|Home | About | Journals | Submit | Contact Us | Français|
As vitamin D deficiency is considered to be more common in regions with little solar ultraviolet light in winter, the aim of the present study was to investigate predictors of vitamin D status by season within a large sample of male smokers from Finland, a country where there is negligible solar ultraviolet light in winter.
Vitamin D (measured by 25-hydroxyvitamin D (25(OH)D) nmol/L) and other serum constituents were assayed and measured anthropometry, and self-reported dietary intake and physical activity (PA) were obtained and analysed by step-wise multiple linear and logistic regression in 2,271 middle-aged Finnish male smokers.
Twenty-seven % of the population in winter and 17% in summer had serum 25(OH)D levels < 25 nmol/L, respectively. In summer, in multiple logistic regression analyses with adjustment for confounding and other predictors, high dietary vitamin D (OR=3.6; 95% CI= 1.5–8.5), some leisure time PA (OR=2.0; 95% CI=1.3–3.1) and having a body mass index (BMI) ≥ 21 kg/m2 compared to < 21 kg/m2 (OR=2.6; 95% CI=1.3–5.0) were associated with 25(OH)D ≥ 25 nmol/L. In winter, additional modifiable factors were occupational PA (OR=1.6; 95% CI=1.1–2.5), and high fish (OR=3.1; 95% CI=1.7–6.2) or poultry consumption (OR=1.7; 95%CI=1.1–2.7). Predictors from linear regression analyses of continuous levels of 25(OH)D were similar to the logistic regression analyses of 25(OH)D < 25 nmol/L.
In this Finnish sample more vitamin D intake, PA and having a BMI ≥ 21 may play important modifiable roles in maintaining an adequate vitamin D status.
The main and natural source of vitamin D is from its photochemical formation in skin by the action of solar UV light (Lips, 2006). Therefore maintenance of adequate vitamin D status in winter is a challenge; especially in higher latitudes where the sun’s angle lowers the intensity of UV wavelengths involved in vitamin D formation. The definition of vitamin D deficiency, indicated by the concentration of the vitamin D metabolite 25-hydroxy-vitamin D (25(OH)D) in blood, has recently been reappraised (Vieth et al., 2007). Although values below 25 nmol/L 25(OH)D have for many years, been considered deficient (Lips, 2006), recent clinical and epidemiological studies have indicated that a 25(OH)D concentration of ≥ 50 nmol/L is required for adequate vitamin D status (Vieth et al., 2007; Holick, 2007). Furthermore, when assessing subjects with cancer, blood concentrations of 25(OH)D ≥80 nmol/L have been considered optimal (Giovanucci, 2008).
Low levels of 25(OH)D have been associated not only with cancer (Giovanucci, et al., 2006; IARC, 2008) but also with increased risk of falling (Bischoff-Ferrari et al., 2004), cardiac disease (Luong & Nguyen, 2006), diabetes (Mathieu & Bandenhoop, 2005), respiratory conditions (Laaski et al., 2007), hypertension (Forman et al., 2008), multiple scelorosis (Cantorna, 2006), and immune disorders (Cantorna, 2006).
Low vitamin D status is of concern in the elderly (Lips et al., 2006), in certain ethnic groups (van der Meer et al., 2008) and in women at risk of osteoporosis (Burgaz et al., 2007). Studies of people with these conditions have shown that exposure to sunlight and, the supply of dietary vitamin D are major determinants of vitamin D status. Together with increased physical activity (PA), these factors have been identified in two large surveys of healthy adults in the USA (Freedman et al., 2007; Giovannucci et al., 2006) while social class, poor health, smoking and alcohol consumption have been reported as additional determinants in large national studies in the UK (Hirani et al., 2009). To our knowledge only one large study has specifically focused on the healthy middle-aged men in Europe (Hintzpeter et al., 2008). Although vitamin D deficiency appears to be more prevalent in Southern than Northern Europe (Ovesen et al., 2003), recent data from a nested case-control study of prostate cancer showed 25(OH)D levels in Finland to be lower than among their Nordic neighbors (Tuohimaa et al., 2004). In addition an early study from Finland reported very low 25(OH)D levels in young men (Lamberg-Allardt et al., 2001). Thus the aim of the present study was to investigate modifiable predictors of vitamin D status in both winter and summer within a large population of middle aged Finnish male smokers.
