Abstract

Accumulating evidence suggests that inadequate vitamin D levels may predispose people to chronic diseases. The authors aimed to investigate whether serum 25-hydroxyvitamin D (25(OH)D) level predicts mortality from cardiovascular disease (CVD). The study was based on the Mini-Finland Health Survey and included 6,219 men and women aged ≥30 years who were free from CVD at baseline (1978–1980). During follow-up through 2006, 640 coronary disease deaths and 293 cerebrovascular disease deaths were identified. Levels of 25(OH)D were determined from serum collected at baseline. Cox's proportional hazards model was used to assess the association between 25(OH)D and risk of CVD death. After adjustment for potential confounders, the hazard ratio for total CVD death was 0.76 (95% confidence interval (95% CI): 0.60, 0.95) for the highest quintile of 25(OH)D level versus the lowest. The association was evident for cerebrovascular death (hazard ratio = 0.48, 95% CI: 0.31, 0.75) but not coronary death (hazard ratio = 0.91, 95% CI: 0.70, 1.18). A low vitamin D level may be associated with higher risk of a fatal CVD event, particularly cerebrovascular death. These findings need to be replicated in other populations. To demonstrate a causal link between vitamin D and CVD, randomized controlled trials are required.

Interest in vitamin D has intensified lately, with a growing body of evidence suggesting that adequate vitamin D status is required for optimal health (1, 2). The importance of vitamin D for bone health has long been acknowledged. Recent evidence suggests that vitamin D can also play a role in reducing the risk of several other diseases, including cardiovascular disease (CVD).

Vitamin D, whether ingested or synthesized in the skin, is metabolized in the human body into 25-hydroxyvitamin D (25(OH)D) and further into the biologically active form, 1,25-dihydroxyvitamin D (1, 2). Receptors for vitamin D have been found in many different cells, including cardiomyocytes and vascular endothelial cells, giving it the potential to have wide-ranging vascular effects (3–6). Evidence from ecologic, animal, and clinical studies also supports a potential beneficial role for vitamin D in the development of CVD (3, 4, 7–10). Furthermore, vitamin D status has been demonstrated to be associated with several established risk factors for CVD (11) and prevalent CVD (12). Epidemiologic evidence, however, is limited and inconclusive; both inverse associations (13–15) and no associations (16) between vitamin D status and CVD risk have been reported. In addition, vitamin D supplementation had no influence on CVD incidence and mortality in the Women's Health Initiative trial (17, 18). In the present study, we extended previous research on vitamin D and CVD by evaluating whether serum 25(OH)D level predicts mortality from coronary and cerebrovascular diseases in a cohort of more than 6,000 Finnish men and women.

MATERIALS AND METHODS

Study population

The Mini-Finland Health Survey was carried out in 1978–1980 in 40 areas of Finland (19). A 2-stage random sample (n = 8,000) was drawn from the population register to represent Finnish men and women aged 30 years or more. A total of 7,217 subjects (90% of the sample) participated in the survey, which included a health examination. Of these, 890 subjects who had a history of (or findings suggestive of) CVD at baseline and 108 subjects who did not have a serum sample available for 25(OH)D analysis were excluded. This resulted in a cohort of 6,219 subjects.

Baseline measurements

Fasting blood samples were taken during a health examination and kept frozen at −20°C until 2003, when 25(OH)D levels were determined by radioimmunoassay (DiaSorin, Inc., Stillwater, Minnesota). The interassay coefficient of variation for 25(OH)D determination was 7.80% at the mean level of 47.3 nmol/L (n = 167) and 9.12% at the level of 131.3 nmol/L (n = 135). The proportion of quality-control samples was 13.5%.

Serum total and high density lipoprotein cholesterol levels were determined with a direct modification of the Liebermann–Burchard method (20) in 1978–1980. The level of high density lipoprotein cholesterol was analyzed from the supernatant of the serum after precipitation of low density lipoprotein cholesterol and very low density lipoprotein cholesterol with magnesium/dextran sulfate. Serum triglyceride levels were analyzed enzymatically (Boehringer Mannheim GmbH, Mannheim, Germany). Low density lipoprotein cholesterol level was calculated according to the Friedewald formula: total cholesterol − high density lipoprotein cholesterol − 0.45 × total triglyceride level. Plasma glucose level was measured using the glucose oxidase method (Boehringer Mannheim) in 1978–1980, and cotinine level was measured by radioimmunoassay (Diagnostic Products Corporation, Los Angeles, California) in 1999.

The health examination included electrocardiographic recordings and blood pressure measurements. Height and weight were measured, and body mass index (weight (kg)/height (m)2) was calculated. Subjects with chronic disease histories, symptoms, or findings suggestive of cardiovascular, respiratory, or musculoskeletal diseases were asked to participate in a standardized physical examination conducted by specially trained physicians. At the end of the examination, the physicians made diagnostic assessments on the basis of all available documents, self-reported disease histories, symptoms, and clinical signs and findings. Angina pectoris was defined as typical chest pain brought on by exertion and relieved by nitroglycerine or rest. Myocardial infarction was defined as a positive history in the medical records, old myocardial infarction found upon electrocardiography, or a typical self-reported history of myocardial infarction treated in a hospital. Stroke was defined as a positive history in the medical records or a typical self-reported history of stroke treated in a hospital.

Information on socioeconomic background, symptoms, diseases, medications, and lifestyle was collected via questionnaires and interviews. Educational level was categorized into 2 groups (low, 0–9 years; high, ≥10 years) and marital status into 4 groups (unmarried, married (including common-law marriage) or in a committed relationship, widowed, divorced). Leisure-time physical activity was assessed with a question about the duration, intensity, and frequency of physical activity, and subjects were classified as inactive, occasionally active, or regularly active. Categories of alcohol consumption (ethanol intake; 0, 1–14.9, or ≥15 g/week) were derived from responses to questionnaire items concerning average weekly consumption of beer, wine, and liquor during the preceding month. Both self-reported information on smoking habits and serum cotinine level were used to generate the smoking variable. Subjects who reported that they had never smoked or had quit smoking and had a serum cotinine level of 100 ng/mL or less were categorized as nonsmokers. The rest of the subjects were divided into tertiles (cutoff points, 405 ng/mL and 767 ng/mL) based on their serum cotinine level.

