- Email: [email protected]
Acute hyperglycaemia and mortality in patients with transient ischaemic attack
Journal of Clinical Neuroscience 17 (2010) 305–307
Contents lists available at ScienceDirect
Journal of Clinical Neuroscience journal homepage: www.elsevier.com/locate/jocn
Clinical Study
Acute hyperglycaemia and mortality in patients with transient ischaemic attack Matthew N.T. Thuy a,b,*, Peter J. Hand a a b
Department of Neurology, Royal Melbourne Hospital, Parkville, Victoria, Australia Department of Medicine, Royal Melbourne Hospital, Parkville, Victoria, Australia
a r t i c l e
i n f o
Article history: Received 23 September 2008 Accepted 6 July 2009
Keywords: Australia Blood glucose Cerebrovascular disorders Hyperglycemia Ischaemic attack Mortality Transient ischaemic attack
a b s t r a c t Acute hyperglycaemia is associated with poorer outcome in stroke, however limited evidence is available regarding its association with transient ischaemic attack (TIA). This study aimed to determine the association between acute hyperglycaemia and mortality in 194 patients with TIA. Mortality data were obtained from a state-wide death registry. No significant association was identified with either multivariate Cox regression (p = 0.65) or Kaplan-Meier analysis (p = 0.85). Because of the low death rate, a larger sample is required to reliably exclude an association. Univariate analysis identified significantly associated variables, including TIA clinical prediction scores (e.g. ABCD and ABCD2). Multivariate analysis identified age, atrial fibrillation and duration P 1 hour as independent significant predictors of mortality. Ó 2009 Elsevier Ltd. All rights reserved.
1. Introduction
2. Methods
Acute hyperglycaemia is an adverse prognostic factor in acute stroke;1 however, limited evidence is available regarding its association with transient ischaemic attack (TIA). The relationship between hyperglycaemia and stroke may be different to that between hyperglycaemia and TIA. Stroke patients suffer a wide range of additional complications (such as infection, poor oral intake, prolonged cerebral ischaemia/infarction) that may interact with hyperglycaemia. Previous studies have suggested that prognosis is improved after correction of hyperglycaemia in intensive care patients2 (although a recent large trial3 has brought this into question) and patients with acute myocardial infarction.4 However, in stroke patients this has not been demonstrated. Short-term correction of mild hyperglycaemia in stroke patients has not been associated with improved outcomes;5 however, that study was underpowered with small overall changes in glucose. If a relationship between hyperglycaemia and outcome in TIA patients were to be established, it would provide important additional prognostic information and a potential therapeutic pathway for future research. The aim of this study was to identify the relationship between acute blood glucose and mortality in patients with TIA.
A retrospective review of non-consecutive TIA patients admitted to the Royal Melbourne Hospital’s Stroke Unit (2001–06) was undertaken. Patients not admitted to the Stroke Unit (e.g. admitted to the emergency department and then discharged) were not reviewed. Patients were identified from the Stroke Unit Database and Department of Neurology discharge summaries. Electronic pathology results were utilised. Inclusion criteria were: (i) TIA according to the definition of the World Health Organization (< 24 hours);6 and (ii) a blood glucose measurement taken during admission. Exclusion criteria (Table 1) were: (i) concurrent terminal illness (e.g. metastatic cancer); (ii) subsequent non-TIA diagnosis on follow up; (iii) TIA P10 days pre-admission; (iv) iatrogenic TIAs; (v) TIAs due to arterial dissection; and (vi) permanent residence outside Victoria. The earliest available blood glucose level was used. One author (MT) was involved in identifying and extracting the clinical information. TIA patients were admitted to the stroke unit because of clinician concern. No specific selection policy for admission was in place between 2001 and 2006. Mortality was determined by a state-wide death registry search (Victorian Registry of Births, Deaths and Marriages). Hyperglycaemia, blood glucose level and other variables were analysed in a multivariate Cox regression to determine independent statistical significance. Kaplan-Meier analysis was also performed to compare data collected from the hyperglycaemic and non-hyperglycaemic patients. Analysis was performed using Statistical Package for the Social Sciences version 16.0 (SPSS, Chicago, IL, USA). Ethical
* Corresponding author. Present address: Austin Hospital, Austin Health, 145 Studley Rd, Heidelberg, Victoria 3084, Australia. Tel.: +61 3 9496 5000. E-mail address: [email protected] (M.N.T. Thuy). 0967-5868/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.jocn.2009.07.089
306
M.N.T. Thuy, P.J. Hand / Journal of Clinical Neuroscience 17 (2010) 305–307
Table 1 Excluded patients with acute hyperglycaemia and transient ischemic attack (TIA) Reason for exclusion
No. patients excluded
Duration >24 hrs Alternative diagnosis or TIA thought unlikely on follow-up No glucose measurement Iatrogenic TIAs or due to carotid/ vertebral dissection Missing notes Accidentally missed Residence outside Victoria TIA P 10 days before admission Metastatic cancer Total
Percentage of patients excluded (%)
41 30
36.0 26.3
23 6
20.2 5.3
4 4 3 2 1 114
3.5 3.5 2.6 1.8 0.9 100.1
Due to rounding, percentages do not add to 100%.
