J Clin Endocrin Metab. First published ahead of print May 9, 2012 as doi:10.1210/jc.

2012-1176

ORIGINAL

ARTICLE C a r e

E n d o c r i n e

A Reverse J-Shaped Association of All-Cause Mortality with Serum 25-Hydroxyvitamin D in General Practice, the CopD Study
D. Durup, H. L. Jrgensen, J. Christensen, P. Schwarz, A. M. Heegaard, and B. Lind
Department of Drug Design and Pharmacology (D.D., A.M.H.), Faculty of Health and Medical Sciences, University of Copenhagen, 2100 Copenhagen, Denmark; Department of Clinical Biochemistry (H.L.J.), Bispebjerg Hospital, 2400 Copenhagen, Denmark; Department of Statistics and Epidemiology (J.C.), the Danish Cancer Society, 2100 Copenhagen Denmark; Research Center of Aging and Osteoporosis (P.S.), Department of Medicine, Glostrup University Hospital and Faculty of Health and Medical Sciences, University of Copenhagen, 2200 Copenhagen, Denmark; and Copenhagen General Practitioners Laboratory (B.L.), 1112 Copenhagen, Denmark

Context: Optimal levels of vitamin D have been a topic of heavy debate, and the correlation between 25-hydroxyvitamin D [25(OH)D] levels and mortality still remains to be established. Objective: The aim of the study was to determine the association between all-cause mortality and serum levels of 25(OH)D, calcium, and PTH. Design and Setting: We conducted a retrospective, observational cohort study, the CopD Study, in a single laboratory center in Copenhagen, Denmark. Participants: Serum 25(OH)D was analyzed from 247,574 subjects from the Copenhagen general practice sector. In addition, serum levels of calcium, albumin-adjusted calcium, PTH, and creatinine were measured in 111,536; 20,512; 34,996; and 189,496 of the subjects, respectively. Main Outcome Measures: Multivariate Cox regression analysis was used to compute hazard ratios for all-cause mortality. Results: During follow-up (median, 3.07 yr), 15,198 (6.1%) subjects died. A reverse J-shaped association between serum level of 25(OH)D and mortality was observed. A serum 25(OH)D level of 50 60 nmol/liter was associated with the lowest mortality risk. Compared to 50 nmol/liter, the hazard ratios (95% confidence intervals) of all-cause mortality at very low (10 nmol/liter) and high (140 nmol/liter) serum levels of 25(OH)D were 2.13 (2.022.24) and 1.42 (1.311.53), respectively. Similarly, both high and low levels of albumin-adjusted serum calcium and serum PTH were associated with an increased mortality, and secondary hyperparathyroidism was associated with higher mortality (P 0.0001). Conclusion: In this study from the general practice sector, a reverse J-shaped relation between the serum level of 25(OH)D and all-cause mortality was observed, indicating not only a lower limit but also an upper limit. The lowest mortality risk was at 50 60 nmol/liter. The study did not allow inference of causality, and further studies are needed to elucidate a possible causal relationship between 25(OH)D levels, especially higher levels, and mortality. (J Clin Endocrinol Metab 97: 0000 0000, 2012)

ISSN Print 0021-972X ISSN Online 1945-7197 Printed in U.S.A. Copyright 2012 by The Endocrine Society doi: 10.1210/jc.2012-1176 Received January 23, 2012. Accepted April 18, 2012.

Abbreviations: CI, Confidence interval; CV%, coefficient of variation percentage; 25(OH)D, 25-hydroxyvitamin D; SHPT, secondary hyperparathyroidism.

J Clin Endocrinol Metab, August 2012, 97(8):0000 0000

jcem.endojournals.org

Copyright (C) 2012 by The Endocrine Society

Durup et al.

25(OH)D and Mortality in General Practice

J Clin Endocrinol Metab, August 2012, 97(8):0000 0000

itamin D insufficiency (25-hydroxyvitamin D [25(OH)D] 50nmol/liter)isprevalent(13)andhasbeensuggested to be involved in various diseases such as diabetes, cardiovascular disease, depression, immune system diseases, and certain cancers (4 8). The biomarker used for the determination of vitamin D status is 25(OH)D, rather than the biologically active hormone 1,25-dihydroxyvitamin D (9). The association between 25(OH)D levels and all-cause mortality has been estimated in several studies (3, 10 23). Most of the studies have investigated the association between low levels of 25(OH)D and mortality and found an increased risk (1215, 19, 20, 22, 23); some studies suggested an inverse relation (3, 11, 21), whereas other studies did not find any association with low levels of 25(OH)D and mortality risk (10, 16, 18). Although the preponderance of focus has been directed toward vitamin D insufficiency, the U.S. Food and Nutrition Board at the Institute of Medicine (IOM) has indicated that higher concentrations of 25(OH)D (below toxicity levels) may also be of concern; this was based on a reevaluation of data in the literature, including some of the above-mentioned studies (24). Two studies have investigated the association between mortality and higher levels of 25(OH)D and found both high and low levels of 25(OH)D to be associated with increased risk of overall mortality (14, 17). An optimal level of vitamin D is a topic of heavy debate among health care professionals, and further investigations are needed to provide evidence of how 25-hydroxyvitamin D levels relate to mortality. In this context, there is a special need for studies that also evaluate higher levels of 25(OH)D. To address this issue, a database containing serum 25(OH)D measurements from 247,574 subjects from the general practice sector was used to determine the association between serum levels of 25(OH)D and mortality, including higher as well as lower levels. Additionally, the study investigated the association between mortality and serum levels of albumin-adjusted serum calcium and PTH because abnormal serum levels of PTH and calcium, as well as vitamin D, have been reported to be associated with increased mortality (16, 18, 2528).