Definite hypertension was defined as systolic blood pressure ≥170 mm Hg and diastolic blood pressure ≥100 mm Hg or the use of antihypertensive medication. Of the remaining members of the study population, those with systolic blood pressure ≥160 mm Hg and diastolic blood pressure ≥95 mm Hg were considered to have mild hypertension and those with systolic blood pressure <140 mm Hg and diastolic blood pressure <90 mm Hg were considered to be normotensive. All other subjects were considered to have borderline hypertension. Diabetes mellitus was defined as a self-reported history of diabetes that had been diagnosed and treated by a physician or a fasting plasma glucose level ≥6.7 mmol/L. In Finland, which is situated geographically between 60°N latitude and 70°N latitude, biologically effective ultraviolet B irradiation for production of vitamin D by the skin is provided by sunshine during the summer months only; therefore, serum 25(OH)D levels are higher between June and September than during the rest of the year. Because the baseline examinations were conducted in different seasons, the subjects were divided into 2 seasonal groups, winter (October–May) and summer (June–September).

Follow-up data

Incident cases of fatal CVD were identified through linkage with Statistics Finland, using the Finnish nationwide individual identification number as the identity link. The Eighth, Ninth, and Tenth revisions of the International Classification of Diseases (ICD-8, ICD-9, and ICD-10, respectively) were used for coding the causes of death. Deaths with ICD codes 410–414 (ICD-8 and ICD-9) and I20–I25 (ICD-10) were classified as coronary heart disease deaths. Cerebrovascular deaths included those due to subarachnoid hemorrhage (ICD-8 and ICD-9 code 430, ICD-10 code I60), hemorrhagic stroke (ICD-8 and ICD-9 code 431, ICD-10 code I61), ischemic stroke (ICD-8 and ICD-9 codes 433–434, ICD-10 code I63), or other unspecified cerebrovascular causes (ICD-8 codes 435–438, ICD-9 codes 432 and 435–438, ICD-10 codes I6 and I64–I69).

Statistical analyses

The Cox proportional hazards model was used to estimate hazard ratios and 95% confidence intervals for total CVD, coronary heart disease, and cerebrovascular deaths according to quintile of serum 25(OH)D level. The follow-up period was defined as the time from the baseline examination to the date of CVD death, death from other causes, or the end of follow-up (December 31, 2006)—whichever came first. All analyses were adjusted for age (in years, as a continuous variable) and sex (model 1). Multivariable analyses (model 2) also included adjustment for the following a priori potential confounders: marital status, educational level, body mass index, alcohol consumption, smoking, leisure-time physical activity, and season of baseline examination. We defined 2 additional models, one of which further included serum levels of high and low density lipoprotein cholesterol (model 3) and another that also included blood pressure and diabetes (model 4).

In secondary analyses, the subjects who died of CVD within the first 4 years of follow-up or were aged 70 years or older at baseline were excluded. In addition, interactions between vitamin D and the potential effect-modifying factors (blood pressure, smoking, age, sex, body mass index, season of baseline examination, serum total, high and low density lipoprotein cholesterol levels, leisure-time physical activity, alcohol consumption and diabetes) were assessed in the main multivariable model (model 2). Because the results of these secondary analyses were essentially similar for total CVD, coronary heart disease, and cerebrovascular mortality, only the results for total CVD are presented. The vitamin D–stroke association was also analyzed according to subtypes of stroke (hemorrhagic and ischemic); these analyses were conducted using tertiles of serum 25(OH)D because of the small number of cases. In addition, subjects were divided into 2 categories using commonly applied cutoff points for low (<50 nmol/L) and high (≥50 nmol/L) levels of vitamin D (1).

RESULTS

The study population consisted of 2,817 men and 3,402 women with a mean age of 49.4 years (standard deviation (SD), 13.6) at baseline. The mean serum 25(OH)D level was 43.4 nmol/L (SD, 19.7) (45.7 nmol/L (SD, 20.3) in men and 41.5 nmol/L (SD, 18.9) in women), with 67.6% of the population having a level less than 50 nmol/L. Older and highly educated subjects were more likely to have a higher vitamin D status than younger and less-educated subjects, respectively (Table 1). Heavy smoking, high alcohol consumption, low leisure-time physical activity, high body mass index, diabetes, hypertension, and a poor serum lipid profile were also associated with low serum 25(OH)D level.

Table 1.

Baseline Characteristics of Participants According to Quintile of Serum 25-Hydroxyvitamin D Level, Mini-Finland Health Survey, Finland, 1978–1980a

