David Levy, MD, FRCP
There has been a recent resurgence of interest in the management of diabetes in hospitalized patients. Guidelines in the USA are well-established, and in the UK the ‘ThinkGlucose’ campaign has highlighted the importance of good diabetes care in hospital. Note that good diabetes care does not simply mean good (i.e. ‘tight’) glucose control: the criteria for judging this are controversial and there is little RCT evidence, though an abundance of less robust data. Minimizing prescribing errors (especially of insulin), ensuring awareness of diabetes in its entirety for the whole hospital team, aiming where possible for autonomy in self-care and insulin administration, quality improvement and audit are all critical components of good inpatient diabetes care. Equally important is tailoring the level of intensity of glucose management to the patient’s clinical need, and this requires clinical acumen and insight….
Many cohort and registry studies have found that new-onset hyperglycemia in hospitalized patients not known to have diabetes carries a high hospital mortality, higher even than patients with known diabetes, although the findings are not consistent. ‘Stress hyperglycemia’ (hospital/acute illness-related hyperglycemia) – hyperglycemia in people without known diabetes that settles once the acute admission is over – is a concept with a long history  and there are many factors that may contribute to vascular harm and a poor prognosis in this state. Now that tentative agreement has been reached on the HbA1c diagnosis of diabetes (see Chapter 1), both newly diagnosed and stress hyperglycemia will be easier to diagnose and differentiate, for example using a combination of blood glucose and HbA1c measurements.
- New-onset diabetes: fasting glucose > 7.0 mmol/L (126 mg/dL) or random glucose > 11.1 mmol/L (200 mg/dL) or HbA1c > 6.5% (48 mmol/ mol); one measure preferably repeated for confirmation.
- ‘Stress hyperglycemia’: blood glucose levels as for new-onset diabetes, but HbA1c < 6.5% (48 mmol/mol).
In specific conditions, for example acute coronary syndromes (ACS) and intensive care, the assumption that ‘lower is better’ has recently been challenged. However, the basis for good perioperative glycemic control is reasonably well founded. These are all areas in which best practice is likely to continue to change rapidly.
Acute Coronary Syndromes (ST-segment elevation acute myocardial infarction, non-ST-segment elevation acute myocardial infarction and unstable angina)
The epidemiology is clear: ACS patients with diabetes have a higher rate of all adverse outcomes (heart failure, renal failure, cardiogenic shock and death). In the USA, mortality rates are higher during admission, at 30 days and at 1 year, although some of these differences may be historical and some related to underuse of proven secondary prevention measures in diabetic patients (e.g. beta-blockers). In the UK between 1995 and 2003, although 30-day mortality improved, and to a similar degree as in non-diabetic patients, the late outcomes (18-month mortality) did not change . Mortality after percutaneous coronary intervention (PCI) undertaken for any reason in diabetic patients is about two-thirds higher than in non-diabetic subjects for at least 4 years after the procedure, highlighting the need for intensive secondary prevention measures (see below) .
Several factors are likely to be responsible for the increased risk in diabetes.
- Classical (textbook) ischemic symptoms are less marked or absent, presumably due to somatic and autonomic neuropathy. Think of cardiac ischemia in a diabetic patient who describes shortness of breath (diastolic dysfunction), nausea/vomiting, sweating, even non-specific upper body discomfort. Patients are unlikely to use textbook terminology if English is not their first language.
- Hospital presentation is later and coronary interventions delayed.
- The extent of infarction is probably no greater than in non-diabetic patients, but coronary artery disease is more diffuse and more advanced, and associated risk factors more marked.
Prevalence of Diabetes in ACS Patients and Identification of Diabetes Status
Glucose intolerance is common in patients with myocardial infarction. In a Scandinavian study (2002), prevalence rates immediately after the event and some months later were similar : normal glucose tolerance, 50%; IGT, 30%; diabetes, 20%. Nevertheless, random admission glucose values were unremarkable, typical of pre-diabetes, with a mean blood glucose of only 6.1 mmol/L (110 mg/dL). This increased to 6.9 mmol/L (124 mg/dL) in a mixed-ethnicity group in the UK, comprising 30% South Asians, but with a similar proportion of diabetes and IGT (30% each). Diagnostic HbA1c measurements may help in this situation, but the lability of glucose measurements in stressed medical conditions warns against adoption of simple numerical limits. For example in the Scandinavian study mean admission HbA1c was only 5.0% (31 mmol/mol), yet a level of 5.3% (34 mmol/mol) strongly predicted diabetes on the OGTT – compare the proposed lower limit of 5.7% (39 mmol/mol) for ‘pre-diabetes’.
Glycemic Control in ACS – DIGAMI and Beyond
Glucose/insulin/potassium (GIK) infusions in the acute phase of myocardial infarction have a long history; the metabolic reasoning includes converting myocardial fatty acid metabolism to more efficient glucose metabolism. Translating the experimental rationale into evidence-based clinical practice has been long and inconclusive. The first DIGAMI study (1997)  showed that 24-hour GIK treatment after admission with transmural myocardial infarction in patients with blood glucose above 11.1 mmol/L (200 mg/dL), followed by 3 months of basal bolus subcutaneous insulin treatment, reduced fatal infarctions by 11%, but there was no effect on total reinfarction rate. Those with newly presenting and known diabetes gained similar benefit, and younger people without heart failure fared particularly well. DIGAMI is still widely cited as justification for this complicated regimen, but it was conducted early in the thrombolysis era; very few had PCI, post-infarction prophylaxis was relatively rudimentary, and BARI 2D (though in patients with stable advanced coronary artery disease) could not find prognostic advantage in the use of insulin. It is difficult to draw conclusions on glycemic control from the follow-up, DIGAMI 2 (2005) , which attempted to separate out the contributions of acute intravenous insulin and longer-term subcutaneous insulin. The study was underpowered and terminated early, and although glycemic targets were not met, mean admission HbA1c was already low (7.3%, 56 mmol/mol), no significant separation occurred between the three groups, and overall glycemic control did not change. Perhaps the most important finding was that cardiac outcomes were at least as good as, if not better than, a concurrent non-diabetic registry group, and vigorous management of all factors (of which glycemic control was one, but only one), including PCI in 40%, is likely to explain the excellent outcome in this otherwise high-risk group.
Further doubt was cast on GIK by the huge CREATE-ECLA study (2007) which recruited about 20,000 patients with ST-segment elevation acute myocardial infarction (STEMI) around the world . Mortality or serious cardiovascular end points were no different in those with and without diabetes, although blood glucose levels were relatively high (8–10 mmol/L, 144–180 mg/dL). An analysis including CREATE-ECLA and another GIK study, OASIS-6 GIK, raised concerns that a higher rate of death or heart failure in the first 2 days after STEMI might be related to hyperglycemia, hyperkalemia and fluid overload as a result of the GIK infusion. The usual plea is that maintaining strict normoglycemia (4–6 mmol/L, 72–108 mg/dL) might reveal cardiac benefits of GIK treatment not apparent in studies using less rigorous glycemic control, but the practical difficulties and potential risks (particularly hypoglycemia) of achieving this either in a large RCT or in individual patients using current technology are prohibitive. The practical approach should be to maintain glycemic control in ACS patients at levels recommended in ambulatory practice (e.g. < 10 mmol/L, 180 mg/dL), which will mean a temporary intravenous insulin infusion for some. Although the level at which it should be initiated is not known, the DIGAMI threshold of 11.1 mmol/L, still widely used, is arbitrary.