- Email: [email protected]
The longitudinal variability of serum 25(OH)D levels
European Journal of Internal Medicine 23 (2012) e106–e111
Contents lists available at SciVerse ScienceDirect
European Journal of Internal Medicine journal homepage: www.elsevier.com/locate/ejim
Original article
The longitudinal variability of serum 25(OH)D levels Walid Saliba a,⁎, Ofra Barnett a, Nili Stein a, Anne Kershenbaum a, Gad Rennert a, b a Department of Community Medicine and Epidemiology, Carmel Medical Center, Clalit Health Services, and Bruce Rappaport Faculty of Medicine, Technion-Israel Institute of Technology, Haifa, Israel b Department of Epidemiology and Disease Prevention, Office of the Chief Physician, Clalit Health Services Headquarters, Tel Aviv, Israel
a r t i c l e
i n f o
Article history: Received 16 October 2011 Received in revised form 4 January 2012 Accepted 7 February 2012 Available online 17 March 2012 Keywords: Vitamin D 25(OH)D Variability Seasonality
a b s t r a c t Background: The extent to which a single serum 25(OH)D measurement represents long-term vitamin D status remains unclear. This study aims to assess the variability of serum 25(OH)D between tests taken at different time intervals. Methods: Using the computerized database of the largest healthcare provider in Israel, we identified subjects in whom a serum 25(OH)D test was performed on at least two different occasions between January 2008 and September 2011 (n = 188,771). For these subjects we selected the first and the last dated tests, then we identified those who were not treated with supplements during the last 6 months before the first and before the last test (n = 94,418). Of these we analyzed subjects in whom the first and the last tests were performed in the same month of the year (n = 8881). Results: The mean serum 25(OH)D level at the first test was 51.7 ± 24.0 nmol/L and was 56.7 ± 24.7 at the last test (P b 0.001); the overall correlation was 0.63 (P b 0.001). For vitamin D status in two categories (b 50 versus ≥ 50 nmol/L), the percentage of agreement between the first and last tests was 74.4%, and was 50.8% for vitamin D status in four categories (b 30, 30–49.9, 50–74.9, and ≥75 nmol/L). The correlation decreased with increasing time between the tests ranging from 0.83 for tests done at the same year to 0.55 after 3 years. The more the first levels were higher or lower, the more likely subjects remain in their first category (≥ 50 versus b 50 nmol/L). Conclusions: Long-term month specific serum 25(OH)D levels are relatively stable. © 2012 European Federation of Internal Medicine. Published by Elsevier B.V. All rights reserved.
1. Introduction Vitamin D is important in supporting and maintaining normal mineral and bone health [1]. In recent years, new data linking vitamin D with extra-skeletal health has emerged [2]. Although, studies have shown conflicting evidence regarding the extra-skeletal manifestations of vitamin D deficiency, low serum 25(OH)D levels have been found to be associated with increased risk for mortality and morbidity from a variety of chronic diseases including diabetes mellitus [3,4], cardiovascular diseases [5,6], cancer [7,8], and all cause of mortality [9,10]. Most of the studies linking vitamin D with extra-skeletal health are based on single measurement of serum 25(OH)D. This approach is based on the assumption that serum 25(OH)D levels are stable over time, while several factors may influence serum 25(OH)D levels. Vitamin D production by skin exposure to sun UVB irradiation is the main source of vitamin D [1,2]. Very few foods naturally contain or are fortified with vitamin D [2]. Hence, oral intake of vitamin D ⁎ Corresponding author at: Department of Community Medicine and Epidemiology, Carmel Medical Center, 7 Michal St., Haifa 34362, Israel. Tel.: + 972 4 825 0474; fax: + 972 4 834 4358. E-mail address: [email protected] (W. Saliba).
from foodstuffs contributes to a lesser degree to vitamin D status particularly in countries where fortified food and supplements are uncommon. As exposure to UVB irradiation is the major determinant of serum 25(OH)D levels, this may result in great variability in serum 25(OH)D levels depending on the season of the year and outdoor activity [1,11,12]. Serum 25(OH)D levels also decrease with increasing age [11,13]. Hence, a single time measurement of 25(OH)D may not be representative of long term vitamin D status. Little is known about the longitudinal variability of serum 25(OH) D levels [14–16]. We thought that the Clalit Health Services (CHS) database may provide a suitable platform that may help to clarify this issue. 2. Materials and methods 2.1. Study population and data collection We used data from the database of Clalit Health Services (CHS) which is a non-for-profit health maintenance organization (HMO) covering more than half of the Israeli population. The database, which includes data of laboratory tests results, and use of medications, was searched for all available serum 25(OH)D test results from January 2008 until September 2011 (1,347,349 tests in
0953-6205/$ – see front matter © 2012 European Federation of Internal Medicine. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.ejim.2012.02.002
W. Saliba et al. / European Journal of Internal Medicine 23 (2012) e106–e111
709,186 subjects which constitutes more than 18% of the CHS population). To ensure consistency only tests that were performed in the two largest laboratories, performing more than 62.2% of the tests, were selected for this analysis (839,043 tests in 444,729 subjects). Of these we selected subjects who had at least two tests during the study period (n = 188,771). In subjects who performed more than two tests during this period, the results of the first and the most recent tests were used. We excluded 1625 subjects; age less than 10 years, reported serum 25(OH)D level less than 10 nmol/L, and repeated testing in different laboratories (n = 187,146 after exclusion). In addition, to eliminate the effect of treatment with vitamin D supplement on the variability of serum 25(OH)D level, we excluded 92,728 subjects who were treated with vitamin D supplements during the last six months before the first and before the last tests (n = 94,418 after exclusion). To eliminate the effect of seasonality which is known to be an important contributor to the variability in serum 25(OH)D levels we selected subjects in whom the first and the last tests were performed in the same month of the year (n = 8881). 2.2. Statistical analysis The following analyses were performed to assess the association between the baseline and last serum 25(OH)D measurements: 1. Pearson's correlation coefficient; 2. Location and spread statistics of the difference between the two tests; 3. Location and spread statistics of the absolute difference between the two tests; 4. Two classification of vitamin D status were used based on serum 25(OH)D levels (b50, and ≥50 nmol/L) and (b30, 30 to 49.9, 50 to 74.9, and ≥75 nmol/L). The percentage of agreement between the categories of the two tests was calculated. For all analyses, P value of less than 0.05 for the 2-tailed tests was considered statistically significant. All statistical analyses were performed using SPSS 18.0 (SPSS Inc., Chicago). 2.3. 25(OH)D assay 25(OH)D was tested in two central laboratories. Both laboratories used the LIAISON® 25-OH Vitamin D Total Assay (DiaSorin), a competitive two-step chemiluminescence assay. The measuring range is 4.0– 150 ng/mL (10–375 nmol/L), the analytical sensitivity b 1.0 ng/mL (2.5 nmol/L), and the functional sensitivityb 4.0 ng/mL (10 nmol/L). The intra-assay imprecision is up to 5%, and the inter-assay precision is up to 15%. The specificity for 25-OH Vitamin D2 is 104%, and for 25-OH Vitamin D3 is 100%. Performance characteristics of the Vitamin D assay were checked in the method evaluation process done by “Clalit Health Services” and were compatible to the manufacturer-generated data. The accuracy of the measurements in the individual laboratory is confirmed by in-house daily QC monitoring and by periodic external QC program (DEQAS). 3. Results 3.1. Baseline characteristics of the study population Of the 8881 included subjects 6310 (61.1%) were females and 7895 (88.9%) were Jews. The mean age was 56.1 ± 17.6 years at the time of the first test (Table 1). 4290 (48.3%) of the tests were performed in winter–spring while 4591 (51.7%) of the tests were performed in summer–autumn. The mean of serum 25(OH)D on the first tests results were higher in summer–autumn, 55.1 ± 24.1 nmol/L, as compared to winter–spring, 48.0 ± 23.3 nmol/L (P b 0.001), and the mean of serum 25(OH)D on the last tests results were 58.1 ± 23.6 nmol/L and 54.3 ± 25.6 nmol/L respectively (P b 0.001).