 Serum 25-Hydroxyvitamin D Quintile
 
P for Trendb 
 1 (n = 1,258) 2 (n = 1,202) 3 (n = 1,284) 4 (n = 1,222) 5 (n = 1,253) 
Range of serum25-hydroxyvitamin D values, nmol/L (median)       
    Men 5.0–28.0 (23.0) 29.0–37.0 (33.0) 38.0–47.0 (42.0) 48.0–61.0 (54.0) 62.0–180.0 (72.0)  
    Women 4.0–25.0 (21.0) 26.0–33.0 (30.0) 34.0–43.0 (38.0) 44.0–55.0 (49.0) 56.0–151.0 (67.0)  
Mean age, years (SD) 53.9 (15.1) 48.9 (13.8) 48.4 (13.1) 48.8 (12.8) 47.0 (12.2) <0.001 
Season, %      <0.001 
    Winter (October–May) 96.7 94.8 88.0 74.6 52.8  
    Summer (June–September) 3.3 5.2 12.0 25.4 47.2  
Education, %      <0.001 
    Low 71.4 68.8 65.4 62.8 60.8  
    High 28.6 31.2 34.6 37.2 39.2  
Marital status, %      0.19 
    Unmarried 14.9 10.3 9.5 10.8 9.1  
    Marriedc 66.9 72.1 76.1 76.7 77.7  
    Widowed 12.3 11.5 9.0 8.1 8.5  
    Divorced 5.8 6.1 5.4 4.4 4.7  
Smokingd, %      <0.001 
    Nonsmoker 66.2 68.4 70.1 72.2 72.7  
    Smoker with low cotinine level 8.8 10.2 10.4 9.7 11.0  
    Smoker with average cotinine level 11.9 10.1 10.0 9.6 8.5  
    Smoker with high cotinine level 13.0 11.3 9.5 8.5 7.9  
Diabetes at baselinee, % 7.0 4.7 3.9 3.5 2.5 <0.001 
Blood pressuref, %      0.035 
    Normotensive 36.5 35.8 34.9 37.0 40.5  
    Borderline hypertension 43.7 43.9 45.0 43.5 41.2  
    Mild hypertension 3.6 4.4 4.5 4.1 4.2  
    Definite hypertension 16.2 16.0 15.7 15.5 14.1  
Leisure-time physical activity, %      <0.001 
    Inactive 47.5 38.8 33.1 29.3 23.6  
    Occasionally active 42.4 48.0 49.5 53.2 53.8  
    Regularly active 10.0 13.2 17.5 17.5 22.6  
Mean body mass indexg (SD) 25.6 (4.5) 26.0 (4.3) 26.1 (3.9) 25.7 (4.0) 25.2 (3.6) <0.001 
Mean alcohol consumption (ethanol intake), g/week (SD) 42.9 (108.0) 40.2 (91.3) 50.9 (107.2) 53.5 (125.9) 51.5 (111.4) 0.001 
Mean serum total cholesterol level, mmol/L (SD) 6.74 (1.39) 6.89 (1.35) 6.96 (1.34) 6.98 (1.28) 6.93 (1.35) <0.001 
Mean serum high density lipoprotein cholesterol level, mmol/L (SD) 1.65 (0.40) 1.66 (0.38) 1.71 (0.40) 1.75 (0.40) 1.80 (0.39) <0.001 
Mean serum low density lipoprotein cholesterol level, mmol/L (SD) 4.34 (1.26) 4.56 (1.23) 4.57 (1.21) 4.61 (1.15) 4.54 (1.23) <0.001 
Mean serum triglyceride level, mmol/L (SD) 1.65 (1.65) 1.50 (0.86) 1.49 (0.86) 1.37 (0.73) 1.31 (0.62) <0.001 
 Serum 25-Hydroxyvitamin D Quintile
 
P for Trendb 
 1 (n = 1,258) 2 (n = 1,202) 3 (n = 1,284) 4 (n = 1,222) 5 (n = 1,253) 
Range of serum25-hydroxyvitamin D values, nmol/L (median)       
    Men 5.0–28.0 (23.0) 29.0–37.0 (33.0) 38.0–47.0 (42.0) 48.0–61.0 (54.0) 62.0–180.0 (72.0)  
    Women 4.0–25.0 (21.0) 26.0–33.0 (30.0) 34.0–43.0 (38.0) 44.0–55.0 (49.0) 56.0–151.0 (67.0)  
Mean age, years (SD) 53.9 (15.1) 48.9 (13.8) 48.4 (13.1) 48.8 (12.8) 47.0 (12.2) <0.001 
Season, %      <0.001 
    Winter (October–May) 96.7 94.8 88.0 74.6 52.8  
    Summer (June–September) 3.3 5.2 12.0 25.4 47.2  
Education, %      <0.001 
    Low 71.4 68.8 65.4 62.8 60.8  
    High 28.6 31.2 34.6 37.2 39.2  
Marital status, %      0.19 
    Unmarried 14.9 10.3 9.5 10.8 9.1  
    Marriedc 66.9 72.1 76.1 76.7 77.7  
    Widowed 12.3 11.5 9.0 8.1 8.5  
    Divorced 5.8 6.1 5.4 4.4 4.7  
Smokingd, %      <0.001 
    Nonsmoker 66.2 68.4 70.1 72.2 72.7  
    Smoker with low cotinine level 8.8 10.2 10.4 9.7 11.0  
    Smoker with average cotinine level 11.9 10.1 10.0 9.6 8.5  
    Smoker with high cotinine level 13.0 11.3 9.5 8.5 7.9  
Diabetes at baselinee, % 7.0 4.7 3.9 3.5 2.5 <0.001 
Blood pressuref, %      0.035 
    Normotensive 36.5 35.8 34.9 37.0 40.5  
    Borderline hypertension 43.7 43.9 45.0 43.5 41.2  
    Mild hypertension 3.6 4.4 4.5 4.1 4.2  
    Definite hypertension 16.2 16.0 15.7 15.5 14.1  
Leisure-time physical activity, %      <0.001 
    Inactive 47.5 38.8 33.1 29.3 23.6  
    Occasionally active 42.4 48.0 49.5 53.2 53.8  
    Regularly active 10.0 13.2 17.5 17.5 22.6  
Mean body mass indexg (SD) 25.6 (4.5) 26.0 (4.3) 26.1 (3.9) 25.7 (4.0) 25.2 (3.6) <0.001 
Mean alcohol consumption (ethanol intake), g/week (SD) 42.9 (108.0) 40.2 (91.3) 50.9 (107.2) 53.5 (125.9) 51.5 (111.4) 0.001 
Mean serum total cholesterol level, mmol/L (SD) 6.74 (1.39) 6.89 (1.35) 6.96 (1.34) 6.98 (1.28) 6.93 (1.35) <0.001 
Mean serum high density lipoprotein cholesterol level, mmol/L (SD) 1.65 (0.40) 1.66 (0.38) 1.71 (0.40) 1.75 (0.40) 1.80 (0.39) <0.001 
Mean serum low density lipoprotein cholesterol level, mmol/L (SD) 4.34 (1.26) 4.56 (1.23) 4.57 (1.21) 4.61 (1.15) 4.54 (1.23) <0.001 
Mean serum triglyceride level, mmol/L (SD) 1.65 (1.65) 1.50 (0.86) 1.49 (0.86) 1.37 (0.73) 1.31 (0.62) <0.001 

