Journal of Clinical Neuroscience 22 (2015) 498–503
Contents lists available at ScienceDirect
Journal of Clinical Neuroscience journal homepage: www.elsevier.com/locate/jocn
Clinical Study
Association between hemoglobin A1C levels and clinical outcome in ischemic stroke patients with or without diabetes Chunyan Lei a, Bo Wu a,b,⇑, Ming Liu a,b, Yanchao Chen a a b
Stroke Clinical Research Unit, Department of Neurology, West China Hospital, Sichuan University, No. 37 Guo Xue Xiang, Chengdu 610041, Sichuan Province, China State Key Laboratory of Human Disease Biotherapy and Ministry of Education, West China Hospital, Sichuan University, Sichuan Province, China
a r t i c l e
i n f o
Article history: Received 5 June 2014 Accepted 25 August 2014
Keywords: Acute ischemic stroke Diabetes Glycosylated hemoglobin Mortality Poor outcome
a b s t r a c t Individuals with diabetes are at a higher risk of suffering stroke, but whether chronic hyperglycemia affects clinical outcomes after stroke is unclear. We examined whether chronic hyperglycemia, measured in terms of hemoglobin A1C (HbA1C) levels, influences clinical outcomes after acute ischemic stroke. We prospectively and consecutively included patients admitted within 7 days of ischemic stroke onset. Demographic and clinical information and outcomes were analyzed separately for patients with or without diabetes in order to identify associations with HbA1C tercile. A total of 1351 patients without diabetes and 526 with diabetes were included. The risk of mortality and poor outcome showed a tendency to increase with increasing HbA1C tercile in both groups. Rates of mortality and poor outcome were significantly higher in the third tercile than in the first tercile. In contrast, rates of mortality and poor outcome in the second tercile were not significantly different from those in the first tercile. The adjusted odds ratios of poor outcome and mortality increased with increasing tercile of HbA1C both in patients with diabetes and in patients without diabetes, and these relationships were independent of other confounders. Kaplan–Meier estimates of cumulative mortality also increased with increasing HbA1C in both groups of patients (with diabetes, p = 0.034; without diabetes, p = 0.025). Our results suggest that elevated HbA1C is associated with risk of poor outcome and mortality in ischemic stroke patients with or without diabetes. Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction Diabetes mellitus is an established risk factor for coronary heart disease and ischemic stroke, most likely because of the poor glycemic control which is characteristic of the disease [1,2]. In fact, chronic hyperglycemia is associated with elevated risk of cardiovascular events and all-cause mortality [3–6], and a systematic review concluded that intensive glycemic control can decrease the risk of some cardiovascular events [7]. Elevation of glycated hemoglobin A1C (HbA1C) is a well-known marker of chronic hyperglycemia, and a study in Japan has shown that HbA1C levels on admission are a significant independent predictor of short-term neurological and functional outcomes [8]. Therefore we investigated whether HbA1C levels might help predict outcomes in stroke patients with or without diabetes. These risk estimates may be misleading for one or both groups, however, since the distribution of HbA1C levels is roughly normal in individuals without diabetes, but is much wider and skewed toward higher values in individuals with diabetes [4]. This highlights the ⇑ Corresponding author. Tel.: +86 189 8060 2142. E-mail address: [email protected] (B. Wu). http://dx.doi.org/10.1016/j.jocn.2014.08.030 0967-5868/Ó 2014 Elsevier Ltd. All rights reserved.
