Insulin remains one of the cornerstones of early severe hyperkalemia management. Insulin works via a complex process to temporarily shift potassium intracellularly. Though insulin certainly lowers plasma potassium concentrations, we often underestimate the hypoglycemic potential of a 10 unit IV insulin dose in this setting. The purpose of this post is to highlight the need for proper supplemental glucose and blood glucose monitoring when treating hyperkalemia with insulin.
Incidence of Hypoglycemia
One of my favorite articles on the management of hyperkalemia was written by Dr. Weisberg in Critical Care Medicine.1 A 10 unit dose of IV regular insulin has an onset of action of about 5-10 minutes, peaks at 25-30 minutes, and lasts 2-3 hours (the Weisberg article actually lists subcutaneous kinetics). Herein lies the problem in that IV dextrose only lasts about an hour (at most). Allon et al reported up to 75% of hemodialysis patients with hyperkalemia developed hypoglycemia at 60 minutes after insulin administration.2 A retrospective review of 219 hyperkalemic patients reported an 8.7% incidence of hypoglycemia after insulin treatment.3 More than half of the hypoglycemic episodes occurred with the commonly used regimen of 10 units of IV insulin with 25 gm of dextrose. A more recent study of 221 end-stage renal disease patients who received insulin for treatment of hyperkalemia reported a 13% incidence of hypoglycemia.4
The overall incidence of hypoglycemia appears to be ~10%, but could be higher.
Risk Factors for Developing Hypoglycemia
The study by Apel et al identified three factors associated with a higher risk of developing hypoglycemia:
- No prior diagnosis of diabetes [odds ratio (OR) 2.3, 95% confidence interval (CI) 1.0–5.1, P = 0.05]
- No use of diabetes medication prior to admission [OR 3.6, 95% CI 1.2–10.7, P = 0.02]
- A lower pretreatment glucose level
- In mg/dL: mean 104 ± 12 mg/dL vs 162 ± 11 mg/dL, P = 0.04
- In mmol/L: mean 5.8 ± 0.7 mmol/L vs 9.0 ± 0.6 mmol/L, P = 0.04
Renal dysfunction in and of itself may also be a risk factor for developing hypoglycemia. Some evidence suggests that insulin is metabolized by the kidneys to some extent. Furthermore, patients with acute kidney injury (AKI) have clinically relevant changes in insulin metabolism, as evidenced by increased hypoglycemic events and lower insulin requirements upon developing AKI.5
Strategies for Avoiding Hypoglycemia
Preventing hypoglycemia is important. Some clinicians use up to 20 units of IV regular insulin as the hypokalemic effect is dose dependent.6 Here is a suggested strategy for administering enough dextrose to counter the initial insulin bolus of 10 or 20 units. It is loosely based on the Rush University protocol.4
|Initial [glucose]||Initial Dextrose Dose||Supplemental Dextrose **||Glucose Monitoring|
|> 200 mg/dL (> 11.1 mmol/L)||None||25 gm (50 mL) of D50 if blood glucose < 70 mg/dL||Hourly up to 3 hours|
|100-200 mg/dL (5.6-11.1 mmol/L)||25 gm (50 mL) of D50||25 gm (50 mL) of D50 if blood glucose < 70 mg/dL||Hourly up to 3 hours|
|< 100 mg/dL (< 5.6 mmol/L)||50 gm (100 mL) of D50* OR
25 gm (50 mL) of D50 + D10 infusion 250 mL/hr for first hour
|25 gm (50 mL) of D50 if blood glucose < 70 mg/dL||q 30 minutes for first hour, then hourly up to 3 hours|
D50 = dextrose 50%; D10 = dextrose 10%
* There are drawbacks to administering 100 mL (50 gm) of D50. Dr. Adam Spaulding (@PharmERAtom) discusses D50 vs. D10 for severe hypoglcyemia in the ED. 50 mL (25 gm) of D50 + infusion of D10 might prevent rebound hypoglycemia and hypertonicity issues with an initial 100 mL (50 gm) dose of D50.
** A supplemental D10 infusion may be needed beyond one hour depending on blood glucose concentrations.
What about using dextrose only?
Theoretically, administering dextrose should stimulate insulin release (if the patient has a functioning pancreas) and thereby lower serum potassium concentrations. Early reports recommend against this technique for two reasons.7
- Endogenous insulin levels are unlikely to rise to the level necessary for a therapeutic effect
- There is a risk of exacerbating the hyperkalemia by inducing hypertonicity.
On the flip side, a more recent randomized, crossover study was conducted in 10 chronic hemodialysis patients who were prone to hyperkalemia.8 Administration of 10 units of insulin with 100 mL of 50% glucose (50 g) was compared with the administration of 100 mL of 50% glucose only. Infusion of a glucose-only bolus caused a clinically significant decrease in serum potassium by 0.5 mEq/L without any episodes of hypoglycemia. The insulin/glucose group saw a drop in potassium of 0.83 mEq/L at 60 minutes.
Bottom Line: Based on conflicting data, and considering that many patients have ESRD plus diabetes, this approach has potential but may not be ready for mainstream use. Also, hyperglycemia may be undesirable in some patients.
Take Home Points
- The hypoglycemic effects of IV insulin last longer than a bolus of dextrose. More than one dextrose dose is often needed when treating hyperkalemia.
