High frequency of vitamin D deficiency in current pregnant Japanese women associated with UV avoidance and hypo-vitamin D diet

Background As a consequence of indoor occupations and reduced exposure to sunlight, concerns have been raised that vitamin D deficiency is widespread in developed countries. Vitamin D is known to be associated with increased risks of morbidity and mortality in various diseases. Objective To investigate the serum vitamin D status and its relation with life-style factors in pregnant Japanese women. Methods Among a cohort for 3,327 pregnant women who participated in an the adjunct study of the Japan Environment and Children's Study during 2011–2013, in which data were obtained on various life-style factors, including both dietary intake of vitamin D and frequency of UV exposure, this study consisted of 1,592 pregnant women, from whom 2,030 serum samples were drawn in Jan, Apr, Jul, and Oct, and the association between serum 25(OH)D level and life-style factors were analyzed using linear mixed models. Results Serum 25(OH)D levels were less than 20ng/mL in 1,486 of 2,030 samples (73.2%). There was an obvious seasonal change, with serum 25(OH)D levels of less than 20 ng/mL in 89.8% and 47.8% of samples in spring (April) and autumn (October), respectively. Both the frequency spent under sunlight and dietary intake of vitamin D were significantly associated with serum 25(OH)D level. An increase in sunlight exposure of more than 15 min for 1 to 2 days per week in non-winter, or dietary intake of 2 μg/day of vitamin D resulted in an elevation of 1 ng/mL in serum 25(OH)D levels. Conclusion These findings indicate that vitamin D deficiency is very severe in Japanese pregnant women, especially those rarely exposed to sunlight. The benefits of UV rays should also be informed of when its risk is alerted, and clinicians should propose the adequate UV exposure level.


Introduction
Vitamin D is a fat-soluble secosteroid with well-established effects on calcium homeostasis. More recently, vitamin D has also been recognized to interact with a nuclear receptor in various other organs [1] and its deficiency is associated with increased risks of morbidity and mortality in various diseases including cardiovascular, malignant, and autoimmune diseases [2,3]. Accumulating evidence suggests that vitamin D deficiency during pregnancy may cause complications such as preeclampsia [3][4][5], although its implications and the underlying mechanisms are not fully understood. And it is even hypothesized that vitamin D deficiency in the fetal period leads to an increased risk of allergic diseases, multiple sclerosis, and cardiovascular diseases in later life [2,[6][7][8].
As a consequence of indoor occupations and reduced exposure to sunlight, concerns have been raised that vitamin D deficiency is widespread in developed countries [2,3]. In Japan, because fish is a primary component of the traditional diet, the risk of vitamin D deficiency is rarely discussed. However, studies indicate that younger people consume less fish [9][10][11][12], and while females appear to be at higher risk of vitamin D deficiency because they tend to avoid direct sunlight exposure to prevent skin-tanning, may be malnourished from maintaining a lean proportion [13]. The importance of monitoring the vitamin D status in Japan has only recently been demonstrated [14,15].
Accordingly, the present study aimed to examine the serum vitamin D status of pregnant Japanese women and to estimate the impact of lifestyle factors on vitamin D levels in a population-based cohort.

Study design
This was cross-sectional sub-study comprising pregnant Japanese women enrolled in the adjunct study of the Japan Environment and Children's Study (JECS) to examine the effects of desert dust exposure on allergic diseases in pregnant women and their children in three areas in Japan; Kyoto (N 35˚), Toyama (N 36˚), and Tottori (N 35.5˚) [16]. The study protocol was approved by ethic comities in Kyoto University, University of Toyama, and Tottori University, and was registered at UMIN000010826 [16].
The details of the study design and protocol have been previously reported [16]. In brief, the JECS is a community based national birth cohort study [17,18], and the JECS participants