e107
Table 1 Characteristics of the subjects at the time of the first and at the time of the last serum 25(OH)D tests (n = 8881). Variable
At the time of the first test
Gender Males Females
2571 (28.9%) 6310 (71.1%)
Ethnicity Jews Arabs
7895 (88.9%) 966 (11.1%)
Age (years) Mean ± SD Median (IQR)
56.1 ± 17.6 58.0 (45.0–69.0)
Season of the test Winter–Spring Summer–Autumn
4290 (48.3%) 4591 (51.7%)
At the time of the last test
57.6 ± 17.7 59.0 (46.0–71.0)
Serum 25(OH)D (nmol/L) Mean ± SD 51.7 ± 24.0 Median (IQR) 50.3 (33.8–66.7)
56.7 ± 24.7 55.3 (39.0–71.9)
Serum 25(OH)D b 50 nmol/L ≥ 50 nmol/L
3643 (41.0%) 5238 (59.0%)
4406 (49.6%) 4475 (50.4%)
3.2. The variability of serum 25(OH)D The mean serum 25(OH)D level was 51.7 ± 24.0 nmol/L at the time of the first test and 56.7 ± 24.7 nmol/L at the time of the last test (Table 1) with mean difference of 5.0 nmol/L (95% CI 4.5–5.4) nmol/L. The Pearson's r correlation coefficient was 0.63 (P b 0.001). The absolute difference between the first and last test was ≤5.0 nmol/L in 25% of subjects, b11.1 nmol/L in 50% of subjects while 75% of subjects had an absolute difference ≤20.8 nmol/L (Table 2). For vitamin D status in two categories (b50 versus ≥50 nmol/L), the percentage of agreement between the results of the first test and the results of the last test was 74.4%, and was 50.8% for vitamin D status in four categories (b30, 30–49.9, 50–74.9, and ≥75 nmol/L) (Table 2). 49.6% of subjects with first serum 25(OH)D level b30 nmol/L remained within this range at the last test, and 78.8% had serum 25(OH)D level b50 nmol/L. 78% of subjects with first serum 25(OH) D 50–74.9 nmol/L had levels ≥50 nmol/L at the last test. While
Table 2 Measurements of association between the first and the last serum 25(OH)D tests by gender. Statistics
All (n = 8881)
Males (n = 2571)
Females (n = 6310)
Pearson's coefficient Differencea (nmol/L) Mean (SD) 95% CI for the mean Median (IQR)
0.63
0.64
0.62
5.0 (21.0) 4.5–5.4 3 (− 7.4–15.1)
3.4 (20.5) 2.6–4.2 1.8 (− 8.7–13.8)
5.6 (21.2) 5.0–6.1 3.4 (− 6.9–15.8)
14.9 (14.5) 14.4–15.5 11.0 (5.0–20.0)
15.6 (15.4) 15.2–16.0 11.0 (5.0–21.1)
75.0
74.1
52.7
50.1
Absolute differenceb (nmol/L) Mean (SD) 15.4 (15.2) 95% CI for the mean 15.1–15.7 Median (IQR) 11.1 (5.0–20.8) Agreement measure (%) 2 categories of serum 74.4 25(OH)Dc 4 categories of serum 50.8 25(OH)Dd a b c d
Difference between the last and the first serum 25(OH)D test results. The absolute difference between the last and the first serum 25(OH)D test results. Serum 25(OH)D in two categories (b50, and ≥50 nmol/L). Serum 25(OH)D in four categories (b30, 30 to 49.9, 50 to74.9, and ≥75 nmol/L).
e108
W. Saliba et al. / European Journal of Internal Medicine 23 (2012) e106–e111
94.6% of subjects with the first serum 25(OH)D ≥75 nmol/L had levels ≥50 nmol/L at the last test (Fig. 1). Stratified analyses by the time interval between the first and the last test revealed that the mean difference between the two tests increased gradually with increasing time between the tests and ranged between 3 nmol/L for tests performed at the same year to 9.8 nmol/L for tests performed 3 years apart (Table 3). The Pearson's correlation coefficient also decreased gradually with increasing time between tests and ranged between 0.83 for tests performed at the same year to 0.55 after 3 years (Table 3). In addition, the percentage of agreement between the two tests decreased with increasing time interval between the first and the last test (Table 3). 3.3. The variability of serum 25(OH)D by gender The mean serum 25(OH)D levels were higher in males as compared to females (P b 0.001), 55.0 ± 24.0 nmol/L and 50.4 ± 23.5 nmol/L respectively at the time of the first test, and 58.4 ± 24.3 nmol/L and 56.0 ± 24.8 nmol/L respectively at the time of the last test. The mean difference between the two tests was slightly higher in females as compared to males, 5.6 nmol/L (95% CI 5.0–6.1) and 3.4 nmol/L (95% CI 2.6–4.2) respectively. The Pearson's r correlation coefficient between the first and the last test was 0.64 in males, and 0.62 in females. The absolute difference between the two tests and the measurement of agreement was similar between males and females (Table 2). Stratified analyses by the time interval between the first and the last test revealed that the pattern of the variability measurements was similar between males and females. However, the percentage of agreement between the two tests decreased with increasing time interval between the two tests among females while it was relatively stable among males after 1–3 years (Table 3). 3.4. The variability of serum 25(OH)D in special subgroups of women The menopausal state was not available for the female population of this study. Thus for the purpose of this study, in order to evaluate the effect of the menopausal state on the variability of serum
25(OH)D, we defined postmenopausal state as age greater than 55 years (n = 3335). The Pearson's r correlation coefficient between the first and the last test was 0.66 for premenopausal women, and 0.58 for postmenopausal women. The other measurements of variability were similar between the two groups (Table 4). To evaluate the effect of current usage of oral contraceptives on the variability of serum 25(OH)D levels in premenopausal women (n = 2975), we defined current users as those who had filled at least one prescription for oral contraceptives during the last 8 weeks before the first and/or before the last serum 25(OH)D tests (n = 225). Current users of oral contraceptives had a slightly lower correlation coefficient between the first and the last test 0.60 as compared to 0.66 among non current users. However, the other measurements of variability were similar between the two groups.
4. Discussion Long-term month specific serum 25(OH)D measurements were relatively stable. The coefficient of correlation and the percentage of agreement between month specific tests decreased with increasing time interval between the tests. In addition, the mean difference between the two tests increased with the time interval between the two tests. These differences were statistically significant, but do not seem to have a clinical significance. Whether periods longer than 3 years between tests are associated with greater and clinically significant variability cannot be answered from this study. Moreover, our study revealed that long-term serum 25(OH)D measurements have an acceptable variability among males and females, and among premenopausal and postmenopausal women, with only minor and clinically non significant differences between the groups. Rejnmark et al. [16] assessed the variability of serum 25(OH)D level in 187 postmenopausal women. They found that among those classified in the lower tertile at baseline, 40% remained in this tertile at subsequent tests. And 32% of those classified in the highest baseline tertile remained in the highest tertile in subsequent tests. The authors of the study concluded that it seems questionable to use a single estimate as a predictor of individual vitamin D status [16]. Hofmann et al.
100% 5.7%
9.7%
90%
24.2%
Serum 25(OH)D level at the last test
15.5%
80% 33.4% 58.7%
70% 29.2%
60% 50%
53.8%
40% 44.2%
30% 49.6%
35.9%
20% 20.0%
10% 12.6%
2.0%
0%
4.5% 0.9%
Serum 25(OH)D level at the first test 50-74.9 nmol/L
30-49.9 nmol/L
<30 nmol/L
Fig. 1. Categories of serum 25(OH)D level at the first test and the corresponding categories at the last test.
