Factors affecting vitamin D status in different populations in the city of São Paulo, Brazil: the São PAulo vitamin D Evaluation Study (SPADES)

Hypovitaminosis D is well documented, mainly among elderly populations in countries from the northern hemisphere [1–4]. For the elderly, hypovitaminosis D stimulates the parathyroid gland, both directly and indirectly, leading to secondary hyperparathyroidism, bringing consequences such as bone mass loss and an increased risk for fractures [3]. For the young population this correlation is not clear. The concentrations of 25-hydroxyvitamin D [25(OH)D] are influenced by several factors, such as the age, skin tone, latitude, time of the day, season, the weather and many others. However, some authors evidenced that, different from what is expected, people from southern European countries have lower concentrations of 25(OH)D than people from northern countries [1, 5]. These data suggest that habit-related factors also influence 25(OH)D concentrations and these have not been investigated in countries where sunlight is abundant, such as Brazil.

This study was named SPADES (The S ão PA ulo Vitamin D E valuation S tudy) and its main objective was to recognize the factors that contribute to determining blood 25(OH)D concentrations in four different population groups in the city of São Paulo, Brazil (23°34’S, subtropical weather).

The blood samples were taken after an eight-hour fast. Ionized calcium, creatinine, albumin, 25(OH)D and intact parathyroid hormone (PTH) were measured. All serum samples were collected into refrigerated tubes, processed in refrigerated centrifuges, and frozen at −20°C until measured. The ionized calcium was measured after centrifugation.

The 25(OH)D concentrations were determined by an immunoradiometric assay (Nichols Institute Diagnostics, San Juan Capristrano, CA, USA). Intra-assay coefficient of variation was 4.8%, and inter-assay coefficient of variation was 16.0% for the lowest values (mean: 35.5 nmol/L) and 3.0% for the highest control (mean: 154.0 nmol/L).

To certify that the radiation conditions were similar through the years, we compared ultraviolet radiation during 4 consecutive years, and the UV year-pattern was repetitive (p = 0,444, unpublished data).

The exponential formula that predicts PTH values from 25(OH)D was created using the software Origin 5.0 (Microcal Inc, Northampton, Massachusetts, USA). This same program was used for generating Figure 1.

The laboratory results obtained for the group as a total and also separated by gender and subgroup of origin are described in Table 1. Men presented higher creatinine, creatinine clearance and 25(OH)D values, when compared to women, but the sex ratio between groups differs significantly; there are more women in NH (72.3%) and CD (69.1%) groups than in the PA (52.5%) and YOUNG (55.5%) groups. We observed that people in YOUNG group had a lower BMI and higher albumin and creatinine clearance values when compared to the other groups.

Only 6.3% of the total population studied had made use of vitamin D supplements (dose varying from 200 to 400 IU a day) and these had a 25(OH)D average blood concentration of 69.1 ± 34.9 nmol/L, while the group that had not taken a vitamin D supplement showed a significantly lower blood concentration (54.8 ± 33.4 nmol/L, p = 0.018).

Using the Pearson´s linear correlation and considering only individuals who were not using vitamin D supplements, significant correlations were found between 25(OH)D and age (r = −0.43, p = 0.001), PTH (r = −0.38, p = 0.001), albumin (r = +0.32, p = 0.001), creatinine clearence (r = +0.31, p = 0.001), ionized calcium (r = +0.30, p = 0.001), month of the year (r = −0.29, p = 0.001) and BMI (r = −0.21, p = 0.001). Creatinine did not show any significant correlation (r = +0.04, p = 0.203). Using a model of multiple linear regression, setting the age as the predictive variable and adjusting by ethnicity and gender, the following correlations were still significant: PTH (p = 0.01), ionized calcium (p = 0.048) and month of the year (p = 0.01).

Considering the PTH values, significant correlations were found with the following variables: age (r = +0.37, p = 0.001), 25(OH)D (r = −0.38, p = 0.001), albumin (r = −0.30, p = 0.001), creatinine clearence (r = −0.29, p = 0.001), creatinine (r = +0.13, p = 0.003), ionized calcium (r = −0.25, p = 0.001) and BMI (r = −0.13, p = 0.003). Using a model of multiple linear regression, setting the age as the predictive variable, 25(OH)D (p = 0.001), creatinine (p = 0.001) and ionized calcium (p = 0.011) were still significant.