Abbreviation: SD, standard deviation.

a

Data are presented as mean values (SD) or percentages, adjusted for age (years, as a continuous variable) and sex (age was adjusted for sex only), unless otherwise noted.

b

P for trend from a linear model, adjusted for age (years, as a continuous variable) and sex.

c

Included persons in common-law marriages and committed relationships.

d

Nonsmokers included persons who were never smokers or ex-smokers and had serum cotinine levels ≤100 ng/mL; the rest of the subjects were divided into tertiles based on their serum cotinine level.

e

Diabetes mellitus was defined as a self-reported history of diabetes that had been diagnosed and treated by a physician or a fasting plasma glucose level ≥6.7 mmol/L.

f

Definite hypertension was defined as systolic blood pressure ≥170 mm Hg and diastolic blood pressure ≥100 mm Hg or use of antihypertensive medication. Of the remaining members of the study population, those with systolic blood pressure ≥160 mm Hg and diastolic blood pressure ≥95 mm Hg were considered to have mild hypertension and those with systolic blood pressure <140 mm Hg and diastolic blood pressure <90 mm Hg were considered to be normotensive. All other subjects were considered to have borderline hypertension.

g

Weight (kg)/height (m)2.

During a median follow-up period of 27.1 years (range, 9 days–28.9 years), 640 coronary heart disease deaths (358 in men and 282 in women) and 293 cerebrovascular disease deaths (122 in men and 171 in women) were identified. Cerebrovascular events included 175 ischemic strokes, 43 hemorrhagic strokes, 22 subarachnoid hemorrhages, and 53 other unspecified cerebrovascular events.

There was an inverse association between serum 25(OH)D level and total CVD mortality when results were adjusted for age and sex only (for the highest quintile in model 1 vs. the lowest, hazard ratio (HR) = 0.71, 95% confidence interval (CI): 0.58, 0.87; P for trend < 0.001 (Table 2)). Adjustment for potential confounders, including also marital status, educational level, body mass index, alcohol consumption, smoking, leisure-time physical activity, and season of baseline examination, only slightly attenuated the association (in model 2, HR = 0.76, 95% CI: 0.61, 0.95; P for trend = 0.005). Further adjustment for serum high and low density lipoprotein cholesterol levels (in model 3, HR = 0.75, 95% CI: 0.59, 0.94; P for trend = 0.004) and diabetes and blood pressure (in model 4, HR = 0.80, 95% CI: 0.64, 1.01; P for trend 0.012) did not notably change the results. Moreover, the results remained essentially similar after the exclusion of subjects who died of CVD during the first 4 years of follow-up (in model 2, HR = 0.76, 95% CI: 0.60, 0.96; P for trend = 0.008) or were 70 years of age or older at baseline (in model 2, HR = 0.79, 95% CI: 0.61, 1.03; P for trend = 0.029). No statistically significant interaction was observed between 25(OH)D level and age (P for interaction = 0.95), alcohol consumption (P = 0.53), body mass index (P = 0.49), diabetes (P = 0.32), blood pressure (P = 0.81), leisure-time physical activity (P = 0.30), season of baseline examination (P = 0.75), sex (P = 0.74), smoking (P = 0.24), or serum total (P = 0.12), high density lipoprotein (P = 0.23), or low density lipoprotein (P = 0.13) cholesterol level.

Table 2.

Hazard Ratios for Cerebrovascular and Coronary Disease Mortality According to Quintile of Baseline Serum 25-Hydroxyvitamin D Level, Mini-Finland Health Survey, Finland, 1978–1980

 Serum 25-Hydroxyvitamin D Quintile
 
P for Trend 
 1a 2
 
3
 
4
 
5
 
 (HR = 1) HR 95% CI HR 95% CI HR 95% CI HR 95% CI 
Total no. of cardiovascular disease deaths 254 194 164 171 150  
    Model 1 (adjusted for age and sex) 1.0 1.00 0.82, 1.20 0.77 0.63, 0.94 0.83 0.68, 1.00 0.71 0.58, 0.87 <0.001 
    Model 2 (multivariable model)b 1.0 1.04 0.86, 1.26 0.81 0.66, 1.00 0.86 0.70, 1.06 0.76 0.61, 0.95 0.005 
No. of cerebrovascular deaths 92 52 68 48 33  
    Model 1 (adjusted for age and sex) 1.0 0.77 0.55, 1.09 0.93 0.68, 1.28 0.69 0.48, 0.98 0.47 0.31, 0.70 <0.001 
    Model 2 (multivariable model)b 1.0 0.80 0.57, 1.14 0.97 0.70, 1.35 0.69 0.48, 1.00 0.48 0.31, 0.75 0.002 
No. of coronary deaths 162 142 96 123 117  
    Model 1 (adjusted for age and sex) 1.0 1.12 0.89, 1.40 0.69 0.54, 0.89 0.91 0.72, 1.15 0.83 0.65, 1.06 0.037 
    Model 2 (multivariable model)b 1.0 1.17 0.93, 1.48 0.73 0.56, 0.95 0.95 0.74, 1.22 0.91 0.70, 1.18 0.20 
Range of serum 25-hydroxyvitamin D values, nmol/L (median)           
    Men 5.0–28.0 (23.0) 29.0–37.0 (33.0) 38.0–47.0 (42.0) 48.0–61.0 (54.0) 62.0–180.0 (72.0)  
    Women 4.0–25.0 (21.0) 26.0–33.0 (30.0) 34.0–43.0 (38.0) 44.0–55.0 (49.0) 56.0–151.0 (67.0)  
 Serum 25-Hydroxyvitamin D Quintile
 
