Epidemiological studies in various populations have raised public health concerns because of the high prevalence of vitamin D (VitD) insufficiency and deficiency (1–9). Since it plays an important role in human physiology, low serum VitD concentrations may pose a health risk (2, 10). Vitamin D is a fat-soluble vitamin involved in calcium and phosphorus homeostasis and is therefore critical for bone health (11). Poor VitD status leads to rickets in children and osteomalacia in adults. Other studies have linked VitD deficiency to immune system function, non-skeletal health problems, disease development, and lower life expectancy, although causal evidence from randomized controlled trials is still limited (12, 13).

Sources of VitD for the human body are dietary intake and skin exposure to ultraviolet B (UVB) radiation with wavelengths of 290–315 nm. Exposure to sunlight is thought to account for up to 90% of the body’s VitD supply, but during the winter months sun exposure in most European countries does not result in the formation of VitD due to the low intensity of UVB radiation (14–16). The amount of VitD formed in the skin by UVB radiation depends on various environmental and personal factors, such as latitude, season, time of day, skin pigmentation, age, or use of sunscreen (17–23). It has already been established that constitutive skin pigmentation and/or skin phototype is an important predictor of VitD status when comparing people from different ethnic backgrounds (24–29), but how different levels of constitutive pigmentation affect VitD status in Caucasians has not yet been well described. To date, there has been only one study focusing on Caucasians, and its results contrast with studies of different ethnic backgrounds, where darker skin types seem to be more prone to deficiency (30).

Vitamin D is formed after UVB exposure to the skin; it enters the bloodstream in a complex with a D-binding protein and is later hydroxylated in the liver to 25-hydroxyvitamin D [25(OH)D], the most common biomarker for assessing VitD status. This form is later hydroxylated in the kidneys to 1, 25 hydroxyvitamin D, which is the active form of VitD (31). The excess VitD produced is stored in adipose tissue and/or partially degraded (32). Melanin pigment, which is crucial for skin color, may interfere with VitD photoproduction in the skin because it acts as a natural filter of UVB radiation in the epidermis, and thus competes for UVB photons with 7-dehydrocholesterol (7-DHC), a substrate for VitD synthesis (32).

Although there is no consensus on the recommended serum concentration of 25(OH)D, the definition of VitD deficiency generally includes serum concentrations of 25(OH)D that fall below 30 nmol/L (10–12 ng/mL) (33–36); concentrations below 50 nmol/L (20 ng/mL) are considered insufficient (33, 35–37), whereas concentrations above 75 nmol/L (30 ng/mL) are considered optimal by the Endocrine Society (34), In Slovenia (latitude 45° to 46° north) and in countries with similar or higher latitudes, UVB radiation in winter (November–April) is insufficient to stimulate VitD synthesis (3). In Slovenia a high prevalence of low VitD status with remarkable seasonal variations has been reported (1), highlighting the need for effective public health interventions.

In foods, VitD is mainly found in the form of ergocalciferol (D2) and cholecalciferol (D3) (38) and their derivatives. Foods may be a natural source of VitD or may be enriched with this vitamin. There are few natural dietary sources of VitD (e.g., fatty fish, fish liver oil, and sun-dried mushrooms), and they are not commonly consumed (5, 7). As a result, most VitD intake is achieved by foods that contain rather small amounts of VitD but are consumed more frequently, such as eggs and cheese (7, 39, 40). On the other hand, food fortification and bioaddition can be important sources of VitD, which may be voluntary or mandatory (41). The term fortification usually describes the process of adding D2 or D3 before or at the end of food processing (42), whereas the term bioaddition describes the process of increasing naturally occurring VitD during food production by feeding animals with VitD-enriched feed (for meat and eggs) or by irradiating fungi or yeast with UVB (42). Other sources of this vitamin include VitD-containing prescription medicines and food supplements.