approval was obtained from the Royal Melbourne Hospital ethics committee. Hyperglycaemia was defined as a serum glucose P 7.0 mmol/L, as a previous meta-analysis7 of hyperglycaemia and stroke outcome found studies generally defined hyperglycaemia at a blood concentration of between P 6 mmol/L and 8 mmol/L. History and new diagnoses of hypertension and diabetes were identified from the medical records and discharge summaries. Variables of interest not found in the record or summary were analysed as not being present.
3. Results Eligible participants (194 patients) were identified from 308 medical records, and 114 patients were excluded (Table 1). The median patient age was 74 years (interquartile range 65–81 years), 52% were male, and 27.8% had hyperglycaemia according to our definition. Of the TIAs, 51% lasted P 1 hour. Of the patients, 72% had hypertension, 45% had hypercholesterolaemia, 31% had a history of smoking, 22% had diabetes and 21% had atrial fibrillation (AF). A total of 34 (17.5%) patients died; the median follow-up was 826 days (interquartile range 374–1308 days). Many of the causes of death were not stroke or vascular-related (Table 2). Three patients (1.5%) had a stroke while inpatients. Twenty patients (10.3%) had carotid stenoses of P 50% on ultrasound or angiography. The following variables had a statistically significant univariate association with mortality: age, duration P1 hour, AF, history of ischaemic heart disease, history of stroke, history of peripheral vascular disease and history of TIA. Validated clinical prediction scores by Johnston et al.8 (California P 3, relative risk [RR] 3.3, p = 0.014), Rothwell et al.9 (ABCD scale score P 5, RR 2.6, p = 0.01) and Johnston and Rothwell et al.10 (ABCD2 scale score P 6 versus 0–3, RR
Table 2 Cause of death of patients with acute hyperglycaemia and transient ischemic attack Cause of death Ischaemic heart disease Cancer Stroke or stroke-related complications Infection Other causes Total
No. of patients deceased 9 8 6 6 5 34
Due to rounding, percentages do not add to 100%.
Percentage of patients deceased (%) 26.4 23.5 17.6 17.6 14.7 99.8%
Table 3 Adjusted multivariate Cox regression analysis of the independent association of various factors with transient ischaemic attack
Acute hyperglycaemia (P7.0 mmol/L) Age (per 10 years) TIA duration P 1 hour Atrial fibrillation Past IHD Past stroke Past TIA Past PVD
RR
95% CI
p value
1.21 1.97 2.87 3.20 1.59 1.74 2.32 1.77
0.54–2.69 1.89–2.05 1.32–6.25 1.46–7.03 0.77–3.31 0.80–3.75 0.99–5.42 0.63–4.99
0.647 0.002 0.008 0.004 0.214 0.161 0.052 0.281
CI = confidence interval, IHD = ischaemic heart disease, PVD = peripheral vascular disease, RR = relative risk, TIA = transient ischaemic attack.