P for Trend 
 1a 2
 
3
 
4
 
5
 
 (HR = 1) HR 95% CI HR 95% CI HR 95% CI HR 95% CI 
Total no. of cardiovascular disease deaths 254 194 164 171 150  
    Model 1 (adjusted for age and sex) 1.0 1.00 0.82, 1.20 0.77 0.63, 0.94 0.83 0.68, 1.00 0.71 0.58, 0.87 <0.001 
    Model 2 (multivariable model)b 1.0 1.04 0.86, 1.26 0.81 0.66, 1.00 0.86 0.70, 1.06 0.76 0.61, 0.95 0.005 
No. of cerebrovascular deaths 92 52 68 48 33  
    Model 1 (adjusted for age and sex) 1.0 0.77 0.55, 1.09 0.93 0.68, 1.28 0.69 0.48, 0.98 0.47 0.31, 0.70 <0.001 
    Model 2 (multivariable model)b 1.0 0.80 0.57, 1.14 0.97 0.70, 1.35 0.69 0.48, 1.00 0.48 0.31, 0.75 0.002 
No. of coronary deaths 162 142 96 123 117  
    Model 1 (adjusted for age and sex) 1.0 1.12 0.89, 1.40 0.69 0.54, 0.89 0.91 0.72, 1.15 0.83 0.65, 1.06 0.037 
    Model 2 (multivariable model)b 1.0 1.17 0.93, 1.48 0.73 0.56, 0.95 0.95 0.74, 1.22 0.91 0.70, 1.18 0.20 
Range of serum 25-hydroxyvitamin D values, nmol/L (median)           
    Men 5.0–28.0 (23.0) 29.0–37.0 (33.0) 38.0–47.0 (42.0) 48.0–61.0 (54.0) 62.0–180.0 (72.0)  
    Women 4.0–25.0 (21.0) 26.0–33.0 (30.0) 34.0–43.0 (38.0) 44.0–55.0 (49.0) 56.0–151.0 (67.0)  

Abbreviations: CI, confidence interval; HR, hazard ratio.

a

Reference category.

b

Adjusted for age (as a continuous variable), sex, marital status (unmarried, married or in a committed relationship, widowed, divorced), education (low, high), body mass index (weight (kg)/height (m)2; continuous variable), alcohol consumption (ethanol intake, g/week; continuous variable), smoking (nonsmoker or smoker with a low, average, or high cotinine level), leisure-time physical activity (inactive, occasionally active, regularly active), and season of baseline examination (winter, summer).

An inverse association was found between serum 25(OH)D level and mortality from cerebrovascular disease (for the highest quintile in model 1 vs. the lowest, HR = 0.47, 95% CI: 0.31, 0.70; P for trend < 0.001 (Table 2)). Results from multivariable analyses were rather similar (in model 2, HR = 0.48, 95% CI: 0.31, 0.75; P for trend = 0.002), indicating that there was little confounding by the covariates. Further adjustment for serum high and low density lipoprotein cholesterol concentrations (in model 3, HR = 0.50, 95% CI: 0.32, 0.77; P for trend = 0.003) and diabetes and blood pressure (in model 4, HR = 0.52, 95% CI: 0.33, 0.80; P for trend = 0.004) did not change the results. In the analyses conducted according to subtype of stroke, the multivariable adjusted hazard ratios (for the highest tertile in model 2 vs. the lowest) for hemorrhagic and ischemic stroke were 0.61 (95% CI: 0.26, 1.46; P for trend = 0.29) and 0.60 (95% CI: 0.38, 0.93; P for trend = 0.050), respectively (Table 3).

Table 3.

Hazard Ratios for Various Subtypes of Stroke According to Tertile of Baseline Serum 25-Hydroxyvitamin D Level, Mini-Finland Health Survey, Finland, 1978–1980

 Serum 25-Hydroxyvitamin D Tertile
 
P for Trend 
 1a 2
 
3
 
 (HR = 1) HR 95% CI HR 95% CI 
No. of fatal hemorrhagic strokes 18 15 10  
    Model 1 (adjusted for age and sex) 1.0 0.86 0.43, 1.71 0.57 0.26, 1.23 0.15 
    Model 2 (multivariable model)b 1.0 0.95 0.46, 1.95 0.61 0.26, 1.46 0.29 
No. of fatal ischemic strokes 78 65 32  
    Model 1 (adjusted for age and sex) 1.0 1.06 0.76, 1.47 0.57 0.38, 0.87 0.017 
    Model 2 (multivariable model)b 1.0 1.11 0.79, 1.56 0.60 0.38, 0.93 0.050 
Range of serum 25-hydroxyvitamin D values, nmol/L (median)       
    Men 5.0–34.0 (27.0) 35.0–51.0 (42.0) 52.0–180.0 (64.0)  
    Women 4.0–31.0 (24.0) 32.0–47.0 (39.0) 48.0–151.0 (59.0)  
 Serum 25-Hydroxyvitamin D Tertile
 
P for Trend 
 1a 2
 
3
 
 (HR = 1) HR 95% CI HR 95% CI 
No. of fatal hemorrhagic strokes 18 15 10  
    Model 1 (adjusted for age and sex) 1.0 0.86 0.43, 1.71 0.57 0.26, 1.23 0.15 
    Model 2 (multivariable model)b 1.0 0.95 0.46, 1.95 0.61 0.26, 1.46 0.29 
No. of fatal ischemic strokes 78 65 32  
    Model 1 (adjusted for age and sex) 1.0 1.06 0.76, 1.47 0.57 0.38, 0.87 0.017 
    Model 2 (multivariable model)b 1.0 1.11 0.79, 1.56 0.60 0.38, 0.93 0.050 
Range of serum 25-hydroxyvitamin D values, nmol/L (median)       
    Men 5.0–34.0 (27.0) 35.0–51.0 (42.0) 52.0–180.0 (64.0)  
    Women 4.0–31.0 (24.0) 32.0–47.0 (39.0) 48.0–151.0 (59.0)  

Abbreviations: CI, confidence interval; HR, hazard ratio.

a

Reference category.

b

Adjusted for age (as a continuous variable), sex, marital status (unmarried, married or in a committed relationship, widowed, divorced), education (low, high), body mass index (weight (kg)/height (m)2; continuous variable), alcohol consumption (ethanol intake, g/week; continuous variable), smoking (nonsmoker or smoker with a low, average, or high cotinine level), leisure-time physical activity (inactive, occasionally active, regularly active), and season of baseline examination (winter, summer).

