Nonfasting Serum Glucose and the Risk of Fatal Stroke in Diabetic and Nondiabetic Subjects
18-Year Follow-up of the Oslo Study
Background and Purpose The association between nonfasting serum glucose and stroke mortality for diabetic and nondiabetic subjects is presented for participants of the Oslo Study.
Methods The study started in 1972; of 16 209 men aged 40 to 49 years, 16 172 had no previous history of stroke and 151 were known to be diabetic. Five diabetic and 80 nondiabetic subjects died of stroke during the 18 years of follow-up, giving a rate ratio of 7.87 (95% confidence interval [CI], 2.48 to 19.14). The rate of mortality for all causes in diabetic subjects was more than five times that of those who were nondiabetic.
Results Nonfasting serum glucose was a predictor of fatal stroke in all participants (diabetic subjects included) without a history of stroke in age-adjusted univariate analysis. The relative risk was 1.13 (CI, 1.03 to 1.25) by increase of 1 mmol/L of serum glucose according to results of proportional hazards regression analysis. Accordingly, relative risk for nondiabetic subjects was 1.02 (CI, 0.83 to 1.26) with no linear trend. The rate ratio of the fifth quintile to the rest was 1.57 (CI, 0.94 to 2.56) for all participants and 1.28 (CI, 0.72 to 2.18) for nondiabetics.
Conclusions There was an interaction between glucose level and body mass index versus stroke for all participants but not for nondiabetic subjects, with an increased risk for men with above-median values of glucose and body mass index. Analysis of nondiabetic subjects failed to show glucose as a definite predictor of fatal stroke.
Diabetes promotes cerebral atherosclerosis and may increase hypertension and hyperlipidemia and coronary heart disease, which are risk factors for stroke.3 4 5 6 7 8 9 10 Because of metabolic changes, diabetic persons have an increased risk of thrombosis and increased blood viscosity.5 11
Hyperglycemia produces increased anaerobic metabolism, raised lactic acid production in ischemic brain tissue, and cellular acidosis. Hyperglycemia also can result as a stress response after stroke. Presence of diabetes or hyperglycemia may affect the degree of stroke outcome.5 The International Collaborative Group study of 15 populations12 did not find asymptomatic hyperglycemia to be a risk factor for coronary heart disease. Later, the Caerphilly and Speedwell study13 indicated that it should not be neglected as a risk factor for coronary heart disease and the Whitehall study1 for both coronary heart disease and stroke.
Previous analysis of stroke incidence and mortality in the Oslo Study6 after 12 years of follow-up of healthy men with no known symptoms or diseases of cardiovascular origin or diabetes showed that glucose level was not a significant risk factor in incidence or mortality of stroke or myocardial infarction, except in total mortality. The present analysis after 18 years of follow-up includes men with symptoms or diseases of cardiovascular origin or diabetes and is focused on the diabetic-nondiabetic dimension.
The purpose of this study was to analyze and present the independence and interactions of nonfasting serum glucose values with other lifestyle factors as they relate to stroke mortality for diabetic and nondiabetic subjects.
Subjects and Methods
In the Oslo Study, all men in Oslo aged 40 to 49 years and a 7% random sample of men aged 20 to 39 years were invited to a screening examination that started in May 1972. At present, an 18-year follow-up on mortality has been completed for the men aged 40 to 49 years. Data on causes of death have been supplied by the Central Bureau of Statistics. The last day of observation was December 31, 1990.
The International Classification of Diseases (ICD, 8th and 9th revisions) codes of 430 through 438 were used to identify stroke mortality. The specific diagnosis of stroke through this 18-year follow-up period was not confirmed on a regular basis by diagnostic methods such as computed tomography (CT). Only since 1985 through 1988, after being introduced around 1976, has this examination been performed on a more regular basis because of improved CT capacity.
All men who had experienced a previous stroke (n=36) were excluded from the risk-factor analysis. This risk-factor study cohort included 16 172 men. Information on diabetes was self reported. Therefore, at baseline men were recorded with hyperglycemia without having been previously diagnosed as diabetic. These men may have undergone treatment during the follow-up period, thus reducing their risk of stroke. At screening, 153 men were reported to have diabetes of types I or II. Information on their treatment regimen was not recorded. Two men with diabetes had experienced a stroke, leaving 151 diabetic subjects to be analyzed. Stroke mortality for 9706 nonparticipants was also recorded.