Dietary intake of VitD is often far below recommended levels – both globally and in Slovenia (40, 43, 44). The daily reference value (DRV) for VitD intake (without endogenous synthesis) for the adult population is 15 μg/day as recommended by the European Food Safety Authority (EFSA), and 20 μg recommended by the German, Austrian, and Swiss Nutrition Association (D-A-CH) (36, 45). In contrast, the Nutritional Reference Value (NRV) specified in the EU food labeling regulations for VitD is 5 μg. However, we should note that EFSA sets DRV for the VitD as adequate intake, so that 97.5% of population avoid deficiency (46), but not insufficiency.

In the light of the alarming prevalence of regional VitD deficiency (1, 40, 44), we designed a study with aim to examine the factors affecting VitD status in the Caucasian population in winter and summer, focusing on objectively measured skin parameters and the consideration of a wide range of personal and lifestyle indicators. The results of present study will be implemented into new policies for public health interventions.

A total of 292 participants completed the study, of whom 38% were male and 62% were female. Enrolment took place in 2019 (70.2%) and 2020 (29.8%). The mean age (standard deviation) of participants at enrolment was 38.2 (±11.8) years, with a median of 37 years. Overall, 33.6% of the participants were overweight or obese, and the mean BMI was 24.1 (±SD 3.3). Insufficient or deficient wintertime VitD status (January) was found in 63.4% of the participants, with mean 25(OH)D concentration 44.5 nmol/L (±SD 17.2). In the summertime 50% of participants had below sufficient status, and the mean 25(OH)D concentration was 80.0 nmol/L (±SD 24.8). The mean (±SD) dietary intake of VitD was 2.9 ± 1.7 μg (median 2.5 μg), with the maximum intake of 12.72 μg per day. Other characteristics of the population are shown in Tables 1, 2.

Factors affecting VitD status in the winter and summer seasons were examined using binary regression analysis. We calculated the seasonal OR, OR using a winter model (Model W), and two summer models (Models S1 and S2). In the summer Model S1, we added objectively quantifiable measurements of the difference in the melanin index between constitutive and facultative skin pigmentation as an indicator of sun exposure level. In the summer Model S2, we added self-reported variables: number of holidays, how often and what SPF factor they use, hours spent in full sun, and whether they wear long sleeves as a protective measure against the sun.

Significant differences in crude ORs for insufficient VitD status in winter were found for the following variables: BMI (higher in obese individuals); seasonal differences in 25(OH)D concentrations (higher in individuals with greater differences); and ITA (lower in tan and intermediate subjects) (Table 3). Daily VitD intakes greater than 5 μg showed a nonsignificant trend (p < 0.1) of lower OR intake of insufficiency. Model W included all the variables used in the crude calculation except IPAQ and showed similar OR to the crude calculations. The highest ORs was found for BMI and the lowest for ITA°. Obese participants were 7.92 (95% CI: 2.11, 29.67) times more likely to be VitD insufficient than normal or underweight participants, while individuals in the ‘intermediate’ and ‘tan’ ITA categories were significantly less likely to have VitD insufficiency than individuals in the ‘light’ ITA category (which is the most common in Slovenia).

For summer, in addition to the crude analysis, two models were configured, one based on measured variables (Model S1), and another based on self-reported variables (Model S2). Sex was significant only in the crude analyses, but not when adjusted with other factors (Models S1 and S2) (Table 3). On the contrary, daily VitD intake was not significant as crude value, but adjusted with other factors in Model S1 and S2, became significant (p < 0.05). In both summer models, the year of observation influenced VitD status. The participants observed in 2020, which included the first summer of the Covid-19 pandemic, were approximately twice as likely to have sub-optimal VitD status. In Model S1, the difference in melanin index (MI) between the inner and outer upper arm was a significant variable (analyzed as a continuous variable); at the same time ITA remained significant in individuals with ‘tan’ skin [0.29 (95% CI: 0.14, 0.63)], when compared to reference ‘light’ skin. The odds ratio for BMI remained highest for the obese subjects at OR 4.4 (95% CI: 1.46, 13.28). The second summer model (Model S2) included variables that were significant in the crude analyses. The analyses showed that the odds ratios were consistent with Model S1 for BMI (OR 5.33; 95% CI: 1.71, 16.59) and the ITA for ‘tan’ (OR 0.33; 95% CI: 0.15, 0.73). In this model, the largest odds ratio for deficiency of VitD was attributed to wearing long sleeves as a protective measure against the sun. Those who reported wearing long sleeves were 5.98 (95% CI: 1.33, 27.03) times more likely to have suboptimal VitD status. The number of hours spent in full sun was also significant.