The Alpha-Tocopherol, Beta-Carotene Cancer Prevention (ATBC) Study was a randomized, double-blind, placebo-controlled primary prevention trial undertaken to determine whether supplementation with α-tocopherol, β-carotene, or both would reduce the incidence of lung and other cancers in male smokers. The rationale, design, and methods of the study, as well as the characteristics of the participants, have been previously described in detail (Anonymous, 1994). Participants were 29,133 male smokers (five or more cigarettes per day at entry), mean age of 58± 5 (50–69) years residing in southwestern Finland (latitude 60 °N–63 ° N). The exclusion criteria included history of cancer or other serious diseases; the use of vitamin E, vitamin A, or β-carotene supplements in excess of predefined doses; and treatment with anticoagulant agents. The current cross-sectional study includes 2271 control subjects from a number of case-control studies nested within ATBC (Faupel-Badger et al., 2007; Freedman et al., 1999; Lim et al., 2009; Stolzenberg-Solomon et al., 2006; Tangrea et al., 1997) as well as an ongoing multi-site study of vitamin D that consists of a subsample of the ATBC participants (personal communication SJ Weinstein)
Through baseline questionnaires, study participants provided self-reported information on general health characteristics, medical, dental, smoking and dietary history. Height, weight, and blood pressure were measured by trained nurses.
Diet was assessed by a validated self-administered comprehensive food frequency questionnaire (Pietinen et al., 1988) in which frequency and portion size of 276 food items consumed over the previous 12 months were reported; nutrient intake was assessed by linkage with the food composition database of the Finnish National Public Health Institute (Pietinen et al., 1988). Vitamin D fortification in Finnish food was minimal at baseline as it was only in 2003 when fortification of milk and sour milk was initiated (Valimaki et al., 2007). As vitamin supplement intake was minimal (11%), vitamin D intake was a combined estimate assessed from both food sources and supplements.
Fasting serum samples were collected during the pre-randomization baseline visit, and stored at −70 °C. Serum cholesterol levels were determined by CHOD-PAP methods (Boehringer-Mannheim) and alpha-tocopherol, retinol and beta-carotene were determined by HPLC (Anonymous, 1994). All serum 25(OH) D data from the ATBC substudies (Faupel-Badger et al., 2007; Freedman et al., 1999; Lim et al., 2009; Stolzenberg-Solomon et al., 2006; Tangrea et al., 1997) were assayed using radioimmunoassay methods, with the exception of one sub-study (Faupel-Badger et al., 2007) that also used the OCTEIA direct ELISA kit (IDS, Inc;) and the recent multi-site investigation (personal communication SJ Weinstein) which used a chemiluminescence assay (DiaSorin, Inc, Stillwater, MN). Intra-batch and inter-batch coefficients of variation for all assays ranged from 5.3%–16.5% and 8.4%–16.5%, respectively (Faupel-Badger et al., 2007; Freedman et al., 1999; Lim et al., 2009; Stolzenberg-Solomon et al., 2006; Tangrea et al; personal communication SJ Weinstein) Indicator variables identifying the sub-studies were included in all models to adjust for sub-study effects.
The categories of serum 25(OH)D were defined based on cut points for deficiency in the literature; i.e. deficient status(<25nmol/L) (Lips et al., 2006), marginal status (<37 nmol/L) (Vieth et al., 2007), <adequate status < 50 nmol/L (Holick, 2007) and for cancer risk < 80nmol/l (Giovannucci et al., 2006). For each individual, daily intakes of vitamin D (in μg/day) and calcium (in mg/day) from both diet and supplements were calculated from dietary assessment and supplement use. Information from the date of blood draw was categorized according to two seasons for stratification purposes (summer: weeks 20–45 & winter: weeks 0–19 & 46–52 (Figure 1)). Information on leisure time physical activity was defined using the question ‘How often do you engage in leisure time PA’ and was dichotomized to ever/never. Information on “on-the job” physical activity was dichotomised to jobs involving heavy or moderate lifting or not. BMI was calculated by dividing weight (in kg) by squared height (in meters). Information on vocational status was included in the analyses as education after high-school either < or ≥ 2 years. Residential status was defined both by size of the municipal area where the subject lived and whether the area had some sea coast as a border.
