Hyperglycemia After SAH
Predictors, Associated Complications, and Impact on Outcome
Background and Purpose— Hyperglycemia is common after subarachnoid hemorrhage (SAH). The extent to which prolonged hyperglycemia contributes to in-hospital complications and poor outcome after SAH is unknown.
Methods— We studied an inception cohort of 281 SAH patients with an initial serum glucose level obtained within 3 days of SAH onset and who had at least 7 daily glucose measurements between SAH days 0 and 10. We defined mean glucose burden (GB) as the average peak daily glucose level >5.8 mmol/L (105 mg/dL). Hospital complications were recorded prospectively, and 3-month outcome was assessed with the modified Rankin scale.
Results— The median GB was 1.8 mmol/L (33 mg/dL). Predictors of high-GB included age ≥54 years, Hunt and Hess grade III–V, poor Acute Physiology and Chronic Health Evaluation (APACHE)-2 physiological subscores, and a history of diabetes mellitus (all P≤0.001). In a multivariate analysis, GB was associated with increased intensive care unit length of stay (P=0.003) and the following complications: congestive heart failure, respiratory failure, pneumonia, and brain stem compression from herniation (all P<0.05). After adjusting for Hunt-Hess grade, aneurysm size, and age, GB was an independent predictor of death (odds ratio, 1.10 per mmol/L; 95% CI, 1.01 to 1.21; P=0.027) and death or severe disability (modified Rankin scale score of 4 to 6; odds ratio, 1.17 per mmol/L; 95% CI 1.07 to 1.28, P<0.001).
Conclusions— Hyperglycemia after SAH is associated with serious hospital complications, increased intensive care unit length of stay, and an increased risk of death or severe disability.
The detrimental effect of hyperglycemia has been well studied in acute vascular syndromes. In ischemic stroke, hyperglycemia occurs in 20% to 40% of patients and is associated with infarct expansion, worse functional outcome, longer hospital stays, higher medical costs, and an increased risk of death.1–3 Both elevated serum glucose levels and diabetes are independently associated with symptomatic hemorrhage after intravenous or intra-arterial thrombolysis for acute cerebral infarction.4–6 In medical and surgical critically ill patients, hyperglycemia is common even in nondiabetic patients and has been associated with increased morbidity and mortality.7 Intensive insulin therapy in medical and surgical intensive care unit (ICU) patients has been shown to reduce the incidence of sepsis, acute renal failure, blood transfusions, critical-illness polyneuropathy, length of ICU stay, and mortality.8–10
Although hyperglycemia is common after subarachnoid hemorrhage (SAH), previous studies have shown an inconsistent relationship between elevated glucose levels on admission and outcome.11–13 To our knowledge, this study is the first to systematically evaluate the cumulative effects of prolonged hyperglycemia in the ICU on hospital complications and outcome after SAH.
The Columbia University SAH Outcomes Project prospectively enrolled 580 patients with spontaneous SAH admitted to the Neurologic ICU, between July 1, 1996, and May 1, 2002. Patients were included in the present analysis if they met 3 criteria: (1) an admission serum glucose value was recorded within SAH days 0 to 3; (2) ≥7 daily glucose values were recorded between SAH days 0 to 10; and (3) a 3-month modified Rankin scale (mRS)14 score or, in its absence, 14-day mRS score was available. SAH day 0 refers to the calendar day of hemorrhage onset.
The study was approved by the hospital Institutional Review Board, and, in all of the cases, written informed consent was obtained from the patient or a surrogate. The diagnosis of SAH was established on the basis of admission computed tomographic (CT) scans or by xanthochromia of the cerebrospinal fluid. Exclusions included secondary SAH from trauma, arteriovenous malformation, or other causes, age <18 years, or admission >14 days after SAH onset. The management of these patients has been described in detail previously.15 Patients who underwent craniotomy received 6 mg of dexamethasone every 6 hours perioperatively. All of the patients were given normal saline without dextrose and supplemental 5% albumin infusions to maintain central venous pressure ≥5 mm Hg. Enetral feeding was initiated on the day after admission and was held 12 hours before procedures. Glucose values >9.9 mmol/L (180 mg/dL) were treated with subcutaneous sliding scale insulin every 4 to 6 hours. Insulin infusion protocols were not routinely used unless glucose levels consistently exceeded 13.3 mmol/L (240 mg/dL) or if the patient became ketotic.
