General characteristics of participant women
A total of 410 fertile women between the ages 22 and 44 years attended the clinics at 3-6 months before they became pregnant were included in this study, of which 100% were Han nationality. The median BMI of them was 20.7 kg/m2 with a range of 19.5-23.0 kg/m2 (as 25-75%). Two hundred and forty (58.5%) women reported having one or more histories of pregnancy, and only 80 (19.5%) had delivered a previous child. One hundred and seventeen (28.5%) women underwent voluntary pregnancy termination. Three hundred and forty (82.9%) subjects were working women who were indoor job holders or office-based workers, while 70 (17.1%) were non-working individuals or housewives. A total of 128 (31.2%) serum samples were taken in spring, 102 (24.9%) in summer, 88 (21.5%) in autumn, and 92 (22.4%) in winter.
The serum levels of 25(OH)D among preconceptional fertile women
The mean serum 25(OH)D for the cohort was 14.7 ± 5.4 ng/mL. The concentrations ranged from a low of 4.5 ng/mL among young age women between 22 to 30 years to a high of 36.0 ng/mL among women aged 31 to 44 years. There was no effect of the gravidity, induced abortions, and BMI of the woman on the serum 25(OH)D level. There was however a statistically significant influence of the 25(OH)D level and season, number of deliveries, woman’s employment status, and age of women (All p < 0.05). Table 1 shows the mean of 25(OH)D levels for specified subgroups. Women aged 22-30 years had significantly lower 25(OH)D level compared with those between 31-44 years (14.3 ± 5.0 ng/mL vs 15.6 ± 6.0 ng/mL, P<0.05). The mean vitamin D level was significantly lower in working women compared to non-working individuals or housewives (14.3 ± 5.2 ng/mL vs 16.5 ± 5.8 ng/mL, P<0.01). Women who had delivered a previous child had higher 25(OH)D levels in comparison to women without a history of childbirth (15.9 ± 5.7 ng/mL vs 14.4 ± 5.3 ng/mL, P<0.05). Mean serum 25(OH)D concentrations varied by month and season of sample collection. The 25(OH)D concentrations were the lowest in winter among that in spring, summer, and autumn (all P<0.001). The range of 25(OH)D concentration was from 18.0 ng/mL in July and August to 11.4 ng/mL in January (Fig.1).
Table 1. Association of serum 25(OH)D level with demographic and reproductive history
Characteristics
|
Demographic and
reproductive history (n)
|
Serum 25(OH)D
(mean± SD ng/mL)
|
t/F
|
p
|
Age
|
22-30 years (277)
31-44 years (133)
|
14.3± 5.0
15.6± 6.0
|
2.203
|
0.029
|
BMI (kg/m2)
|
Underweight (46)
Normal (293)
Overweight and Obese (71)
|
13.7± 4.5
14.9 ± 5.4
14.4 ± 5.7
|
1.273
|
0.281
|
Gravidity
|
No (170)
One and more (240)
|
14.1± 5.1
15.1 ± 5.6
|
1.813
|
0.071
|
Parity
|
No(330)
One (80)
|
14.4 ± 5.3
15.9 ± 5.7
|
2.290
|
0.023
|
Numbers of
induced abortion
|
No (293)
One (85)
Two and more (32)
|
14.7 ± 5.7
14.7 ± 5.2
15.2 ± 6.2
|
0.143
|
0.866
|
Employment status
|
Working (340)
Non-working (70)
|
14.3 ±5.2
16.5 ± 5.8
|
3.127
|
0.002
|
Season
|
Spring (128)
Summer (102)
Autumn (88)
Winter (92)
|
14.5 ± 5.1#
16.6 ± 5.0
15.9 ± 5.6
11.8 ± 4.8*
|
16.417
|
0.000
|
*Significantly lower compared to spring, summer, and autumn (all p<0.01).
#Significantly lower compared to summer (p<0.01).
For the stepwise multiple linear regression analysis (Table 2), the three variables, season, woman’s employment status, and parity are correlated to concentrations of serum 25(OH)D of the women [R2 = 0.113, F = 17.306, p = 0.000]. The winter, working women as well as nulliparous women contributed to low serum 25(OH)D concentrations among the women. Furthermore, woman’ age and BMI was no longer a significant predictor of vitamin D after adjusting for all other factors.