As the aim of this study was to identify predictors of 25(OH)D levels, initial data screening was performed in order to identify statistically significant and biologically meaningful variables associated with continuous and categorical vitamin D status. T-tests for continuous variables and chi-square tests for categorical variables were used to determine statistical significance with two-sided p-values less than 0.01. Those unadjusted factors found to be significant were then included in a forward stepwise multiple linear regression analysis in order to ascertain the independent predictors of serum 25(OH)D. This and future models included age and smoking as they are known to be related to 25(OH)D levels (Lips, 2006; Holick, 2007; IARC, 2008). In addition as α-tocopherol was adjusted for serum cholesterol in the data analysis, cholesterol was also included as a confounder. Total dietary energy was also added as a confounder in order to adjust individual dietary intakes for total energy intake.
In the linear regression serum 25(OH)D natural log-transformed to normalize its distribution. The variables which remained significant (p<0.01) after stepwise linear regression modeling, were then included in two other linear regression models, stratified by season (summer and winter). Categorical logistic regression analyses (25 nmol/L 25(OH)D) were also performed stratified by season, using the same predictor and confounding variables as in the linear regression. All dietary intake and serum nutrient variables were categorized into quartiles of individuals. Linear trends of ordered categorical variables (e.g., categories of BMI and dietary vitamin D) were assessed using the continuous values of the variable and using a likelihood ratio test (Breslow & Day, 1980). All statistical analyses were performed using the SPSS 15 statistical package.
Serum 25(OH)D levels varied by week of blood collection, with the highest levels during the summer and autumn months and the lowest levels during winter and spring (Figure 1). Twenty-seven % of the population in winter and 17% in summer had serum 25(OH)D < 25 nmol/L respectively. These percentages were 75% and 62% respectively, for a cut-point of < 50 nmol/L 25(OH)D
In Table 1 unadjusted factors significantly (p<0.01) associated with higher 25(OH)D levels were: those whose blood was drawn in the summer months, living in coastal or urban areas, higher vocational achievements, more leisure time physical activity (PA), BMI ≥21, higher intake of fish, poultry, alcoholic spirits, vitamin D (including supplements), protein, polyunsaturated fatty acids, N-3 fats from fish, eicosapentaenoic acids, β-carotene and higher serum levels of retinol, β-carotene and α-tocopherol (cholesterol-adjusted). Missing more than ten teeth and dietary calcium intake were significantly associated with lower 25(OH)D levels.
The factors which remained significant when entered into the stepwise multiple linear regression (adjusting for confounders) against natural log-transformed 25(OH)D nmol/L were: summer season of blood collection, coastal residence, greater leisure time PA, BMI ≥ 21 kg/m2, greater dietary intake of fish and poultry, higher vitamin D intake, and higher serum α-tocopherol (cholesterol-adjusted), retinol and β-carotene (all positive) and missing teeth (negative). When stratified by season, differences emerged in the relative value of these predictive factors (log-transformed: total r2 = 0.290; summer r2 = 0.345 and winter r2 = 0.243)
In a further investigation of seasonal difference Table 2 presents results of logistic regression analyses of the factors which are associated with 25(OH)D deficiency (i.e., the odds of being ≥25 nmol/L 25(OH)D). In summer, having no versus leisure time physical activity (OR=2.0; 95% CI = 1.3–3.1), a BMI ≥ 21 kg/m2 versus < 21 kg/m2 (OR= 2.6; 95% CI = 1.3–5.0) and a high versus low vitamin D intake (OR=3.6; 95% CI = 1.5–8.5) were the strongest factors associated with lack of vitamin D deficiency. In winter, the association with dietary vitamin D intake was similar to summer values but the effect of having some leisure time PA was less evident and increased BMI had no effect (OR ≥25 nmol/L 25(OH)D: a high versus low vitamin D intake = 3.6; 95% CI = 1.8–7.3; having some versus no leisure physical activity OR = 1.3; 95% CI = 1.0–1.9; BMI ≥ 21 kg/m2 versus < 21 kg/m2= 1.0; 95% CI = 0.5–1.9). Although not a significant predictor over the whole year (Table 1), occupational activity which was of greater intensity versus light activity was associated with ≥25 nmol/L 25OHD, especially in winter (OR = 1.6; 95% CI = 1.1–2.5). Another difference in winter was that higher consumption of fish or poultry (adjusted for vitamin D intake), and having higher serum levels of retinol and α-tocopherol, were more strongly associated with vitamin D deficiency than in the summer months.