Glucose Burden Calculation
We recorded the highest daily level obtained between SAH days 0 and 10 and defined daily glucose burden (GB) as the excess of this value >5.8 mmol/L (105 mg/dL). We chose this reference value because it is the upper limit of normal for our laboratory and because of recent data indicating that tight glucose control to approximately this level is associated with improved survival in critically ill patients.9 When a daily maximum glucose value equal to or below the reference value was recorded, a GB of zero was assigned for that day. Mean GB was calculated by averaging all of the daily GBs measured between SAH days 0 to 10. Admission GB was calculated using the highest glucose value obtained during the first calendar day of admission, referenced to 5.8 mmol/L.
Secondary analyses relating GB to 3-month outcome were also performed using 2 higher reference values. The reference glucose value of 7.8 mmol/L (140 mg/dL) was selected based on a receiver-operating characteristic analysis of serum glucose measurements to identify a level that most accurately distinguished favorable (mRS 1 to 3) from unfavorable (mRS 4 to 6) outcome at 3 months (sensitivity 0.69 and specificity 0.73). We also examined the reference point of 11.1 mmol/L (200 mg/dL), which is generally accepted as a level of severe hyperglycemia.
Clinical and Radiographic Assessment
We recorded baseline demographic data, past medical and social history, clinical features at SAH onset, clinical status on admission, and admission and follow-up CT findings as described previously.15–18 Admission clinical status was evaluated with the Hunt-Hess scale16 and the APACHE-2 physiological subscore17 (see supplemental Table I, available online at http://stroke.ahajournals.org). Prospectively recorded hospital complications included the following: respiratory failure requiring mechanical ventilation, sepsis/bacteremia, congestive heart failure, mean fever burden >38°C between SAH days 0 to 10, anemia (hemoglobin <9.0 mg/L) treated with blood transfusion, brain stem compression from herniation, treated hydrocephalus, aneurysm rebleeding, symptomatic vasospasm, and cerebral infarction. Criteria defining these variables have been described previously.19
Survival and functional outcome were assessed at 3 months using the mRS (0=full recovery, 6=death); poor outcome was defined as death or severe disability (mRS score 4 to 6). When 3-month mRS scores were not available, the 14-day mRS was analyzed according to the principle of last observation carried forward. We also assessed high-level instrumental disability with the Lawton Instrumental Activities of Daily Living (IADL) scale (scored 8=fully independent and 30=completely dependent),20 Quality of Life with the Sickness Impact Profile Total score (scored 0=no dysfunction and 100=maximal dysfunction),21 and global cognitive function with the Telephone Interview of Cognitive Status (scored 51=best and 0=worst).22
We evaluated admission predictors of GB, associations of GB with concurrent hospital complications, and associations of GB with 3-month outcome. Continuous variables were dichotomized based on clinical cut points or median values. χ2 analysis was used to evaluate categorical variables. Student t test was used between normally distributed continuous variables. Candidate demographic and admission variables in the univariate analysis were used to create a multivariable model for independent predictors of GB using generalized linear regression. Significant hospital complications associated with GB on univariate analysis were then added individually to this model to calculate adjusted probability values.
Logistic regression was used to identify independent admission predictors of dichotomized 3-month outcome measures (mRS, Lawton IADL, Telephone Interview of Cognitive Status, and Sickness Impact Profile total score). Admission and mean GB were then added individually to these models to calculate adjusted odds ratios for the strength of association of hyperglycemia with each aspect of outcome. Tests for interactions were performed for all of the significant variables in the multivariable models. Continuous nonnormally distributed dependent variables were transformed using square-root transformation to perform linear regression analysis. Significance was set at the P≤0.05 level for all of the analyses.