Results of screening and the methods of the Oslo Study have been reported elsewhere.14 In short, blood samples from subjects in the nonfasting state were taken to record levels of total serum cholesterol, triglycerides, and glucose. Height, weight, and systolic and diastolic blood pressures were measured. From a questionnaire, information on history of cardiovascular diseases and diabetes, symptoms of angina and/or atherosclerosis obliterans of the lower limbs, smoking habits, and physical activity at work and at leisure was collected. The smoking criterion used in this analysis was daily cigarette smoking (yes or no). Physical activity at work and at leisure was recorded at four levels of activity. Body mass index (BMI) was calculated as the ratio of body weight in kilograms to height in meters squared. All results are based on baseline measurements. The men have not been reexamined.
The nonfasting serum glucose levels have been adjusted for time since the last meal by multiplying each person’s glucose level (xt1) by the ratio of the total group mean to the mean of the time period concerned for each individual as follows: t1=xt1(Xt1−tn/Xt1), where Xt1−tn is the mean over all time periods, and Xt1 is the mean of the time period concerned for each individual.15 The traditional units of glucose have been converted to SI units as follows: 0.056×value in mg/dL=value in mmol/L.
Rates of mortality are presented as number of cases per 1000 person-years. The rate ratio and its 95% confidence interval (CI) are used for group comparisons.16
Age-adjusted proportional hazards regression analyses (Cox) were carried out for analysis of prediction, independence, and interactions of risk factors in stroke mortality. The majority of the analyses were performed using bmdp statistical software.17
Five diabetic and 80 nondiabetic subjects died of stroke during the follow-up period. The rate of fatal stroke (Table 1⇓) of nondiabetics was 0.29. Diabetic subjects had a rate of 2.3, which is 7.9 times that of nondiabetics. Nonparticipants had 2.57 times the rate of stroke compared with nondiabetic subjects.
The analysis by type of stroke was limited because of the above-mentioned lack of CT data. The number of strokes of uncertain cause was relatively high (31% of all). The diabetic subjects of this study were assigned the diagnosis of unspecified stroke. Analysis of the present data shows glucose level to predict neither hemorrhagic nor thromboembolic stroke but unspecified stroke.
Diabetic subjects had an increased risk of myocardial infarction of 4.17 compared with nondiabetics (Table 2⇓). The rate ratio for all cardiovascular diseases was found to be 4.6. No significant rate ratio between the groups was found for cancer. Diabetes was registered as cause of death for 12 diabetic and 2 nondiabetic subjects. With regard to the remaining causes of death, the diabetic subjects had a significantly increased risk. Diabetic subjects had over five times the mortality of all causes of death compared with nondiabetics.
The mean value of glucose was much higher in diabetic than nondiabetic subjects (Table 3⇓). The standard deviation was noticeably larger also. Cases of stroke in men with a history of stroke showed a higher mean level of glucose than noncases. This was not noticeable among men without previous stroke.
The rate of stroke by quintile of serum glucose values in all participants did not show a linear increase in risk for stroke with increasing levels of serum glucose (test of trend, NS) (Table 4⇓). The rate ratio of the fifth to first quintile was 1.44 (95% CI, 0.73 to 2.96), and the rate ratio of the fifth quintile to the rest was 1.57 (95% CI, 0.94 to 2.56). The risk-by-quintile values of glucose differed between diabetic and nondiabetic subjects (Fig 1⇓) in that diabetics had events only in the 10th decile of glucose distribution (>6.72 mmol/L). Age-adjusted proportional hazards regression analysis (Table 5⇓) showed that glucose level was a significant risk factor for all participants but not when diabetic subjects were excluded from the analysis. Further analysis (results not shown) showed that diastolic blood pressure had some confounding effect but treatment of hypertension did not.
Glucose level was found to interact with BMI in analyses of all participants but not significantly for nondiabetic subjects alone. No other interactions were found among the risk factors measured (see above). The risk for fatal stroke, presented as the negative log value of the survivor function of low and high values of BMI (dichotomized by the median value of 24.5 kg/m2) (Fig 2⇓), showed that high values of glucose (above a median of 5.65 mmol/L) were related to increased risk of stroke only in high-BMI men.
Diabetes was found to be an independent risk factor for stroke after adjusting for age, glucose level, diastolic blood pressure, daily cigarette smoking, and physical exercise at leisure in a multivariate analysis (Table 6⇓).