4.9, p = 0.013, ABCD 4–5 versus 0–3, RR 3.6, p = 0.042) were all highly predictive of subsequent mortality. Neither hyperglycaemia (RR 0.93, 95% confidence interval [CI] 0.43–2.0, p = 0.846) nor blood glucose (RR 1.03 per mmol/L increase, 95% CI 0.87–1.22, p = 0.731) were significant predictors of mortality. Other non-significant variables were sex, hypertension, diabetes, cholesterol, smoking, earliest blood pressure reading >140/90 mmHg, speech or motor symptoms. Hyperglycaemia and significant predictors of mortality (except the clinical prediction scores) were incorporated into a multivariate Cox regression analysis (Table 3). Only age, TIA duration P1 hour and AF retained independent significant associations. The clinical prediction scores, when incorporated, did not retain independent significance. They are not included as they are correlated with two of the variables. Hyperglycaemic and non-hyperglycaemic groups were compared in a Kaplan-Meier curve analysis (Supplementary Fig. 1). There were no significant differences in survival (Mantel-Cox, p = 0.846).
4. Discussion This retrospective study attempted to determine the association between acute hyperglycaemia and mortality in 194 patients with TIA who were admitted to a stroke unit. No significant association was found in this sample of patients, including in a multivariate model. This may be unsurprising, given the low overall death rate and the subsequent treatment of diabetes. The death rate was particularly low early after TIA (1.55% at 30 days, 3.09% at 90 days), which is consistent with at least one previous well-known study.8 Given the low death rate, a much larger sample is required to detect a significant difference in mortality due to the presence of hyperglycaemia. The small sample size and because many deaths were not due to neurological or vascular causes might explain the weak impact of other important risk factors like hypertension, blood pressure > 140/90 mmHg, diabetes and smoking. Despite this, some variables had statistically significant univariate associations. These were age, duration >1 hour, AF, past history of ischaemic heart disease, peripheral vascular disease, stroke and TIA, and the recently validated clinical prediction scores, California, ABCD and ABCD2. Further multivariate analysis found age, AF and duration >1 hour to be independently associated with mortality, with other variables no longer independently significant. It is unsurprising that age and AF were independently associated with mortality. Older persons are probably more likely to die due to higher rates of pre-existing disease and increased debility. Regarding AF, many patients (44%) died of ischaemic heart disease or stroke, with AF possibly suggesting either pre-existing cardiovascular disease or high subsequent risk of stroke. It is more difficult to explain why a duration > 1 hour correlates
M.N.T. Thuy, P.J. Hand / Journal of Clinical Neuroscience 17 (2010) 305–307
independently with mortality. It may be that longer TIAs are more likely to represent true vascular events as opposed to TIA mimics. One small study found that TIA duration >1 hour was significantly correlated with cerebral infarction on MRI and thromboembolic sources (such as from the carotid arteries or heart).11 A past history of vascular events was not independently significant and this may be because such patients tended to be older. In a larger sample, these variables may be independently statistically significant. The clinical prediction scores for subsequent stroke (California, ABCD, ABCD2) were also highly correlated with mortality but were not independently significant. This is probably because they correlate with both age and clinical duration. The correlation of these scores with mortality is a useful finding of relevance to clinical practice. The ability of these scores to predict future stroke was not tested in this study. Inpatient stroke occurrence and death from stroke were not analysed due to the small number of patients. Collection of information about recurrent vascular events in the community and independent self care (using the Modified Rankin Scale) was attempted; however, after careful consideration, these data were not published. A future paper may investigate possible clinical prediction of carotid stenosis in TIA patients. 5. Conclusion In this sample, hyperglycaemia at TIA diagnosis is not significantly associated with mortality in patients admitted into a stroke unit. Given the low death rate, a much larger sample may be required to reliably exclude any significant association. Acknowledgements We are very grateful to the following: Professor S. Davis for support and critical feedback; Royal Melbourne Hospital’s Health Information Services for retrieval of records; Victorian Department
307
of Births, Deaths and Marriages for patient mortality data; Marnie Collins for statistical analysis and review and Professor E. Storey for critical comments.