need to assess the prognostic value of HbA1C separately for individuals with or without diabetes; this is particularly true for individuals with chronic hyperglycemia who have not been diagnosed with diabetes. In addition, the literature has not adequately addressed the question of whether elevated HbA1C also affects long-term clinical outcomes after stroke. To examine these questions, the present study aimed to take advantage of a large-scale, long-term prospective study of acute ischemic stroke patients in China in order to explore whether chronic hyperglycemia, reflected in serum levels of HbA1C, is associated with functional outcomes or mortality. HbA1C levels are a more robust glycemic index than glucose testing because they reflect the average blood glucose level over the previous few months and they are less affected by meals eaten immediately before testing [5,9–14]. 2. Subjects and methods 2.1. Patients and evaluation This study was conducted using prospective data from the Chengdu Stroke Registry, which has been described previously
C. Lei et al. / Journal of Clinical Neuroscience 22 (2015) 498–503
[15]. This registry project was approved by the Scientific Research Department of West China Hospital, Sichuan University; the study protocol conformed to local and international standards of research ethics. In the present study, we included patients admitted within 7 days of ischemic stroke onset from January 2009 to 31 July 2012. Informed consent was obtained from subjects or their guardians. Patients who refused to participate in the Registry or who refused to participate in follow-up were excluded. Patients were also excluded if data on their HbA1C levels within 24 hours after hospital admission were unavailable.
499
without diabetes (72.0%). At baseline, patients with diabetes were older and had higher systolic blood pressure, fasting glucose levels, and hypertension rates than did patients without diabetes (Table 1). During follow-up, 78 deaths from all causes occurred among patients with diabetes, including 51 deaths from cardiovascular causes. Among patients without diabetes, 193 deaths from all causes occurred, including 126 deaths from cardiovascular causes. When we analyzed all patients in our cohort together, we did not observe a significant association between HbA1C and risk of any of the post-stroke outcomes that we analyzed (data not shown). Therefore we performed analyses separately on patients with or without diabetes.
2.2. Data collection and follow-up 3.1. Mortality and poor outcome among patients with diabetes Baseline information was collected at admission on age, sex, initial stroke severity assessed using the National Institutes of Health Stroke Scale, and stroke risk factors identified in previous reports (hypertension, diabetes mellitus, and current smoking and alcohol intake) [15]. Outcome measures were mortality and poor outcome, as measured using the modified Rankin Scale (mRS). Hypertension was defined as current use of antihypertensive medications, or systolic blood pressure P140 mmHg and/or diastolic blood pressure P90 mmHg. Diabetes mellitus was defined as the use of antidiabetic medication, a fasting serum glucose level P7.0 mmol/L, a non-fasting glucose concentration P11.1 mmol/L, or a self-reported physician diagnosis. Patients who had smoked P1 cigarette/day for at least 1 year were classified as currently smoking. Patients who had consumed P50 ml of alcohol/day for more than 1 year were classified as having an excessive alcohol intake. Poor outcome was defined as a mRS score of 3–6. Experienced stroke neurologists blinded to relevant patient data assessed mRS scores. Outcome data were collected by telephone every 3 months after discharge until July 2013. Participant deaths were confirmed by direct telephone contact with family members. HbA1C was measured by high-pressure liquid chromatography separation of hemoglobin fractions certified by an equivalent program in China, with the normal reference range taken to be 4.0–6.0%. 2.3. Statistical analysis All statistical analyses were performed using the Statistical Package for the Social Sciences version 16 (SPSS, Chicago, IL, USA). Results were reported as percentages, mean ± standard deviation, or odds ratios with 95% confidence intervals, as appropriate. The v2 or Fisher exact test were used to compare categorical data, while Student’s t-test and the Mann–Whitney U test were used to compare continuous data. Two-sided values of p < 0.05 were considered statistically significant. Variables identified as significant in the univariate analysis (p < 0.05) were used in forward stepwise multivariate regression to examine their ability to independently predict mortality and poor outcome. Cumulative survival and recurrence rates were estimated by the Kaplan–Meier product limit method, and log rank tests were used to assess the significance of inter-group differences. 3. Results Between January 2009 and 31 July 2012, 2351 patients were admitted our center and enrolled in the Chengdu Stroke Registry. We considered all these patients for enrollment in the present study; we excluded 294 for lack of HbA1C testing, leaving a total of 2057 patients. Of these, 180 (8.8%) were lost to follow-up; their baseline characteristics did not differ significantly from those of the 1877 patients in the final analysis (Fig. 1, Table 1). The final group comprised 526 patients with diabetes (28.0%) and 1351