W. Saliba et al. / European Journal of Internal Medicine 23 (2012) e106–e111
e109
Table 3 Measurements of association between the first and the last serum 25(OH)D tests by time interval between tests and stratified by gender. All (n = 8881) Statistics
Pearson's coefficient Differencea (nmol/L) Mean (SD) 95% CI for the mean Median (IQR) Absolute differenceb (nmol/L) Mean (SD) 95% CI for the mean Median (IQR) Agreement measure (%) 2 categories of serum 25(OH)Dc 4 categories of serum 25(OH)Dd
Time interval between the two serum 25(OH)D tests Same year n = 387
After 1 year n = 4930
After 2 years n = 2682
After 3 years n = 882
0.83
0.63
0.60
0.55
3.0 (14.7) 1.6–4.5 0.9 (− 5.2–9.2)
4.3 (20.6) 3.8–4.9 2.7 (− 7.5–14.4)
4.7 (21.5) 3.9–5.5 2.6 (− 8.4–15.2)
9.8 (23.8) 8.3–11.4 7.5 (− 4.9–21.4)
10.7 (10.5) 9.6–11.7 7.3 (3.2–14.4)
14.9 (14.8) 14.5–15.3 10.9 (4.9–20.2)
15.9 (15.2) 15.3–16.4 11.4 (5.2–21.1)
18.6 (17.9) 17.4–19.8 13.2 (6.1–25.6)
81.9 64.9
75.2 51.0
72.9 49.9
71.7 47.1
Males (n = 2571) Statistics
Pearson's coefficient Differencea (nmol/L) Mean (SD) 95% CI for the mean Median (IQR) Absolute differenceb (nmol/L) Mean (SD) 95% CI for the mean Median (IQR) Agreement measure (%) 2 categories of serum 25(OH)Dc 4 categories of serum 25(OH)Dd
Time interval between the two serum 25(OH)D tests Same year n = 146
After 1 year n = 1559
After 2 years n = 706
After 3 years n = 160
0.83
0.64
0.60
0.55
3.3 (15.3) 0.77–5.8 1.3 (− 5.7–10.0)
3.2 (20.1) 2.2–4.2 1.9 (− 8.7–14.0)
3.2 (21.4) 1.6–4.8 1.4 (− 10.0–13.4)
7.0 (24.9) 3.1–10.9 5.5 (− 7.2–19.1)
11.4 (10.8) 9.6–13.1 8.0 (3.8–15.3)
14.7 (14.0) 14.0–15.4 11.0 (5.0–19.9)
15.5 (15.1) 14.3–16.6 11.4 (5.2–20.0)
18.4 (18.0) 15.6–21.2 13.4 (6.2–24.3)
80.1 65.7
74.9 51.7
74.2 52.5
75.6 51.8
Females (n = 6310) Statistics
Pearson's coefficient Differencea (nmol/L) Mean (SD) 95% CI for the mean Median (IQR) Absolute differenceb (nmol/L) Mean (SD) 95% CI for the mean Median (IQR) Agreement measure (%) 2 categories of serum 25(OH)Dc 4 categories of serum 25(OH)Dd a b c d
Time interval between the two serum 25(OH)D tests Same year n = 241
After 1 year n = 3371
After 2 years n = 1976
After 3 years n = 772
0.83
0.62
0.61
0.55
2.9 (14.3) 1.1–4.7 0.8 (− 5.1–8.7)
4.9 (20.8) 4.2–5.6 3.1 (− 6.9–14.6)
5.2 (21.5) 4.3–6.2 3.0 (− 7.9–16.1)
10.5 (23.6) 8.8–12.2 7.9 (− 4.4–22.0)
10.3 (10.4) 8.9–11.6 7.1 (3.1–13.6)
15.1 (15.2) 14.6–15.6 10.9 (4.8–20.5)
16.0 (15.3) 15.4–16.7 11.4 (5.2–21.7)
18.6 (17.8) 17.3–20.0 13.1 (6.0–25.8)
83.0 64.3
75.2 50.6
72.3 48.9
70.9 45.9
Difference between the last and the first serum 25(OH)D test results. The absolute difference between the last and the first serum 25(OH)D test results. Serum 25(OH)D in two categories (b50, and ≥50 nmol/L). Serum 25(OH)D in four categories (b30, 30 to 49.9, 50 to 74.9, and ≥75 nmol/L).
[14] studied 29 subjects from the PLCO screening trial. Each of these subjects had 3 serum 25(OH)D tests taken at the same month; at baseline, after 1 year, and at 5 years. The intra-class correlation coefficient (ICC) was 0.71 (95% CI: 0.63–0.77). Similar to our findings the correlation between the baseline values and subsequent values decreased with increased time between the tests; after 1 year was 0.65, and after 5 years was 0.53 [14]. Jorde et al. [15] studied 2668 subjects and found that depending on the method of adjusting for season, the correlation between two measurements of serum 25(OH)D taken 14 years apart ranged between 0.42 and 0.52. However, in a subgroup of 94 subjects the correlation was 0.8 between tests taken after 12 months [15]. In both studies [14,15] the authors concluded that a single time measurement of serum 25(OH)D level may be an appropriate approach for the assessment of long term vitamin D status.
The Institute of Medicine (IOM)'s document on Dietary Reference Intakes (DRIs) for Vitamin D and calcium states that persons are at risk of deficiency at serum 25OHD below 30 nmol/L. Some, but not all, persons are potentially at risk for inadequacy at serum 25OHD levels between 30 and 50 nmol/L, and practically all persons are sufficient at serum 25OHD 50 nmol/L and above. Serum 25OHD above 75 nmol/L is not consistently associated with increased benefit [17,18]. This is in line with our recent findings that a threshold of 50 nmol/L was sufficient for serum PTH suppression and prevention of secondary hyperparathyroidism [19]. When the threshold of 50 nmol/L was used to indicate sufficiency for levels ≥50 nmol/L, and deficiency or inadequacy for levels b50 nmol/L we noted that the more the first tests level were higher or lower the more likely subjects remain in their first category (≥50 versus b50 nmol/L) at the last test (Fig. 1). Similar findings were revealed by Jorde et al.
e110
W. Saliba et al. / European Journal of Internal Medicine 23 (2012) e106–e111
Table 4 Measurements of association between the first and the last serum 25(OH)D tests among female by age group (n = 6310). Statistics
Pearson's coefficient Differenceb (nmol/L) Mean (SD) 95% CI for the mean Median (IQR) Absolute differencec (nmol/L) Mean (SD) 95% CI for the mean Median (IQR) Agreement measure (%) 2 categories of serum 25(OH)Dd 4 categories of serum 25(OH)De a b c d e
Age groupa ≤55 years (n = 2975)
>55 years (n = 3335)
0.66
0.58
5.0 (20.5) 4.3–5.8 3.1 (− 7.0–14.9)
6.0 (21.9) 5.3–6.8 3.7 (− 6.9–16.8)
15.0 (14.7) 14.5–15.6 10.7 (4.8–20.6)
16.1 (16.0) 15.5–16.6 11.4 (5.1–21.7)
74.9 51.5
73.4 48.8
This age cut point was used as a proxy for the menopausal state. Difference between the last and the first serum 25(OH)D test results. The absolute difference between the last and the first serum 25(OH)D test results. Serum 25(OH)D in two categories (b 50, and ≥ 50 nmol/L). Serum 25(OH)D in four categories (b30, 30 to 49.9, 50 to74.9, and ≥75 nmol/L).
