U/kg) in both groups. Forty patients were included. No differences were found:
Houshyar et al. 2015
This is very similar to the Doshi trial, being an RCT with forty patients.4 Patients were randomized to early basal glargine versus glargine initiated after closure of the anion gap. These authors also used weight-based glargine dosing, albeit at a higher dose (0.4 U/kg) in both groups.
No differences were found in the duration of acidosis (the primary endpoint). Hyperglycemia following discontinuation of the insulin infusion was less common in patients treated with early basal glargine (35% of measurements vs. 51% showed a glucose >8.3 mM or >150 mg/dL, p =0.046). This validates the original study by Hsia et al. above.2
new study: Rappaport et al. 2019
This is a retrospective study comparing early versus late initiation of basal insulin at the University of Vermont Medical Center.5 I have an intellectual conflict of interest due to my involvement. However, the vast bulk of work on the publication was performed by our outstanding pharmacy team (especially Stephen Rappaport).
In 2014, based upon newly released British DKA guidelines, we began the practice of resuming patients’ home-dose basal insulin. Upon admission with DKA, patients would be treated promptly with their prior dose of basal insulin (typically glargine). This is somewhat different from previous studies described above (which used weight-based dosing).2–4
This is a retrospective study comparing patients treated with home-dose basal insulin prior to DKA resolution, compared to patients who were initiated on basal insulin later on (typically after closure of the anion gap). The main exclusion criteria were patients not admitted to the medical ICU, patients treated with a dose of insulin different from their home dose, or patients with a new diagnosis of diabetes:
The patient groups were similar, as shown below. The main differences were that patients treated with early basal insulin were mostly treated between 2014-2015, whereas late initiation of basal insulin was mostly performed in 2012-2013.
Most outcomes didn’t reach statistical significance (table below). However, every single outcome trends in a direction favoring early initiation of basal insulin. Early basal insulin correlated with reduced duration of the insulin infusion by three hours, which makes perfect sense (this equals the time required to order basal insulin and wait two hours for it to take effect). This time difference was too small to translate into a significant difference in ICU or hospital length-of-stay. Patients treated with early basal insulin tended to have smoother transition off the insulin infusion, thus requiring fewer measurements of fingerstick glucose (echoing data from Houshyar et al. and Hsia et al).2,4
Based on the retrospective correlational nature of this study, it cannot prove anything. However, it does support the idea that early home-dose basal insulin is safe and feasible. We have continued to have favorable results with this, which remains our routine practice today.
technical notes: tips about early home-dose basal insulin
(1) Give the patient’s full 24-hour basal insulin prescription as a single dose.
Determine the total amount of basal insulin that the patient is on over a 24-hour period. For patients on daily glargine or detemir, this is simply their home dose. For patients are on twice-daily glargine, this will include two doses of glargine. For patients on an insulin pump, the basal insulin can be calculated by integrating the basal rate of insulin over a 24-hour period (of course, in such patients, the pump should be discontinued).
The full dose of basal insulin should be provided in the form of a once-daily insulin dose (e.g. glargine or detemir). Administration of reduced doses will lead to confusion subsequently, as it will be unclear whether the patient has received basal insulin.
(2) Basal insulin doesn’t need to be administered immediately upon arrival, but ideally should be given well in advance of closure of the anion gap.
There is no necessity to provide basal insulin immediately upon admission. Ideally, basal insulin should be provided well in advance of discontinuation of the insulin infusion (e.g. >4 hours before) to facilitate a smooth transition.
- For a patient with mild DKA, prompt basal insulin may be helpful (in anticipation that the insulin infusion will be required for only a few hours).
- For a patient with severe DKA, immediate basal insulin is less important (because the patient will likely be on the insulin infusion for >6-10 hours).
(3) Pay close attention to the administration times of basal insulin in electronic orders.
Ideally, basal insulin should be administered roughly every 24 hours. When ordering “daily” insulin in an electronic medical record, this often defaults to a specific time (e.g. every day at 9 AM). If the first dose of basal insulin was provided at 9 PM, then this could result in two doses being spaced only 12 hours apart (causing doses to overlap, thereby increasing the risk of hypoglycemia). This problem can usually be avoided by ordering the insulin “q24hr” instead of “daily.” Also, double-check the scheduled administration times to ensure that the order is correct.
(4) Be wary of patients with acute physiologic stress.
