Abstract
Background: Hyperglycemia at the time of acute ischemic stroke has been linked to worse outcome in both human and animal studies.
Objective: To describe the prevalence and severity of hyperglycemia on hospital admission among acute ischemic stroke patients, to examine the independent relationship of admission hyperglycemia to all-cause mortality, and to document the inpatient management of hyperglycemia.
Methods: Patients hospitalized with acute ischemic stroke at one hospital from July 1993 to June 1998 (n = 656) were identified. Demographic data, diagnoses, and blood glucose (BG) values were retrieved from the electronic medical record system. Admission stroke severity, fingerstick BG results, and new diabetes diagnoses were obtained by chart review. Hyperglycemia was defined as admitting random serum BG ≥ 130 mg/dL. Hazard ratios (HR) for 30-day, 1-year, and 6-year mortality were calculated using multivariable Cox regression models.
Results: Hyperglycemia at admission to hospital was present in 40% of patients with acute stroke. Patients with hyperglycemia were more often women and more likely to have prior diagnoses of diabetes and heart failure. Almost all of these patients remained hyperglycemic during their hospital stay (mean BG = 206 mg/dL), and 43% received no inpatient hypoglycemic drugs. Hyperglycemic patients had longer hospital stay (7 vs 6 days, p = 0.015) and higher inpatient hospital charges ($6,611 vs $5,262, p < 0.001). Hyperglycemia independently increased the risk for death at 30 days (HR 1.87, p ≤ 0.01), 1 year (HR 1.75, p ≤ 0.01), and 6 years after stroke (HR 1.41, p ≤ 0.01).
Conclusions: Admitting hyperglycemia was common among patients with acute ischemic stroke and was associated with increased short- and long-term mortality and with increased inpatient charges. Inpatient blood glucose management was suboptimal in this hospital. A trial of intensive treatment of hyperglycemia should be considered.
Hyperglycemia has been reported to augment acute ischemic brain injury in many human and animal studies.1-10⇓⇓⇓⇓⇓⇓⇓⇓⇓ Possible mechanisms of this effect include increased brain tissue acidosis, increased blood–brain barrier permeability, and increased hemorrhagic transformation of the infarct. Whether acute hyperglycemia independently affects patient outcomes or whether this effect primarily reflects the effects of increased stroke severity or poor glycemic control is also debated. Although some studies have not found an effect of acute blood glucose (BG) on subsequent stroke outcomes,11-13⇓⇓ many of these studies have been limited by small size, lack of control for other important clinical factors, and relatively short patient follow-up periods. Furthermore, even if hyperglycemia is accepted as adversely affecting stroke outcome, little data exist about the type or effectiveness of commonly used treatment(s) for hyperglycemia in acute ischemic stroke patients. The goals of this study were: 1) to document the prevalence and severity of acute hyperglycemia in a large cohort of patients with acute ischemic stroke, 2) to examine the independent effect of acute hyperglycemia on short- and long-term all-cause mortality, and 3) to document the type and effectiveness of inpatient treatments for hyperglycemia among patients with acute ischemic stroke.
Methods.
Data were obtained from the Regenstrief Medical Record System (RMRS), an electronic medical record system that captures patient level laboratory, clinical, and utilization data for inpatient and outpatient encounters at Wishard Hospital, an inner-city public teaching hospital in Indianapolis, IN.14 Patients admitted to the hospital with acute ischemic stroke between July 1, 1993 and June 30, 1998 were identified using primary position discharge ICD-9 codes 434 and 436,15 stroke diagnosis was confirmed by chart review in all patients, and follow-up data were collected through June 30, 1999. We extracted the following data from patients’ electronic medical records: demographics, inpatient and outpatient diagnoses before stroke, all blood glucose values obtained at the time of presentation for acute stroke and throughout the stroke-related hospitalization, diabetes medications (insulin in any form or route and oral hypoglycemic drugs) administered during the 6 months before the index stroke and during the hospital stay, glycosylated hemoglobin values obtained within 6 months of stroke admission, and health services utilization and charges during the hospitalization. All-cause mortality was determined using hospital discharge data stored in the RMRS and death certificate files from the Indiana State Department of Health.
Additional data were obtained by reviewing patients’ paper hospital charts, including retrospective assessment of stroke severity on admission using the NIH Stroke Scale (NIHSS),16 bedside fingerstick blood glucose assessments (number of tests and values), and new diagnoses of diabetes. New diabetes diagnosis was defined as recording a diagnosis of type I or II diabetes during the index admission or discharge, with a diabetic medication in a patient with no history of diabetes or record of having taken a diabetic medication. NIHSS at the time of admission with stroke could be estimated in only 66% of patients because of absent written medical records. Most of these records had been sent to long-term storage and were irretrievable.