even with longer time between tests; they found that 75% of those with serum 25(OH)D levels b30 nmol/L had serum 25(OH)D levels b50 nmol/L 14 years later. Similarly, 87% of subjects with serum 25(OH)D levels >79 nmol/L had serum 25(OH)D levels >50 nmol/L 14 years later [16]. Interestingly, in our study the serum 25(OH)D levels at the time of the last test were significantly higher than those at the time of the first test. This may partially be explained by the increasing interest of the public in improving vitamin D status by changing their habits, like more frequent exposure to sunlight and acquiring vitamin D supplements over the counter. This hypothesis may be further supported by our recent finding of increasing interest of the general public in assessing their vitamin D status [11]. The results of our study and other studies [14,15] suggest that month specific serum 25(OH)D levels are relatively stable and may be a suitable estimate of the corresponding long term season specific level. However, most studies that assess the association of vitamin D with a clinical outcome report the estimate of risk for an outcome for a certain individual as a function of his serum 25(OH)D level based on a single measurement. It follows that two subjects who have the same serum 25(OH)D level but their measurements were taken in different seasons are classified at the same risk for an outcome which is not true because their season specific levels will be different. Similarly, the same subject that is classified at high risk for an outcome based on a single measurement of serum 25(OH)D during winter would have been classified at a lower risk if serum 25(OH)D measurement had been performed in summer and vice versa. As the season specific serum 25(OH)D levels are relatively stable, it is likely that the mean of two serum 25(OH)D levels at two seasons (winter– spring and summer–autumn) will also be stable, and may provide a better estimate of vitamin D status to estimate the risk of future clinical outcome. Several prospective cohort studies that assessed serum 25(OH)D levels with clinical outcomes did indeed adjust for seasonality by using quartiles of serum 25(OH)D within each month of blood drawings [20,21]. This approach assumes that individuals remain in the same serum 25(OH)D quartile through the year. Month specific quartiles should be determined in order to estimate the individual specific risk for an outcome by using this approach. Alternatively, although not within the scope of this study, serial serum 25(OH)D measurements can be analyzed in the framework of time-dependent covariates and thus address some of the limitations of using only single measurement. Our study included a very large number of subjects with a wide range of age while two out of the three previous studies included either a small number of subjects [14,16], or were limited to subjects
aged 55–70 years [14] or postmenopausal women [16]. Another strength of this study is that unlike other studies [14,15] blood samples were not stored for years and serum 25(OH)D tests were performed shortly after blood collection. Nevertheless, this study is a database-based and the study population may be selective and not representative of the general population. However, during the period of this study there was an increasing interest of the general public in assessing vitamin D status; hence, 25(OH)D measurement was not limited to high risk population. Although we excluded subjects who were treated with vitamin D supplements during the last six months before the first or before the last test by crossing our data with the computerized medications database, this does not exclude the possibility that some subjects acquired their supplements over the counter. In conclusion, month specific measurements are relatively stable and have an acceptable variability over a 3 years period. The mean of season's values may provide a better estimate of vitamin D status in order to estimate future clinical outcomes. Learning points • Long-term month specific serum 25(OH)D levels are relatively stable at least over a period of 3 years. • The correlation between repeated tests decreased with increasing time between the tests. • The more the baseline serum 25(OH)D levels are higher or lower the more likely subjects remain in their baseline category (≥50 versus b50 nmol/L). Conflict of interest No conflict of interest or financial disclosures were reported by the authors of this paper. References [1] Holick MF. Vitamin D deficiency. N Engl J Med 2007;357:266–81. [2] Holick MF, Binkley NC, Bischoff-Ferrari HA, Gordon CM, Hanley DA, Heaney RP, et al. Endocrine Society. Evaluation, treatment, and prevention of vitamin D deficiency: an Endocrine Society clinical practice guideline. J Clin Endocrinol Metab 2011;96:1911–30. [3] Wolden-Kirk H, Overbergh L, Christesen HT, Brusgaard K, Mathieu C. Vitamin D and diabetes: its importance for beta cell and immune function. Mol Cell Endocrinol 2011 [Epub ahead of print]. [4] Mattila C, Knekt P, Männistö S, Rissanen H, Laaksonen MA, Montonen J, et al. Serum 25-hydroxyvitamin D concentration and subsequent risk of type 2 diabetes. Diabetes Care 2007;30:2569–70.
W. Saliba et al. / European Journal of Internal Medicine 23 (2012) e106–e111 [5] Giovannucci E, Liu Y, Hollis BW, Rimm EB. 25-hydroxyvitamin D and risk of myocardial infarction in men: a prospective study. Arch Intern Med 2008;168: 1174–80. [6] Leu M, Giovannucci E. Vitamin D: epidemiology of cardiovascular risks and events. Best Pract Res Clin Endocrinol Metab 2011;25:633–46. [7] Ma Y, Zhang P, Wang F, Yang J, Liu Z, Qin H. Association between vitamin D and risk of colorectal cancer: a systematic review of prospective studies. J Clin Oncol 2011, doi:10.1200/JCO.2011.35.7566. [8] Engel P, Fagherazzi G, Boutten A, Dupré T, Mesrine S, Boutron-Ruault MC, et al. Serum 25(OH) vitamin D and risk of breast cancer: a nested case–control study from the French E3N cohort. Cancer Epidemiol Biomarkers Prev 2010;19: 2341–50. [9] Melamed ML, Michos ED, Post W, Astor B. 25-hydroxyvitamin D levels and the risk of mortality in the general population. Arch Intern Med 2008;168:1629–37. [10] Ford ES, Zhao G, Tsai J, Li C. Vitamin D and all-cause mortality among adults in USA: findings from the National Health and Nutrition Examination Survey Linked Mortality study. Int J Epidemiol 2011;40:998–1005. [11] Saliba W, Rennert HS, Kershenbaum A, Rennert G. Serum 25(OH)D concentrations in sunny Israel. Osteoporos Int 2011 [Epub ahead of print]. [12] Pearce SH, Cheetham TD. Diagnosis and management of vitamin D deficiency. BMJ 2010;340:142–7. [13] Yetley EA. Assessing the vitamin D status of the US population. Am J Clin Nutr 2008;88:558S–64S. [14] Hofmann JN, Yu K, Horst RL, Hayes RB, Purdue MP. Long-term variation in serum 25-hydroxyvitamin D concentration among participants in the prostate, lung,
[15]
[16]
[17]
[18]
[19] [20]
[21]
e111
colorectal, and ovarian cancer screening trial. Cancer Epidemiol Biomarkers Prev 2010;19:927–31. Jorde R, Sneve M, Hutchinson M, Emaus N, Figenschau Y, Grimnes G. Tracking of serum 25-hydroxyvitamin D levels during 14 years in a population-based study and during 12 months in an intervention study. Am J Epidemiol 2010;171:903–8. Rejnmark L, Lauridsen AL, Brot C, Vestergaard P, Heickendorff L, Nexo E, et al. Vitamin D and its binding protein Gc: long-term variability in peri- and postmenopausal women with and without hormone replacement therapy. Scand J Clin Lab Invest 2006;66:227–38. National Academy of Science, Institute of Medicine. Dietary references intake for calcium and vitamin D. http://www.nap.edu/catalog.php?record_id=13050 accessed 1 September 2011. Ross AC, Manson JE, Abrams SA, Aloia JF, Brannon PM, Clinton SK, et al. The 2011 report on dietary reference intakes for calcium and vitamin D from the Institute of Medicine: what clinicians need to know. J Clin Endocrinol Metab 2011;96: 53–8. Saliba W, Barnett O, Rennert HS, Lavi I, Rennert G. The relationship between serum 25(OH)D and parathyroid hormone levels. Am J Med 2011;124:1165–70. Dobnig H, Pilz S, Scharnagl H, Renner W, Seelhorst U, Wellnitz B, et al. Independent association of low serum 25-hydroxyvitamin D and 1,25-dihydroxyvitamin d levels with all-cause and cardiovascular mortality. Arch Intern Med 2008;168: 1340–9. Hutchinson MS, Grimnes G, Joakimsen RM, Figenschau Y, Jorde R. Low serum 25hydroxyvitamin D levels are associated with increased all-cause mortality risk in a general population: the Tromsø study. Eur J Endocrinol 2010;162:935–42.