Estimates from linear regression analyses (Appendix 1) were similar to all the above for both summer and winter. Figure 2 presents a non-linear association between BMI and 25(OH)D. It appears in these data that this distribution approximates an inverted U shape with either very low or very high BMI measures associated with lower levels of 25(OH)D (< 40 nmol/L).
In this sample of Finnish male smokers, lifestyle contributors to higher 25(OH)D levels in summer were high dietary vitamin D intake, increased leisure time PA, and a BMI ≥ 21 kg/m2. In winter a greater intensity of occupational PA and greater fish and chicken consumption were associated with increased 25(OH)D levels. Higher vitamin D status was also positively associated in those whose blood draw was in summer, had higher vocational achievements, had more teeth and lived near the coast.
Mean 25(OH)D levels in our study were comparable to those in most other reported Finnish populations (Laaksi et al., 2007; Tuohaimaa et al., 2004; Lamberg-Allardt et al., 2001; Knekt et al., 2008; Valimaki et al., 2004) but slightly lower than in other Nordic countries (Holvik et al., 2006), Germany (Hintzpeter et al., 2008) and Great Britain (Hypponen et al., 2008).
Our observation that 25(OH)D serum levels are higher in summer and autumn than in winter and spring is consistent with other studies where season of blood draw was ascertained (Hintzpeter et al., 2008; Freedman et al., 2007; Jones et al., 2005).
Dietary vitamin D has consistently been reported as a determinant of vitamin D status in Nordic and northern European countries (Burgaz et al., 2007; Hintzpeter et al., 2008; Oveson et al., 2003; Tuohimaa et al., 2004; Chapuy et al., 1997; Huotari et al., 2008). In our data, both consumption of fish and total vitamin D intake (including supplements) were predictors of biochemical vitamin D status; this appears to be especially important for fish consumption in the winter months. Predictors may be quite different in populations in which the diet is based less on fish products than is the typical Finnish diet. Also, in winter, poultry but not dairy products (this study was undertaken before vitamin D was added to milk products in Finland), was a predictor of vitamin D status. This could be due to the common practice of supplementation of poultry feed of battery bred chickens with vitamin D; cholecalciferol is incorporated into poultry feed at a concentration of 62.4ug/kg (Mattila et al., 1999). Thus it is feasible that the consumption of meat from intensively reared broiler chickens would be supplying more dietary vitamin D than meat from animals not receiving such dietary supplements. It should be noted that after 2003 when vitamin D was added to milk products in Finland, the mean winter-time serum 25(OH)D concentrations increased by 50% (Valimaki et al., 2007).
Our data seem to indicate an unadjusted association between higher 25(OH)D values and lower dietary consumption of calcium. However it must be noted that these differences disappear once adjustment for confounding is taken into account (especially poultry and fish); we cannot suggest any mechanistic explanation for the negative association.
We also found leisure time PA to be a contributor to vitamin D status, consistent with other studies that measured PA (Scragg & Camargo, 2008). This association has often been attributed to PA being a surrogate for sun exposure; however, in the few studies in which both exposures were measured simultaneously (Jones et al., 2005; Brock et al., 2007; Scragg et al., 1992), the PA-vitamin D relationship persisted, independently of sun exposure. It is interesting that occupational PA which was regarded as of high intensity was associated with deficiency (≥25 nmol/L 25(OH)D), especially in winter. Maybe some aspect of exercise is contributing to the maintenance of vitamin D status, other than by increasing exposure of skin to sunlight, as suggested by Bell et al., (1988). Further investigation into the role of PA independent of sunlight exposure in vitamin D metabolism would appear warranted
Our findings of lower (<40 nmol/l 25(OH)D) among men with very low BMI are also of interest (Figure 2). This inverted U-shaped relationship was also recently reported in two large studies of English men (Hirani et al., 2009) and German men (Hintzpeter et al., 2008), as well as in a smaller study of young Finnish men (Laaksi et al., 2007). These findings need further investigation. BMI is not just an indication of fat mass, especially at low levels. As BMI rises from very low levels to normal range, there may be an increase in lean tissue mass, which is mainly muscle. One explanation for the low BMI 25(OH)D relationship may be just that low 25(OH)D may be a correlate for poor health or illness. The low 25(OH)D and obesity relationship has been explained by “trapping” of the vitamin D parent compound, cholecalciferol, in adipose tissue (Looker, 2007; Wortsman et al., 2008).