Patients were classified as hyperglycemic on admission with stroke if admitting blood glucose was ≥130 mg/dL. Patients without an admitting blood glucose value were excluded. Comparisons of differences between patients with and without admission hyperglycemia were performed with Student’s t-tests or χ2 tests. Multivariable modeling of time to death associated with admission hyperglycemia was performed with Cox regression models. Separate models were constructed for all-cause 30-day, 1-year, and 6-year mortality. Independent variables included patient demographics, other medical diagnoses at time of stroke, and admission hyperglycemia as a dichotomous variable. All analyses were done using SAS software (SAS Institute, Cary, NC). This study was approved by the Indiana University Institutional Review Board.
Results.
Sample characteristics.
Using ICD-9 codes, we identified 671 patients with acute ischemic stroke; ischemic stroke diagnosis was confirmed in 656 (98%), of whom 643 (98%) had an admission blood glucose value. Table 1 shows other demographic and risk factor characteristics of the sample. Age range was 19 to 95 years with a mean age of 65 years; slightly more than half were women and 55% were black. Admission hyperglycemia was present in 258 (40%) patients, and 163 (25%) had admission blood glucose ≥ 180 mg/dL. Patients with hyperglycemia on admission were more often women, were more likely to have diagnoses of diabetes and chronic heart failure, and had slightly higher (worse) NIHSS scores. Nearly one third of the euglycemic patients had diabetes. Forty-seven (18%) of the patients with hyperglycemia had no prior evidence of diabetes; of these, five (11%) were newly diagnosed with diabetes during their stroke admission.
Demographics
Inpatient BG management.
An average of 6 additional serum BG tests were obtained for 175 (68%) patients with hyperglycemia (table 2). Blood glucose was monitored by fingerstick testing only (no serum BG tests) in 36 (14%) of the patients with hyperglycemia, with an average of eight fingerstick tests per patient. All hyperglycemic patients had at least one serum or fingerstick BG obtained during their stroke admission. Hyperglycemic patients’ serum BG remained high during their hospital stay with mean BG in this group of 206 ± 5 mg/dL. Ninety percent of the patients with admission hyperglycemia had a mean BG ≥ 130 mg/dL during their hospital stay. As expected, patients with hyperglycemia with a prior diagnosis of diabetes had higher BG during their hospitalization (mean BG 228 vs 138 mg/dL, p < 0.001), more BG tests (mean 8 vs 5 tests, p = 0.01), and were more likely to have mean BG ≥ 130 mg/dL during their hospital stay (84% vs 34%, p < 0.001, data not shown) than patients with hyperglycemia without diabetes. Only 147 (44%) of the 334 diabetic patients had glycosylated hemoglobin values recorded in the 6 months before or after stroke; of these, 111 (76%) had elevated glycosylated hemoglobin levels.
Inpatient blood glucose (BG) monitoring and control
In-hospital treatment of hyperglycemia was suboptimal (table 3): 43% of hyperglycemic patients received no diabetes medications during their hospital stay, 10% received only oral hypoglycemics, and 13% received only sliding scale insulin (no concomitant fixed dose subcutaneous insulin or oral hypoglycemics). Mean serum BG was lowest in patients not receiving diabetes medications.
In-hospital treatment in patients with admission hyperglycemia (n = 258)
Patient outcomes.
Patients with admission hyperglycemia had longer inpatient stays (7.2 vs 6.0 days, p = 0.015) and higher median inpatient charges ($6,611 vs $5,262, p < 0.001, table 4). More patients with hyperglycemia died within 30 days and within 1 year after stroke, although in-hospital mortality did not differ among groups. With controls for age, sex, race, and chronic conditions, admission hyperglycemia independently increased the hazard of dying at 30 days, 1 year, and 6 years after stroke (1 year data shown in table 5, figure). Variables significant in the 30-day and 6-year multivariable survival models were (HR, 95% CI): 30-day mortality—hyperglycemia (1.87, 1.05 to 3.32), age (1.04 per year, 1.02 to 1.07), and prior diagnosis of hypertension (0.48, 0.25 to 0.9); 6-year mortality—hyperglycemia (1.41, 1.02 to 1.94), age (1.04 per year, 1.03 to 1.06), and history of myocardial infarction (1.49, 1.04 to 2.14).
Patient outcomes by BG group
One-year survival analysis
Figure. Six-year survival curve. Solid line indicates subjects with admission BG < 130 mg/dL. Dashed line indicates subjects with admission BG ≥ 130 mg/dL. Time is shown as days since index ischemic stroke.
Discussion.
Compared with patients with euglycemia, patients with hyperglycemia on admission had significantly increased risk for death at 30 days, 1 year, and 6 years after stroke. Importantly, the effect of hyperglycemia on subsequent risk for death was independent of the risk conferred by a diagnosis of diabetes, other chronic cardiovascular conditions, and age. Hyperglycemia was not associated with an increased risk for in-hospital death. This suggests that the association of hyperglycemia with worse patient outcomes is not merely a reflection of the phenomenon of stress-induced hyperglycemia that can occur in patients with severe stroke.