The mathematical equation which predicts the PTH value, based on the 25(OH)D concentration is expressed by the formula PTH = 10 + 104.24.e^{-(vitD-12.5)/62.36}. We also defined the concentration of 75.0 nmol/L as the cut-off value for which 25(OH)D concentrations did not correlate with PTH levels (Figure 1, Table 2). Within the groups, 91.5% from NH; 87.6% from CD; 48.5% from PA and 40.3% from YOUNG fell below this concentration cut-off.

Within each age group, as age advances, progressively lower 25(OH)D and higher PTH concentrations were observed. 25OHD concentrations lower than 75.0 nmol/L were found for 85.7% of those older than 71 years, while the incidence of secondary hyperparathyroidism was 57% among those (Table 3).

Several reports have previously shown that the prevalence of hypovitaminosis D is high among the elderly in the United States and Europe (< 30 ng/mL), but there are few data from countries in South America, where sunlight is abundant throughout the year, such as Brazil [1–4]. The abundant sunlight would make us believe this was a problem restricted to countries of higher latitudes. However, Lips et al. and van der Wielen et al. demonstrated that countries from southern Europe show higher incidence of hypovitaminosis D when compared to northern countries [1–5]. Some recent reviews showed that the vitamin D deficiency is common in South Asia and the Middle East [13, 14].

The city of São Paulo is crossed by the Tropic of Capricorn (23°S) and presents relatively mild winters, with temperatures rarely achieving 0°C (32°F). It is a large urban center, being the largest one in South America, with a predominant Caucasian population. It has more than 10 million inhabitants and a large concentration of high-rise buildings and industries. During winter there is a significant concentration of air pollutants which can absorb solar UVB radiation and, consequently, reducing vitamin D formation [15]. Our elderly population has been increased similar to the pattern of developed countries. Between 1980 and 2003, there was an increment of 8.7 years in the life expectancy at birth, which nowadays is 71.3 years in Brazil.

Age was the variable that most significantly correlated with 25(OH)D plasma concentrations, as was demonstrated by a multiple regression model applied. Younger individuals had higher 25(OH)D concentrations (Table 3). For this reason, in the multiple regression vitamin D status analysis, we chose to use age as the predictive variable, adjusted by gender and ethnicity. Among the other parameters evaluated in the model, ionized calcium, PTH and month of the year, which are other variables that present known correlation with 25(OH)D, maintained significance [16, 17].

It is known that the 25(OH)D plasma concentrations follow a seasonal variation, demonstrated in places of more temperate weather [16, 18, 19]. Our data clearly demonstrate the existence of a seasonal pattern for 25(OH)D concentrations throughout the year that follows the ultraviolet radiation distribution and does not change over time. This phenomenon was evident in all the groups, except for the NH one (Table 4). In this group, although the group age was similar to the one in CD, the mean of the values in summer was not higher than that obtained in winter. This suggests that sunlight exposure was not enough for this group. On the other hand, it was observed that the subjects from CD reached values 34% higher in the summer, almost reaching the increment rate observed for young people, which was 47%.

Gannagé-Yared et al. demonstrated that there are influences from life habits on 25(OH)D concentration (in this study the use of veil was evaluated) [20]. Other authors demonstrated that the seasonal variation also exerts influences on bone remodeling markers, bone mass and fractures [19, 21]. In the same way, the individuals from PA group demonstrated a well defined seasonal variation when prospectively evaluated, except for the ones older than 71 years old who presented a pattern similar to that of the NH and CD groups [9]. An interesting fact is that the group PA had some sunlight exposure during their physical activities at least twice a week for 1 h (in this group the use of sunscreen was evaluated and no significant variation was found between users and non-users). These data suggest that if more frequently exposed to sunlight, the individuals younger than 70 years old can achieve an adequate vitamin D status [22–24]. This is probably one of the advantages of living in lower latitudes, where sunlight is more abundant.