Of 576 enrolled patients, 7 were excluded because of unavailable outcome data, 209 because of the absence of a baseline glucose value between SAH days 0 to 3, and 79 because of the lack of ≥7 daily glucose values between SAH days 0 to 10. Of the remaining 281 patients who were included in the analysis, 72% were poor grade (Hunt-Hess III–V), and 10% had a history of diabetes (supplemental Table I). Excluded patients were overrepresented in the Hunt-Hess grades I and V subclasses. Excluded patients also had less SAH and intraventricular hemorrhage, a lower rate of diabetes, and shorter ICU stays.
The median peak glucose value in the study population was 7.6 mmol/L (range, 3.2 to 40.5 mmol/L), and the median GB exceeding 5.8 mmol/L was 1.8 mmol/L (range, 0.1 to 12.9 mmol/L). All of the patients had a positive GB >5.8 mmol/L, 95% had a burden >7.8 mmol/L, and 25% had a burden >11.1 mmol/L. The GB was highest on SAH day 1, and diabetic patients sustained higher glucose levels than nondiabetic patients (Figure).
Generalized multiple linear regression (R2=0.43 for the entire model; P<0.001) identified a history of diabetes mellitus, age ≥54 years, an APACHE-2 physiological subscore ≥5 (P=0.001), and Hunt-Hess grade III–V (P=0.001) as independent predictors of GB (Table 1). When both diabetes and APACHE-2 subscore ≥5 were present, there was a synergistic increase in GB (P=0.017).
Several hospital complications had significant univariate associations with GB, with the strongest associations noted for congestive heart failure and respiratory failure requiring mechanical ventilation (Table 2). Both of these complications retained their significant association with GB, as did brain stem compression from herniation and pneumonia, after controlling for admission predictors. Steroids were administered for postsurgical edema prophylaxis to 74% of patients (207 of 281). There was no increase in GB in patients who received steroids compared with those who did not. Increased ICU length of stay was significantly associated with GB (F=11.21; P=0.001; degrees of freedom=1).
A 14-day mRS was carried forward in 70 of the 281 study patients. The mean 14-day mRS score of these patients was not significantly different from those who were alive with 90-day follow-up. Admission Hunt-Hess grade, aneurysm size, and age were identified as significant independent admission predictors of death and death or severe disability using multiple logistic regression (Nagelkerke R2=0.37; P=0.001). When added to this multivariate model, peak admission glucose level was not predictive of death or severe disability (Table 3). By contrast, mean GB >5.8 mmol/L was predictive of death alone and death or severe disability, after controlling for the above risk factors. GB referenced to 7.8 mmol/L was also predictive of death or severe disability but not mortality as a single end point. Both admission and mean GB referenced to all 3 of the cut points independently predicted loss of independence in IADLs after controlling for level of education, admission Hunt-Hess grade, and race/ethnicity (admission model Nagelkerke R2=0.21; P=0.001). By contrast, no aspect of GB was predictive of cognitive impairment or quality of life.
Our study demonstrates the negative impact of sustained hyperglycemia after SAH. In contrast to prior studies that have analyzed admission and mean hospital glucose levels,12,13 the calculation of GB in our study allowed for a more comprehensive evaluation of the extent of extreme levels of hyperglycemia over time. GB >5.8 mmol/L was independently associated with death, severe disability, and loss of high-level functional independence at 3 months. These findings suggest that the use of intensive insulin therapy to achieve tight control of glucose to normal levels in patients with SAH may be important.