The results of 18 years of follow-up in the Oslo Study on mortality confirm previous findings that diabetic persons are at increased risk for fatal stroke. The diabetic subjects in our study had a high mean level of glucose with a large variation in their glucose levels at the screening, indicating varying degrees of diabetes control. They had a total mortality rate five times higher than that of nondiabetic subjects. Two thirds of the deaths were due to cardiovascular diseases and diabetes. The rate ratio was higher for stroke than for myocardial infarction in our study. The Whitehall Study1 found an increased risk among diabetic subjects, but the numbers were low (three stroke cases). Mortel et al3 found diabetes to be second to hypertension as a risk factor for stroke mortality. We found diabetes to be an independent risk factor. Yano et al4 of the Honolulu Heart Program found an increased mortality rate of stroke in diabetic subjects of 10.4 per 1000 compared with 8.2 per 1000 normoglycemic subjects over 9 years of follow-up. Results of 16 years of follow-up in the Framingham study7 showed that diabetic subjects developed more cerebrovascular events than nondiabetics. In their cohort, seven diabetic subjects died from this cause compared with the expected 1.9 (men and women). The 10-year follow-up of the National Health and Nutrition Examination Survey I confirms diabetes to be a risk factor for hemorrhagic stroke in men and women and for thromboembolic stroke in women.9
The risk-factor analysis showed that nondiabetic subjects of the fifth quintile (>6.20 mmol/L) seemed to be at slightly increased risk for stroke compared with the rest, although this was not of statistical significance. Additional analysis of asymptomatic hyperglycemia (glucose levels above the 95th percentile [6.6 mmol/L]) gave a nonsignificant rate ratio of 1.4 (95% CI, 0.7 to 2.7). Cox proportional analysis confirmed this finding. Only when diabetic subjects were included in the age-adjusted analysis was glucose a significant risk factor. Nonfasting glucose level was not found to be an independant risk factor in multivariate analysis when adjusted for age, diabetes, diastolic blood pressure, daily cigarette smoking, and physical exercise at leisure.
In this study, glucose was measured in the nonfasting state, which reflects greater intrasubject variation than if glucose is measured in the fasting state or after a glucose load. The glucose-load method would have better distinguished the individual’s ability to metabolize glucose. This may in part explain the lack of a stronger association between glucose and stroke mortality. In another study that measured glucose in the nonfasting state, Janghorbani et al8 found asymptomatic hyperglycemia to be a significant risk factor for stroke mortality. However, in multivariate analysis this association could not be found among men, only for women.
Analysis of all participants showed an interaction of glucose level with BMI with regard to stroke risk. We found this association only among men with glucose values above the median (5.56 mmol/L). This association can be supported by pathophysiological mechanisms of the lipid metabolism.18 19 Holmen11 states that increased weight caused by abdominal fat in particular is an important factor in the development of insulin resistance. Possible confounding may have occurred by nonrecorded predictors of atherosclerosis, such as high-density and low-density lipoproteins.
In conclusion, this study showed nonfasting serum glucose to be a risk factor for stroke mortality in men with BMI above the median value. The intrasubject variation when glucose level is recorded in the nonfasting state may be less predictive than measurements by the oral glucose-load method. The increased risk found for all participants was carried mainly by the diabetic subjects.
We want to thank Dr Bent Indredavik at the Regional Hospital of Trondheim for valuable help.
- Received December 22, 1994.
- Revision received February 7, 1995.
- Accepted February 21, 1995.
- Copyright © 1995 by American Heart Association
Fuller JH, Shipley MJ, Rose G, Jarrett RJ, Keen H. Mortality from coronary heart disease and stroke in relation to degree of glycaemia: the Whitehall study. BMJ. 1983;287:867-870.
Kannel WB, McGee DL. Diabetes and glucose tolerance as risk factors for cardiovascular disease: the Framingham Study. Diabetes Care. 1979;2:120-126.
Lund Håheim L, Holme I, Hjermann I, Leren P. Risk factors of stroke incidence and mortality: a 12-year follow-up of the Oslo Study. Stroke. 1993;24:1484-1489.
Garcia MJ, McNamara PM, Gordon T, Kannell WB. Morbidity and mortality in diabetics in the Framingham population: sixteen years follow-up study. Diabetes. 1974;23:105-111.
White LR, Cornoni-Huntley JC, Huntley RR, Feldman JJ. Health Status and Well-being of the Elderly. New York, NY: Oxford University Press; 1990.
Steiner G. Diabetes and atherosclerosis: an overview. Diabetes. 1981;30(suppl 2):1-7.
Stensvold I, Tverdal A, Urdal P, Graff-Iversen S. Non-fasting serum triglyceride concentration and mortality from coronary heart disease and any cause in middle aged Norwegian women. BMJ. 1993;307:1318-1322.
Kleinbaum DD, Kupper LL, Morgenstern H. Epidemiologic Research. New York, NY: Van Nostrand Reinhold; 1982.
Dixon WJ, ed. BMDP Statistical Software. Berkeley, Calif: University of California Press; 1990.
Bjorntorp P. Obesity and adipose tissue distribution as risk factors for the development of disease: a review. Infusionsther Transfusionsmed. 1990;17:24-27.