Statistical analysis
Average 25(OH)D levels among groups were estimated using the linear mixed model analysis, with intra-individual variation by repeated measurements accounted for. For comparison of variables among more than two groups, p values were adjusted by Dunnett's method.
A uni-variate model was applied for each factor, followed by multi-variate analysis incorporating all variables with p values of <0.1 in the univariate models. Backward elimination method was applied to construct the final model.
The entire cohort dataset (Fig 1) was used for analysis of seasonal 25(OH)D changes. The reasonable answer dataset, which excluded subjects with total calories on the FFQ of less than 50% and more than 150% of predicted values (Fig 1) was used for the analysis that include dietary intake of vitamin D, and the full answer dataset (Fig 1) was used for sensitivity analysis.
All analyses were performed using SAS software, version 9.3 (SAS Institute), and two-sided P<0.05 were considered statistically significant.

Subjects
Of the 6,340 serum samples from 3,495 pregnant women who participated before May 2013 in the adjunct study, 2,030 samples which were collected in Apr, Jul, Oct, and Jan during 2012 to 2013 from 1,592 pregnant women, were included in this sub-study, as illustrated in Fig 1. Table 1 summarizes the characteristics of the study cohort. All subjects were pregnant with an age range from teenagers to over 45 years, and various socioeconomic backgrounds. Subject characteristics were similar to those reported by the Japanese government in 2012, except that the proportion of current smokers was lower in the study cohort (1.6-2.7%) than that in the government report (12.8% for women in their 20s and 16.6% for women in their 30s) [25,26]. The proportion of subjects with an education level up to junior high school completion was lower (1.6-3.3%) compared with the government report (6.0%) [27,28]. Overall, the study cohort was considered to be a good representation of pregnant women in Japan.

Frequency of UV exposure and vitamin D status
The mean serum 25(OH)D level was 16.7± 6.99 ng/mL (range: <5-71 ng/mL). Vitamin D deficiency, defined as less than 20ng/mL, was present in 73.2% (1,486 of 2,030 samples), and 10.8% (219 of 2,030 samples) had less than 10 ng/mL, which is defined as severe vitamin D deficiency. The distribution showed a clear seasonal change (Fig 2, Table 2), and 87.7% (880 of 1,003 samples) had less than 20ng/mL in winter and spring. This trend was observed even among women who reported sunlight exposure for at least 15 minutes on more than five days a week (Fig 3). However, at the end of summer (October), the mean 25(OH)D level of the group was much higher compared with that of subjects with least exposed to sunlight (Fig 3), and 61.5% (88 of 143 samples) of subjects who reported sunlight exposure for at least 15 minutes on more than five days a week achieved 20 ng/mL, while 34.6% (27 of 78 samples) of subjects with least exposed to sunlight achieved 20 ng/mL. Furthermore, vitamin D levels were evaluated in subjects employed in agriculture/fishery, who work outside during their daily lives. As expected, this group showed higher 25(OH)D levels, especially in autumn (LS mean 33.6 ng/mL, 95%CI; 17.8-63.6), although the number of subjects employed in agriculture/fishery was very few (7 samples on 6 subjects) and a statistical significance was not achieved. Unexpectedly, the subjects whose self-reported skin type was fair tended to have a lower 25 (OH)D levels than other skin types in this relatively homogenous population ( Table 2).  An increase in the frequency of sunlight exposure to at least 15 minutes for 1 to 2 days per week resulted in elevations in 25(OH)D levels of approximately 1 ng/mL in non-winter and 0.5 ng/mL in winter (Fig 3).

Vitamin D status and dietary vitamin D intake
The mean dietary intake of vitamin D (energy-adjusted) in Japanese pregnant women was estimated to be 5.5±2.8 μg/day. As few as 22.6% of subjects consumed above 7.0 μg/day, described as the "adequate intake" per day for Japanese pregnant women [29]. Nobody exceeded 100μg/ day, described as the "tolerable upper intake" per day in the guideline.
The amount of dietary vitamin D intake was significantly associated with the 25(OH)D level (Tables 2 and 3, Fig 4), and 198 (90.8%) of 218 samples from subjects with above 7.0 μg/ day (energy-adjusted) had above 10 ng/mL even in winter and spring, although the majority (174/218, 79.8%) of these subjects did not achieve 20 ng/mL. The average daily intake of vitamin D in women with above 20 ng/mL of 25(OH)D in winter and spring was 6.5μg/day.
An increase of 1μg/day dietary vitamin D intake led to an elevation of approximately 0.5 ng/mL in serum 25(OH)D (Fig 4).