Contents lists available at SciVerse ScienceDirect
European Journal of Internal Medicine journal homepage: www.elsevier.com/locate/ejim
Original article
The longitudinal variability of serum 25(OH)D levels Walid Saliba a,⁎, Ofra Barnett a, Nili Stein a, Anne Kershenbaum a, Gad Rennert a, b a Department of Community Medicine and Epidemiology, Carmel Medical Center, Clalit Health Services, and Bruce Rappaport Faculty of Medicine, Technion-Israel Institute of Technology, Haifa, Israel b Department of Epidemiology and Disease Prevention, Office of the Chief Physician, Clalit Health Services Headquarters, Tel Aviv, Israel
a r t i c l e
i n f o
Article history: Received 16 October 2011 Received in revised form 4 January 2012 Accepted 7 February 2012 Available online 17 March 2012 Keywords: Vitamin D 25(OH)D Variability Seasonality
a b s t r a c t Background: The extent to which a single serum 25(OH)D measurement represents long-term vitamin D status remains unclear. This study aims to assess the variability of serum 25(OH)D between tests taken at different time intervals. Methods: Using the computerized database of the largest healthcare provider in Israel, we identified subjects in whom a serum 25(OH)D test was performed on at least two different occasions between January 2008 and September 2011 (n = 188,771). For these subjects we selected the first and the last dated tests, then we identified those who were not treated with supplements during the last 6 months before the first and before the last test (n = 94,418). Of these we analyzed subjects in whom the first and the last tests were performed in the same month of the year (n = 8881). Results: The mean serum 25(OH)D level at the first test was 51.7 ± 24.0 nmol/L and was 56.7 ± 24.7 at the last test (P b 0.001); the overall correlation was 0.63 (P b 0.001). For vitamin D status in two categories (b 50 versus ≥ 50 nmol/L), the percentage of agreement between the first and last tests was 74.4%, and was 50.8% for vitamin D status in four categories (b 30, 30–49.9, 50–74.9, and ≥75 nmol/L). The correlation decreased with increasing time between the tests ranging from 0.83 for tests done at the same year to 0.55 after 3 years. The more the first levels were higher or lower, the more likely subjects remain in their first category (≥ 50 versus b 50 nmol/L). Conclusions: Long-term month specific serum 25(OH)D levels are relatively stable. © 2012 European Federation of Internal Medicine. Published by Elsevier B.V. All rights reserved.
1. Introduction Vitamin D is important in supporting and maintaining normal mineral and bone health [1]. In recent years, new data linking vitamin D with extra-skeletal health has emerged [2]. Although, studies have shown conflicting evidence regarding the extra-skeletal manifestations of vitamin D deficiency, low serum 25(OH)D levels have been found to be associated with increased risk for mortality and morbidity from a variety of chronic diseases including diabetes mellitus [3,4], cardiovascular diseases [5,6], cancer [7,8], and all cause of mortality [9,10]. Most of the studies linking vitamin D with extra-skeletal health are based on single measurement of serum 25(OH)D. This approach is based on the assumption that serum 25(OH)D levels are stable over time, while several factors may influence serum 25(OH)D levels. Vitamin D production by skin exposure to sun UVB irradiation is the main source of vitamin D [1,2]. Very few foods naturally contain or are fortified with vitamin D [2]. Hence, oral intake of vitamin D ⁎ Corresponding author at: Department of Community Medicine and Epidemiology, Carmel Medical Center, 7 Michal St., Haifa 34362, Israel. Tel.: + 972 4 825 0474; fax: + 972 4 834 4358. E-mail address: [email protected] (W. Saliba).
from foodstuffs contributes to a lesser degree to vitamin D status particularly in countries where fortified food and supplements are uncommon. As exposure to UVB irradiation is the major determinant of serum 25(OH)D levels, this may result in great variability in serum 25(OH)D levels depending on the season of the year and outdoor activity [1,11,12]. Serum 25(OH)D levels also decrease with increasing age [11,13]. Hence, a single time measurement of 25(OH)D may not be representative of long term vitamin D status. Little is known about the longitudinal variability of serum 25(OH) D levels [14–16]. We thought that the Clalit Health Services (CHS) database may provide a suitable platform that may help to clarify this issue. 2. Materials and methods 2.1. Study population and data collection We used data from the database of Clalit Health Services (CHS) which is a non-for-profit health maintenance organization (HMO) covering more than half of the Israeli population. The database, which includes data of laboratory tests results, and use of medications, was searched for all available serum 25(OH)D test results from January 2008 until September 2011 (1,347,349 tests in
0953-6205/$ – see front matter © 2012 European Federation of Internal Medicine. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.ejim.2012.02.002
W. Saliba et al. / European Journal of Internal Medicine 23 (2012) e106–e111
709,186 subjects which constitutes more than 18% of the CHS population). To ensure consistency only tests that were performed in the two largest laboratories, performing more than 62.2% of the tests, were selected for this analysis (839,043 tests in 444,729 subjects). Of these we selected subjects who had at least two tests during the study period (n = 188,771). In subjects who performed more than two tests during this period, the results of the first and the most recent tests were used. We excluded 1625 subjects; age less than 10 years, reported serum 25(OH)D level less than 10 nmol/L, and repeated testing in different laboratories (n = 187,146 after exclusion). In addition, to eliminate the effect of treatment with vitamin D supplement on the variability of serum 25(OH)D level, we excluded 92,728 subjects who were treated with vitamin D supplements during the last six months before the first and before the last tests (n = 94,418 after exclusion). To eliminate the effect of seasonality which is known to be an important contributor to the variability in serum 25(OH)D levels we selected subjects in whom the first and the last tests were performed in the same month of the year (n = 8881). 2.2. Statistical analysis The following analyses were performed to assess the association between the baseline and last serum 25(OH)D measurements: 1. Pearson's correlation coefficient; 2. Location and spread statistics of the difference between the two tests; 3. Location and spread statistics of the absolute difference between the two tests; 4. Two classification of vitamin D status were used based on serum 25(OH)D levels (b50, and ≥50 nmol/L) and (b30, 30 to 49.9, 50 to 74.9, and ≥75 nmol/L). The percentage of agreement between the categories of the two tests was calculated. For all analyses, P value of less than 0.05 for the 2-tailed tests was considered statistically significant. All statistical analyses were performed using SPSS 18.0 (SPSS Inc., Chicago). 2.3. 25(OH)D assay 25(OH)D was tested in two central laboratories. Both laboratories used the LIAISON® 25-OH Vitamin D Total Assay (DiaSorin), a competitive two-step chemiluminescence assay. The measuring range is 4.0– 150 ng/mL (10–375 nmol/L), the analytical sensitivity b 1.0 ng/mL (2.5 nmol/L), and the functional sensitivityb 4.0 ng/mL (10 nmol/L). The intra-assay imprecision is up to 5%, and the inter-assay precision is up to 15%. The specificity for 25-OH Vitamin D2 is 104%, and for 25-OH Vitamin D3 is 100%. Performance characteristics of the Vitamin D assay were checked in the method evaluation process done by “Clalit Health Services” and were compatible to the manufacturer-generated data. The accuracy of the measurements in the individual laboratory is confirmed by in-house daily QC monitoring and by periodic external QC program (DEQAS). 3. Results 3.1. Baseline characteristics of the study population Of the 8881 included subjects 6310 (61.1%) were females and 7895 (88.9%) were Jews. The mean age was 56.1 ± 17.6 years at the time of the first test (Table 1). 4290 (48.3%) of the tests were performed in winter–spring while 4591 (51.7%) of the tests were performed in summer–autumn. The mean of serum 25(OH)D on the first tests results were higher in summer–autumn, 55.1 ± 24.1 nmol/L, as compared to winter–spring, 48.0 ± 23.3 nmol/L (P b 0.001), and the mean of serum 25(OH)D on the last tests results were 58.1 ± 23.6 nmol/L and 54.3 ± 25.6 nmol/L respectively (P b 0.001).