In model 1, with adjustment for age and sex only, an inverse association was found between serum 25(OH)D level and the risk of coronary heart disease death (for the highest quintile vs. the lowest, HR = 0.83, 95% CI: 0.65, 1.06; P for trend = 0.037 (Table 2)). After adjustment for potential confounders, the hazard ratios were no longer statistically significant (in model 2, HR = 0.91, 95% CI: 0.70, 1.18; P for trend = 0.20). No single risk factor was responsible for the attenuation. Further inclusion of serum high and low density lipoprotein cholesterol concentrations in the model did not change the results (in model 3, HR = 0.87, 95% CI: 0.67, 1.14; P for trend = 0.12). In the model 4 that also included diabetes and blood pressure, the hazard ratio for the highest quintile versus the lowest was 0.96 (95% CI: 0.73, 1.25; P for trend = 0.28).

In further analysis based on the cutoff value of 50 nmol/L for serum 25(OH)D level, the multivariable adjusted hazard ratio (for high vitamin D category in model 2 vs. low vitamin D category) for total CVD death was 0.88 (95% CI: 0.75, 1.03). For mortality from cerebrovascular disease and coronary heart disease, the corresponding hazard ratios were 0.58 (95% CI: 0.42, 0.79) and 1.04 (95% CI: 0.86, 1.25), respectively.

DISCUSSION

This cohort study provided evidence that a low circulating level of vitamin D may predict a higher risk of CVD death. The observed association was particularly striking for mortality from cerebrovascular disease; subjects in the highest quintile of serum 25(OH)D level had less than half the risk of cerebrovascular death as those in the lowest quintile.

Epidemiologic evidence on the association between vitamin D and the risk of CVD is limited. In the Health Professionals Follow-up Study, men with a high circulating level of vitamin D had half the risk of myocardial infarction as men with vitamin D insufficiency (14). Similarly, a study of German adults who were undergoing elective cardiac catheterization showed a 2-fold risk of CVD death among persons in the lowest quartile of baseline vitamin D level compared with those in the highest quartile (13). In addition, among participants in the Framingham Offspring Study cohort, vitamin D deficiency was associated with an increased risk of CVD (relative risk = 1.62, 95% CI: 1.11, 2.36) (15). However, in contrast to our findings, the association was observed only in hypertensive subjects, not in those without hypertension (15). Moreover, in a recent cohort study based on data from the Third National Health and Nutrition Examination Survey, Melamed et al. (16) could not find a statistically significant association between vitamin D status and CVD mortality in the general population. Note, however, that data on coronary heart disease and cerebrovascular events were pooled in the Third National Health and Nutrition Examination Survey (16) and in most earlier studies (13, 15), whereas we analyzed deaths from coronary heart disease and cerebrovascular disease separately. Indeed, our findings suggest that vitamin D might have a more important role in the prevention of cerebrovascular disease, especially ischemic stroke, than in the etiology of coronary heart disease. There is no obvious explanation for this finding, and confirmation in other studies is required. However, although coronary heart disease and cerebrovascular disease share important risk factors, the impact of some risk factors (e.g., blood lipids and blood pressure) varies (21). This may imply different underlying mechanisms for these diseases and offers a possible explanation for the differences in our results for cerebrovascular disease and coronary heart disease.

Although the exact mechanisms by which an adequate vitamin D status may protect against CVD are not fully understood, experimental studies indicate that vitamin D is one of the most potent hormones for suppressing the renin-angiotensin system and thus for regulating blood pressure (4). The vascular effects of vitamin D also include inhibition of thrombosis (3) and arterial calcification (22). Furthermore, several types of cells, including vascular smooth muscle cells and lymphocytes, express receptors for vitamin D and have the ability to convert circulating 25(OH)D to 1,25-dihydroxyvitamin D, which in turn can reduce the proliferation of lymphocytes and the production of cytokines (5, 6). Because there is increasing evidence that systemic inflammation plays an important role in the development of atherosclerosis (23), the antiinflammatory properties of vitamin D warrant further exploration. Vitamin D deficiency, on the other hand, increases the secretion of parathyroid hormone, which has been shown to contribute to pathologic changes in the cardiovascular system (24). Further studies are required to clarify the vitamin D–CVD association and the mechanisms behind the association. Data from supplementation trials of the effects of vitamin D on bone health could provide useful information.

The mean serum 25(OH)D values in our study were approximately of the same order of magnitude as those previously found in Finland (25) but somewhat lower than those generally observed in other European (13, 26) and American (12) populations. There is no absolute consensus as to what the optimal range for serum 25(OH)D levels should be. However, relatively high concentrations of 25(OH)D (>75 nmol/L) are required to maintain normal parathyroid hormone levels, and even higher concentrations (≥83–121 nmol/L) are suggested to be desirable for cancer prevention (27). Although optimal levels for cardiovascular protection may differ from those, it is noteworthy that the values in our cohort were substantially lower than those previously thought to be sufficient (27).

The main strengths of the present study lay in the prospective design and the fairly large nationally representative population sample. In addition, information on CVD mortality was obtained from the nationwide mortality register, which is based on death certificates and has been shown to have reasonably good validity (28, 29). A further strength of the study was the information on CVD and its risk factors at baseline from the physician's examination. However, while the detailed data on multiple CVD risk factors allowed adjustment for potential confounders, we cannot rule out the possibility of residual confounding. It can be speculated that persons with chronic illness may have reduced serum vitamin D levels because of their limited exposure to sunlight and inadequate dietary intake of vitamin D. This raises the possibility that low vitamin D status is only a nonspecific indicator of chronic illness rather than a direct contributor to disease pathogenesis. It is important to acknowledge potential confounding by dietary factors, as we did not have information on vitamin D intake from diet and supplements. The major dietary source of vitamin D is fatty fish, consumption of which is suggested to be protective against CVD because of its n-3 polyunsaturated fatty acid content (30). A recent meta-analysis, however, did not show a definite effect of omega-3 fatty acids on CVD events (31).

