Patients at greatest risk for adverse hospital outcomes may be those without a previous history of diabetes, which emphasizes the importance of treating inpatient hyperglycemia regardless of the cause. A retrospective review of medical records of more than 2,000 critically ill patients showed significant increases in mortality among patients with new hyperglycemia, that is, those without previously diagnosed diabetes. The mortality rate was about eight times higher among patients with hyperglycemia as those with normal glucose levels (P < 0.01) and about five times higher than patients with diagnosed diabetes (P < 0.01).14 In the aforementioned retrospective study of 250,000 admissions, the odds of mortality were significantly higher for patients with no previous history of diabetes than for those with diagnosed diabetes (P < 0.01). Compared with normoglycemic patients, those without diagnosed diabetes had a 35% increased risk of death if their glucose was 111–145 mg/dL (OR 1.35; 95% CI 1.30–1.41). Successively increasing glucose ranges were associated with a doubled, tripled, and quadrupled mortality risk (146–199 mg/dL: OR 2.14, 95% CI 2.04–2.24; 200–299 mg/dL: OR 2.91, 95% CI 2.71–3.11; >300 mg/dL: OR 4.04, 95% CI 3.44–4.75).18 Observational studies such as these can only be considered hypothesis generating. Indeed, because the degree of illness will be associated with the level of stress hyperglycemia, studies such as these are particularly prone to influence by unmeasured confounders. It is only through randomized clinical trials that one may know whether glucose control actually improves the risk of adverse outcomes in hospitalized patients with hyperglycemia.
Effects of Glycemic Control on Inpatient Outcomes
Both retrospective and prospective controlled studies have brought us to our current understanding of optimal glucose control for both critically ill and noncritically ill patients. In the DIGAMI study, 1-year mortality significantly decreased by 29% (P = 0.027) in patients with diabetes randomly assigned to insulin-glucose infusion for the first 24 h after acute MI compared with patients given standard therapy of the time, in which insulin was given during the first 24 h only if it was deemed clinically necessary.25 Other early controlled trials comparing tight glucose control to standard treatment approaches also demonstrated significant reductions in mortality and morbidity among both ICU and noncritical inpatient populations. In an often-cited 2001 prospective, randomized clinical trial involving 1,548 surgical ICU patients in Belgium, intensive insulin therapy to maintain glycemia between 80 and 110 mg/dL significantly reduced mortality risk by 32% (P < 0.04) compared with standard treatment of the time, in which insulin was given only when patients’ blood glucose exceeded 215 mg/dL, with the goal of maintaining blood glucose values of 180-200 mg/dL. In addition, intensive insulin therapy also significantly reduced the duration of ICU stays and ventilatory support, need for dialysis, and episodes of septicemia.1 These results were supported by a retrospective analysis of ICU patients who had undergone coronary artery bypass grafting (CABG). The investigators compared hospital records from two time periods during which different glucose control approaches had been used. During the earlier period, subcutaneous (SQ) insulin had resulted in a mean glucose value of 213 mg/dL among CABG patients, whereas a later protocol using insulin infusion resulted in a mean glucose value of 177 mg/dL. Insulin infusion reduced the risk of death by 57% (OR 0.43; P = 0.001).19
Further investigations, however, were not able to confirm the benefit of near normalization of glucose and raised concerns regarding the risk of hypoglycemia with this approach. In 2003, the same Belgian investigators published results from a second clinical trial, showing that tight glucose control did not significantly reduce mortality in the medical (as opposed to surgical) ICU, except among patients whose ICU stays exceeded 5 days.26 Subsequently, the NICE-SUGAR study highlighted the dangers of hypoglycemia that accompany tight glucose control. This international investigation compared 90-day mortality in a cohort of 6,104 patients who were admitted to medical and surgical ICUs at 42 different hospitals where they were assigned randomly to glycemic targets of 81–108 mg/dL and 144–180 mg/dL. Rates of severe hypoglycemia were ~15 times greater among intensively treated patients (OR 14.7, 95% CI 9.0–25.9; P < 0.001), and mortality was 14% higher in the same group compared with patients whose glucose was less intensively controlled (P = 0.02).9 A meta-analysis of 26 randomized controlled trials involving 13,567 ICU patients (including the NICE-SUGAR cohort) supported the NICE-SUGAR finding that the risk of hypoglycemia is too great to justify near-normal glucose values, especially in medical ICU patients. In the pooled analysis of studies reporting hypoglycemia, the relative risk of hypoglycemia was 6.0 (95% CI 4.5–8.0) for intensive insulin therapy compared with conventional glucose control. Meanwhile, the overall relative risk of death was 0.93 (95% CI 0.83–1.04).8 As a result of these findings, recommendations for target glucose levels were relaxed from the euglycemic range to the values shown in Table 1.1.
Protocols for Glucose Management
A major goal of treating hyperglycemia in the hospital is patient safety, because overtreatment and undertreatment of hyperglycemia represents major quality concerns. Well-defined, validated protocols for the management of hyperglycemia will include provisions for glucose monitoring and the treatment of hypoglycemia as well as guidance on matching insulin administration to blood glucose levels and nutrition (either meals or enteral or parenteral nutrition) in a dynamic fashion.
Changing patient circumstances also drive modifications to insulin regimens in the hospital. These include transitions from ICU to noncritical care settings, which call for changes from IV infusion to SQ injections of insulin; nutrition therapy transitions between enteral or parenteral therapy and solid foods; or perioperative glycemic control. Patients admitted for diabetic hyperglycemic crises (diabetes ketoacidosis or hyperglycemic hyperosmolar state) also will require insulin therapy along with close monitoring of blood glucose values to reduce the risk of hypoglycemia. Of course, these patients also require extensive management decisions related to fluids and electrolytes, beyond mere glycemic control.
Monitoring the patient’s glycemic status falls to point-of-care (POC) capillary blood glucose meters, which provide nearly instantaneous results and have become the standard measurement technique at the hospital bedside. Caution is required in interpreting the results from POC meters in patients who have anemia, polycythemia, or hypoperfusion or who use certain medications. Newer technologies, including continuous glucose monitoring, are under study.
Ongoing education of hospital personnel in these protocols is essential not only to ensure proper implementation but also to gain support of those involved in the care of inpatients with hyperglycemia, including the hospital administration. Evidence supporting the cost-effectiveness of a rational systems approach to inpatient glycemic management will help persuade administrators to provide necessary financial and operational support.27,28
Status of Glycemic Control in the Hospital Setting
The health-care community at large now generally accepts that both hyperglycemia and hypoglycemia are markers of poor clinical outcomes, and many institutions have made important strides to improve glycemia at their facilities. Multiple barriers persist, however, and the frequency of poor glycemic control remains high. In an analysis of a database containing information on 70,000 admissions of patients with diabetes, an HbA1c was recorded for only 18% of cases.29 The authors of this study found that, when A1C was measured, a value >8% prompted a change in antihyperglycemic regimen for only two-thirds of patients (64%). Additionally, several studies have documented failures to reliably follow hypoglycemia management protocols, with long delays in glucose retesting after hypoglycemic events, poor documentation of the hypoglycemic and subsequent treatment, and long intervals before hypoglycemia resolution.30–32 The root of the problem may be in poor communication and coordination between health-care teams,30,33 but knowledge gaps also appear to contribute. In a recent survey of health-care professionals working in an urban, community teaching hospital, only about half of questions regarding best practices for managing inpatient hyperglycemia were answered correctly by physicians, nurses, and dietitians (mean scores of 53%, 52%, and 48%, respectively). Pharmacists performed