Among patients with diabetes, the first tercile of HbA1C proportion was 633.5% (4.7–6.7 mmol/L); the second tercile, 33.5–66.8% (6.8–8.2 mmol/L); and the third tercile, >66.8% (8.3–17.6 mmol/L). There was a significant trend towards greater mortality rates with increasing HbA1C tercile (p = 0.047). Post hoc analysis showed mortality rate to be significantly higher in the third tercile than in the first, both at 3 months (p = 0.045) and at 1 year (p = 0.041). However, the mortality rate did not differ significantly between the second and first terciles, either at 3 months (p = 0.511) or 1 year (p = 0.537). The frequencies of poor outcome, based on mRS score, differed with HbA1C tercile at both 3 months and 1 year (Table 2). Post hoc analysis showed that poor outcome was significantly more frequent in the third tercile than in the first, at both 3 months (p = 0.012) and 1 year (p = 0.034). 3.2. Mortality and poor outcome among patients without diabetes Among patients without diabetes, the first tercile of HbA1C proportion was 633.4% (3.9–5.5 mmol/L); the second tercile, 33.4– 66.7% (5.6–5.8 mmol/L); and the third tercile, >66.7% (5.9– 6.3 mmol/L). Increasing HbA1C tercile was associated with significant increases in both 3 month and 1 year mortality rates (Table 2). Mortality rates at both time points were significantly higher in the third tercile than in the first (p = 0.009). In contrast, mortality rates in the second tercile were not significantly different from those in the first (p = 0.073). As with mortality, increasing HbA1C tercile was associated with significant increases in poor outcome at both 3 months and 1 year (Table 2). The rate of poor outcome was significantly higher in the third tercile than in the first (3 months, p = 0.017; 1 year, p = 0.046). The first and second terciles showed similar rates of poor outcome (p = 0.514). We conducted multivariable analysis in which we adjusted for other potential confounding variables. Taking the lowest HbA1C tercile as a reference, we found that the odds ratio for poor outcome were higher in the second tercile and even higher in the third tercile, both among patients with diabetes and among those without diabetes (Table 3). Both groups of patients showed similar Kaplan–Meier estimates of cumulative mortality by the end of followup (p = 0.451). However, estimated mortality was significantly higher in the second and third terciles than in the first, both for patients with diabetes (p = 0.034) and for patients without (p = 0.025; Fig. 2). Moreover, HbA1C levels associated with poor outcome were significantly higher than those associated with good outcome in both groups (Table 4). 4. Discussion This large-scale, prospective study showed that poor outcome and mortality increased with increasing HbA1C tercile when patients with or without diabetes were considered separately. This
500
C. Lei et al. / Journal of Clinical Neuroscience 22 (2015) 498–503
Fig. 1. Flow diagram showing the selection process for study participants. HbA1C = hemoglobin A1C.
Table 1 Demographic and clinical characteristics of ischemic stroke patients with or without diabetes Characteristic Age (years) Male, n (%) Hypertension, n (%) Current smoking, n (%) Excessive alcohol intake, n (%) NIHSS score Systolic BP (mmHg) Diastolic BP (mmHg) Fasting glucose (mmol/L) Total cholesterol (mmol/L) Triglycerides (mmol/L) HDL (mmol/L) LDL (mmol/L) TOAST subtype, n (%) Large artery atherosclerosis Small artery occlusion Cardioembolism Other determined Undetermined Stroke-related complications, n (%) Respiratory tract infection Urinary tract infection Intestinal infection Epilepsy Electrolyte disturbance Gastrointestinal bleeding
With diabetes n = 526
Without diabetes n = 1351
p value
65.22 ± 3.98 309 (58.75%) 342 (65.02%) 87 (16.54%) 155 (29.47%) 6.11 ± 6.00 147.39 ± 22.52 84.28 ± 13.81 9.36 ± 4.00 5.51 ± 20.27 2.77 ± 13.58 1.21 ± 0.47 2.77 ± 1.01
62.10 ± 2.77 855 (63.28%) 656 (48.56%) 305 (22.58%) 462 (34.20%) 6.58 ± 6.35 142.82 ± 50.47 83.55 ± 15.00 5.92 ± 2.77 4.68 ± 8.19 1.98 ± 16.28 1.31 ± 0.50 2.84 ± 7.08
Contents lists available at ScienceDirect
Journal of Clinical Neuroscience journal homepage: www.elsevier.com/locate/jocn
Clinical Study
Association between hemoglobin A1C levels and clinical outcome in ischemic stroke patients with or without diabetes Chunyan Lei a, Bo Wu a,b,⇑, Ming Liu a,b, Yanchao Chen a a b