A further limitation of this study was the use of a single measurement of vitamin D. It can be questioned whether serum 25(OH)D level measured at a single point in time reflects only recent exposure rather than long-term exposure. Nevertheless, in 1 study, the correlation coefficient for correlation between 2 measurements of vitamin D taken 3 years apart appeared to be moderately high, 0.70 (32), suggesting that a single serum measurement of this compound could be a useful tool in epidemiologic studies. Such measurement, however, fails to take into account the intraindividual seasonal variation in serum 25(OH)D levels. Inclusion of the season of baseline examination as a potential confounder in the model or the use of it as an effect-modifying factor did not substantially change our results. Because serum samples were stored at −20°C up to 25 years before the determination of vitamin D level, a change in vitamin D concentrations during storage is a potential concern. However, vitamin D metabolites in blood stored at 24°C for up to 72 hours have been shown to remain intact (33), and only a minimal decline is observed for plasma 25(OH)D level for up to 4 years of storage at −20°C (34). Although the evidence suggests that 25(OH)D is a stable compound (33–35), we cannot rule out the possibility that levels might have changed during storage at −20°C. Finally, the use of mortality rather than incidence data was a potential limitation of the study.

In conclusion, our results suggest that a low circulating level of vitamin D may be associated with a higher risk of fatal CVD events. Although a possible causal link between vitamin D and CVD is biologically plausible, further investigations from different populations with repeated measurements of vitamin D are warranted. To demonstrate a causal link between vitamin D status and the risk of CVD, randomized controlled clinical trials are required. Because CVD remains the leading cause of death in most developed countries, identification of new CVD risk factors (such as vitamin D) is an area of much interest, both scientifically and among the lay public. Our findings may have profound public health implications; while the prevalence of suboptimal vitamin D levels has been observed to be high worldwide, vitamin D status can be rather inexpensively and easily improved through supplementation or lifestyle measures.

Abbreviations

    Abbreviations
  • CI

    confidence interval

  • CVD

    cardiovascular disease

  • HR

    hazard ratio

  • ICD

    International Classification of Diseases

  • 25(OH)D

    25-hydroxyvitamin D

  • SD

    standard deviation

Author affiliation: National Institute for Health and Welfare, Helsinki, Finland.

This work was supported in part by the Social Insurance Institution of Finland.

Conflict of interest: none declared.

References

1.
Holick
MF
Vitamin D deficiency
N Engl J Med
 , 
2007
, vol. 
357
 
3
(pg. 
266
-
281
)
2.
Michos
ED
Melamed
ML
Vitamin D and cardiovascular disease risk
Curr Opin Clin Nutr Metab Care
 , 
2008
, vol. 
11
 
1
(pg. 
7
-
12
)
3.
Aihara
K
Azuma
H
Akaike
M
, et al.  . 
Disruption of nuclear vitamin D receptor gene causes enhanced thrombogenicity in mice
J Biol Chem.
 , 
2004
, vol. 
279
 
34
(pg. 
35798
-
35802
)
4.
Li
YC
Kong
J
Wei
M
, et al.  . 
1,25-Dihydroxyvitamin D(3) is a negative endocrine regulator of the renin-angiotensin system
J Clin Invest
 , 
2002
, vol. 
110
 
2
(pg. 
229
-
238
)
5.
Merke
J
Hofmann
W
Goldschmidt
D
, et al.  . 
Demonstration of 1,25(OH)2 vitamin D3 receptors and actions in vascular smooth muscle cells in vitro
Calcif Tissue Int
 , 
1987
, vol. 
41
 
2
(pg. 
112
-
114
)
6.
Rigby
WF
Denome
S
Fanger
MW
Regulation of lymphokine production and human T lymphocyte activation by 1,25-dihydroxyvitamin D3. Specific inhibition at the level of messenger RNA
J Clin Invest
 , 
1987
, vol. 
79
 
6
(pg. 
1659
-
1664
)
7.
Poole
KE
Loveridge
N
Barker
PJ
, et al.  . 
Reduced vitamin D in acute stroke
Stroke
 , 
2006
, vol. 
37
 
1
(pg. 
243
-
245
)
8.
Scragg
R
Jackson
R
Holdaway
IM
, et al.  . 
Myocardial infarction is inversely associated with plasma 25-hydroxyvitamin D3 levels: a community-based study
Int J Epidemiol
 , 
1990
, vol. 
19
 
3
(pg. 
559
-
563
)
9.
Zittermann
A
Schleithoff
SS
Koerfer
R
Putting cardiovascular disease and vitamin D insufficiency into perspective
Br J Nutr
 , 
2005
, vol. 
94
 
4
(pg. 
483
-
492
)
10.
Zittermann
A
Schleithoff
SS
Tenderich
G
, et al.  . 
Low vitamin D status: a contributing factor in the pathogenesis of congestive heart failure?
J Am Coll Cardiol
 , 
2003
, vol. 
41
 
1
(pg. 
105
-
112
)
11.
Martins
D
Wolf
M
Pan
D
, et al.  . 
Prevalence of cardiovascular risk factors and the serum levels of 25-hydroxyvitamin D in the United States: data from the Third National Health and Nutrition Examination Survey
Arch Intern Med
 , 
2007
, vol. 
167
 
11
(pg. 
1159
-
1165
)
12.
Kendrick
J
Targher
G
Smits
G
, et al.  . 
25-Hydroxyvitamin D deficiency is independently associated with cardiovascular disease in the Third National Health and Nutrition Examination Survey
Atherosclerosis
 , 
2009
, vol. 
205
 
1
(pg. 
255
-
260
)
13.
Dobnig
H
Pilz
S
Scharnagl
H
, et al.  . 
Independent association of low serum 25-hydroxyvitamin D and 1,25-dihydroxyvitamin D levels with all-cause and cardiovascular mortality
Arch Intern Med
 , 
2008
, vol. 
168
 
