Hospital Pharmacy - April 2018 - 122

122
received a CII. Patients were excluded if they received the
CII for less than 24 hours or had incomplete data in their
electronic record. If multiple infusions were given during a
single hospitalization, all infusions were included.
Data collected from the institutions electronic medical
record included age, gender, race, weight, height, total body
surface area (TBSA), length of stay (LOS), acute physiology
and chronic health evaluation (APACHE) II score, survival,
days of mechanical ventilation, type of burn injury, presence
of infection, type of nutrition, average carbohydrate intake,
hours of CII, point of care glucoses, average insulin received,
hours in which blood glucose was less than 70, 60, and 40
mg/dL. The LOS was collected for the time in the intensive
care unit and for total time hospitalized. Laboratory values
collected included albumin, prealbumin, C-reactive protein,
serum creatinine, creatinine clearance, hemoglobin A1c,
white blood cell count, and blood glucose. All laboratory values were assessed at the start of the first CII. Renal function
was assessed during the entirety of the CII, using the acute
kidney injury network (AKIN) classification, to determine
whether acute kidney injury or renal failure contributed to
incidence of hypoglycemia.
Attributable cause of each hypoglycemic event was classified using 4 standard definitions: (1) a protocol violation
that patient lacked concomitant tube feeds, total parenteral
nutrition, or dextrose infusion during the CII; (2) renal failure; (3) glucose variability, defined as a change in blood glucose of greater than 50 mg/dL in 1 hour; or (4) other CII
protocol violation. Hypoglycemic events that could not be
explained by the any of the previous 4 explanations were
classified as unknown.
SigmaPlot 11.2 was used for data analysis. Univariable
analysis was used to compare patients with and without
hypoglycemia. Independent variables with a P ≤ .1 were
included in a multivariable logistic regression. During
regression modeling, P < .05 was considered significant.
Nominal data were analyzed by Fisher exact test. Continuous
variables were compared using either Mann-Whitney U test
or t test, depending on normal distribution. Shapiro-Wilk
test was used to test for normal distribution. Analysis of
variance was utilized to compare glucose variability between
protocols.

Results
Thirty-eight patients were admitted to the burn center and
started on CII between January 1, 2013, and October 31,
2014. Six patients were excluded for CII less than 24 hours
(n = 5) and missing data in the electronic medical record (n
= 1). The average patient was a 52-year-old Caucasian
male, weighing 95 kg with a 33% TBSA burn with an
APACHE II score of 20. Hypoglycemia was present in 87
of 6540 total hours of CII therapy (1.1%) or 2.7 hours per
patient. Of the 32 included patients, two-thirds experienced
at least 1 serum glucose less than 70 mg/dL and half less

Hospital Pharmacy 53(2)
than 60 mg/dL. The most commonly assessed reasons for
the hypoglycemic episodes were protocol violations (47%)
and glucose variability (30%). Patients with hypoglycemia
(<60 mg/dL) were older, had a history of diabetes, lower
albumin and serum creatinine (Table 1), longer CII duration
and LOS, and lower average carbohydrate intake (Table 2).
After multivariable logistic regression, only history of diabetes remained a statistically significant risk factor for
hypoglycemia with an odds ratio of 15.4 (95% confidence
interval: 2.5-95.1).
Overall, 4 different CII protocols were prescribed. During
37% of the hypoglycemic episodes, patients were receiving
a modified CII usually reserved for patients with renal failure (Protocol 1). Forty-eight percent of episodes occurred
when patients received a standard CII (Protocol 2), regardless of renal function. Protocol 1 reduced the amount of
renal failure patients experiencing hypoglycemia from 91%
to 62%. All of the protocols had a high glucose variability,
as assessed by hours / day within goal range (13.1 ± 2.5,
14.1 ± 3.1, 14.3 ± 2.4, 9.8; P = .282) (Figure 1). Protocols 0
and 3 had lower rates of hypoglycemia, but also lower time
within the goal range.

Discussion
In this study, the authors investigated patients receiving CII
to determine incidence of hypoglycemia. The authors
attempted to define each episode with an ultimate goal of
future improvement. During the study period, 50% of
included patients had an episode of clinically significant
hypoglycemia (<60 mg/dL). The amount of patients experiencing hypoglycemia is concerning due to the association of
hypoglycemia with morbidity and mortality in critically ill
patients.4
The high incidence of hypoglycemia was associated with
longer LOS and longer time on CII. It is difficult to determine
whether the extended stay was due to hypoglycemic episodes
or the underlying severity of the burn injury. The study was
not powered to perform regression analysis on predictors of
increased LOS and the study was retrospective, thus causality
cannot be determined. Factoring in the highly dynamic nature
of patients with burn injury, increased insulin resistance, a
past medical history of diabetes and obesity, and incidence of
renal failure, glycemic control becomes difficult to achieve
on a consistent basis. There was nearly 3 times the number of
patients with a history of diabetes in the group that experienced hypoglycemic events. Patients with a history of diabetes have also been shown to sustain a higher percentage of full
thickness burns, operations, infections, and longer LOS over
nondiabetics.9 The purpose of this study was not to analyze
effects on LOS, but identify areas to potentially reduce glucose variability and improve glycemic control
Renal function plays an important role in the regulation of
insulin in the body. Approximately 50% of systemic insulin is
cleared through the kidneys.10 The modified protocol used in

Table of Contents for the Digital Edition of Hospital Pharmacy - April 2018