Stroke Clinical Research Unit, Department of Neurology, West China Hospital, Sichuan University, No. 37 Guo Xue Xiang, Chengdu 610041, Sichuan Province, China State Key Laboratory of Human Disease Biotherapy and Ministry of Education, West China Hospital, Sichuan University, Sichuan Province, China
a r t i c l e
i n f o
Article history: Received 5 June 2014 Accepted 25 August 2014
Keywords: Acute ischemic stroke Diabetes Glycosylated hemoglobin Mortality Poor outcome
a b s t r a c t Individuals with diabetes are at a higher risk of suffering stroke, but whether chronic hyperglycemia affects clinical outcomes after stroke is unclear. We examined whether chronic hyperglycemia, measured in terms of hemoglobin A1C (HbA1C) levels, influences clinical outcomes after acute ischemic stroke. We prospectively and consecutively included patients admitted within 7 days of ischemic stroke onset. Demographic and clinical information and outcomes were analyzed separately for patients with or without diabetes in order to identify associations with HbA1C tercile. A total of 1351 patients without diabetes and 526 with diabetes were included. The risk of mortality and poor outcome showed a tendency to increase with increasing HbA1C tercile in both groups. Rates of mortality and poor outcome were significantly higher in the third tercile than in the first tercile. In contrast, rates of mortality and poor outcome in the second tercile were not significantly different from those in the first tercile. The adjusted odds ratios of poor outcome and mortality increased with increasing tercile of HbA1C both in patients with diabetes and in patients without diabetes, and these relationships were independent of other confounders. Kaplan–Meier estimates of cumulative mortality also increased with increasing HbA1C in both groups of patients (with diabetes, p = 0.034; without diabetes, p = 0.025). Our results suggest that elevated HbA1C is associated with risk of poor outcome and mortality in ischemic stroke patients with or without diabetes. Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction Diabetes mellitus is an established risk factor for coronary heart disease and ischemic stroke, most likely because of the poor glycemic control which is characteristic of the disease [1,2]. In fact, chronic hyperglycemia is associated with elevated risk of cardiovascular events and all-cause mortality [3–6], and a systematic review concluded that intensive glycemic control can decrease the risk of some cardiovascular events [7]. Elevation of glycated hemoglobin A1C (HbA1C) is a well-known marker of chronic hyperglycemia, and a study in Japan has shown that HbA1C levels on admission are a significant independent predictor of short-term neurological and functional outcomes [8]. Therefore we investigated whether HbA1C levels might help predict outcomes in stroke patients with or without diabetes. These risk estimates may be misleading for one or both groups, however, since the distribution of HbA1C levels is roughly normal in individuals without diabetes, but is much wider and skewed toward higher values in individuals with diabetes [4]. This highlights the ⇑ Corresponding author. Tel.: +86 189 8060 2142. E-mail address: [email protected] (B. Wu). http://dx.doi.org/10.1016/j.jocn.2014.08.030 0967-5868/Ó 2014 Elsevier Ltd. All rights reserved.
need to assess the prognostic value of HbA1C separately for individuals with or without diabetes; this is particularly true for individuals with chronic hyperglycemia who have not been diagnosed with diabetes. In addition, the literature has not adequately addressed the question of whether elevated HbA1C also affects long-term clinical outcomes after stroke. To examine these questions, the present study aimed to take advantage of a large-scale, long-term prospective study of acute ischemic stroke patients in China in order to explore whether chronic hyperglycemia, reflected in serum levels of HbA1C, is associated with functional outcomes or mortality. HbA1C levels are a more robust glycemic index than glucose testing because they reflect the average blood glucose level over the previous few months and they are less affected by meals eaten immediately before testing [5,9–14]. 2. Subjects and methods 2.1. Patients and evaluation This study was conducted using prospective data from the Chengdu Stroke Registry, which has been described previously
C. Lei et al. / Journal of Clinical Neuroscience 22 (2015) 498–503
[15]. This registry project was approved by the Scientific Research Department of West China Hospital, Sichuan University; the study protocol conformed to local and international standards of research ethics. In the present study, we included patients admitted within 7 days of ischemic stroke onset from January 2009 to 31 July 2012. Informed consent was obtained from subjects or their guardians. Patients who refused to participate in the Registry or who refused to participate in follow-up were excluded. Patients were also excluded if data on their HbA1C levels within 24 hours after hospital admission were unavailable.