e107
Table 1 Characteristics of the subjects at the time of the first and at the time of the last serum 25(OH)D tests (n = 8881). Variable
At the time of the first test
Gender Males Females
2571 (28.9%) 6310 (71.1%)
Ethnicity Jews Arabs
7895 (88.9%) 966 (11.1%)
Age (years) Mean ± SD Median (IQR)
56.1 ± 17.6 58.0 (45.0–69.0)
Season of the test Winter–Spring Summer–Autumn
4290 (48.3%) 4591 (51.7%)
At the time of the last test
57.6 ± 17.7 59.0 (46.0–71.0)
Serum 25(OH)D (nmol/L) Mean ± SD 51.7 ± 24.0 Median (IQR) 50.3 (33.8–66.7)
56.7 ± 24.7 55.3 (39.0–71.9)
Serum 25(OH)D b 50 nmol/L ≥ 50 nmol/L
3643 (41.0%) 5238 (59.0%)
4406 (49.6%) 4475 (50.4%)
3.2. The variability of serum 25(OH)D The mean serum 25(OH)D level was 51.7 ± 24.0 nmol/L at the time of the first test and 56.7 ± 24.7 nmol/L at the time of the last test (Table 1) with mean difference of 5.0 nmol/L (95% CI 4.5–5.4) nmol/L. The Pearson's r correlation coefficient was 0.63 (P b 0.001). The absolute difference between the first and last test was ≤5.0 nmol/L in 25% of subjects, b11.1 nmol/L in 50% of subjects while 75% of subjects had an absolute difference ≤20.8 nmol/L (Table 2). For vitamin D status in two categories (b50 versus ≥50 nmol/L), the percentage of agreement between the results of the first test and the results of the last test was 74.4%, and was 50.8% for vitamin D status in four categories (b30, 30–49.9, 50–74.9, and ≥75 nmol/L) (Table 2). 49.6% of subjects with first serum 25(OH)D level b30 nmol/L remained within this range at the last test, and 78.8% had serum 25(OH)D level b50 nmol/L. 78% of subjects with first serum 25(OH) D 50–74.9 nmol/L had levels ≥50 nmol/L at the last test. While
Table 2 Measurements of association between the first and the last serum 25(OH)D tests by gender. Statistics
All (n = 8881)
Males (n = 2571)
Females (n = 6310)
Pearson's coefficient Differencea (nmol/L) Mean (SD) 95% CI for the mean Median (IQR)
0.63
0.64
0.62
5.0 (21.0) 4.5–5.4 3 (− 7.4–15.1)
3.4 (20.5) 2.6–4.2 1.8 (− 8.7–13.8)
5.6 (21.2) 5.0–6.1 3.4 (− 6.9–15.8)
14.9 (14.5) 14.4–15.5 11.0 (5.0–20.0)
15.6 (15.4) 15.2–16.0 11.0 (5.0–21.1)
75.0
74.1
52.7
50.1
Absolute differenceb (nmol/L) Mean (SD) 15.4 (15.2) 95% CI for the mean 15.1–15.7 Median (IQR) 11.1 (5.0–20.8) Agreement measure (%) 2 categories of serum 74.4 25(OH)Dc 4 categories of serum 50.8 25(OH)Dd a b c d
Difference between the last and the first serum 25(OH)D test results. The absolute difference between the last and the first serum 25(OH)D test results. Serum 25(OH)D in two categories (b50, and ≥50 nmol/L). Serum 25(OH)D in four categories (b30, 30 to 49.9, 50 to74.9, and ≥75 nmol/L).
e108
W. Saliba et al. / European Journal of Internal Medicine 23 (2012) e106–e111
94.6% of subjects with the first serum 25(OH)D ≥75 nmol/L had levels ≥50 nmol/L at the last test (Fig. 1). Stratified analyses by the time interval between the first and the last test revealed that the mean difference between the two tests increased gradually with increasing time between the tests and ranged between 3 nmol/L for tests performed at the same year to 9.8 nmol/L for tests performed 3 years apart (Table 3). The Pearson's correlation coefficient also decreased gradually with increasing time between tests and ranged between 0.83 for tests performed at the same year to 0.55 after 3 years (Table 3). In addition, the percentage of agreement between the two tests decreased with increasing time interval between the first and the last test (Table 3). 3.3. The variability of serum 25(OH)D by gender The mean serum 25(OH)D levels were higher in males as compared to females (P b 0.001), 55.0 ± 24.0 nmol/L and 50.4 ± 23.5 nmol/L respectively at the time of the first test, and 58.4 ± 24.3 nmol/L and 56.0 ± 24.8 nmol/L respectively at the time of the last test. The mean difference between the two tests was slightly higher in females as compared to males, 5.6 nmol/L (95% CI 5.0–6.1) and 3.4 nmol/L (95% CI 2.6–4.2) respectively. The Pearson's r correlation coefficient between the first and the last test was 0.64 in males, and 0.62 in females. The absolute difference between the two tests and the measurement of agreement was similar between males and females (Table 2). Stratified analyses by the time interval between the first and the last test revealed that the pattern of the variability measurements was similar between males and females. However, the percentage of agreement between the two tests decreased with increasing time interval between the two tests among females while it was relatively stable among males after 1–3 years (Table 3). 3.4. The variability of serum 25(OH)D in special subgroups of women The menopausal state was not available for the female population of this study. Thus for the purpose of this study, in order to evaluate the effect of the menopausal state on the variability of serum
25(OH)D, we defined postmenopausal state as age greater than 55 years (n = 3335). The Pearson's r correlation coefficient between the first and the last test was 0.66 for premenopausal women, and 0.58 for postmenopausal women. The other measurements of variability were similar between the two groups (Table 4). To evaluate the effect of current usage of oral contraceptives on the variability of serum 25(OH)D levels in premenopausal women (n = 2975), we defined current users as those who had filled at least one prescription for oral contraceptives during the last 8 weeks before the first and/or before the last serum 25(OH)D tests (n = 225). Current users of oral contraceptives had a slightly lower correlation coefficient between the first and the last test 0.60 as compared to 0.66 among non current users. However, the other measurements of variability were similar between the two groups.
4. Discussion Long-term month specific serum 25(OH)D measurements were relatively stable. The coefficient of correlation and the percentage of agreement between month specific tests decreased with increasing time interval between the tests. In addition, the mean difference between the two tests increased with the time interval between the two tests. These differences were statistically significant, but do not seem to have a clinical significance. Whether periods longer than 3 years between tests are associated with greater and clinically significant variability cannot be answered from this study. Moreover, our study revealed that long-term serum 25(OH)D measurements have an acceptable variability among males and females, and among premenopausal and postmenopausal women, with only minor and clinically non significant differences between the groups. Rejnmark et al. [16] assessed the variability of serum 25(OH)D level in 187 postmenopausal women. They found that among those classified in the lower tertile at baseline, 40% remained in this tertile at subsequent tests. And 32% of those classified in the highest baseline tertile remained in the highest tertile in subsequent tests. The authors of the study concluded that it seems questionable to use a single estimate as a predictor of individual vitamin D status [16]. Hofmann et al.
100% 5.7%
9.7%
90%
24.2%
Serum 25(OH)D level at the last test
15.5%
80% 33.4% 58.7%
70% 29.2%
60% 50%
53.8%
40% 44.2%
30% 49.6%
35.9%
20% 20.0%
10% 12.6%
2.0%
0%
4.5% 0.9%
Serum 25(OH)D level at the first test 50-74.9 nmol/L
30-49.9 nmol/L
<30 nmol/L
Fig. 1. Categories of serum 25(OH)D level at the first test and the corresponding categories at the last test.