Hyperglycemia in the setting of acute neurological injury is felt to be attributed, in part, to a catecholamine surge and generalized stress response.2,23 We found that poor Hunt-Hess grade and elevated APACHE-2 physiological subscores on admission were independently predictive of GB, thus providing a link among the severity of neurological injury, the overall extent of physiological derangement, and hyperglycemia. Other predictors of GB included older age and a history of diabetes mellitus. The systemic inflammatory response syndrome, autonomic nervous system instability, hypothalamic-pituitary axis dysfunction, and neurogenic cardiopulmonary injury have also been observed after SAH, indicating that hyperglycemia reflects only 1 aspect of a broad range of homeostatic derangements.24
Controversy exists regarding the impact of hyperglycemia on outcome after SAH. Lanzino et al13 found an association between admission glucose levels and death or severe disability after SAH, but this association was not significant after controlling for other outcome predictors. Similarly, Dorhout Mees et al12 found that admission glucose level was not an independent predictor of outcome after SAH. By contrast, a smaller study by Alberti et al7 did find admission glucose to be a predictor of outcome after SAH after controlling for clinical grade and CT findings, but only Hunt-Hess grade IV and V patients were assessed.
In our study, increased GB showed a stronger association with a combined end point of death or severe disability than with mortality alone. This suggests that hyperglycemia may be primarily associated with disabling but nonfatal complications, such as ICU neuropathy, nosocomial infections, and impaired wound healing.8–10 Strict glucose control has also been linked to reductions in intracranial pressure, duration of mechanical ventilation, and seizures in critically ill neurological patients.10 The observed association between hyperglycemia and increased ICU length of stay suggests that physical deconditioning may directly contribute to functional disability in affected patients. The lack of association between GB >11.1 mmol/L and death or severe disability was likely because of underpowering, as only 25% of patients had a burden over this level.
The strong association between hyperglycemia and various hospital complications provides a plausible mechanism for how hyperglycemia may contribute to poor outcome after SAH. However, it must be emphasized that it is impossible to determine whether hyperglycemia is a cause or an effect of these complications. The association of GB with pneumonia may be explained in part by hyperglycemia-related immunosuppression or by the increased risk of respiratory failure treated with mechanical ventilation. The association of hyperglycemia with congestive heart failure and brain stem herniation may reflect a secondary stress response resulting from these complications.
Others have reported an association between hyperglycemia and an increased risk of symptomatic vasospasm after SAH.11,25 We failed to confirm this association for reasons that are unclear. Although others have also found an association between hyperglycemia and the magnitude of brain edema in ischemic stroke and intracerebral hemorrhage26,27 we also failed to find an association between GB and global cerebral edema on admission.
A number of limitations of our study should be mentioned. Both poor and good Hunt-Hess grade patients were underrepresented, because they were unlikely to stay in the ICU long enough to accrue the 7 daily glucose values needed for study inclusion. Calculation of a true “area under the curve” based on a regular schedule of frequent glucose measurements, rather than using the single highest value on any given day, would have added precision to our study, but at the expense of simplicity. Seventy-four percent of our patients received steroids, and glucose levels >9.9 mmol/L were routinely treated with subcutaneous insulin. It is possible that the observed association between GB and outcome might have been weaker had we not used steroids or stronger if insulin had not been used. Finally, a 14-day mRS score was carried forward in 25% of patients in lieu of a 3-month score because of incomplete follow-up.
In summary, GB after SAH is associated with an increased frequency of complications, prolonged ICU length of stay, and an increased risk of death or functional disability. Clinical trials are needed to assess the impact of strict normoglycemic management in SAH patients.
This work was supported by a Grant-in-Aid from the American Heart Association to S.A.M. (#9750432N).
- Received July 15, 2005.
- Accepted July 27, 2005.
Baird TA, Parsons MW, Phanh T, Butcher KS, Desmond PM, Tress BM, Colman PG, Chambers BR, Davis SM. Persistent poststroke hyperglycemia is independently associated with infarct expansion and worse clinical outcome. Stroke. 2003; 34: 2208–2214.
Capes SE, Hunt D, Malmberg K, Pathak P, Gerstein HC. Stress hyperglycemia and prognosis of stroke in nondiabetic and diabetic patients: a systematic overview. Stroke. 2001; 32: 2426–2432.