499
without diabetes (72.0%). At baseline, patients with diabetes were older and had higher systolic blood pressure, fasting glucose levels, and hypertension rates than did patients without diabetes (Table 1). During follow-up, 78 deaths from all causes occurred among patients with diabetes, including 51 deaths from cardiovascular causes. Among patients without diabetes, 193 deaths from all causes occurred, including 126 deaths from cardiovascular causes. When we analyzed all patients in our cohort together, we did not observe a significant association between HbA1C and risk of any of the post-stroke outcomes that we analyzed (data not shown). Therefore we performed analyses separately on patients with or without diabetes.
2.2. Data collection and follow-up 3.1. Mortality and poor outcome among patients with diabetes Baseline information was collected at admission on age, sex, initial stroke severity assessed using the National Institutes of Health Stroke Scale, and stroke risk factors identified in previous reports (hypertension, diabetes mellitus, and current smoking and alcohol intake) [15]. Outcome measures were mortality and poor outcome, as measured using the modified Rankin Scale (mRS). Hypertension was defined as current use of antihypertensive medications, or systolic blood pressure P140 mmHg and/or diastolic blood pressure P90 mmHg. Diabetes mellitus was defined as the use of antidiabetic medication, a fasting serum glucose level P7.0 mmol/L, a non-fasting glucose concentration P11.1 mmol/L, or a self-reported physician diagnosis. Patients who had smoked P1 cigarette/day for at least 1 year were classified as currently smoking. Patients who had consumed P50 ml of alcohol/day for more than 1 year were classified as having an excessive alcohol intake. Poor outcome was defined as a mRS score of 3–6. Experienced stroke neurologists blinded to relevant patient data assessed mRS scores. Outcome data were collected by telephone every 3 months after discharge until July 2013. Participant deaths were confirmed by direct telephone contact with family members. HbA1C was measured by high-pressure liquid chromatography separation of hemoglobin fractions certified by an equivalent program in China, with the normal reference range taken to be 4.0–6.0%. 2.3. Statistical analysis All statistical analyses were performed using the Statistical Package for the Social Sciences version 16 (SPSS, Chicago, IL, USA). Results were reported as percentages, mean ± standard deviation, or odds ratios with 95% confidence intervals, as appropriate. The v2 or Fisher exact test were used to compare categorical data, while Student’s t-test and the Mann–Whitney U test were used to compare continuous data. Two-sided values of p < 0.05 were considered statistically significant. Variables identified as significant in the univariate analysis (p < 0.05) were used in forward stepwise multivariate regression to examine their ability to independently predict mortality and poor outcome. Cumulative survival and recurrence rates were estimated by the Kaplan–Meier product limit method, and log rank tests were used to assess the significance of inter-group differences. 3. Results Between January 2009 and 31 July 2012, 2351 patients were admitted our center and enrolled in the Chengdu Stroke Registry. We considered all these patients for enrollment in the present study; we excluded 294 for lack of HbA1C testing, leaving a total of 2057 patients. Of these, 180 (8.8%) were lost to follow-up; their baseline characteristics did not differ significantly from those of the 1877 patients in the final analysis (Fig. 1, Table 1). The final group comprised 526 patients with diabetes (28.0%) and 1351