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.jocn.2009.07.089. References 1. Baird T, Parsons MW, Barber PA, et al. The influence of diabetes mellitus and hyperglycaemia on stroke incidence and outcome. J Clin Neuroscience 2002;9:618–26. 2. Van den Berghe G, Wouters P, Weekers F, et al. Intensive insulin therapy in critically ill patients. N Engl J Med 2001;345:1359–68. 3. The NICE-SUGAR Study Investigators Finfer S, Chittock DR, et al. Intensive versus conventional glucose control in critically ill patients. N Engl J Med 2009;360:1283–97. 4. Malmberg K, Ryden L, Efendic S, et al. Randomized trial of insulin-glucose infusion followed by subcutaneous insulin treatment in diabetic patients with acute myocardial infarction (DIGAMI study): effects on mortality at 1 year. J Am Coll Cardiol 1995;26:57–65. 5. Gray CS, Hildreth AJ, Sandercock PA, et al. Glucose-potassium-insulin infusions in the management of post-stroke hyperglycaemia: the UK Glucose Insulin in Stroke Trial (GIST-UK). Lancet Neurol 2007;6:397–406. 6. World Health Organization. Cerebrovascular disorders: a clinical and research classification. Offset publication No 43 Geneva: World Health Organization 1978. 7. Capes S, Hunt D, Malmberg K, et al. Stress hyperglycaemia and prognosis of stroke in diabetic and non-diabetic patients: a systematic overview. Stroke 2001;32:2426–32. 8. Johnston SC, Gress DR, Browner WS, et al. Short-term prognosis after emergency deparment diagnosis of TIA. JAMA 2000;284:2901–6. 9. Rothwell PM, Giles MF, Flossmann E, et al. A simple score (ABCD) to identify individuals at high early risk of stroke after transient ischaemic attack. Lancet 2005;366:29–36. 10. Johnston S, Rothwell P, Nguyen-Huynh M, et al. Validation and refinement of scores to predict very early stroke risk after transient ischaemic attack. Lancet 2007;369:283–92. 11. Kimura K, Minematsu K, Yasaka M, et al. The duration of symptoms in transient ischemic attack. Neurology 1999;52:976–80.
Contents lists available at ScienceDirect
Journal of Clinical Neuroscience journal homepage: www.elsevier.com/locate/jocn
Clinical Study
Acute hyperglycaemia and mortality in patients with transient ischaemic attack Matthew N.T. Thuy a,b,*, Peter J. Hand a a b
Department of Neurology, Royal Melbourne Hospital, Parkville, Victoria, Australia Department of Medicine, Royal Melbourne Hospital, Parkville, Victoria, Australia
a r t i c l e
i n f o
Article history: Received 23 September 2008 Accepted 6 July 2009
Keywords: Australia Blood glucose Cerebrovascular disorders Hyperglycemia Ischaemic attack Mortality Transient ischaemic attack
a b s t r a c t Acute hyperglycaemia is associated with poorer outcome in stroke, however limited evidence is available regarding its association with transient ischaemic attack (TIA). This study aimed to determine the association between acute hyperglycaemia and mortality in 194 patients with TIA. Mortality data were obtained from a state-wide death registry. No significant association was identified with either multivariate Cox regression (p = 0.65) or Kaplan-Meier analysis (p = 0.85). Because of the low death rate, a larger sample is required to reliably exclude an association. Univariate analysis identified significantly associated variables, including TIA clinical prediction scores (e.g. ABCD and ABCD2). Multivariate analysis identified age, atrial fibrillation and duration P 1 hour as independent significant predictors of mortality. Ó 2009 Elsevier Ltd. All rights reserved.
