Effect of a Standardized Treatment Panel on Hypoglycemic Events in Hospitalized Acute Hyperkalemic Patients Treated With Intravenous Regular Insulin

Abstract

Purpose: Regular insulin is a commonly utilized treatment option for acute hyperkalemia. Despite its benefit, hypoglycemia and associated morbidity/mortality remain important concerns. This institution recently created a treatment panel to standardize regular insulin dosing (0.1 unit/kg) and blood glucose (BG) monitoring in patients with acute hyperkalemia. The purpose of this study is to investigate whether the order panel reduces hypoglycemic events in adults treated with intravenous (IV) regular insulin for hyperkalemia and to determine the effect the treatment panel has on regular insulin dosing, serum potassium, BG monitoring, and dextrose supplementation. Methods: This retrospective study was performed at a single academic medical center. Adults receiving IV regular insulin for acute hyperkalemia were included if BG was assessed prior to and following regular insulin administration. Primary outcome was hypoglycemia within 4 hours of regular insulin administration. Secondary outcomes were the change from baseline serum potassium, frequency of severe hypoglycemia, BG checks within 30 minutes prior to and within 4 hours following insulin administration, regular insulin dosing, and administration of dextrose 50% in water (D50W) following regular insulin administration. Hypoglycemia and severe hypoglycemia were defined as a BG concentration of <70 mg/dL and <50 mg/dL, respectively. Results: One hundred sixty-five patients were included; 75 using the treatment panel and 90 not. Patients using the treatment panel received a lower median (interquartile range [IQR]) regular insulin dose (.10 [0.09-0.10 unit/kg] vs 0.11 [0.09-0.14 unit/kg], P = .004) and had more frequent BG checks during the 4 hours following regular insulin administration (median [IQR]: 4 [3-5] vs 2 [1-3], P < .001). Hypoglycemia (13.3% vs 27.8%, P = .024) and severe hypoglycemia (2.7% vs 11.1%, P = .038) occurred less frequently with the treatment panel. Similar decreases in serum potassium were noted following IV regular insulin administration. Conclusions: Acute hyperkalemic patients utilizing a standardized treatment panel for the dosing and monitoring of IV regular insulin experienced fewer hypoglycemic and severe hypoglycemic episodes and had similar potassium lower effects. The treatment panel decreased regular insulin dosing and increased BG monitoring prior to and following regular insulin administration.

Introduction

Intravenous (IV) regular insulin has been historically utilized at a fixed dose of 10 units when treating acute hyperkalemia.^1,2 Though effective, this frequently causes hypoglycemia and a single hypoglycemia episode is associated with increased mortality risk and longer lengths of hospital stay.^3-9 Some are now advocating for the utilization of weight-based (0.1 units/kg) or reduced dose (5 units) regular insulin based on evidence suggesting lower hypoglycemia rates and similar effects on serum potassium.^6-9 Hypoglycemia, however, still occurs in up to 20% receiving weight-based or reduced dose IV regular insulin, and a multimodal strategy with standardized monitoring and treatment may prevent some hypoglycemic events and allow for rapid treatment of hypoglycemia when it occurs.^6-9

Upstate University Hospital recently transitioned to a multimodal treatment panel for the dosing and monitoring of IV regular insulin in acute hyperkalemia. Prior to the treatment panel, patients received a fixed 10-unit IV regular insulin dose, and blood glucose (BG) monitoring and dextrose supplementation were at the discretion of the ordering provider. The treatment panel contained the following orders to standardize regular insulin dosing, BG monitoring, and dextrose supplementation: (1) check BG within 30 minutes prior to IV regular insulin administration, (2) administer 0.1 units/kg IV regular insulin up to a maximum of 10 units, (3) administer with regular insulin 25 mL of IV dextrose 50% in water (D50W) regardless of previous BG, (4) check BG every 30 minutes for 4 occurrences and every 1 hour for 2 occurrences following regular insulin administration, and (5) administer 25 mL of IV D50W for any repeat BG < 70 mg/dL.

We theorized the treatment panel would decrease hypoglycemic events and have a similar effect on serum potassium in acutely hyperkalemic adults treated with IV regular insulin. Secondarily, we theorized the treatment panel would decrease regular insulin dosing while increasing BG monitoring before and after regular insulin administration.