- One ‘amp’ (50 mL, 25 gm) of D50 is not enough to counteract the hypoglycemic effect of insulin in patients with normoglycemia to start.
- Make sure to check glucose at the hour mark after administering IV insulin. Insulin’s peak effect occurs at about 60 minutes and this is when hypoglycemia has most often been reported in the literature.
- Order sets are key to preventing this common medication error.
Dr. Scott Weingart and I discuss this issue on his EMCrit podcast 101: Avoiding Resuscitation Medication Errors – Part 1.
Editor’s note: Reference 1 indicates that the kinetics for IV regular insulin are as follows: onset 20 minutes, duration 4-6 hours. That is actually the subcutaneous insulin kinetics. The corrected IV kinetics are reflected in the post.
Expert Peer Review
Thank you for the opportunity to review this recommendation. Hypoglycemia as an unintended consequence of hyperkalemia treatment with insulin only makes a bad situation worse. The incidence of hypoglycemia and severe hypoglycemia (fingerstick <40 mg/dL) in the treatment of hyperkalemia is surprisingly high.
With respect to the treatment algorithm, I suspect that hourly blood sugar measurements for the first three hours will be sufficient for all but the highest risk patients. Most of the patients identified as hypoglycemic in the prior studies were identified as such by one hour, and often asymptomatic.
As the recommendations stand, they may lead to some iatrogenic hyperglycemia (since the majority of patients do not become hypoglycemic after receiving insulin for hyperkalemia). Depending on concomitant disease processes, hyperglycemia also may be undesirable.
The take-home points are particularly useful, and an important reminder to remain vigilant after the initial therapy has been delivered.
Anecdotally, many clinicians use 5 units of regular insulin IV to treat hyperkalemic patients who are at high risk for hypoglycemia. I was interested to note that episodes of severe hypoglycemia still occurred in two patients receiving 5 units in the Schafers study.
Kavita, thank you for peer-reviewing our post prior to publication. In thinking about the table and considering your thoughts on risk of hyperglycemia, I made some adjustments to it in order to be a little less aggressive with blood glucose monitoring and glucose supplementation. The new suggestions should still provide safe monitoring of glucose, especially at the 60 minute mark, while avoiding unnecessary hyperglycemia.
Expert Peer Review
Thank you for this insightful and pragmatic post on what is undoubtedly an overlooked complication of hyperkalemia management. Overall, this is a very concise, yet complete post. Please see below some of my thoughts/suggestions.
- While not specifically cited as a risk factor in the Apel study, I would argue that the presence of underlying renal dysfunction (AKI, ESRD) also be acknowledged as a potential risk factor in the post (Apel, et al obviously didn’t recognize it as a risk factor because the entire study population was ESRD). This rationale is based upon the fact that 79% of patients with hypoglycemia in the Schafer’s study had AKI or ESRD, and similarly, 9 of 12 dialysis patients in the Allon study had hypoglycemia (75%). To this same point, evidence exists that insulin is actually metabolized by the kidneys to some extent (the kidneys do contain various CYP isoenzymes); patients with AKI have clinically relevant changes in insulin metabolism, as evidenced by increased hypoglycemic events and lower insulin requirements upon developing AKI (Nutrition 2011;27:766-72).
- Might there be any direct/indirect evidence to support utilizing the rapid acting insulin analogs for treatment of hyperkalemia to better match the duration of the IV dextrose (and hence potentially mitigate the complication of hypoglycemia)? This seems plausible given a shorter time to peak and duration of action.
- Given the kinetics of both insulin and dextrose, it seems that one might be able to simply time the initial dose of dextrose to be given ~30 minutes AFTER the dose of insulin as a potential alternative solution to the first take home point. Perhaps this may be a unique strategy worth mentioning in your post.
- There is actually some evidence to suggest that a ‘dextrose-only’ treatment approach to hyperkalemia may be effective. This essentially takes advantage of the natural, pancreatic secretion of insulin that occurs in response to exogenous dextrose administration. The following study, albeit small, found that this approach significantly reduced serum K while potentially mitigating risk of hypoglycemia (reduced from ~6 meq/L on average → ~5.5 meq/L). Consider including this evidence in your post, perhaps as a potential alternative for the management of mild hyperkalemia in the patient with normo/hypoglycemia at baseline. (PMID: 24576893)
Thank you so much for your insightful comments.
- I have updated the post to reflect the insulin metabolism issue with renal dysfunction and listed renal dysfunction as a potential risk factor for developing hypoglycemia. I included the reference you suggested.
- Although the rapid-acting insulin idea does make sense, I haven\'t seen any data on that so I didn\'t include it in the post.
- This suggestion also makes a lot of sense, but I do have concerns not giving the glucose up front. Even in a critically ill hyperkalemic patient, it is impossible to predict what resources will be available or what other sick patients may be on their way in to the ED. Even when we schedule labs to be drawn or medications given at a specific time, it often doesn\'t occur when we hope due to external ED environment circumstances. Therefore, I\'m not confident EDs could implement that technique as suggested and I worry that if the patient had delayed glucose administration, serious hypoglycemia could result. Instead, I think using a D10 infusion over an hour makes sense and takes this issue out of the equation. The post does recommend a D10 infusion as an option.
- Thanks for bringing up the dextrose-only idea. Early data suggested this actually worsens hyperkalemia. However, the newer RCT you cited found it could help. I added a section on dextrose only to address the conflicting data and provide recommendations.