LANDMARK STUDIES

Hyperglycemia in the Hospitalized Patient

Reviewed by Yasser Ousman, MD, CDE

STUDY Hyperglycemia was present in 38% In this study, investigators divided Umpierrez GE, Isaacs SD, Bazargan N, of admitted patients; 26% had a known patients into three groups: those with a You X, Thaler LM, Kitabchi AE: history of diabetes, and 12% had no his- known history of diabetes, those with Hyperglycemia: an independent marker tory of diabetes before the admission. new hyperglycemia, and those with nor- of in-hospital mortality in patients with Newly discovered hyperglycemia moglycemia. Review of patients’ charac- undiagnosed diabetes. J Clin Endocrinol was associated with higher in-hospital teristics revealed that there were no sig- Metab 87:978Ð982, 2002 mortality rate (16%) compared to those nificant differences among the three patients with a history of diabetes (3%) patient groups in terms of mean age, sex, SUMMARY and patients with normoglycemia race, blood pressure, or admission ward Introduction. Hyperglycemia at time of (1.7%). (medicine or surgery). admission has been associated with Patients with new hyperglycemia As expected, diabetic patients and increased hospital mortality in critically had a longer average hospital stay and a patients with new hyperglycemia had ill patients; however, it is not known higher admission rate to the intensive significantly higher admission blood whether hyperglycemia in patients care unit (ICU). They were less also like- glucose levels than did those with nor- admitted to general hospital wards is ly to be discharged to home, frequently moglycemia. Compared with known also associated with poor outcomes. requiring transfer to a transitional care diabetic patients and patients with nor- unit or nursing home facility. moglycemia, those with new hyper- Design. A retrospective review of med- glycemia were more likely to be admit- ical records of 2,030 consecutive adult Conclusion. In-hospital hyperglycemia ted to the ICU. patients admitted to a community teach- is a common finding and is a marker of Total mortality was significantly ing hospital from July 1, 1998, to poor outcomes and higher mortality, higher in patients with new hyper- October 20, 1998. more so in patients with no known his- glycemia (16%) than in diabetic patients tory of diabetes than in those with a (3%) and patients with normoglycemia Aims. The goals of this study were to known history of diabetes or in those (1.7%). The difference in mortality determine the prevalence of in-hospital with normoglycemia. between the first and third groups is hyperglycemia and to determine the sur- striking: nearly 10 times as many deaths vival (primary endpoint) and functional COMMENTARY among new hyperglycemic patients than outcome (secondary endpoint) of The findings of this important study by among normoglycemic ones. The patients with hyperglycemia with and Umpierrez et al. complement and con- increased risk of in-hospital mortality in without diabetes. firm data from previous studies that patients with new hyperglycemia New hyperglycemia was defined as addressed the relationship between remained highly significant after adjust- an admission or in-hospital fasting glu- admission or in-hospital hyperglycemia ment for age, body mass index, sex, cose ≥126 mg/dl or more or a random and patients’ clinical outcomes. hypertension, coronary artery disease, blood glucose ≥200 mg/dl on two or Hyperglycemia has been associated with presence of infection, renal failure, and more determinations in patients who did poor outcomes in patients with and ICU admission. not have a known history of diabetes. without known diabetes who are admit- The investigators also examined the ted to the hospital for myocardial infarc- clinical characteristics of the deceased Results. Of the 2,030 admitted patients, tion, congestive heart failure, and stroke. patients in the three groups. Deceased 144 patients (7%) were excluded A direct relationship has also been patients in the new hyperglycemic group because no blood glucose measurement found between in-hospital glucose levels tended to be younger than those in the was recorded during the hospital stay. and the risk of post-operative infection known diabetes group and those in the Data from the remaining 1,886 patients in patients with diabetes undergoing car- normoglycemic group. The causes of were analyzed. diopulmonary bypass surgery. death were similar in all three groups.

CLINICAL DIABETES • Volume 20, Number 3, 2002 147 LANDMARK STUDIES

Mean random blood glucose was highest nosed diabetes. We know that on average are admitted to general hospital wards. in deceased patients with known dia- there is a 7- to 10 year delay between the Meanwhile, admitting physicians and betes. onset of diabetes and its diagnosis.1 It is physicians taking care of inpatients need Patients with new hyperglycemia also clear that new hyperglycemia in to address and treat hyperglycemia more had a longer mean hospital stay (9 days) patients who are admitted to a hospital is aggressively. Upon discharge and recov- compared to patients with known dia- a marker of an acute illness and does ery from acute illness, patients without betes (5.5 days) and those with normo- predict poor outcomes. This has been known diabetes but who had hyper- glycemia (4.5 days). They also were less called “stress hyperglycemia.” glycemia upon admission or during their likely to be discharged from the hospital The second question is whether hospital stay should be reevaluated and without spending time in a transitional treatment of stress hyperglycemia affects screened for diabetes, and therapeutic care unit or nursing home facility. Only patient survival. The answer to this ques- measures should be initiated. 56% of patients with new hyperglycemia tion is partially known. Malmberg et al.2 went straight home compared with 74% have shown that such an intervention in REFERENCES of patients with known diabetes and 84% diabetic patients admitted with myocar- 1Harris MI, Klein R, Welborn TA, Knuiman of normoglycemic patients. dial infarction results in significant MW: Onset of NIDDM occurs at least 4 to 7 Another interesting finding was that reduction in mortality. More recently, years before clinical diagnosis. Diabetes Care 15:815Ð819, 1992 new hyperglycemia was frequently left van den Berghe et al.3 demonstrated that 2Malmberg K, Ryden L, Efendic S, Herlitz J, untreated. Only 13% of patients had tight glucose control achieved through Nicol P, Waldenstrom A, Wedel H, Welin L: Ran- orders for a diabetic diet; 2% were pre- use of intravenous insulin infusion in domized trial of insulin-glucose infusion fol- lowed by subcutaneous insulin treatment in dia- scribed oral hypoglycemic agents; 6% mechanically ventilated patients in a sur- betic patients with acute myocardial infarction received scheduled insulin regimens; and gical intensive care unit resulted in a sig- (DIGAMI study): effect on mortality at 1 year. J 35% received sliding-scale insulin. nificant reduction in patients’ in-hospital Am Coll Cardiol 26:57Ð65, 1995 As with previous studies addressing mortality and morbidity. The reduction 3van den Berghe G, Wouters P, Weekers F, Verwaest C, Bruyninckx F, Schetz M, Vlasselaers the relationship between admission or in mortality was seen in patients with or D, Ferdinande P, Lauwers P, Bouillon R: Inten- in-hospital hyperglycemia and patients’ without known diabetes. sive insulin therapy in the surgical intensive care unit. N Engl J Med 345:1359Ð1367, 2001 clinical outcomes, this study raises two Based on the data provided by questions. Umpierrez et al., the logical next step First, what is the clinical significance would be to conduct a prospective, ran- Yasser Ousman, MD, CDE, is the associ- of hyperglycemia in inpatients with no domized trial that examines the impact ate director of the diabetes team at known history of diabetes? It is clear that of tight or improved glucose control on Washington Hospital Center in Washing- some of these patients do have undiag- morbidity and mortality in patients who ton, D.C.

148 Volume 20, Number 3, 2002 • CLINICAL DIABETES