W. Saliba et al. / European Journal of Internal Medicine 23 (2012) e106–e111
e109
Table 3 Measurements of association between the first and the last serum 25(OH)D tests by time interval between tests and stratified by gender. All (n = 8881) Statistics
Pearson's coefficient Differencea (nmol/L) Mean (SD) 95% CI for the mean Median (IQR) Absolute differenceb (nmol/L) Mean (SD) 95% CI for the mean Median (IQR) Agreement measure (%) 2 categories of serum 25(OH)Dc 4 categories of serum 25(OH)Dd
Time interval between the two serum 25(OH)D tests Same year n = 387
After 1 year n = 4930
After 2 years n = 2682
After 3 years n = 882
0.83
0.63
0.60
0.55
3.0 (14.7) 1.6–4.5 0.9 (− 5.2–9.2)
4.3 (20.6) 3.8–4.9 2.7 (− 7.5–14.4)
4.7 (21.5) 3.9–5.5 2.6 (− 8.4–15.2)
9.8 (23.8) 8.3–11.4 7.5 (− 4.9–21.4)
10.7 (10.5) 9.6–11.7 7.3 (3.2–14.4)
14.9 (14.8) 14.5–15.3 10.9 (4.9–20.2)
15.9 (15.2) 15.3–16.4 11.4 (5.2–21.1)
18.6 (17.9) 17.4–19.8 13.2 (6.1–25.6)
81.9 64.9
75.2 51.0
72.9 49.9
71.7 47.1
Males (n = 2571) Statistics
Pearson's coefficient Differencea (nmol/L) Mean (SD) 95% CI for the mean Median (IQR) Absolute differenceb (nmol/L) Mean (SD) 95% CI for the mean Median (IQR) Agreement measure (%) 2 categories of serum 25(OH)Dc 4 categories of serum 25(OH)Dd
Time interval between the two serum 25(OH)D tests Same year n = 146
After 1 year n = 1559
After 2 years n = 706
After 3 years n = 160
0.83
0.64
0.60
0.55
3.3 (15.3) 0.77–5.8 1.3 (− 5.7–10.0)
3.2 (20.1) 2.2–4.2 1.9 (− 8.7–14.0)
3.2 (21.4) 1.6–4.8 1.4 (− 10.0–13.4)
7.0 (24.9) 3.1–10.9 5.5 (− 7.2–19.1)
11.4 (10.8) 9.6–13.1 8.0 (3.8–15.3)
14.7 (14.0) 14.0–15.4 11.0 (5.0–19.9)
15.5 (15.1) 14.3–16.6 11.4 (5.2–20.0)
18.4 (18.0) 15.6–21.2 13.4 (6.2–24.3)
80.1 65.7
74.9 51.7
74.2 52.5
75.6 51.8
Females (n = 6310) Statistics
Pearson's coefficient Differencea (nmol/L) Mean (SD) 95% CI for the mean Median (IQR) Absolute differenceb (nmol/L) Mean (SD) 95% CI for the mean Median (IQR) Agreement measure (%) 2 categories of serum 25(OH)Dc 4 categories of serum 25(OH)Dd a b c d
Time interval between the two serum 25(OH)D tests Same year n = 241
After 1 year n = 3371
After 2 years n = 1976
After 3 years n = 772
0.83
0.62
0.61
0.55
2.9 (14.3) 1.1–4.7 0.8 (− 5.1–8.7)
4.9 (20.8) 4.2–5.6 3.1 (− 6.9–14.6)
5.2 (21.5) 4.3–6.2 3.0 (− 7.9–16.1)
10.5 (23.6) 8.8–12.2 7.9 (− 4.4–22.0)
10.3 (10.4) 8.9–11.6 7.1 (3.1–13.6)
15.1 (15.2) 14.6–15.6 10.9 (4.8–20.5)
16.0 (15.3) 15.4–16.7 11.4 (5.2–21.7)
18.6 (17.8) 17.3–20.0 13.1 (6.0–25.8)
83.0 64.3
75.2 50.6
72.3 48.9
70.9 45.9
Difference between the last and the first serum 25(OH)D test results. The absolute difference between the last and the first serum 25(OH)D test results. Serum 25(OH)D in two categories (b50, and ≥50 nmol/L). Serum 25(OH)D in four categories (b30, 30 to 49.9, 50 to 74.9, and ≥75 nmol/L).
[14] studied 29 subjects from the PLCO screening trial. Each of these subjects had 3 serum 25(OH)D tests taken at the same month; at baseline, after 1 year, and at 5 years. The intra-class correlation coefficient (ICC) was 0.71 (95% CI: 0.63–0.77). Similar to our findings the correlation between the baseline values and subsequent values decreased with increased time between the tests; after 1 year was 0.65, and after 5 years was 0.53 [14]. Jorde et al. [15] studied 2668 subjects and found that depending on the method of adjusting for season, the correlation between two measurements of serum 25(OH)D taken 14 years apart ranged between 0.42 and 0.52. However, in a subgroup of 94 subjects the correlation was 0.8 between tests taken after 12 months [15]. In both studies [14,15] the authors concluded that a single time measurement of serum 25(OH)D level may be an appropriate approach for the assessment of long term vitamin D status.
The Institute of Medicine (IOM)'s document on Dietary Reference Intakes (DRIs) for Vitamin D and calcium states that persons are at risk of deficiency at serum 25OHD below 30 nmol/L. Some, but not all, persons are potentially at risk for inadequacy at serum 25OHD levels between 30 and 50 nmol/L, and practically all persons are sufficient at serum 25OHD 50 nmol/L and above. Serum 25OHD above 75 nmol/L is not consistently associated with increased benefit [17,18]. This is in line with our recent findings that a threshold of 50 nmol/L was sufficient for serum PTH suppression and prevention of secondary hyperparathyroidism [19]. When the threshold of 50 nmol/L was used to indicate sufficiency for levels ≥50 nmol/L, and deficiency or inadequacy for levels b50 nmol/L we noted that the more the first tests level were higher or lower the more likely subjects remain in their first category (≥50 versus b50 nmol/L) at the last test (Fig. 1). Similar findings were revealed by Jorde et al.
e110
W. Saliba et al. / European Journal of Internal Medicine 23 (2012) e106–e111
Table 4 Measurements of association between the first and the last serum 25(OH)D tests among female by age group (n = 6310). Statistics
Pearson's coefficient Differenceb (nmol/L) Mean (SD) 95% CI for the mean Median (IQR) Absolute differencec (nmol/L) Mean (SD) 95% CI for the mean Median (IQR) Agreement measure (%) 2 categories of serum 25(OH)Dd 4 categories of serum 25(OH)De a b c d e
Age groupa ≤55 years (n = 2975)
>55 years (n = 3335)
0.66
0.58
5.0 (20.5) 4.3–5.8 3.1 (− 7.0–14.9)
6.0 (21.9) 5.3–6.8 3.7 (− 6.9–16.8)
15.0 (14.7) 14.5–15.6 10.7 (4.8–20.6)
16.1 (16.0) 15.5–16.6 11.4 (5.1–21.7)
74.9 51.5
73.4 48.8
This age cut point was used as a proxy for the menopausal state. Difference between the last and the first serum 25(OH)D test results. The absolute difference between the last and the first serum 25(OH)D test results. Serum 25(OH)D in two categories (b 50, and ≥ 50 nmol/L). Serum 25(OH)D in four categories (b30, 30 to 49.9, 50 to74.9, and ≥75 nmol/L).