In our study, it was possible to observe the correlation that exists between PTH and 25(OH)D, which was expressed through an exponential equation (Figure 1), the same way it has been described [20, 25, 26]. The concentration from which the correlation between 25(OH)D and PTH was lost was > 75.0 nmol/L (Table 2), a value similar to previous reports [27]. It means that above this concentration, sufficiency of vitamin D does not interfere with PTH secretion. We observed that the lower concentrations of 25(OH)D are associated with secondary hyperparathyroidism and that was more frequent in NH and CD (Tables 1, 4 and 3). Among the young individuals evaluated in this study, we found a higher 25(OH)D mean. However, for a good proportion of the samples the values were below the sufficiency limit, but besides that, no correlation with PTH was observed (r = −0.018, p = 0.876). Among young individuals PTH does not seem to be a good parameter to indirectly evaluate vitamin D status [10].

For the multiple regression analysis, setting PTH level as the depending variable and age as the adjusted predictive variable, the evaluated parameters that maintained the significance in the model were 25(OH)D, creatinine and ionized calcium, variables which are known to exert influence on PTH levels [28, 29]. Table 3 suggested that PTH levels increased with age, possibly due to several factors acting together, such as vitamin D insufficiency/deficiency, renal function decrease (1α-hydroxylation) and the influence of prescription drugs on vitamin D metabolism [29, 30]. Another important factor is the decreased ingestion and intestinal absorption of calcium [31–33].

The primary objective of hypovitaminosis D correction is promoting the maximum intestinal calcium absorption, reducing the frequency of falls and correcting the secondary hyperparathyroidism [27]. According to Bischoff-Ferrari et al., the supplementation with at least 700 IU of vitamin D/d was enough to reduce the risk of vertebral and non-vertebral fractures [34]. Besides that, there are other non-skeletal effects which have been attributed to vitamin D status, such as the prevention of cancer and autoimmune diseases [4]. In spite of all this knowledge, the number of individuals who received vitamin D supplements was very low. When the mean 25(OH)D concentration was evaluated for this group, we observed that although it was higher than the mean value for individuals who did not receive supplementation, it was still below the ideal limit (73% of these individuals presented concentrations < 75.0 nmol/L) [3, 4]. Therefore the vitamin D doses these individuals received (200–400 IU/d) were not enough for the prevention of hypovitaminosis D. The Endocrine Practice Guidelines recommends adults need 1,500-2,000 IU vitamin D/d to prevent vitamin D deficiency [35].

These results point out to the need for governmental public health action, which could implement measures such as vitamin D supplement distribution to nursing homes, food supplementation and medical education about the high prevalence of this condition among the elderly, considering the associated risk for fractures.

Positive aspects of our paper include that we evaluated a large number of adults from different populations collected in the same city and analyzed by the same method. It was possible to demonstrate that hypovitaminosis D is a public health problem even in Brazil. However what is still lacking is the relationship with vitamin D status with the amount of sun exposure.

Due to the high degree of miscegenation of the Brazilian population, it is always difficult to classify them ethnically. Nevertheless, most of the evaluated individuals were white (84.4%). According to the 2000 Brazilian Demographic Census for the city of São Paulo, the white population is the majority (67.0% of population, data from IBGE- Instituto Brasileiro de Geografia e Estatística). However, our data cannot be extrapolated to all different regions of the country, because, besides different latitudes and weather conditions, the population presents different genetic backgrounds and habits, which are important factors for the serum vitamin D status.

In addition, several other factors potentially affect vitamin D status such as genetic factors, adiposity, dietary intake, intestinal malabsorption and impaired hydroxylation in the liver and kidneys.

We found very low 25(OH)D concentrations for the elderly when compared to younger people. This finding is possibly due to age and habit-related sunlight exposure differences. The existence of a seasonal variation over 25(OH)D concentrations is evident, as well as the influence of age is. Secondary hyperparathyroidism was frequent among the older individuals and was associated with 25(OH)D levels, creatinine and ionized calcium. According to our data, PTH values tend to plateau above 75.0 nmol/L. We found that 73.3% of the sample population studied is below this value.

SSM – M.D., Ph.D. at the UNIFESP (Bone Metabolism section), and Professor at the Santa Casa de São Paulo Medical School

MLC – M.D., Ph.D. and Professor at the UNIFESP (Bone Metabolism section)