To our knowledge, this is the first report of residence near to the coast being a predictor of higher vitamin D status in a population study although geographical location or latitude have been reported to be associated with vitamin D status (Giovannuci et al., 2006; Freedman et al., 2007), but not consistently (Ovesen et al., 2003; Binkley et al., 2007; Hagenau et al. 2009; Kimlin & Schallhorn, 2004). One explanation for the fact that Northern Europeans have higher vitamin D levels than their compatriots in the in the South is their high fatty fish consumption (van der Meer et al., 2008). Another reason may be that Northern Europeans expose more unprotected skin to the sun (Burgaz et al., 2007; Ovesen et al., 2003). Both factors could also be associated with coastal residence; alternatively, the latter could be an independent surrogate for intensity or amount of sun exposure. One hypothesis is that involvement in sports such as sailing and fishing could increase vitamin D by sunlight reflection from water.
The finding that lack of teeth is a predictor of vitamin D deficiency confirms other reports of vitamin D status and dental health (Grant, 2008). This finding could be explained by either the role of vitamin D in calcium metabolism or dentition being a correlate of poor health and lower social status, as has been reported recently in the UK (Hirani et al., 2009). Another explanation may be that lack of teeth is a marker for lower immunity (Cantorna, 2006).
The suggestive relationship between low vitamin D status and lack of vocational achievement, may reflect the role of socioeconomic status found in recent data from the UK that found that low income/material deprivation was associated with low vitamin D status in two nationally representative samples (Hirani et al., 2009). When vocational achievement was stratified by season, it appeared that higher levels were slightly more related to vitamin D status in summer, possibly an indication of differences in diets or of summer travel.
Our observed associations of vitamin D levels with serum retinol and α-tocopherol are relatively new and may be chance findings. A small study in Spain has also found low levels of α-tocopherol associated with low 25(OH)D but they found high blood retinol levels to be associated with low 25(OH)D, a result inconsistent with our findings (Mata-Grandos at al., 2008). Both these results are surprising as blood retinol concentrations in adults are fairly constant and only fall when liver reserves of retinol are nearly depleted. As α-tocopherol and β-carotene concentrations are mainly related to dietary intake (and in the case of α-tocopherol, serum cholesterol) of foods, one hypothesis may be that may be that maybe that these serum values are markers for foods high in vitamin D (eg liver or cod-liver oil). As serum was obtained prior to randomization into the trial, no impact of the α-tocopherol and β-carotene supplements would be expected.
Our study is limited, as many other studies have been, by not having a direct measure of sun exposure or indoor or outdoor PA patterns. Week of blood draw was used as a proxy measure of the effect of sun exposure, and data was stratified by winter and summer season. Despite these measures we cannot exclude misclassification with respect to sun-exposure and possible bias of effect estimates in the regression models because of residual confounding. A further limitation may be the possible measurement errors in the dietary assessment. Another limitation, as with all studies using 25(OH)D levels, is that there are still problems in comparing different laboratory measures of 25(OH)D across studies (Lips, 2006; Millen & Bodnar, 2008); however we have included indicator variables to adjust for laboratory effects in the statistical analyses Strengths of the present investigation are its large sample size and having individually measured BMI and detailed and validated dietary intake.
In conclusion, this study takes advantage of a window of opportunity to study data available from a country with low winter sunlight when vitamin D had not been routinely added to any basic foodstuff. These analyses indicate that higher fish or dietary vitamin D intake, greater PA, and maintenance of adequate BMI were factors with the potential to improve low vitamin D status. The interesting inter-relationship among these factors in times of limited sun exposure needs further investigation.
This research was supported in part by the Intramural Research Program of the NIH and the National Cancer Institute. Additionally, this research was supported by U.S. Public Health Service contracts N01-CN-45165, N01-RC-45035, and N01-RC-37004 from the National Cancer Institute, Department of Health and Human Services.
All authors reviewed the final manuscript before submission, and none had a conflict of interest with regard to this work.