We also examined parameters related to interindividual differences in serum 25(OH)D concentration between winter and summer. Figure 1 presents a plot of summer and winter serum 25(OH)D concentration, while the results of the repeated-measures ANCOVA analyses are shown in Table 4. Statistically significant predictors in the objective model C1 were BMI, year of observation, and ITA*ΔMI, while sex was non-significant with p = 0.057. In the subjective model C2, predictors of seasonal difference in serum 25(OH)D concentration were BMI, sex, ITA, and number of holiday days.

Public health interest in vitamin D is increasing due to its versatile body functions and the high prevalence of deficiency worldwide (3, 7). For example, a nationally representative Slovenian Nutrihealth study showed that approximately 80% of adults have insufficient VitD status in winter and 60% have suboptimal VitD status in summer (1).

While several studies have addressed VitD status in population-based cross-sectional studies [see a very recent review by Cashman on this topic (61)], few studies have followed seasonal changes in serum 25(OH)D concentration with a prospective cohort study design, and have generally focused on specific population subgroups or had limited sample sizes. The largest prospective cohort study was conducted by MacDonald et al., (62) who studied 314 Caucasian postmenopausal women and emphasized the importance of BMI. Andersen et al. (63) examined seasonal changes in VitD status in 54 Danish adolescent girls and 52 elderly women, focusing on both sun habits and VitD supplementation. Pittaway et al. (64) studied 91 older subjects (71% women, aged 60–85 years) in Tasmania and also noted the importance of VitD supplementation; supplement use was associated with mostly smaller seasonal variations in serum 25(OH)D concentration. Brustad et al. (65) followed VitD status in 60 Norwegian adults (73% women, age 20–60 years) with bi-monthly observation time and reported that high dietary VitD intake largely masks seasonal variations. On the other hand, Wolman et al. (66) included only subjects who supplemented VitD, but their study was very specifically focused on elite ballet dancers (n = 19; 65% women); the authors reported notable seasonal differences in VitD status, but did not focus on skin parameters. Our literature search did not identify any prospective cohort studies addressing seasonal changes in VitD status in the general adult population and using objective methods to measure skin color and pigmentation. VitD supplementation is a known major parameter linked to serum 25(OH)D concentration (67), therefore we wanted to focus on other parameters related to VitD status, and to exclude the effects of supplementation by study design.

The current study therefore used a prospective cohort design with measurements during the winter and summer seasons in subjects who did not supplement with VitD. Although VitD also occurs naturally in foods, the average daily dietary intake of this vitamin in Slovenian adults is only 2.9 μg (40) – much lower than typically supplemented dosages (25 μg) (68, 69). Nevertheless, we also followed the dietary intake of VitD with foods using a validated food frequency questionnaire (51). Serum 25(OH)D concentration was measured as a marker of VitD status, as were other characteristics, such as skin pigmentation and sun behavior. The strength of the study also lies in the use of objective measures of skin color and pigmentation. Using ITA, we were able to account for different levels of constitutive skin color in the Caucasian population. Given the study objectives, we recruited subjects with a wide variability in the parameters studied, and therefore did not use a nationally representative sampling approach. Overall, 62% of the study participants were women. BMI was self-reported, and 66% of participants fell into the normal/underweight category, 25% into the overweight category, and 6% into the obese category, the mean BMI of the entire population was 24.1 (SD ± 3.8).

