Liberal Glucose Control in ICU Patients With Diabetes: A Before-and-After Study.

OBJECTIVES: To assess the feasibility, biochemical efficacy, and safety of liberal versus conventional glucose control in ICU patients with diabetes. DESIGN: Prospective, open-label, sequential period study. SETTING: A 22-bed mixed ICU of a tertiary hospital in Australia. PATIENTS: We compared 350 consecutive patients with diabetes admitted over 15 months who received liberal glucose control with a preintervention control population of 350 consecutive patients with diabetes who received conventional glucose control. INTERVENTIONS: Liberal control patients received insulin therapy if glucose was greater than 14 mmol/L (target: 10-14 mmol/L [180-252 mg/dL]). Conventional control patients received insulin therapy if glucose was greater than 10 mmol/L (target: 6-10 mmol/L [108-180 mg/dL]). MEASUREMENTS AND MAIN RESULTS: We assessed separation in blood glucose, insulin requirements, occurrence of hypoglycemia (blood glucose ≤ 3.9 mmol/L [70 mg/dL]), creatinine and white cell count levels, and clinical outcomes. The median (interquartile range) time-weighted average blood glucose concentration was significantly higher in the liberal control group (11.0 mmol/L [8.7-12.0 mmol/L]; 198 mg/dL [157-216 mg/dL]) than in the conventional control group (9.6 mmol/L [8.5-11.0 mmol/L]; 173 mg/dL [153-198 mg/dL]; p < 0.001). Overall, 132 liberal control patients (37.7%) and 188 conventional control patients (53.7%) received insulin in ICU (p < 0.001). Hypoglycemia occurred in 6.6% and 8.6%, respectively (p = 0.32). Among 314 patients with glycated hemoglobin A1c greater than or equal to 7%, hypoglycemia occurred in 4.1% and 9.6%, respectively (p = 0.053). Trajectories of creatinine and white cell count were similar in the groups. In multivariable analyses, we found no independent association between glucose control and mortality, duration of mechanical ventilation, or ICU-free days to day 30. CONCLUSIONS: In ICU patients with diabetes, during a period of liberal glucose control, insulin administration, and among patients with hemoglobin A1c greater than or equal to 7%, the prevalence of hypoglycemia was reduced, without negatively affecting serum creatinine, the white cell count response, or other clinical outcomes. (Trial Registration: Australian New Zealand Clinical Trials Registry; ACTRN12615000216516).

Liberal Versus Conventional Glucose Targets in Critically Ill Diabetic Patients: An Exploratory Safety Cohort Assessment.

OBJECTIVES: To assess the feasibility, safety, and impact on relative hypoglycemia of liberal versus conventional blood glucose concentration targets in critically ill diabetic patients. DESIGN: Prospective, open-label, sequential-period exploratory study. SETTING: A 22-bed multidisciplinary ICU of a tertiary care hospital in Australia. PATIENTS: Eighty adult diabetic patients, 40 from the conventional before period and 40 from the liberal after period. INTERVENTIONS: Blood glucose concentration targets were 6-10 mmol/L during the before period and 10-14 mmol/L during the after period. MEASUREMENTS AND MAIN RESULTS: We used admission glycated hemoglobin to estimate premorbid baseline blood glucose concentration. We defined glycemic distance as the difference between blood glucose concentration in ICU and baseline blood glucose concentration. During the first 48 ICU hours, we recorded absolute (blood glucose concentration, < 3.9 mmol/L) and relative (glycemic distance, > 30% below baseline) hypoglycemia rates, insulin administration, and outcomes. The groups had similar baseline characteristics. We observed a negative glycemic distance in 248 of 488 blood glucose concentrations (50.8%) during the before period and 164 of 485 (33.8%) during the after period (p < 0.001). We detected relative hypoglycemia in 20 (50.0%) and nine (22.5%) patients in the before and after periods, respectively (p = 0.01). On day 1, 50.0% and 16.7% received insulin in the before and after periods (p = 0.007). ICU and hospital length of stay and mortality were similar between groups. CONCLUSIONS: In a safety cohort of critically ill diabetic patients, a blood glucose concentration target of 10-14 mmol/L resulted in fewer episodes of negative glycemic distance or relative hypoglycemia and reduced insulin administration compared with a target of 6-10 mmol/L.

Laboratory alerts to guide early intensive care team review in surgical patients: A feasibility, safety, and efficacy pilot randomized controlled trial.

AIM: Common blood tests can help identify patients at risk of death, unplanned intensive care unit (ICU) admission, or rapid response team (RRT) call. We aimed to test whether early ICU-team review triggered by such laboratory tests (lab alert) is feasible, safe, and can alter physiological variables, clinical management, and clinical outcomes. METHODS: In prospective pilot randomized controlled trial in surgical wards of a tertiary hospital, we studied patients admitted for >24 h. We applied a previously validated risk assessment tool to each set of common laboratory tests to identify patients at risk and generate a "lab-alert". We randomly allocated such lab-alert patients to receive early ICU-team review (intervention) or usual care (control). RESULTS: We studied 205 patients (males 54.1%; average age 79 years; 103 randomized to intervention and 102 to usual care). Intervention patients were more likely to trigger RRT activation during their first lab-alert (10.7 vs. 2.0%; P < 0.001) but less likely to receive an allied health referral (18.0% vs. 24.5%; p = 0.007). They were less likely to trigger RRT activation in the 24-h before subsequent alerts (18.4 vs. 22.4%; p = 0.008) and less likely to generate further alerts (204 vs. 320; p < 0.001), but more likely to receive a not for resuscitation or endotracheal intubation status in the 24-h before subsequent alerts (26.6 vs. 17.3%; p = 0.05). Mortality at 24 h was 1.9% for the intervention group vs. 2.9% in the control group (p = 0.63). Finally, overall mortality was 19.4% for intervention patients vs. 23.5% for control patients (p = 0.50). CONCLUSION: Among surgical patients, lab alerts identify patients with a high mortality. Lab alert-triggered interventions are associated with more first alert-associated RRT activations; more changes in resuscitation status toward a more conservative approach; fewer subsequent alert-associated RRT activations; fewer subsequent alerts, and decreased allied health interventions (ANZCTRN12615000146594).