1. Introduction
2. Methods
Acute hyperglycaemia is an adverse prognostic factor in acute stroke;1 however, limited evidence is available regarding its association with transient ischaemic attack (TIA). The relationship between hyperglycaemia and stroke may be different to that between hyperglycaemia and TIA. Stroke patients suffer a wide range of additional complications (such as infection, poor oral intake, prolonged cerebral ischaemia/infarction) that may interact with hyperglycaemia. Previous studies have suggested that prognosis is improved after correction of hyperglycaemia in intensive care patients2 (although a recent large trial3 has brought this into question) and patients with acute myocardial infarction.4 However, in stroke patients this has not been demonstrated. Short-term correction of mild hyperglycaemia in stroke patients has not been associated with improved outcomes;5 however, that study was underpowered with small overall changes in glucose. If a relationship between hyperglycaemia and outcome in TIA patients were to be established, it would provide important additional prognostic information and a potential therapeutic pathway for future research. The aim of this study was to identify the relationship between acute blood glucose and mortality in patients with TIA.
A retrospective review of non-consecutive TIA patients admitted to the Royal Melbourne Hospital’s Stroke Unit (2001–06) was undertaken. Patients not admitted to the Stroke Unit (e.g. admitted to the emergency department and then discharged) were not reviewed. Patients were identified from the Stroke Unit Database and Department of Neurology discharge summaries. Electronic pathology results were utilised. Inclusion criteria were: (i) TIA according to the definition of the World Health Organization (< 24 hours);6 and (ii) a blood glucose measurement taken during admission. Exclusion criteria (Table 1) were: (i) concurrent terminal illness (e.g. metastatic cancer); (ii) subsequent non-TIA diagnosis on follow up; (iii) TIA P10 days pre-admission; (iv) iatrogenic TIAs; (v) TIAs due to arterial dissection; and (vi) permanent residence outside Victoria. The earliest available blood glucose level was used. One author (MT) was involved in identifying and extracting the clinical information. TIA patients were admitted to the stroke unit because of clinician concern. No specific selection policy for admission was in place between 2001 and 2006. Mortality was determined by a state-wide death registry search (Victorian Registry of Births, Deaths and Marriages). Hyperglycaemia, blood glucose level and other variables were analysed in a multivariate Cox regression to determine independent statistical significance. Kaplan-Meier analysis was also performed to compare data collected from the hyperglycaemic and non-hyperglycaemic patients. Analysis was performed using Statistical Package for the Social Sciences version 16.0 (SPSS, Chicago, IL, USA). Ethical
* Corresponding author. Present address: Austin Hospital, Austin Health, 145 Studley Rd, Heidelberg, Victoria 3084, Australia. Tel.: +61 3 9496 5000. E-mail address: [email protected] (M.N.T. Thuy). 0967-5868/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.jocn.2009.07.089
306
M.N.T. Thuy, P.J. Hand / Journal of Clinical Neuroscience 17 (2010) 305–307
Table 1 Excluded patients with acute hyperglycaemia and transient ischemic attack (TIA) Reason for exclusion
No. patients excluded
Duration >24 hrs Alternative diagnosis or TIA thought unlikely on follow-up No glucose measurement Iatrogenic TIAs or due to carotid/ vertebral dissection Missing notes Accidentally missed Residence outside Victoria TIA P 10 days before admission Metastatic cancer Total
Percentage of patients excluded (%)
41 30
36.0 26.3
23 6
20.2 5.3
4 4 3 2 1 114
3.5 3.5 2.6 1.8 0.9 100.1
Due to rounding, percentages do not add to 100%.
approval was obtained from the Royal Melbourne Hospital ethics committee. Hyperglycaemia was defined as a serum glucose P 7.0 mmol/L, as a previous meta-analysis7 of hyperglycaemia and stroke outcome found studies generally defined hyperglycaemia at a blood concentration of between P 6 mmol/L and 8 mmol/L. History and new diagnoses of hypertension and diabetes were identified from the medical records and discharge summaries. Variables of interest not found in the record or summary were analysed as not being present.