even with longer time between tests; they found that 75% of those with serum 25(OH)D levels b30 nmol/L had serum 25(OH)D levels b50 nmol/L 14 years later. Similarly, 87% of subjects with serum 25(OH)D levels >79 nmol/L had serum 25(OH)D levels >50 nmol/L 14 years later [16]. Interestingly, in our study the serum 25(OH)D levels at the time of the last test were significantly higher than those at the time of the first test. This may partially be explained by the increasing interest of the public in improving vitamin D status by changing their habits, like more frequent exposure to sunlight and acquiring vitamin D supplements over the counter. This hypothesis may be further supported by our recent finding of increasing interest of the general public in assessing their vitamin D status [11]. The results of our study and other studies [14,15] suggest that month specific serum 25(OH)D levels are relatively stable and may be a suitable estimate of the corresponding long term season specific level. However, most studies that assess the association of vitamin D with a clinical outcome report the estimate of risk for an outcome for a certain individual as a function of his serum 25(OH)D level based on a single measurement. It follows that two subjects who have the same serum 25(OH)D level but their measurements were taken in different seasons are classified at the same risk for an outcome which is not true because their season specific levels will be different. Similarly, the same subject that is classified at high risk for an outcome based on a single measurement of serum 25(OH)D during winter would have been classified at a lower risk if serum 25(OH)D measurement had been performed in summer and vice versa. As the season specific serum 25(OH)D levels are relatively stable, it is likely that the mean of two serum 25(OH)D levels at two seasons (winter– spring and summer–autumn) will also be stable, and may provide a better estimate of vitamin D status to estimate the risk of future clinical outcome. Several prospective cohort studies that assessed serum 25(OH)D levels with clinical outcomes did indeed adjust for seasonality by using quartiles of serum 25(OH)D within each month of blood drawings [20,21]. This approach assumes that individuals remain in the same serum 25(OH)D quartile through the year. Month specific quartiles should be determined in order to estimate the individual specific risk for an outcome by using this approach. Alternatively, although not within the scope of this study, serial serum 25(OH)D measurements can be analyzed in the framework of time-dependent covariates and thus address some of the limitations of using only single measurement. Our study included a very large number of subjects with a wide range of age while two out of the three previous studies included either a small number of subjects [14,16], or were limited to subjects
aged 55–70 years [14] or postmenopausal women [16]. Another strength of this study is that unlike other studies [14,15] blood samples were not stored for years and serum 25(OH)D tests were performed shortly after blood collection. Nevertheless, this study is a database-based and the study population may be selective and not representative of the general population. However, during the period of this study there was an increasing interest of the general public in assessing vitamin D status; hence, 25(OH)D measurement was not limited to high risk population. Although we excluded subjects who were treated with vitamin D supplements during the last six months before the first or before the last test by crossing our data with the computerized medications database, this does not exclude the possibility that some subjects acquired their supplements over the counter. In conclusion, month specific measurements are relatively stable and have an acceptable variability over a 3 years period. The mean of season's values may provide a better estimate of vitamin D status in order to estimate future clinical outcomes. Learning points • Long-term month specific serum 25(OH)D levels are relatively stable at least over a period of 3 years. • The correlation between repeated tests decreased with increasing time between the tests. • The more the baseline serum 25(OH)D levels are higher or lower the more likely subjects remain in their baseline category (≥50 versus b50 nmol/L). Conflict of interest No conflict of interest or financial disclosures were reported by the authors of this paper. References [1] Holick MF. Vitamin D deficiency. N Engl J Med 2007;357:266–81. [2] Holick MF, Binkley NC, Bischoff-Ferrari HA, Gordon CM, Hanley DA, Heaney RP, et al. Endocrine Society. Evaluation, treatment, and prevention of vitamin D deficiency: an Endocrine Society clinical practice guideline. J Clin Endocrinol Metab 2011;96:1911–30. [3] Wolden-Kirk H, Overbergh L, Christesen HT, Brusgaard K, Mathieu C. Vitamin D and diabetes: its importance for beta cell and immune function. Mol Cell Endocrinol 2011 [Epub ahead of print]. [4] Mattila C, Knekt P, Männistö S, Rissanen H, Laaksonen MA, Montonen J, et al. Serum 25-hydroxyvitamin D concentration and subsequent risk of type 2 diabetes. Diabetes Care 2007;30:2569–70.
W. Saliba et al. / European Journal of Internal Medicine 23 (2012) e106–e111 [5] Giovannucci E, Liu Y, Hollis BW, Rimm EB. 25-hydroxyvitamin D and risk of myocardial infarction in men: a prospective study. Arch Intern Med 2008;168: 1174–80. [6] Leu M, Giovannucci E. Vitamin D: epidemiology of cardiovascular risks and events. Best Pract Res Clin Endocrinol Metab 2011;25:633–46. [7] Ma Y, Zhang P, Wang F, Yang J, Liu Z, Qin H. Association between vitamin D and risk of colorectal cancer: a systematic review of prospective studies. J Clin Oncol 2011, doi:10.1200/JCO.2011.35.7566. [8] Engel P, Fagherazzi G, Boutten A, Dupré T, Mesrine S, Boutron-Ruault MC, et al. Serum 25(OH) vitamin D and risk of breast cancer: a nested case–control study from the French E3N cohort. Cancer Epidemiol Biomarkers Prev 2010;19: 2341–50. [9] Melamed ML, Michos ED, Post W, Astor B. 25-hydroxyvitamin D levels and the risk of mortality in the general population. Arch Intern Med 2008;168:1629–37. [10] Ford ES, Zhao G, Tsai J, Li C. Vitamin D and all-cause mortality among adults in USA: findings from the National Health and Nutrition Examination Survey Linked Mortality study. Int J Epidemiol 2011;40:998–1005. [11] Saliba W, Rennert HS, Kershenbaum A, Rennert G. Serum 25(OH)D concentrations in sunny Israel. Osteoporos Int 2011 [Epub ahead of print]. [12] Pearce SH, Cheetham TD. Diagnosis and management of vitamin D deficiency. BMJ 2010;340:142–7. [13] Yetley EA. Assessing the vitamin D status of the US population. Am J Clin Nutr 2008;88:558S–64S. [14] Hofmann JN, Yu K, Horst RL, Hayes RB, Purdue MP. Long-term variation in serum 25-hydroxyvitamin D concentration among participants in the prostate, lung,
[15]
[16]
[17]
[18]
[19] [20]
[21]
e111
colorectal, and ovarian cancer screening trial. Cancer Epidemiol Biomarkers Prev 2010;19:927–31. Jorde R, Sneve M, Hutchinson M, Emaus N, Figenschau Y, Grimnes G. Tracking of serum 25-hydroxyvitamin D levels during 14 years in a population-based study and during 12 months in an intervention study. Am J Epidemiol 2010;171:903–8. Rejnmark L, Lauridsen AL, Brot C, Vestergaard P, Heickendorff L, Nexo E, et al. Vitamin D and its binding protein Gc: long-term variability in peri- and postmenopausal women with and without hormone replacement therapy. Scand J Clin Lab Invest 2006;66:227–38. National Academy of Science, Institute of Medicine. Dietary references intake for calcium and vitamin D. http://www.nap.edu/catalog.php?record_id=13050 accessed 1 September 2011. Ross AC, Manson JE, Abrams SA, Aloia JF, Brannon PM, Clinton SK, et al. The 2011 report on dietary reference intakes for calcium and vitamin D from the Institute of Medicine: what clinicians need to know. J Clin Endocrinol Metab 2011;96: 53–8. Saliba W, Barnett O, Rennert HS, Lavi I, Rennert G. The relationship between serum 25(OH)D and parathyroid hormone levels. Am J Med 2011;124:1165–70. Dobnig H, Pilz S, Scharnagl H, Renner W, Seelhorst U, Wellnitz B, et al. Independent association of low serum 25-hydroxyvitamin D and 1,25-dihydroxyvitamin d levels with all-cause and cardiovascular mortality. Arch Intern Med 2008;168: 1340–9. Hutchinson MS, Grimnes G, Joakimsen RM, Figenschau Y, Jorde R. Low serum 25hydroxyvitamin D levels are associated with increased all-cause mortality risk in a general population: the Tromsø study. Eur J Endocrinol 2010;162:935–42.