Table 1 Protocol-directed insulin infusion improve perioperative hyperglycaemia in critical care

Zimmerman et al. (2004)16

168 postoperative cardiothoracic surgical intensive care patients

A nurse-driven insulin infusion protocol was developed and implemented in postoperative cardiothoracic surgical intensive care patients with or without diabetes.

80 to 150 mg/dl (4.4 to 8.3 mmol/l)

Every 1 to 4 hours

12 patients (7.1%) <40 mg/dl (2.2 mmol/l)

Findings showed percentage and time of blood glucose measurements within the tight glycaemic control range (control 47% vs. protocol 61%; P = 0.001),

This before-and-after cohort study used two periods of measurement: a 6-month baseline period prior to the initiation of the insulin

28 patients (16.7%) <65 mg/dl (3.6 mmol/l)

Area under curve (AUC) of glucose exposure >150 mg/dl (8.3mmol/l) vs. time for the first 24 hours of the insulin infusion (control 28.4 vs. protocol 14.8; P < 0.001), median time to blood glucose <150 mg/dl (8.3mmol/l) (control 9.4 h vs. protocol 2.1h; P < 0.001), and percentage blood glucose <65 mg/dl (3.6 mmol/l) as a marker for hypoglycaemia (control 9.8% vs. protocol 16.7%; NS).

infusion protocol (control group, n = 174) followed by a 6-month intervention period, in which the protocol was used (protocol group, n = 168).

Tamaki et al. (2008) 17

40 cardiac surgery patients

The Yale insulin infusion protocol was modified by taking into consideration the characteristics of Japanese diabetics and the hospital environment.

80 to 140 mg/dl (4.4 to 7.8 mmol/l)

Every 30 min to every 2 hours

Blood glucose values <60 mg/dl (3.3mmol/l) 0.5% ± 5.9%

Analyses of 1,656 blood glucose measurements during insulin infusion revealed that the percentage of samples that showed achievement of target blood glucose level (80 to 140 mg/dl (4.4 to 7.8 mmol/l)) was higher under protocol (78 ± 15%, n = 870) than control (57 ± 23%, n = 786, P < 0.0001).

The modified protocol was tested in 40 type-2 diabetic patients after elective open-heart surgery, compared with 35 type-2 diabetic patients under empirical blood glucose control.

On the other hand, the fraction of samples with blood glucose <60 mg/dl (3.3 mmol/l) was comparable in the two groups (protocol: 0.5 ± 5.9‰, control: 5.1 ± 18.5‰).

None of the patients with hypoglycaemia showed significant clinical adverse effects.

Caddell et al. (2010) 18

100 cardiovascular surgery patients

Prospective data were gathered on 100 consecutive cardiovascular surgery patients managed with standard insulin infusion protocol and 100 patients managed with an insulin-resistance-guided protocol. Clinical characteristics and glycaemic indices were analyzed for the two groups. Primary outcomes included: percentage of time spent in the target range; number of hypoglycaemic and hyperglycaemic episodes; time to achievement of target blood glucose concentration; and total daily dose of insulin required.

80 to 110 mg/dl (4.4 to 6.1 mmol/l)

Hourly

<70 mg/dl (3.9 mmol/l): 0.12 event per patient

The insulin-resistance guided protocol resulted in significant improvements, including increased percentage of time spent in the normoglycaemic range (82.5% vs. 65.8%, P < 0.001), reduced rate of hypoglycaemic episodes (0.12 vs. 0.99, P < 0.01), reduced rate of hyperglycaemic episodes (capillary blood glucose >126 mg/dl (7 mmol/l): 4.8 vs. 8.2, P < 0.01), and reduced time to the first measurement in the target range. Total daily dose of insulin was mildly increased, but failed to reach statistical significance (92.48 vs. 82.64 units, P = 0.32).

<40 mg/dl (2.2 mmol/l): 0.04 event per patient

Leibowitz et al. (2010) 31

203 cardiac surgery patients

Patients with diabetes mellitus or random blood glucose >150 mg/dl were treated in the intensive care unit with intravenous insulin, followed by a multi-injection protocol consisting of four glargine-aspart insulin injections in the ward, with a glycaemic target of 110 to 150 mg/dl (6.1 to 8.3 mml/l).

110 to 150 mg/dl (6.1 to 8.3 mmol/l)

Every 20 min to every 4 hours

3% patients with blood glucose <60mg/dl (3.3 mmol/l)

During the intervention, mean blood glucose ± SD was 151 ± 19 mg/dl (8.4 ± 1.1 mmol/l) and 157 ± 32 mg/dl (8.7 ± 1.8 mmol/l) in the intensive care unit and ward, respectively, vs. 166 ± 27 mg/dl (9.2 ± 1.5 mmol/l) and 184 ± 46 mg/dl (10.2 ± 2.6 mmol/l) during the control period (P < 0.0001). The incidence of hypoglycaemia (blood glucose less than 60 mg/dl) was low and similar in the two groups (2.5% control vs. 3% intervention). Intensive insulin treatment decreased the risk for infection from 11% to 5% (56% risk reduction, P = 0.018), mainly by reducing the incidence of graft harvest site infection (6.9% vs. 2.5%, P = 0.034). The incidence of atrial fibrillation after coronary artery bypass graft surgery decreased from 30% to 18% (39% risk reduction; P = 0.042).