Methods

This single-center, retrospective chart review met criteria for exemption from the institutional review board. Upstate University Hospital is a 472-bed tertiary academic medical center and level 1 trauma center that has more than 70 000 annual emergency department visits. Adults age 18 years or older ordered IV regular insulin between November 2016 and August 2017 were identified via a query of the electronic medical record. The results were then chronologically randomized and the primary author, who was blinded to treatment allocation, assigned patients into blocks of 5. These blocks were then randomly reviewed and patients were reviewed for inclusion if they had a serum potassium >5 mmol/L and documentation of acute hyperkalemia in the medical record or in the medication order. Patients were excluded if they did not have a baseline BG or a repeat BG within 4 hours of regular insulin administration. Blocks of 5 were reviewed until 165 were included, based on an estimate that 161 would be required to achieve 90% statistical power, assuming a 30% baseline rate for the primary outcome and an expected 15% reduction with the treatment panel; these estimates were based on institutional data and previous literature.^6-8

All data were collected using a standardized data collection form by a single reviewer trained in data abstraction. The corresponding author intermittently reviewed collected data for consistency and accuracy. Collected data included patient demographics, concurrent orders for subcutaneous insulin or oral antidiabetic medications, regular insulin dosing, potassium concentration prior to regular insulin, potassium concentration following regular insulin, creatinine clearance (CrCl) estimated via the Cockcroft-Gault equation, baseline BG, number of repeat BG values within 4 hours of regular insulin administration, lowest BG value recorded within 4 hours of regular insulin administration, D50W supplementation within 30 minutes prior to and 4 hours following regular insulin administration and administration of oral glucose solution from the automated dispensing cabinet within 30 prior to and 4 hours following regular insulin administration.

The primary outcome was the frequency of hypoglycemic episodes within 4 hours of IV regular insulin administration. The secondary outcomes were (1) the decrease from baseline serum potassium; (2) frequency of severe hypoglycemia episodes within 4 hours following regular insulin administration; (3) frequency of BG checks within the 30 minutes prior to regular insulin administration; (4) administered regular insulin dose; (5) frequency of co-administered D50W with regular insulin; (6) number of BG checks within the 4 hours following regular insulin administration; and (7) rescue dextrose or glucose administration within 4 hours of regular insulin administration. Hypoglycemia and severe hypoglycemia were defined as a BG < 70 mg/dL and < 50 mg/dL, respectively; these values are institutional definitions and are similar to previous publications.^4,10

All data were presented using descriptive statistics, including mean and standard deviation (SD), median and interquartile range (IQR), and number (n) and percentage (%). Categorical variables were compared using chi-square test for independence or Fisher exact test. Continuous variables were compared using the Mann-Whitney U test or Student t test. All tests were 2-tailed and a P value < .05 was considered statistically significant. All statistical tests were performed using SPSS version 23.0 (IBM Corp; Armonk, New York).

Results

One hundred sixty-six patients with acute hyperkalemia were identified; 1 patient was excluded as they had no BG testing. In total, 165 were included and 75 used the treatment panel and 90 did not. Demographics can be found in Table 1. No significant differences were found in age, weight, gender, CrCl, history of diabetes mellitus, baseline BG, baseline serum potassium, concurrent use of subcutaneous insulin, or concurrent use of oral antidiabetic medications. Figure 1 compares the frequency of hypoglycemic and severe hypoglycemic events. Hypoglycemic (treatment panel: 13.3% vs nontreatment panel: 27.8%, P = .024) and severe hypoglycemic events (treatment panel: 2.7% vs nontreatment panel: 11.1%, P = .038) occurred less commonly with the treatment panel.

Table 1.

Demographics.

	Ordered without panel (n = 90)	Ordered with panel (n = 75)	P value
Age (years), mean ± standard deviation (SD)	59.3 ± 16.3	60.7 ± 17.7	.391
Male gender, n (%)	62 (68.9)	44 (58.7)	.173
Weight (kg), median (IQR)	88.0 (72.5-110)	84.4 (75.3-106.6)	.590
Diabetes mellitus, n (%)	45 (50)	28 (38.3)	.103
At least 1 concurrent order for subcutaneous insulin, n (%)	32 (35.6)	24 (32)	.631
At least 1 concurrent order for an oral antidiabetic medication, n (%)	0 (0.0)	1 (1.3)	.455
Creatinine clearance (mL/min), median (IQR)	35.5 (19-66.3)	47 (23-80)	.124
Baseline potassium (mmol/L), median (IQR)	6.1 (5.7-6.7)	6.0 (5.7-6.4)	.505
Baseline Blood Glucose (mg/dL), median (IQR)	123 (93.8-194)	132 (109-173)	.466

Note. IQR = interquartile range.

