Evidence of the Month

Commentaries on both new and classic studies of importance for the treatment of diabetes are posted here monthly.

Frank Nobels Department of Endocrinology and Diabetes O.L.Vrouw Hospital Aalst, Belgium	Click here to download PowerPoint slides of this commentary.
	Click here for the programme evaluation and post-test.

Background
As diabetes mellitus reaches epidemic proportions, the number of hospitalized patients with the disorder also continues to rise. Glycaemic control in hospitalized patients has, however, not been a major therapeutic focus until the publication of the landmark DIGAMI-1 (Diabetes Mellitus, Insulin Glucose Infusion in Acute Myocardial Infarction) (Malmberg, 1997) and Van den Berghe et al (2001, 2006) studies. These prospective, randomized, controlled trials showed, respectively in patients with acute myocardial infarction and in critically ill patients admitted to an intensive care unit, that strict glycaemic control had a major impact on morbidity and mortality. These trials confirm extensive evidence from observational studies indicating that in hospitalized patients with critical illness, hyperglycaemia is associated with an increased risk of complications and mortality (Clement et al, 2004). This also applies beyond the area of critical care to patients admitted to general surgical and medical wards. In these patients, hyperglycaemia is also associated with increased risk of in-hospital complications, prolonged hospital stay, disability after hospital discharge, and even mortality (Clement et al, 2004).

In the critical care setting, a variety of continuous insulin infusion protocols have been shown to be effective in achieving good glycaemic control (Lecomte et al, in press). Few clinical trials have focused on the optimal management of in-patient hyperglycaemia in the non-critical setting. Surveys have shown that the quality of diabetes care on general surgical and medical wards is far from optimal (Sampson et al, 2007). Hyperglycaemia is frequently overlooked and inadequately addressed. Many patients are treated with subcutaneous regular insulin ‘sliding-scales’. These have been heavily criticized, but prospective randomized studies have until now not been performed (Queale et al, 1997). Umpierrez et al provide the first prospective randomized clinical trial that compares the efficacy and safety of sliding scale insulin with basal-bolus insulin.

Methods and Key Results
A total of 130 individuals with Type 2 diabetes, insulin naive, aged 18 to 80 years, and admitted to general medical services were randomized to receive basal-bolus insulin or a standard sliding scale protocol. Blood glucose at admission was 140 to 400 mg/dl (7.8 to 22.2 mmol/l). Individuals without a known history of diabetes, those on insulin before hospitalization, and those needing corticosteroid therapy were excluded. The goal of insulin therapy was to maintain fasting and pre-meal blood glucose levels between 70 and 140 mg/dl (3.9 and 7.8 mmol/l). Oral anti-diabetic drugs were discontinued on admission.

Patients on the basal-bolus regimen started with a total daily insulin dose of 0.4 to 0.5 units/kg, one-half given as glargine once daily and the other half as 3 equally divided glulisine doses before meals. The dose of insulin glargine was increased if the fasting and pre-meal blood glucose measurements were above target and reduced after an episode of hypoglycaemia. Supplemental insulin glulisine was given in addition to the scheduled pre-meal insulin for blood glucose >140 mg/dl (7.8 mmol/l). If a patient was not able to eat, the dose of insulin glargine was given, but the pre-meal insulin glulisine was held until meals were resumed.

Those randomized to sliding scale insulin received regular insulin 4 times daily for glucose levels >140 mg/dl (7.8 mmol/l) (a lower threshold than in sliding scales used in many hospitals). If levels remained persistently >140 mg/dl (7.8 mmol/l), the sliding scale dosing was increased to a more aggressive algorithm. For individuals not able to eat, a less aggressive algorithm was used. If the mean daily blood glucose level was >240 mg/dl (13.3 mmol/l), individuals were switched to the basal-bolus regimen.

The mean admission blood glucose was 229 ± 6 mg/dl (12.7 ± 0.3 mmol/l) and HbA_1c was 8.8 ± 2 %. Individuals treated with the basal-bolus regimen had greater improvement in glycaemic control than those treated with the sliding scale (P<0.01). The blood glucose target of 70 to 140 mg/dl (3.9 to 7.8 mmol/l) was achieved in 66% of individuals in the basal-bolus group and in 38% of those in the sliding scale group. Despite increasing insulin doses, 14% of individuals treated with sliding scale insulin remained with blood glucose >240 mg/dl (13.3 mmol/l). Their glycaemic control rapidly improved after switching to the basal-bolus insulin regimen. The rate of hypoglycaemia was similar in both groups. No blood glucose values <40 mg/dl (2.2 mmol/l) were detected. Length of hospital stay was also similar in both groups. The mean insulin daily dose was significantly higher in the basal-bolus regimen compared with that in the sliding scale treatment group, respectively, 42 ± 2 versus 12.5 ± 2 units (P<0.001).

Clinical Implications
This study confirms the view that sliding scales are an inappropriate means of in-patient blood glucose control. Although several aspects of the study can be criticized (use of glulisine in the basal-bolus group versus regular insulin in the sliding scale group, much lower insulin doses in the sliding scale group, exclusion of important patient groups), the conclusion is logical. Sliding scale regimens that call for insulin only when blood glucose levels are elevated will always alternate too much insulin with too little, promoting a roller coaster pattern of blood glucose fluctuation. Despite heavy criticism, the use of these regimens remains widespread. They are inappropriately used as an easy solution to allow junior physicians and nurses to regulate blood glucose levels, without investing too much effort. In my opinion, the main lesson that should be drawn from the study of Umpierrez et al is that an adequate in-patient blood glucose policy needs a careful organization in place instead of a one-size-fits-all regimen.

Many factors can disturb blood glucose control during hospitalization. Many patients with diabetes are not concentrated in one department, but spread throughout the hospital on wards where physicians and nurses usually lack expertise in diabetes treatment. There is increasing pressure to limit the duration of hospital stay, leaving less time for achieving a good blood glucose equilibrium and for discharge planning. Such a difficult challenge demands a good organization, with standardized order sets and support from the diabetes team. Protocols should give guidance on the following topics: how to measure blood glucose (with quality assurance of point-of-care blood glucose measurements) (Nobels et al, 2004), how to detect diabetes and/or hyperglycaemia, what level of control should be aimed for in different situations, how to treat hypoglycaemia, instructions for the use of intravenous and subcutaneous insulin and for the transition between the 2 regimens, instructions for fasting, when to ask for support of the diabetes team, and discharge planning. Ensuring continuity of care after the hospitalization is very important, so that the opportunity to improve blood glucose control does not get lost at home. Diabetes specialist nurses can fulfill a central role in this policy. They can provide help for the development of the protocols, train nurses and physicians, give support when necessary, give education for selected patients, and provide liaison with primary care.

Since every centre is uniquely organized, it will need to develop its own instructions. Unfortunately, papers with practical advice are scarce (Bhattacharyya et al, 2002; Clement et al, 2004). Since we can learn a lot from each other, hospitals should be encouraged to make their protocols available. Although this is a young research field, with many unanswered questions, especially on the targets that should be reached, on the adverse effects of hypoglycaemia, and on the practical organization of in-hospital blood glucose control, we should not await further research to improve the organization of our in-patient diabetes care.

Lecomte P, Foubert L, Nobels F, et al. Dynamic tight glycemic control during and after cardiac surgery is effective, feasible and safe. Anesth Analg. In press.