Although the observed prevalence of inadequate VitD status should not be generalized to the population, it is interesting to compare the observed prevalence with the nationally representative Nutrihealth (NH) study (1). The mean 25(OH)D concentration in January was 44.5 nmol/ L comparing to 36.7 nmol/L in NM study (winter time), for the September, the mean concentration was 80.0 nmol/L, comparing to 70.4 n mol/L in NH study (summer time)The prevalence of suboptimal VitD status (25(OH)D concentration < 75 nmol/L) in our study was 95.2% in winter (NH: 98.0%) and 50.0% in summer (NH: 62.6%). More notable differences were observed in the prevalence of insufficient VitD status (<50 nmol/L), which was 63.4% in winter (NH: 81.6%) and 5.5% in summer (NH: 25.3%). These differences can be explained not only by the recruitment approach and the study population, but also by the observation periods. While the Nutrihealth study recruited subjects over the entire calendar year, our study focused on two narrow sampling periods – January for the winter season and September for the summer season. This approach was chosen to minimize within-season variability in environmental factors.

Regression analyses modeling was used to identify the most important factors affecting seasonal VitD status in our population group. Large differences in winter and summer VitD status, therefore the analyses were made with different cutoff targets for each season. The cutoff value was set at <50 nmol/L (insufficient status) for winter, and at <75 nmol/L (suboptimal status) for summer, because of the seasonal differences in environmental factors and human behavior, the regression models were constructed as season specific. We used one winter model (Model W), which did not include variables related to sun exposure/protective behavior, and two summer models (Model S1 and S2): the first considering only objectively measured variables, and the second taking into consideration subjectively measured parameters. All the models included the usual variables that may influence VitD status, such as sex, BMI, VitD intake, and skin color category (ITA), along with seasonal differences in 25(OH)D concentration. Considering the differences in environmental and lifestyle factors, the year of observation was also included in all the models. In this regard, it should be noted that the second year of the study coincided with the Covid-19 pandemic and its associated restrictions, which affected the behavior of the subjects during the summer sampling period. Given the prospective study, we also evaluated seasonal differences in serum 25(OH)D concentration, using repeated measures analyses.

Our results show that BMI is a strong predictor of VitD status in both winter and summer periods. Compared with normal/underweight individuals, obese individuals were several times more likely to have insufficient VitD status in winter (model W; OR 7.9), and suboptimal VitD status in summer (Models S1 and S2; OR 4.4 and 5.5, respectively; Table 3). BMI was also confirmed as a significant variable in repeated measures analyses of seasonal changes in serum 25(OH)D concentration (Table 4). BMI has previously been described as a predictor of VitD status (25, 26, 28, 62, 67, 70–72) and of seasonal differences in serum 25(OH)D concentration (62). Several factors could contribute to the lower 25(OH)D concentration in individuals with a higher BMI, including a larger pool for the accumulation of VitD in adipose tissues, and lifestyle covariates (i.e., less outdoor activity due to lower mobility), but the association remains controversial (71, 73). In conclusion, BMI is a strong predictor, but we should interpret results with some caution since BMI was self-reported variable.

In our study, sex was not found to be a significant predictor of VitD status either in winter or in summer (after adjustment for other factors; Table 3), but there was an association with seasonal changes in serum 25(OH)D concentration (Table 4). We could find no other literature with prospective cohort data on this topic, and data from cross-sectional studies vary widely. In some studies, although sex was reported as a significant predictor of VitD status, the direction of the effect may be in the opposite direction (61). However, in the NH study (1) and some other studies (25, 26), lower VitD status was found in women.

Daily dietary intake of VitD (with food) was low (median 2.5 μg) and comparable to epidemiological data for Slovenian adults (NH: 2.7 μg) (40, 44). Also, the maximum intake was 12.72 μg per day, which is well below the recommended intake of 20 μg per day (45). Nevertheless, dietary VitD intake showed tendency (p = 0.057) to be significantly correlated with VitD status as a sole variable. Surprisingly in the summer models, the higher dietary VitD intake was correlated with optimal VitD status (Models S1 and S2). This was unexpected, since usually VitD intake is bigger predictor of VitD status in winter, in the absence of endogenous VitD synthesis which is the major contributor to VitD source (14–16). Literature shows, that supplementation with higher doses of VitD is a predictor of personal VitD status, whereas the effect of lower dietary VitD intake with foods is not consistent across studies (63, 74).