Our study is the first to report BG management and utilization data in patients with hyperglycemia with acute ischemic stroke. Although hyperglycemia was common in our patients, inpatient BG management was suboptimal. Nine of 10 patients with admission hyperglycemia continued to have high BG during their hospital stay, and 43% of hyperglycemic patients received no hypoglycemic medications during their admission. Other studies have also documented poor BG management in the hospital setting, indicating that this problem is not limited to stroke patients.17,18⇓ Hyperglycemic stroke patients spent significantly more time in the hospital and, consequently, had higher inpatient stroke charges. When we removed patients with BG > 500 mg/dL from the cohort, who are more likely to receive IV insulin and be managed in the intensive care unit, the effect of hyperglycemia on length of stay and charges persisted (data not shown).
Hyperglycemia was present in more than 40% of our inner-city patients with acute ischemic stroke and nearly one fourth of patients had BG > 180 mg/dL. Most of the patients with hyperglycemia either had a prior diagnosis of DM or received this diagnosed during their stroke admission; only 16% of patient with hyperglycemia on admission either had transient hyperglycemia not associated with diabetes or were diagnosed with diabetes later during outpatient follow-up. Because our cohort is more than 50% black, this high prevalence of diabetes is partly because of the known increased prevalence of diabetes in blacks.19 However, researchers have found that the prevalence of diabetes in patients with stroke increased significantly between 1970 and 1985,20 and the prevalence of diabetes in the population at large has recently been shown to be rapidly increasing.21 As a result, the proportion of patients with acute ischemic stroke presenting with hyperglycemia may also be expected to increase.
Other studies have reported conflicting effects of hyperglycemia on mortality after ischemic stroke, with some finding increased short-term mortality,1-3,22-26⇓⇓⇓⇓⇓⇓⇓ increased long-term mortality,20 or no effect of acute hyperglycemia on mortality.11-13,27⇓⇓⇓ Few studies report patient follow-up beyond 3 to 6 months after stroke, and many of these studies have not used multivariable survival analyses. Hyperglycemia also has been linked to mortality after myocardial infarction.28 Our results parallel these findings, and also show that hyperglycemia remains an independent risk for mortality even up to 6 years after stroke.
An important question is whether the effect of hyperglycemia during acute stroke is independent of long-term glycemic control. However, because less than half of our patients with diabetes had a glycosylated hemoglobin value within 6 months before or after stroke, our ability to evaluate the impact of glycemic control on mortality is limited. Other studies have reported that the increased mortality seen in hyperglycemic stroke patients is not an effect of poor long-term glycemic control.5,29⇓ Another limitation of this study is the lack of stroke severity estimates on all patients, which limited our ability to evaluate the impact of hyperglycemia on clinical stroke outcomes and stroke-related mortality. Prior analyses of a large, randomized controlled clinical trial sample demonstrated that hyperglycemia was an independent predictor of poor clinical outcomes.6 Finally, our patients have a high prevalence of diabetes and many barriers to optimal glycemic control,30 thus our results may not be representative of other practice situations.
Nonetheless, these data show that hyperglycemia is common, is often inadequately managed, and is associated with a nearly twofold increase in all-cause mortality after acute ischemic stroke. Importantly, this effect on mortality is present 30 days after stroke onset and persists for at least 6 years after stroke. Although the precise mechanisms by which hyperglycemia is associated with increased poststroke mortality are unknown, our data show that this association is present and is independent of demographic and other chronic cardiovascular conditions. Because acute treatment of hyperglycemia is available in most acute care settings, hyperglycemia is an attractive target for acute stroke therapy. There is evidence that hyperglycemia can be rapidly and safely treated in acute stroke,31 and that acute insulin treatment in patients with diabetes with acute myocardial infarctions reduces subsequent mortality32; together with these reports, our data support the need for a trial to determine the efficacy of rapid normalization of hyperglycemia in acute ischemic stroke.
Acknowledgments
Supported by grant D43-TW01082 from the Fogarty International Center, NIH, and was performed in part in the Regenstrief Institute for Health Care; and by a Research Career Award, Health Services Research and Development, Veterans Health Administration (L.S.W.).
Acknowledgment
The authors thank David A. Pérez Martinez, MD, and Jadwiga Pack-Kania, MD, for assistance in reviewing medical records of stroke patients.
- Received September 21, 2001.
- Accepted March 22, 2002.
References
Disputes & Debates: Rapid online correspondence
NOTE: All authors' disclosures must be entered and current in our database before comments can be posted. Enter and update disclosures at http://submit.neurology.org. Exception: replies to comments concerning an article you originally authored do not require updated disclosures.
- Stay timely. Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- 200 words maximum.
- 5 references maximum. Reference 1 must be the article on which you are commenting.
- 5 authors maximum. Exception: replies can include all original authors of the article.
- Submitted comments are subject to editing and editor review prior to posting.