If DKA is the result of medication nonadherence, then resumption of the patient’s home dose of basal insulin should work perfectly. Unfortunately, some patients have DKA due to an acute physiologic stressor (e.g. infection). If this physiologic stressor lingers, then the patient may have an increased requirement for basal insulin until resolution of the acute physiologic stress. Therefore, it cannot necessarily be assumed that the home-dose basal insulin will be sufficient.
Patients with ongoing physiologic stress (e.g. pancreatitis, sepsis) will be harder to bridge off the insulin infusion. Such patients may require an increased basal dose of insulin until their stress resolves. Monitor carefully for rebound hyperglycemia and DKA following discontinuation of the insulin infusion. Patients with sepsis and DKA may require an extra day or two to stabilize before leaving the ICU; that’s OK.
(5) Don’t discontinue the insulin infusion until the patient meets criteria to do so.
Criteria for discontinuation of insulin infusion are typically the following:
- Closure of the anion gap (e.g. uncorrected anion gap <12 mEq/L)
- Resolution of acidosis (e.g. serum bicarbonate >18 mEq/L)
- Patient has received long-acting basal insulin (at least two hours previously)
- Patient feels clinically improved and is ready to eat (although this isn't an absolute requirement for patients with gastrointestinal pathology)
- Glucose is reasonably well controlled (e.g. <250 mg/dL or <14 mM)
The use of basal insulin may create a temptation to discontinue the insulin infusion prematurely (before meeting the above criteria). This should be resisted, because premature cessation of insulin infusion increases the risk of rebound hyperglycemia and recrudescent DKA. In particular, it should be borne in mind that metabolic acidosis increases insulin resistance, so if the patient has persistent acidosis then their basal insulin dose may be inadequate to prevent them from slipping back into DKA (see #2 above).
- Early administration of home-dose basal insulin prescription is suggested by the British DKA guidelines. A recent retrospective study supports the safety and efficacy of this strategy.
- Early administration of basal home-dose insulin may facilitate slightly faster (~3 hours) and smoother discontinuation of the insulin infusion. However, this is unlikely to lead to major differences in length-of-stay.
- Prompt home-dose basal insulin may be of greatest value in a subset of patients with mild DKA, who could potentially receive an insulin infusion for <12 hours and thereby avoid ICU admission. This would be most relevant to healthcare systems wherein mild DKA is resolved within an emergency department or ED-ICU context.
- Future studies of early home-dose basal insulin should arguably focus on patients with mild DKA. Although early home-dose basal insulin is safe and effective in all patients, demonstrating statistical benefit is most likely among patients with mild, uncomplicated DKA.
- IBCC DKA chapter (& additional links in this chapter)
- Early glargine administration at start of DKA treatment (Bryan Hayes, UMEM pearls)
- 1.Kitabchi A, Umpierrez G, Miles J, Fisher J. Hyperglycemic crises in adult patients with diabetes. Diabetes Care. 2009;32(7):1335-1343. https://www.ncbi.nlm.nih.gov/pubmed/19564476.
- 2.Hsia E, Seggelke S, Gibbs J, et al. Subcutaneous administration of glargine to diabetic patients receiving insulin infusion prevents rebound hyperglycemia. J Clin Endocrinol Metab. 2012;97(9):3132-3137. https://www.ncbi.nlm.nih.gov/pubmed/22685233.
- 3.Doshi P, Potter A, De L, Banuelos R, Darger B, Chathampally Y. Prospective randomized trial of insulin glargine in acute management of diabetic ketoacidosis in the emergency department: a pilot study. Acad Emerg Med. 2015;22(6):657-662. https://www.ncbi.nlm.nih.gov/pubmed/26013711.
- 4.Houshyar J, Bahrami A, Aliasgarzadeh A. Effectiveness of Insulin Glargine on Recovery of Patients with Diabetic Ketoacidosis: A Randomized Controlled Trial. J Clin Diagn Res. 2015;9(5):OC01-5. https://www.ncbi.nlm.nih.gov/pubmed/26155506.
- 5.Rappaport S, Endicott J, Gilbert M, Farkas J, Clouser R, McMillian W. A Retrospective Study of Early vs Delayed Home Dose Basal Insulin in the Acute Management of Diabetic Ketoacidosis. J Endocr Soc. 2019;3(5):1079-1086. https://www.ncbi.nlm.nih.gov/pubmed/31069278.