Figure 1.

Hypoglycemic event frequency.

All but 3 repeat serum potassium values were drawn within 6 hours following regular insulin administration; the remaining 3 values were drawn within 12 hours. Sodium polystyrene sulfonate administration was similar between groups (treatment panel: 27/75, 36.0% vs nontreatment panel: 28/90, 31.1%, P = .507). Sodium bicarbonate was administered more frequently in the nontreatment panel group (treatment panel: 6/75, 8% vs nontreatment panel: 20/90, 22.2%, P = .013). Median (IQR) repeat serum potassium was similar following regular insulin administration (treatment panel: 5.4 [5.0-5.9] mmol/L vs nontreatment panel: 5.2 [5.0-5.8] mmol/L, P = .489). Figure 2 shows that the treatment panel and nontreatment panel groups had similar serum potassium decreases following regular insulin administration (treatment panel: −0.6 [–0.3 to −1.1] mmol/L vs nontreatment panel: −0.8 [–0.4 to −1.3]) mmol/L, P = .151).

Figure 2.

Serum potassium decrease following regular insulin administration.

Additional secondary outcomes are shown in Table 2. Patients using the treatment panel were more likely to have their BG checked within 30 minutes prior to regular insulin administration, had a higher median (IQR) number of BG checks within 4 hours following regular insulin administration, and received lower median (IQR) weight-based and non–weight-based regular insulin doses. Frequency of IV D50W or oral glucose did not differ prior to or following regular insulin administration. Frequency of IV D50W or oral glucose were not statistically different in hypoglycemia patients following regular insulin administration (treatment panel: 9/10, 90% vs nontreatment panel: 17/25, 68%, P = .205). Three treatment panel patients received 15 g oral glucose solution following regular insulin administration; 2 received oral glucose solution in combination with IV D50W and 1 received oral glucose alone. No nontreatment panel patient received oral glucose solution. The remaining patients received IV D50W alone. Median (IQR) administered D50W dose did not differ before (treatment panel: 25 [25-25] g vs nontreatment panel: 25 [25-25] g, P = .170) or after (treatment panel: 25 [25-25] g vs nontreatment panel: 25 [25-25] g, P = .381) regular insulin administration. Treatment panel patients had the following indications for IV D50W or oral glucose following regular insulin administration: (1) BG < 70 mg/dL: 9/12, 75%; and (2) BG > 70 mg/dL but < 100 mg/dL and decreasing: 3/12, 25%. Nontreatment panel patients had the following indications for IV D50W or oral glucose following regular insulin administration: (1) BG < 70 mg/dL: 17/23; 73.9%; (2) BG > 70 mg/dL but < 100 mg/dL and decreasing: 3/23, 13% and; (3) unknown indication, repeat BG > than baseline or > 100 mg/dL: 3/23, 13%.

Table 2.

Secondary Outcomes.

	Ordered without panel (n = 90)	Ordered with panel (n = 75)	P value
Blood glucose checked within 30 minutes prior to regular insulin administration, n (%)	30 (33.3)	57 (76.0)	<.001
Regular insulin dose (units), median (IQR)	10 (10-10)	9 (7-10)	<.001
Regular insulin dose (units/kg), median (IQR)	0.11 (0.09-0.14)	0.10 (0.09-0.10)	.004
D50W given concurrently with regular insulin, n (%)	85 (94.4)	73 (97.3)	.457
Blood glucose checks within 4 hours following regular insulin administration, median (IQR)	2 (1-3)	4 (3-5)	<.001
IV D50W or oral glucose following regular insulin administration, n (%)	23 (25.6)	13 (17.3)	.257

Note. IQR = interquartile range.