Other factors associated with 25(OH)D levels
In the univariate linear mixed models with random effect of repeated measurements, the following factors were found to be significantly associated with 25(OH)D level (Table 2); residential location, frequency of UV exposure in daily life, frequency of UV exposure at weekends, month of blood sample collection, pregnancy trimester of blood sample, use of vitamin D supplements, dietary intake of vitamin D, dietary calorie intake, living with children, and smoking habits of subjects and their partners.
In the multivariate model incorporating all these variables, the following were consecutively excluded; frequency of UV exposure at weekends, age, smoking of partner, and smoking. The final model included the following factors; month of blood collection, residential location, pregnancy trimester of blood collection, use of vitamin D supplements, frequency of UV exposure in daily life, and dietary intake of vitamin D (Table 3).
Vitamin D supplementation, multivitamin tablets or calcium tablets, was reported by very few pregnant women (5.1%). Most tablets contained 2.5-5.0 μg/day of vitamin D, and 15μg/ day at most (one case). However, their serum increase of 25(OH)D level were as much as 4.5 ng/mL (Table 3).
Contrary to previous reports from other countries, a univariate model showed that age was negatively associated with 25(OH)D levels in Japanese pregnant women (Table 2). This tendency was lost in the multivariate model or a model including covariates of dietary intake of vitamin D, UV exposure frequency, and living with children. Therefore it is suggested that this was the consequence of confounding effects of less exposure to sunlight and lower dietary consumption of vitamin D among younger populations in Japan [9,30]. Unexpectedly, even after the adjustment for lifestyle and dietary variables, pregnant women living in Toyama, the northern-most of the three locations, exhibited significantly higher 25 (OH)D levels than in the other two locations (Table 3). By examining other characteristics, it was found that only samples taken in winter (January) in Toyama showed higher 25(OH)D levels than in the other two locations (Fig 5). Because Toyama is famous for cold yellowtail and other seafood products in winter, these women may have had a fish-rich diet in winter,  although this was not sufficiently reflected in the answers to the questionnaire on "typical" diet. Furthermore, Toyama had higher snowfall than the other two locations in January 2013 (Toyama: 4cm/day, Tottori: 0.8cm/day, Kyoto: 0cm/day) and some sunlight hours (Toyama: Entire cohort dataset (excluding non-responders for each question).
All values are from univariate linear mixed models, with 25(OH)D natural log-transformed. LS means and 95%CIs are shown as exponentials of log-transformed 25(OH)D.
For variables with more than three groups, p values are adjusted by Dunnett. a P values for trend.