12
(pg. 
1340
-
1349
)
14.
Giovannucci
E
Liu
Y
Hollis
BW
, et al.  . 
25-hydroxyvitamin D and risk of myocardial infarction in men: a prospective study
Arch Intern Med
 , 
2008
, vol. 
168
 
11
(pg. 
1174
-
1180
)
15.
Wang
TJ
Pencina
MJ
Booth
SL
, et al.  . 
Vitamin D deficiency and risk of cardiovascular disease
Circulation
 , 
2008
, vol. 
117
 
4
(pg. 
503
-
511
)
16.
Melamed
ML
Michos
ED
Post
W
, et al.  . 
25-Hydroxyvitamin D levels and the risk of mortality in the general population
Arch Intern Med
 , 
2008
, vol. 
168
 
15
(pg. 
1629
-
1637
)
17.
Hsia
J
Heiss
G
Ren
H
, et al.  . 
Calcium/vitamin D supplementation and cardiovascular events
Circulation
 , 
2007
, vol. 
115
 
7
(pg. 
846
-
854
)
18.
Neuhouser
ML
Wassertheil-Smoller
S
Thomson
C
, et al.  . 
Multivitamin use and risk of cancer and cardiovascular disease in the Women's Health Initiative cohorts
Arch Intern Med
 , 
2009
, vol. 
169
 
3
(pg. 
294
-
304
)
19.
Aromaa
A
Heliövaara
M
Impivaara
O
, et al.  . 
The Execution of Mini-Finland Health Survey: Aims, Methods, and Study Population [in Finnish with English summary]. (Publication no. ML:88)
 , 
1989
Helsinki, Finland
Social Insurance Institution
20.
Carr
JJ
Drekter
IJ
Simplified rapid technic for the extraction and determination of serum cholesterol without saponification
Clin Chem.
 , 
1956
, vol. 
2
 
5
(pg. 
353
-
68
)
21.
Kannel
WB
D'Agostino
RB
Sullivan
L
, et al.  . 
Concept and usefulness of cardiovascular risk profiles
Am Heart J
 , 
2004
, vol. 
148
 
1
(pg. 
16
-
26
)
22.
Watson
KE
Abrolat
ML
Malone
LL
, et al.  . 
Active serum vitamin D levels are inversely correlated with coronary calcification
Circulation
 , 
1997
, vol. 
96
 
6
(pg. 
1755
-
1760
)
23.
Jouni
ZE
Winzerling
JJ
McNamara
DJ
1,25-Dihydroxyvitamin D3-induced HL-60 macrophages: regulation of cholesterol and LDL metabolism
Atherosclerosis
 , 
1995
, vol. 
117
 
1
(pg. 
125
-
138
)
24.
Rostand
SG
Drüeke
TB
Parathyroid hormone, vitamin D, and cardiovascular disease in chronic renal failure
Kidney Int
 , 
1999
, vol. 
56
 
2
(pg. 
383
-
392
)
25.
Tuohimaa
P
Tenkanen
L
Ahonen
M
, et al.  . 
Both high and low levels of blood vitamin D are associated with a higher prostate cancer risk: a longitudinal, nested case-control study in the Nordic countries
Int J Cancer
 , 
2004
, vol. 
108
 
1
(pg. 
104
-
108
)
26.
Ovesen
L
Andersen
R
Jakobsen
J
Geographical differences in vitamin D status, with particular reference to European countries
Proc Nutr Soc.
 , 
2003
, vol. 
62
 
4
(pg. 
813
-
821
)
27.
Bischoff-Ferrari
HA
Giovannucci
E
Willett
WC
, et al.  . 
Estimation of optimal serum concentrations of 25-hydroxyvitamin D for multiple health outcomes
Am J Clin Nutr
 , 
2006
, vol. 
84
 
1
(pg. 
18
-
28
)
28.
Mähönen
M
Salomaa
V
Torppa
J
, et al.  . 
The validity of the routine mortality statistics on coronary heart disease in Finland: comparison with the FINMONICA MI register data for the years 1983–1992
J Clin Epidemiol
 , 
1999
, vol. 
52
 
2
(pg. 
157
-
166
)
29.
Pajunen
P
Koukkunen
H
Ketonen
M
, et al.  . 
The validity of the Finnish Hospital Discharge Register and Causes of Death Register data on coronary heart disease
Eur J Cardiovasc Prev Rehabil
 , 
2005
, vol. 
12
 
2
(pg. 
132
-
137
)
30.
He
K
Song
Y
Daviglus
ML
, et al.  . 
Accumulated evidence on fish consumption and coronary heart disease mortality: a meta-analysis of cohort studies
Circulation
 , 
2004
, vol. 
109
 
22
(pg. 
2705
-
2711
)
31.
Hooper
L
Thompson
RL
Harrison
RA
, et al.  . 
Risks and benefits of omega 3 fats for mortality, cardiovascular disease, and cancer: systematic review
BMJ
 , 
2006
, vol. 
332
 
7544
(pg. 
752
-
760
)
32.
Platz
EA
Leitzmann
MF
Hollis
BW
, et al.  . 
Plasma 1,25-dihydroxy- and 25-hydroxyvitamin D and subsequent risk of prostate cancer
Cancer Causes Control
 , 
2004
, vol. 
15
 
3
(pg. 
255
-
265
)
33.
Lissner
D
Mason
RS
Posen
S
Stability of vitamin D metabolites in human blood serum and plasma
Clin Chem.
 , 
1981
, vol. 
27
 
5
(pg. 
773
-
774
)
34.
Ockè
MC
Schrijver
J
Obermann-de Boer
GL
, et al.  . 
Stability of blood (pro)vitamins during four years of storage at −20 degrees C: consequences for epidemiologic research
J Clin Epidemiol
 , 
1995
, vol. 
48
 
8
(pg. 
1077
-
1085
)
35.
Hollis
BW
Measuring 25-hydroxyvitamin D in a clinical environment: challenges and needs
Am J Clin Nutr
 , 
2008
, vol. 
88
 
suppl 2
(pg. 
507S
-
510S
)