Williams LS, Rotich J, Qi R, Fineberg N, Espay A, Bruno A, Fineberg SE, Tierney WR. Effects of admission hyperglycemia on mortality and costs in acute ischemic stroke. Neurology. 2002; 59: 67–71.
Bruno A, Levine SR, Frankel MR, Brott TG, Lin Y, Tilley BC, Lyden PD, Broderick JP, Kwiatkowski TG, Fineberg SE. Admission glucose level and clinical outcomes in the NINDS r-tPA stroke trial. Neurology. 2002; 59: 669–674.
Demchuk AM, Morgenstern LB, Krieger DW, Linda Chi T, Hu W, Wein TH, Hardy RJ, Grotta JC, Buchan AM. Serum glucose level and diabetes predict tissue plasminogen activator-related intracerebral hemorrhage in acute ischemic stroke. Stroke. 1999; 30: 34–39.
Kase CS, Furlan AJ, Wechsler LR, Higashida RT, Rowley HA, Hart RG, Molinari GF, Frederick LS, Roberts HC, Gebel JM, Sila CA, Schulz GA, Roberts RS, Gent M. Cerebral hemorrhage after intra-arterial thrombolysis for ischemic stroke: The proact ii trial. Neurology. 2001; 57: 1603–1610.
Van den Berghe G, Schoonheydt K, Becx P, Bruyninckx F, Wouters PJ. Insulin therapy protects the central and peripheral nervous system of intensive care patients. Neurology. 2005; 64: 1348–1353.
Charpentier C, Audibert G, Guillemin F, Civit T, Ducrocq X, Bracard S, Hepner H, Picard L, Laxenaire MC. Multivariate analysis of predictors of cerebral vasospasm occurrence after aneurysmal subarachnoid hemorrhage. Stroke. 1999; 30: 1402–1408.
Dorhout Mees SM, van Dijk GW, Algra A, Kempink DR, Rinkel GJ. Glucose levels and outcome after subarachnoid hemorrhage. Neurology. 2003; 61: 1132–1133.
van Swieten JC, Koudstaal PJ, Visser MC, Schouten HJ, van Gijn J. Interobserver agreement for the assessment of handicap in stroke patients. Stroke. 1988; 19: 604–607.
Claassen J, Bernardini GL, Kreiter K, Bates J, Du YE, Copeland D, Connolly ES, Mayer SA. Effect of cisternal and ventricular blood on risk of delayed cerebral ischemia after subarachnoid hemorrhage: The Fisher scale revisited. Stroke. 2001; 32: 2012–2020.
Hijdra A, van Gijn J, Nagelkerke NJ, Vermeulen M, van Crevel H. Prediction of delayed cerebral ischemia, rebleeding, and outcome after aneurysmal subarachnoid hemorrhage. Stroke. 1988; 19: 1250–1256.
Commichau C, Scarmeas N, Mayer SA. Risk factors for fever in the neurologic intensive care unit. Neurology. 2003; 60: 837–841.
Damiano A. Sickness Impact Profile. User’s Manual and Interpretation Guide. Baltimore, MD: The Johns Hopkins University Press; 1996.
Brandt JSM, Folstein M. The telephone interview for cognitive status. Neuropsychiatry Neuropsychol Behav Neurol. 1988; 1: 111–117.
Allport LE, Butcher KS, Baird TA, MacGregor L, Desmond PM, Tress BM, Colman P, Davis SM. Insular cortical ischemia is independently associated with acute stress hyperglycemia. Stroke. 2004; 35: 1886–1891.
Berger L, Hakim AM. The association of hyperglycemia with cerebral edema in stroke. Stroke. 1986; 17: 865–871.
Song EC, Chu K, Jeong SW, Jung KH, Kim SH, Kim M, Yoon BW. Hyperglycemia exacerbates brain edema and perihematomal cell death after intracerebral hemorrhage. Stroke. 2003; 34: 2215–2220.