Discussion
In this study, we showed that Japanese pregnant women are in severe vitamin D deficiency status (10.8% are <10 ng/mL, 73.2% are <20 ng/mL). This was expected from their lifestyles, and consistent with previous urban Japanese studies and other recent Asian studies in which it is reported that lighter skin tones are culturally preferred [14,15,[31][32][33]. The thresholds for 25 (OH)D levels (10 ng/mL for severe deficiency, and 20 ng/mL for deficiency) were derived from non-pregnant populations and an optimal serum level during pregnancy has not been established. However, it is at least suggested that Japanese pregnant women have lower vitamin Vitamin D deficiency in pregnant Japanese women linked to UV avoidance and low vitamin D diet D levels compared with a century ago, at which time the majority of the populations engaged in agriculture or fishery, spending many hours outside every day; among subjects who Vitamin D deficiency in pregnant Japanese women linked to UV avoidance and low vitamin D diet deficiency can cause or contribute to a variety of diseases [34]. The association between vitamin D deficiency and specific morbidities, especially diseases that is increasing in these decades, should be further investigated.
In this study, the average daily vitamin D intake among women with 25(OH)D levels above 20 ng/mL in winter and spring was 6.3 μg/day, which is similar to reports in Norwegian pregnant women (7.0 μg/day) [35]. Although the Japanese Guideline for Nutrition suggest 7.0 μg/ day, and this appears reasonable for Japanese pregnant women based on the results of our study, it is also important to note that 25(OH)D levels above 20 ng/mL were not achieved only by diet for majority of the women in winter and spring. This leads to the proposal that dietary intake of 7.0 μg/day is necessary, but not sufficient to maintain adequate 25(OH)D levels at least 20 ng/mL in Japan.
Although vitamin D supplementation was reported by few pregnant women (5.1%) and most consumed only 100 to 200 IU/day of vitamin D in our study, the serum 25(OH)D levels increased by as much as 4 ng/mL in supplemented women. This figure is consistent with the previous report that showed 100IU of vitamin D increased the 25(OH)D level by 2 to 3 ng/mL in subjects with serum 25(OH)D levels of less than 15 ng/mL [34].
Darker skin is generally a risk factor for a low level of 25(OH)D [34]. However, in our study, comprising subjects of almost uniform ethnicity, the self-reported fair-colored skin had a tendency toward lower 25(OH)D levels. This subpopulation may have avoided sunlight to an extreme due to fear of skin cancer. Among Caucasians, especially those who emigrated to a low latitude area, UV exposure is a definite risk factor for skin cancer development. However, skin cancer mortality is very low in Japanese living in Japan, even in those who were children in the era without UV protection. The skin cancer mortality is 1.2 / 100 thousand Japanese women, while colon cancer mortality is 36.5 / 100 thousand Japanese women in 2017 [36]. The natural skin tone may be adapted to the sunlight in Japan as an evolutionary feature related to island dwelling. Skin production of vitamin D is thought to be accomplished after exposure to moderate sunlight for several (in summer) to several ten (in winter) minutes without causing burns [3,34]. National Institute for Environmental Studies comments that getting exposure to UV ray for several (in summer) to several ten (in winter) minutes in Japan that will never reach 1 MED (Minimal Erythema Dose) for people with skin photo type III (Japanese people), will lead vitamin D synthesis while minimizing its harms [37]. And it provides in real-time the amount of vitamin D synthesized in the body at some locations in Japan on the web [38], based on the logics described by Miyauchi and Nakajima [39]. Individuals should also be informed of the benefits of UV rays when alerted about its risk, with the available information shown above.
Major strengths of this study were a relatively large sample size based on a large population-based birth cohort (from the JECS), and a high response rate for various background questionnaires including both dietary intake of vitamin D and frequency of sunlight exposure, which will contribute to high generalizability. Despite the strengths, this study has some limitations. FFQ and frequency of sunlight exposure were both self-reported, and there may be some mis-categorizations. A uniform questionnaire was used for sunlight exposure throughout a year, which may not be a meaningful measure of differences in sunlight exposure in winter, when it is estimated that at least several ten minutes of sunlight is necessary for vitamin D production in skin, whereas the questionnaire asked the frequency of "at least 15 minute exposure /day" per week. Finally, the present results are applicable to only Japanese pregnant women, as it is known that 25(OH)D levels differ between ethnicities [34].
In conclusion, vitamin D deficiency is very severe in Japanese pregnant women, and lifestyle factors including the frequency of sunlight exposure and dietary intake of vitamin D have a clinically relevant impact on serum levels. This suggests that vitamin D level may be enhanced by changes in lifestyle. Pregnant women should be informed of both the risks and benefits of UV ray. Further investigations are required to establish the impact of vitamin D deficiency on morbidities.