We found that mean 250H-D concentrations varied significantly across the levels of consumption of foods containing vitamin D, sunlight exposure, and any tanning bed exposure as ascertained by our vitamin D-specific questionnaire. We observed positive and significant correlations of a similar magnitude (Spearman r = 0.17 to 0.19) for all of these factors. Stronger positive and significant correlations were noted for vitamin D supplement use and among the small percentage of women who reported tanning bed use. The reported dietary and light exposures captured by our questionnaire accounted for up to 36% of the variation in serum 25(OH)D levels in this analysis. There was a consistent relationship with the variation in 25(OH)D levels explained by food consumption (~3.1%), supplements (~18.9%), sunlight (~2.2%), and tanning bed use (~3.0%) in both crude and adjusted models.
Dietary/supplement and light exposures may be important in chronic disease risk. However, such analyses often rely on recalled information that is almost certainly imperfect. Although the interpretation of the strength of effect sizes is influenced by the biological and clinical context, if we use Cohen’s conventions for a small (R2 = 2%), medium (R2 = 13%), and large (R2 = 26%) strength of association(29), our results suggest that reported diet, sunlight, and tanning bed use have a small level of association and that supplements have a medium to large level of association with 25(OH)D levels. Combined, these variables have a large effect (27.5% − 36.0%) on 25(OH)D levels. Presumably, 25(OH)D levels can also be influenced by environmental factors (e.g., exact latitude, cloud or tree cover, ground surface reflectivity, time of day) to personal influences (e.g., sunscreen use, clothing, hats, standing versus sitting) to activation/absorption and metabolic variation of vitamin D between individuals. Thus, given the imperfect measures of recalled diet and light exposures, and the many other factors that can contribute to vitamin D status, the four variables we assessed in our questionnaire explained a substantial amount of variation in 25(OH)D levels in this population of women.
As expected 25(OH)D serum concentrations were normally distributed with a range from 10.1 to 182.5 nmol/L in this, mainly white (93.2%) population-based sample of English-speaking women in Alberta. As others have recently reported(30, 31), circulating 25(OH)D concentrations were not lower in the oldest age group in our population of women. In this population, 80% of women had mean 25(OH)D concentrations of 50 nmol/L or greater which is considered a sufficient level, at least for bone health(32, 33). Of the 20% of women below 50 nmol/L, only 4.1% were vitamin D deficient at < 30 nmol/L. While the majority of Albertans speak English at home or identify it as their “mother tongue”(34), non-English speakers were excluded in this study and may have different 25(OH)D levels than presented here due to diet, lifestyle factors, or skin pigmentation.
Another limitation of this study is that we only had physical activity data for a subset of women (n = 262), but among these women physical activity was significantly related to 25(OH)D levels. We therefore repeated the analysis presented in Table 3 for this subset of women and adjusted for physical activity. The results were very similar to those reported in the final model in Table 3 without adjustment for physical activity, with the partial R2 for supplements again being the highest (about 18.5%). These results again imply that there is a consistent relationship between 25(OH)D levels and the diet and light variables as measured in our questionnaire regardless of adjustment for other covariates.
There is substantial variation in the ascertainment of vitamin D related foods across studies. We attempted to capture all major sources of vitamin D from food in our study and only included those foods that were consumed by at least 5% of the population in our previous reliability study(16). Yogurt was not found to be a major source of vitamin D from food but yogurt made with vitamin D fortified milk became more common during our study. In another study in Alberta(35), about 20% of breast cancer cases reported consuming yogurt in a 2010 questionnaire and at least half of their consumption was yogurt made with vitamin D fortified milk. Daily consumption of such yogurt could increase vitamin D from food by approximately 1.2 µg per day. This implies that we probably under-estimated vitamin D intake from food for some participants, particularly for the latter years of our study, and this could have contributed to the relatively low correlation of 25(OH)D levels with food.
Among the small proportion of women who reported tanning bed use there was a moderate correlation with 25(OH)D levels. Tanning beds mainly emit UV-A radiation, but there is a small, variable amount of UV-B that can trigger the production of vitamin D. While a comprehensive picture of total light (e.g., UV) exposures, including tanning bed use, is important in research studies, tanning beds are associated with an increased risk for skin cancer (36) and their use as a targeted source for vitamin D is highly controversial.
Outdoor sunlight exposures explained a consistent but small level of the variation in serum 25(OH)D levels. Correlations were similar between food and sunlight with respect to 25(OH)D levels, suggesting that our sunlight ascertainment in the recent past performed at the same level as dietary exposure of vitamin D. Furthermore, when we stratified by lag time, we found the highest correlation with a short lag time (0–1 years, r = 0.26), one that exceeded the correlation with the past year’s diet (r = 0.19). These results suggest that our ascertainment of sunlight is a useful measurement in assessing putative vitamin D exposure, even in a population at that resided predominantly in the north region where UV-B intensity is only sufficient for cutaneous vitamin D production for about half of the year(37). It should be noted that we did not collect sun screen use. In our pilot study, we collected this information but women had a difficult time recalling past use accurately, could not recall the sun protection factor (SPF) used, and confused “suntan lotion” with sun screen based on the reported years of use.
We could not directly assess vitamin D related exposures in the distant past because we did not have distant past serum samples for 25(OH)D measurement. However, we did find that our questionnaire measures of recalled diet and light exposures in the more recent past had significant positive correlations with 25(OH)D levels. Given the imperfect measures of recalled diet and light exposures, and the many other factors that can contribute to vitamin D status, the four variables we assessed in our questionnaire explained a substantial amount of variation in 25(OH)D levels in this population of women in Alberta. These results suggest that our comprehensive dietary and light exposure questionnaire is a reasonable proxy measure of vitamin D status for exposures in the recent past when either 25(OH)D measurements or serum samples are not available for study participants.