3. Results Eligible participants (194 patients) were identified from 308 medical records, and 114 patients were excluded (Table 1). The median patient age was 74 years (interquartile range 65–81 years), 52% were male, and 27.8% had hyperglycaemia according to our definition. Of the TIAs, 51% lasted P 1 hour. Of the patients, 72% had hypertension, 45% had hypercholesterolaemia, 31% had a history of smoking, 22% had diabetes and 21% had atrial fibrillation (AF). A total of 34 (17.5%) patients died; the median follow-up was 826 days (interquartile range 374–1308 days). Many of the causes of death were not stroke or vascular-related (Table 2). Three patients (1.5%) had a stroke while inpatients. Twenty patients (10.3%) had carotid stenoses of P 50% on ultrasound or angiography. The following variables had a statistically significant univariate association with mortality: age, duration P1 hour, AF, history of ischaemic heart disease, history of stroke, history of peripheral vascular disease and history of TIA. Validated clinical prediction scores by Johnston et al.8 (California P 3, relative risk [RR] 3.3, p = 0.014), Rothwell et al.9 (ABCD scale score P 5, RR 2.6, p = 0.01) and Johnston and Rothwell et al.10 (ABCD2 scale score P 6 versus 0–3, RR
Table 2 Cause of death of patients with acute hyperglycaemia and transient ischemic attack Cause of death Ischaemic heart disease Cancer Stroke or stroke-related complications Infection Other causes Total
No. of patients deceased 9 8 6 6 5 34
Due to rounding, percentages do not add to 100%.
Percentage of patients deceased (%) 26.4 23.5 17.6 17.6 14.7 99.8%
Table 3 Adjusted multivariate Cox regression analysis of the independent association of various factors with transient ischaemic attack
Acute hyperglycaemia (P7.0 mmol/L) Age (per 10 years) TIA duration P 1 hour Atrial fibrillation Past IHD Past stroke Past TIA Past PVD
RR
95% CI
p value
1.21 1.97 2.87 3.20 1.59 1.74 2.32 1.77
0.54–2.69 1.89–2.05 1.32–6.25 1.46–7.03 0.77–3.31 0.80–3.75 0.99–5.42 0.63–4.99
0.647 0.002 0.008 0.004 0.214 0.161 0.052 0.281
CI = confidence interval, IHD = ischaemic heart disease, PVD = peripheral vascular disease, RR = relative risk, TIA = transient ischaemic attack.
4.9, p = 0.013, ABCD 4–5 versus 0–3, RR 3.6, p = 0.042) were all highly predictive of subsequent mortality. Neither hyperglycaemia (RR 0.93, 95% confidence interval [CI] 0.43–2.0, p = 0.846) nor blood glucose (RR 1.03 per mmol/L increase, 95% CI 0.87–1.22, p = 0.731) were significant predictors of mortality. Other non-significant variables were sex, hypertension, diabetes, cholesterol, smoking, earliest blood pressure reading >140/90 mmHg, speech or motor symptoms. Hyperglycaemia and significant predictors of mortality (except the clinical prediction scores) were incorporated into a multivariate Cox regression analysis (Table 3). Only age, TIA duration P1 hour and AF retained independent significant associations. The clinical prediction scores, when incorporated, did not retain independent significance. They are not included as they are correlated with two of the variables. Hyperglycaemic and non-hyperglycaemic groups were compared in a Kaplan-Meier curve analysis (Supplementary Fig. 1). There were no significant differences in survival (Mantel-Cox, p = 0.846).