The study cohort (n = 410) consisted of consecutive patients undergoing cardiothoracic surgery. Control patients (n = 207) were admitted during the first 8 months

(CONTROL GROUP)

The intervention group of patients (n = 203) were operated on during the following 8 months.

The main outcome measures were glycaemic control and the rate of postsurgery infection.

Goldberg et al. (2004) 24

118 cardiothoracic intensive care unit patients

A standardized, intensive insulin infusion protocol was used for all patients admitted to two cardiothoracic intensive care unit s. Hourly blood glucose levels, relevant baseline variables, and clinical interventions were collected prospectively from the active hospital chart and cardiothoracic intensive care unit nursing records.

100 to 139 mg/dl (5.6 to 7.7 mmol/l)

Hourly

Five blood glucose values (0.2%) <60 mg/dl (3.3 mmol/l)

The insulin infusion protocol was used 137 times in 118 patients. The median time required to reach target blood glucose levels (100 to 139 mg/dl (5.6 to 7.7 mmol/l)) was 5 hours. Once blood glucose levels decreased below 140 mg/dl, 58% of 2242 subsequent hourly blood glucose values fell within the target range, 73% within a ‘clinically desirable’ range of 80 to 199 mg/dl (4.4 to 11 mmol/l). Only five (0.2%) blood glucose values were less than 60 mg/dl (3.3 mmol/l), with no associated adverse clinical events.

Lowest recorded blood glucose value: 48 mg/dl (2.7 mmol/l)

Lecomte et al. (2008) 25

651 cardiac surgery patients

483 nondiabetics and 168 diabetics scheduled for cardiac surgery with cardiopulmonary bypass were recruited.

80 to 110 mg/dl (4.4 to 6.1 mmol/l)

Every 30 min to every 2 hours

Blood glucose values <60mg/dl (3.3 mmol/l)

18,893 blood glucose measurements were made during and after surgery. During surgery, the mean glucose level in nondiabetic patients was within targeted levels except during (112 ± 17 mg/dl (6.2 ± 0.9mmol/l)) and after rewarming (113 ± 19 mg/dl (6.3 ± 1.1mmol/l)) on cardiopulmonary bypass.

To anticipate rapid perioperative changes in insulin requirement or sensitivity during surgery, a dynamic algorithm presented in tabular form, with rows representing blood glucose ranges and columns representing insulin dosages based on the patients’ insulin sensitivity was developed. The algorithm adjusts insulin dosage based on blood glucose level and the projected insulin sensitivity (for example, reduced sensitivity during cardiopulmonary bypass and normalizing sensitivity after surgery).

In nondiabetic patients: 0.16%

In diabetics, blood glucose was decreased from 121 ± 40 mg/dl (6.7 ± 2.2 mmol/l) at anaesthesia induction to 112 ± 26 mg/dl (6.2 ± 1.4 mmol/l) at the end of surgery (P < 0.05), with 52.9% of patients achieving the target.

In diabetic patients: 0.22%

In the intensive care unit, the mean glucose level was within the targeted range at all times, except for diabetics on arrival at the intensive care unit (113 ± 24 mg/dl (6.3 ± 1.3mmol/l)).

Lowest recorded blood glucose value: 40 mg/dl

Of all blood glucose measurements (operating room and intensive care unit), 68.0% were within the target, with 0.12% of measurements in nondiabetics and 0.18% in diabetics below 60 mg/dl (3.3 mmol/l). Hypoglycaemia <50 mg/dl (2.8mmol/l) was avoided in all but four (0.6%) patients (40 mg/dl (2.2mmol/l) was the lowest observed value).

(2.2mmol/l)

Studer et al. (2010) 26

230 cardiac surgery patients

230 consecutive patients (mean ± SD age: 67 ± 11 years; diabetic patients: n = 62) undergoing cardiac surgery (coronary artery bypass grafting: n = 137; 20% off-pump) or intrathoracic aortic (n = 10) surgery were included.

100 to 139 mg/dl (5.6 to 7.7 mmol/l)

Every 1 to 3 hours

Blood glucose <75 mg/dl (4.2 mmol/l): Postoperative day 1: 12 patients (5.3%)

All patients received postoperative insulin therapy. Patients spent 57.3% and 69.7% of time within the blood glucose target range on postoperative days 1 and 2, respectively. The percentage of time was significantly higher in nondiabetics than in diabetics. Mean blood glucose measurements per patient intraoperatively, on postoperative days 1 and 2 were 4 ± 1, 10 ± 2 and 7 ± 2, respectively. No patient experienced any severe hypoglycaemic events (blood glucose <50 mg/dl (2.8mmol/l)).

Blood glucose control was managed according to an insulin therapy protocol, described by Goldberg et al. [24], in use for 6 months. Insulin infusion rate and frequency of blood glucose monitoring were adjusted according to: (1) the current blood glucose

Postoperative day 2: 7 patients (3.1%).

value; (2) the previous blood glucose value; and (3) the current insulin infusion rate. Efficacy was assessed by the percentage of time spent at the target blood glucose level (100 to 139 mg/dl (5.6 to 7.7mmol/l)) intraoperatively and during the first two postoperative days.

Blood glucose <50 mg/dl) (2.78%): 0 patients (0%)