Among patients with diabetes, the first tercile of HbA1C proportion was 633.5% (4.7–6.7 mmol/L); the second tercile, 33.5–66.8% (6.8–8.2 mmol/L); and the third tercile, >66.8% (8.3–17.6 mmol/L). There was a significant trend towards greater mortality rates with increasing HbA1C tercile (p = 0.047). Post hoc analysis showed mortality rate to be significantly higher in the third tercile than in the first, both at 3 months (p = 0.045) and at 1 year (p = 0.041). However, the mortality rate did not differ significantly between the second and first terciles, either at 3 months (p = 0.511) or 1 year (p = 0.537). The frequencies of poor outcome, based on mRS score, differed with HbA1C tercile at both 3 months and 1 year (Table 2). Post hoc analysis showed that poor outcome was significantly more frequent in the third tercile than in the first, at both 3 months (p = 0.012) and 1 year (p = 0.034). 3.2. Mortality and poor outcome among patients without diabetes Among patients without diabetes, the first tercile of HbA1C proportion was 633.4% (3.9–5.5 mmol/L); the second tercile, 33.4– 66.7% (5.6–5.8 mmol/L); and the third tercile, >66.7% (5.9– 6.3 mmol/L). Increasing HbA1C tercile was associated with significant increases in both 3 month and 1 year mortality rates (Table 2). Mortality rates at both time points were significantly higher in the third tercile than in the first (p = 0.009). In contrast, mortality rates in the second tercile were not significantly different from those in the first (p = 0.073). As with mortality, increasing HbA1C tercile was associated with significant increases in poor outcome at both 3 months and 1 year (Table 2). The rate of poor outcome was significantly higher in the third tercile than in the first (3 months, p = 0.017; 1 year, p = 0.046). The first and second terciles showed similar rates of poor outcome (p = 0.514). We conducted multivariable analysis in which we adjusted for other potential confounding variables. Taking the lowest HbA1C tercile as a reference, we found that the odds ratio for poor outcome were higher in the second tercile and even higher in the third tercile, both among patients with diabetes and among those without diabetes (Table 3). Both groups of patients showed similar Kaplan–Meier estimates of cumulative mortality by the end of followup (p = 0.451). However, estimated mortality was significantly higher in the second and third terciles than in the first, both for patients with diabetes (p = 0.034) and for patients without (p = 0.025; Fig. 2). Moreover, HbA1C levels associated with poor outcome were significantly higher than those associated with good outcome in both groups (Table 4). 4. Discussion This large-scale, prospective study showed that poor outcome and mortality increased with increasing HbA1C tercile when patients with or without diabetes were considered separately. This
500
C. Lei et al. / Journal of Clinical Neuroscience 22 (2015) 498–503
Fig. 1. Flow diagram showing the selection process for study participants. HbA1C = hemoglobin A1C.
Table 1 Demographic and clinical characteristics of ischemic stroke patients with or without diabetes Characteristic Age (years) Male, n (%) Hypertension, n (%) Current smoking, n (%) Excessive alcohol intake, n (%) NIHSS score Systolic BP (mmHg) Diastolic BP (mmHg) Fasting glucose (mmol/L) Total cholesterol (mmol/L) Triglycerides (mmol/L) HDL (mmol/L) LDL (mmol/L) TOAST subtype, n (%) Large artery atherosclerosis Small artery occlusion Cardioembolism Other determined Undetermined Stroke-related complications, n (%) Respiratory tract infection Urinary tract infection Intestinal infection Epilepsy Electrolyte disturbance Gastrointestinal bleeding
With diabetes n = 526
Without diabetes n = 1351
p value
65.22 ± 3.98 309 (58.75%) 342 (65.02%) 87 (16.54%) 155 (29.47%) 6.11 ± 6.00 147.39 ± 22.52 84.28 ± 13.81 9.36 ± 4.00 5.51 ± 20.27 2.77 ± 13.58 1.21 ± 0.47 2.77 ± 1.01
62.10 ± 2.77 855 (63.28%) 656 (48.56%) 305 (22.58%) 462 (34.20%) 6.58 ± 6.35 142.82 ± 50.47 83.55 ± 15.00 5.92 ± 2.77 4.68 ± 8.19 1.98 ± 16.28 1.31 ± 0.50 2.84 ± 7.08