In our study, we used ITA as an objective and quantitative method to classify constitutional skin pigmentation. It has six classes ranging from ‘very light’ to ‘dark’ (56), but because the study was conducted in Slovenia with a Caucasian population, our sample did not include the ‘brown’ and ‘dark’ categories. ITA is used as an objective alternative to the subjective Fitzpatrick skin phototype (58), which is based on self-reported pigmentation and solar reactivity of the skin (50), and has been used as a measure of skin pigmentation in several studies (75). It should be noted that the ITA and the Fitzpatrick phototype (FT) classifications are not completely equivalent; whereas the FT is easier to use in a clinical setting, ITA is preferred for its objective classification of skin type in research (58, 76, 77). The analyses were performed using the base category of ‘light’, which was the most common in our study group. The odds ratios for lower VitD status decreased with darker skin pigment (lower ITA) in both winter and summer (Table 3). The lowest OR for insufficient VitD status (winter Model W) or suboptimal VitD status (summer Models S1 and S2) were observed in the ‘tan’ category: 0.28, 0.29, and 0.33 for Models W, S1, and S2, respectively. These observations add new and interesting findings to the existing knowledge. Previous studies that also considered skin color type compared individuals of different ethnicities and showed that individuals with more pigmented skin living in the same conditions had lower VitD status compared to Caucasians (24–28). To our knowledge, only one study has examined correlations with skin color in the Caucasian population: Dimakopoulos et al. (74) conducted a cross-sectional examination of factors affecting serum 25(OH)D concentrations and showed similar trends – the likelihood of lower VitD status was inversely related to darker skin. The observed results suggest that darker constitutive skin color in the Caucasian population is related to better VitD status, but there is likely a tipping point at which darker constitutive skin color no longer works in favor, but limits the production of 25(OH)D. It should also be noted that darker skin types tend to be less sun-reactive than lighter skin types, i.e., they have lower erythema sensitivity and better tanning ability; consequently, these individuals may be more susceptible to riskier sun habits and less likely to use sun protection measurements, (78) which could also contribute to lower odds for VitD deficiency.

In our study the melanin index was used as a measure of pigmentation; the difference between constitutive and facultative pigmentation was measured as ΔMI to objectively assess the degree of sun exposure of individuals and was included in the objective models S2 and C2. Dosimeters can be used to objectively quantify sun exposure, but their usage in larger and longer-term studies is challenging (63), whereas ΔMI can be easily measured during a single visit without burdening subjects on a daily basis during the research. The ΔMI was found to predict both seasonal change in serum 25(OH)D (Table 4; Model C2), and VitD status in summer (Table 3; Model S2).

To gain further insight, we also examined a set of self-reported variables related to peoples’ behavior and lifestyles (Models S1 and C1). In contrast to some other studies (25, 26, 70, 74), physical activity level was not found to be a significant predictor of VitD status, and therefore was not included in the modelling. On the other hand, the number of summer holiday days and the use of SPF were significant in the crude analyses of OR for summer low VitD status, but not when adjusted with other factors. We could hypothesize that this is also due to the interdependence of these two factors. On the other hand, the number of summer holiday days was strongly associated with individual seasonal changes in serum 25(OH)D concentration (Table 4; Model C2, p < 0.001). This was not the case with the use of SPF (p = 0.14). Some other studies also found no association between the use of SPF and VitD status (79), or sunscreen use was found to be a predictor of serum 25(OH)D, but normal sunscreen use was not associated with insufficient VitD status (80). Another subjectively reported parameter examined in our study was the hours of sun exposure between 10 am and 4 pm during summer, when the intensity of UV-B light is strong enough for VitD biosynthesis (81). This parameter was marginally significantly associated with seasonal changes in serum 25(OH)D concentration (Table 4; Model C2, p = 0.078), but significantly associated with summer VitD status (Table 3, Model S2, p < 0.05). Similar observations had previously been reported by Dimakopoulos et al. (74) This was also true for wearing long sleeves as a precautionary measure, which was identified as a very important parameter associated with VitD status in summer. Wearing long sleeves in summer increased the odds of suboptimal VitD status by 5.98-fold (95% CI: 1.33, 27.03) (Table 3, Model S2). Similarly, the observations of Datta et al. showed that daily self-reported summer clothing habits in summer, together with sun exposure, most influenced 25(OH)D concentration (27). Some other studies also associated body exposure to intense sunlight with VitD status (82, 83). It should be noted that caution should be exercised with increased sun exposure because of the risk of skin cancer, and that there is a fine line between beneficial UVB exposure and health risks, such as sunburn and skin cancer. It should also be noted that acute overexposure to sunbathing is less effective for VitD biosynthesis than more frequent, lower-level exposure (84), which is a viable option for safer sun exposure. However, debates on this topic are conflicting and indicate that there is no safe level of UVB exposure (85, 86).