4. Discussion This retrospective study attempted to determine the association between acute hyperglycaemia and mortality in 194 patients with TIA who were admitted to a stroke unit. No significant association was found in this sample of patients, including in a multivariate model. This may be unsurprising, given the low overall death rate and the subsequent treatment of diabetes. The death rate was particularly low early after TIA (1.55% at 30 days, 3.09% at 90 days), which is consistent with at least one previous well-known study.8 Given the low death rate, a much larger sample is required to detect a significant difference in mortality due to the presence of hyperglycaemia. The small sample size and because many deaths were not due to neurological or vascular causes might explain the weak impact of other important risk factors like hypertension, blood pressure > 140/90 mmHg, diabetes and smoking. Despite this, some variables had statistically significant univariate associations. These were age, duration >1 hour, AF, past history of ischaemic heart disease, peripheral vascular disease, stroke and TIA, and the recently validated clinical prediction scores, California, ABCD and ABCD2. Further multivariate analysis found age, AF and duration >1 hour to be independently associated with mortality, with other variables no longer independently significant. It is unsurprising that age and AF were independently associated with mortality. Older persons are probably more likely to die due to higher rates of pre-existing disease and increased debility. Regarding AF, many patients (44%) died of ischaemic heart disease or stroke, with AF possibly suggesting either pre-existing cardiovascular disease or high subsequent risk of stroke. It is more difficult to explain why a duration > 1 hour correlates
M.N.T. Thuy, P.J. Hand / Journal of Clinical Neuroscience 17 (2010) 305–307
independently with mortality. It may be that longer TIAs are more likely to represent true vascular events as opposed to TIA mimics. One small study found that TIA duration >1 hour was significantly correlated with cerebral infarction on MRI and thromboembolic sources (such as from the carotid arteries or heart).11 A past history of vascular events was not independently significant and this may be because such patients tended to be older. In a larger sample, these variables may be independently statistically significant. The clinical prediction scores for subsequent stroke (California, ABCD, ABCD2) were also highly correlated with mortality but were not independently significant. This is probably because they correlate with both age and clinical duration. The correlation of these scores with mortality is a useful finding of relevance to clinical practice. The ability of these scores to predict future stroke was not tested in this study. Inpatient stroke occurrence and death from stroke were not analysed due to the small number of patients. Collection of information about recurrent vascular events in the community and independent self care (using the Modified Rankin Scale) was attempted; however, after careful consideration, these data were not published. A future paper may investigate possible clinical prediction of carotid stenosis in TIA patients. 5. Conclusion In this sample, hyperglycaemia at TIA diagnosis is not significantly associated with mortality in patients admitted into a stroke unit. Given the low death rate, a much larger sample may be required to reliably exclude any significant association. Acknowledgements We are very grateful to the following: Professor S. Davis for support and critical feedback; Royal Melbourne Hospital’s Health Information Services for retrieval of records; Victorian Department
307
of Births, Deaths and Marriages for patient mortality data; Marnie Collins for statistical analysis and review and Professor E. Storey for critical comments.
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.jocn.2009.07.089. References 1. Baird T, Parsons MW, Barber PA, et al. The influence of diabetes mellitus and hyperglycaemia on stroke incidence and outcome. J Clin Neuroscience 2002;9:618–26. 2. Van den Berghe G, Wouters P, Weekers F, et al. Intensive insulin therapy in critically ill patients. N Engl J Med 2001;345:1359–68. 3. The NICE-SUGAR Study Investigators Finfer S, Chittock DR, et al. Intensive versus conventional glucose control in critically ill patients. N Engl J Med 2009;360:1283–97. 4. Malmberg K, Ryden L, Efendic S, et al. Randomized trial of insulin-glucose infusion followed by subcutaneous insulin treatment in diabetic patients with acute myocardial infarction (DIGAMI study): effects on mortality at 1 year. J Am Coll Cardiol 1995;26:57–65. 5. Gray CS, Hildreth AJ, Sandercock PA, et al. Glucose-potassium-insulin infusions in the management of post-stroke hyperglycaemia: the UK Glucose Insulin in Stroke Trial (GIST-UK). Lancet Neurol 2007;6:397–406. 6. World Health Organization. Cerebrovascular disorders: a clinical and research classification. Offset publication No 43 Geneva: World Health Organization 1978. 7. Capes S, Hunt D, Malmberg K, et al. Stress hyperglycaemia and prognosis of stroke in diabetic and non-diabetic patients: a systematic overview. Stroke 2001;32:2426–32. 8. Johnston SC, Gress DR, Browner WS, et al. Short-term prognosis after emergency deparment diagnosis of TIA. JAMA 2000;284:2901–6. 9. Rothwell PM, Giles MF, Flossmann E, et al. A simple score (ABCD) to identify individuals at high early risk of stroke after transient ischaemic attack. Lancet 2005;366:29–36. 10. Johnston S, Rothwell P, Nguyen-Huynh M, et al. Validation and refinement of scores to predict very early stroke risk after transient ischaemic attack. Lancet 2007;369:283–92. 11. Kimura K, Minematsu K, Yasaka M, et al. The duration of symptoms in transient ischemic attack. Neurology 1999;52:976–80.