Table2. Determinants of 25(OH)D level from a stepwise multiple linear regression
|
Unstandardized coefficients
|
p
|
B
|
S.E
|
Constant
|
12.306
|
0.900
|
0.000
|
Season
|
1.472
|
0.238
|
0.000
|
Employment
|
-1.789
|
0.688
|
0.010
|
Parity
|
1.390
|
0.655
|
0.034
|
Further, the mean serum 25(OH)D concentration differed by season after stratification by age and by women’s employment status. Seasonal variation of 25(OH)D levels was present in working individuals (F=17.013, P = 0.000), with lower concentrations in winter (11.5 ± 4.6) and spring (13.7 ± 4.6) than in summer (16.4 ± 4.9) and autumn (15.9 ± 5.6) (Table 3). This seasonal variation was quantitatively less among women aged 31-44 years when compared with young women aged 22-30 years. For example, the difference between winter and summer mean 25(OH)D levels was 5.1 ng/mL (95% CI: 3.5-6.7 ng/mL) among young women aged 22-30 years, and 3.8 ng/mL (95% CI: 0.9-6.8 ng/mL) among women aged 31-44 years.
Table 3. The mean 25(OH)D levels by season after stratification by age and by women’s employment status (mean± SD ng/mL)
Season
|
Employment status
|
Age
|
Working (n=285)
|
Non-working (n=57)
|
22-30 years
(n=241)
|
31-44 years
(n=101)
|
Spring
|
13.7± 4.6
|
17.3± 5.8
|
14.5 ± 5.0
|
14.7 ± 5.5
|
Summer
|
16.4 ± 4.9
|
17.7 ± 5.5
|
16.1 ± 4.9
|
17.5 ± 5.1
|
Autumn
|
15.9 ± 5.6
|
15.6 ± 5.7
|
15.7 ± 5.1
|
16.3 ± 6.7
|
Winter
|
11.5 ± 4.6
|
13.8 ± 6.0
|
11.0 ± 3.8
|
13.6 ± 6.5
|
F
|
17.013
|
1.384
|
15.655
|
2.664
|
p
|
0.000
|
0.256
|
0.000
|
0.051
|
Vitamin D deficiency in seasons and in terms of maternal variables
A large proportion (346, 84.4%) of women had serum 25(OH)D concentration below 20 ng/mL, indicating that they were deficient in vitamin D. Only 3 (0.7%) women were sufficient with serum 25(OH)D (30 ng/mL or more), 61 (14.9%) were insufficient (20-30 ng/mL). Consistent with the mean 25(OH)D data, 295 (86.8%) working women were vitamin D deficient in contrast to 51 (72.9%) non-working individuals or housewives (χ2=8.523, p=0.004). In line with seasonal variation of vitamin D levels, the percentage of women with 25(OH)D concentrations below 20 ng/mL was significantly higher in winter (n=87, 94.6%) than in spring (n=110, 85.9%), summer (n=80, 78.4%), and autumn (n=69, 78.4%) (χ2=12.602, p=0.006). A higher proportion (n=286, 86.7%) of nulliparous women with 25(OH)D <20 ng/mL was found compared to women (n=60, 75%) who had delivered a previous child (χ2=6.653, p=0.010).
Binary logistic regression analysis was performed after that 25(OH)D was specified as categorical with cut point at 20 ng/mL (deficient < 20 ng/mL, non-deficient ≧20 ng/mL). The results establish that women in winter have significantly elevated OR of 5.00 to develop vitamin D deficiency, although, it is important to note the wide confidence interval (95%CI 1.75–14.25). The nulliparous women have 2.32 higher odds of deficiency than women who had delivered a previous child (95% CI 1.08-3.98). The non-working women or housewives have a significantly lower susceptibility to develop vitamin D deficiency than working women (OR 0.49, 95% CI 0.26 to 0.96) (Table 4). This is consistent with the results of the multiple linear regression analysis described above.
Table 4. Results from binary logistic regression analysis for vitamin D deficiency
|
Subcategory
|
OR
|
p
|
95% Confidence interval
|
Season
|
|
|
0.007
|
|
|
winter
|
5.000
|
0.003
|
1.754-14.250
|
|
Spring
|
1.925
|
0.080
|
0.925-4.005
|
|
Summer
|
1.014
|
0.970
|
0.500-2.055
|
|
Autumn
|
1.00
|
|
|
Employment tatus
|
Non-working
|
0.494
|
0.036
|
0.256-0.955
|
Working
|
1.00
|
|
|
Parity
|
Nulliparous
|
2.320
|
0.030
|
1.076-3.981
|
|
One
|
1.00
|
|
|