We have mentioned several strengths of the study; in particular, the prospective cohort design, the control for VitD intake (exclusion of supplementation), and the use of objective measurements, but we should also point out some limitations of the study. Considering the population in our region, the study population consisted only of Caucasian adults, which means that we did not have access to subjects with dark brown and black skin color. On the other hand, previous studies have highlighted skin phototype as an important predictor of VitD status when comparing different ethnic populations (24–29), but this issue has not been adequately addressed within Caucasian population where differences in skin color are less pronounced. The data on body weight and height (used for calculation of BMI) was self-reported and not measured. We should also mention that VitD intake was not assessed from weighted food records, which are considered the gold standard method. VitD is mainly found in foods that are not consumed on a daily basis, which means that the collection of food records would be needed for several days, imposing a significant burden on the study participants. Therefore, dietary VitD intake was estimated using a validated semi-structured FFQ (51) which required only a single administration. In addition, the method used for objective assessment of summer sun exposure has some limitations. The difference in constitutive and facultative skin pigmentation has been used as an indicator of individual sun exposure, but the development of skin pigmentation varies between different skin color types and may reach a plateau above which an additional UV dose results in little or no additional pigmentation (75). This limitation could be overcome by using UV dosimeters (24, 63, 72), which was not possible in our study because of the large sample size. We should mention that dosimeters are usually used for rather short periods, whereas the half-life of serum 25(OH)D is measured in weeks (87). Long-term use of UV dosimeters would impose a significant burden on the participants, with high risks of noncompliance use during the most intense sun exposure (e.g., sunbathing).

VitD deficiency is highly prevalent around the world. Factors affecting seasonal VitD status have been studied in Caucasian adults. While previous studies have mostly used a cross-sectional design and subjective assessment of skin pigmentation, we conducted a longitudinal cohort study of subjects without VitD supplementation with summer and winter observation periods. Using a range of objective measures, we demonstrated that VitD status is hugely influenced by several personal parameters, limiting the usefulness of generalized population-level recommendations. While seasonal differences and a higher risk of VitD deficiency in winter are well known, we showed that skin pigmentation in both summer and winter is related to VitD status and to seasonal changes in serum 25(OH)D concentration. The general observation was that in the Caucasian population, darker constitutional skin color means lower odds of lower VitD status. Another key factor that should be given special consideration in future public health recommendations is body mass index. The results of the study will be used in the development of a screening tool to identify individuals at higher risk of insufficient VitD status. We used several objective methods to assess skin color and sun exposure behavior that will support further studies. Future research should primarily use objectively measured parameters and wider variability in skin color types.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving humans were approved by the Institutional Ethics Committee of VIST – Faculty of Applied Sciences (protocol code VITAD-01-2018, Approval No. 2018/4-ET-SK on 8th January 2019) and included in the ClinicalTrials.gov register under the entry NCT03818594. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

KŽ and IP: conceptualisation. IP, KŽ, and MH: methodology. MH: validation and formal analysis. MH, TP, IP, and KŽ: investigation. IP and KŽ: resources. MH: data curation and writing—original draft preparation. IP, KŽ, and TP: writing—review and editing. MH: visualization. KŽ and IP: supervision. All the authors have read and agreed to the published version of the manuscript.