Hyperglycemia is highly prevalent among non-critically ill hospitalized patients, and there are multiple challenges to achieving safe and effective glycemic control in the inpatient setting. In this persuasive study by Kyi and colleagues, the investigators sought to test the hypothesis that clinical inertia by primary inpatient teams in consulting inpatient diabetes specialists contributes to less-effective glycemic control in the hospital. Using a parallel cluster randomized design in eight medical/surgical units, the study compared usual care (inpatient diabetes consultations requested by primary admitting team) vs. early identification (within first 24 hours) and reflex management by a specialized diabetes team for all inpatients with hyperglycemia.
The authors found that early detection and management of hyperglycemia by a diabetes management team resulted in a 24% reduction in adverse glycemic days, defined as a blood glucose less than 4 mmol/L (<72 mg/dl) or >15 mmol/l (>270 mg/dl) and greater use of insulin initiation in previously naïve patients (57% vs. 34%, P = 0.001) compared to usual care. While early intervention reduced overt hyperglycemia by 55%, hypoglycemia rates were not significantly reduced.
Since multiple observational studies have linked hyperglycemia to poorer clinical outcomes, the study also evaluated the effect of the intervention on secondary outcomes of hospital-acquired infections, acute kidney injury, acute myocardial infarction, unplanned critical care admission and in-hospital mortality. Over the course of this 24-week trial, there was an 80% reduction in the odds of hospital-acquired infections, but no difference in the other secondary clinical outcomes.
This is an important study because it is the first randomized controlled trial to demonstrate that early detection and intervention in hyperglycemic inpatients is effective at improving glycemic control and suggests that diabetes specialists are underutilized in the hospital. Rushakoff and colleagues used a similar approach to detect patients with extreme hyperglycemia via the electronic medical record and delivered an electronic consultation service, and in an observational study was found to be highly effective at reducing rates of both hyperglycemia and hypoglycemia (Rushakoff RJ, et al. Ann Intern Med. 2017;doi: 10.7326/M16-1413). Moreover, the study adds to evidence supporting the role of glycemic control to prevent hospital-acquired infections.
While the results of this innovative clinical care model were promising, there are several important points to note regarding the generalizability of the study findings. First, this Australian-based study did not follow the U.S. practice of routinely discontinuing non-insulin antihyperglycemic medications at admission for the majority of hospitalized patients. Second, the outcomes were limited to the first 14 days of admission, although this would not be expected to have significant impact on inferences drawn since only a very small minority of inpatients have a length of stay longer than this. Third, although glucose data were available electronically, the hospital in which this study was performed did not have an EMR and structured insulin/glucose order sets, which may have affected the quality of diabetes care provided in the usual care arm. At our institution, for example, a subcutaneous insulin decision-support tool is available for providers to assist them in selection of an initial insulin regimen (Mathioudakis N, et al. J Diabetes Sci Technol. 2018;doi: 10.1177/1932296818798036).
Fourth, there were some differences in the intervention group compared to control group (fewer males, lower baseline HbA1c, less insulin use at baseline, more emergent and surgical admissions). While these differences are likely due to chance alone, a strength of this study was the inclusion of baseline and active periods in both intervention and control arms, which can account for secular trends in hospital care that are not directly related to the intervention itself.
The authors acknowledge that it could be practically challenging to carry out this type of resource-intensive intervention in a real-world setting, and a cost-effectiveness analysis would be needed to better understand whether the incremental increase in resources (eg, number of diabetes specialists need for higher volume of consultations) is offset by savings (eg, reducing hospital-acquired infections).
Nestoras Mathioudakis, MD, MHS
Assistant Professor of Medicine
Clinical Director, Endocrinology, Diabetes, & Metabolism
Johns Hopkins University School of Medicine
Disclosures: Mathioudakis reports no relevant financial disclosures.