Discussion

The implementation of our standardized hyperkalemia treatment panel for the dosing and monitoring of IV regular insulin decreased hypoglycemic and severe hypoglycemic events while maintaining regular insulin’s potassium lower effects. These findings are consistent with at least 4 recent publications, all of which utilized similar order sets or treatment panels. These publications placed heavy emphasis on weight-based or reduced dose regular insulin and 3 directly attributed these dosing strategies with hypoglycemia reductions.^6-8 Our treatment panel was associated with a 0.01 unit/kg or 1-unit reduction in median regular insulin dose. Though statistically significant, this reduction appears lower than 3 of the 4 previously noted studies; McNicholas et al and LaRue et al noted dosage reductions of up to 5 units and Wheeler et al had dosage reductions of approximately 2.5 units based on patient weight.^6-8 As regular insulin’s effect on BG is dose-dependent, it is not unreasonable to expect our treatment panel to have less effect on hypoglycemia. We, however, observed an approximate 15% hypoglycemia reduction, which is similar to the approximant 10% to 15% reduction in these studies.^6-8 Our results, therefore, suggest a factor other than regular insulin dosing could additionally be contributing to our treatment panel’s hypoglycemia reductions. Interestingly, Brown et al observed hypoglycemia reductions with weight-based regular insulin but did not observe an absolute reduction in insulin dosing.⁹

Our only other significant findings were the frequency of BG monitoring before and after regular insulin administration. BG monitoring has an important role in safe insulin use, and recent evidence suggests less frequent hypoglycemic events with more frequent BG monitoring.¹¹ Our treatment panel was associated with increased frequency of preinsulin BG checks and an absolute increase in BG monitoring in the 4 hours following regular insulin administration. One could hypothetically argue that increases in BG monitoring could prevent hypoglycemia, as staff could be more aware of patients approaching hypoglycemia and intervene. Interestingly, 6 patients received IV D50W or oral glucose following regular insulin administration for a decreasing BG that was >70 mg/dL but <100 mg/dL. IV D50W and oral glucose were presumably administered to prevent further progression to hypoglycemia, suggesting staff may intervene if they feel a patient is at risk. We, therefore, feel it is possible our hypoglycemia reductions may have related to increases in staff awareness of decreasing BG. We did not, however, observe an increase in the administration of IV D50W or oral glucose prior to or following regular insulin administration. Though this could be interpreted as contradictory, it does not preclude the above inference, as patients could have received alternative oral glucose sources. Oral glucose solution is the only form of oral glucose routinely documented in our electronic medical record, as it is obtained from pharmacy automated dispensing cabinets. Other forms of oral glucose, such as juices and regular soda, are recommended by hospital policy but not routinely documented in the electronic medical record and patients could have received one of these alternative sources without our knowledge. We, therefore, feel oral glucose could have been administered more frequently, potentially preventing hypoglycemia progression and is worth consideration as there is no other easily identified mechanism for our hypoglycemia reductions. Irrespective, we feel the most logical explanation for our observed reduction in hypoglycemia is a multimodal approach including weight-based IV regular insulin and standardized procedures for BG monitoring and dextrose supplementation following regular insulin administration for acute hyperkalemia.

There are limitations to this study that require discussion. First, our study was conducted at a single center, potentially limiting its generalizability. Second, our study excluded patients if they did not have a baseline BG and at least 1 repeat BG with 4 hours of regular insulin administration. We, therefore, could have missed some data points, though we feel their clinical utility is likely limited given the 4- to 6-hour duration for IV regular insulin. Third, it is possible the nontreatment panel group may have had episodes of unrecognized hypoglycemia, as they had fewer BG checks. Our results, therefore, represent a potentially conservative estimate. Fourth, the retrospective nature of our study with convenience sampling implies the potential for some selection bias. We attempted to minimize this bias by chronologically randomizing our patient list and then reviewing patients in random blocks of 5. Finally, we did not evaluate for renal insufficiency. We, however, found no statistical difference in creatinine clearance between treatment groups.

Conclusion

The implementation of a standardized hyperkalemia treatment panel for dosing and monitoring IV regular insulin led to a decrease in hypoglycemic events without sacrificing regular insulin’s effect on serum potassium. This reduction in hypoglycemic events was most likely related to a reduction in regular insulin dose and an increase in BG monitoring. Our results support the adoption of a standardized treatment panel at institutions aiming to reduce hypoglycemic events related to regular insulin administration in acute hyperkalemia.