study period, the CGPL has analyzed blood samples from a total of 776,954 subjects, of which 247,574 subjects (32%) had serum 25(OH)D analyzed. The CGPL is the sole laboratory that serves the general practitioners in the greater Copenhagen area, and it covers approximately 1.1 million inhabitants. In addition to serum 25(OH)D, serum levels of calcium, albumin-adjusted calcium, PTH, and creatinine were measured in 111,536; 20,512; 34,996; and 189,496 of the subjects, respectively. A total of 94.8% of the subjects had their blood collected either at the CGPL or at the office of their primary health care physician, who then sent the blood samples to the CGPL for further analysis. The rest of the blood samples (5.2%) were taken by the CGPL at institutions/residential care or home visits. The personal identification number (CPR number) is unique to every citizen in Denmark and enables matching of individuals to registers. All mortality information was extracted from the civil registration database on the same day (August 25, 2011) corresponding to end of study. All subjects were followed from the date of blood measurement until the date of emigration, date of death, or August 25, 2011, whichever came first; the median follow-up for all subjects was 3.07 yr (1.59; 5.96; 5th and 95th percentiles). All necessary approvals from the Danish Data Protection Agency were obtained before collecting data (no. 2010-41-4826).

Biochemical analyses 25(OH)D assays

25(OH)D was assessed in serum by two commercially available assays, LIAISON 25(OH)D assay (Diasorin, Saluggia, Italy) and OCTEIA 25(OH)D3 and 25(OH)D2 (Immunodiagnostic Systems, Ltd., Boldon, UK) according to the instructions of the manufacturers. Both assays determine the sum of 25(OH)D3 and 25(OH)D2. For the LIAISON assay, the interserial coefficient of variation percentage (CV%) was 12.5% (at level 41 nmol/liter). For the OCTEIA assay, the interserial CV% was 9% (at level 32 nmol/liter). Results obtained by the OCTEIA assay were adjusted to results obtained by LIAISON using the equation: LIAISON 0.893 OCTEIA 0.48. The equation was determined by parallel analysis of 59 human serum samples during a period of 5 d in August 2007. The LIAISON assay was used after August 19, 2007 and until the end of study. Both assays were subject to external quality control through participation in the Vitamin D External Quality Assessment Scheme (DEQAS; Charing Cross Hospital, London, UK). The assessment scheme included four distributions annually. Each distribution comprised five samples. The results from DEQAS through the entire study period (from 2004 to 2010) confirmed the reliability of the assays, and the results from CGPL deviated less than 15% from the method mean.

Subjects and Methods

Study subjects
The Copenhagen General Practitioners Laboratory (CGPL) serves physicians in the primary care sector of the greater Copenhagen area mainly by conducting a wide range of blood tests. In this study, the CGPL database was accessed, and subjects with a serum 25(OH)D measurement were included. The first and the last subject included in the study had their blood measurements on April 29, 2004, and January 22, 2010, respectively. If a subject had more than one serum 25(OH)D measurement, only the first measurement was used. During the

PTH assay
PTH was determined in serum by the commercially available ADVIA Centaur iPTH kit (Bayer/Siemens, Tarrytown, NY) according to the instructions of the manufacturer (upper limit of the normal range, 7.6 pmol/liter). The interserial CV% was 9.1% (at level 2.76 pmol/liter) and 5.6% (at level 26.2 pmol/ liter). The assay is specific for intact PTH (amino acid 1 84).

J Clin Endocrinol Metab, August 2012, 97(8):0000 0000

jcem.endojournals.org

Calcium assay
Total calcium was determined in serum by the Advia Chemistry System (Bayer/Siemens) using Arsenazo III reagents. Results were traceable to and adjusted to the target values of Reference Serum X, Nordic Society of Clinical Chemistry . The interserial CV% was 1.6% (at level 2.43 mmol/liter) and 1.2% (at level 3.13 mmol/liter).

Albumin assay
Albumin was determined in serum by the commercially available Advia Chemistry System (Bayer/Siemens) albumin kit according to the instructions of the manufacturer. The interserial CV% was 2.1% (at level 31.6 g/liter) and 1.7% (at level 40.6 g/liter).

Albumin-adjusted calcium
Albumin-adjusted calcium was calculated as: total calcium (in mmol/liter) 0.020 (41.3 albumin (g/liter)).

portions, and the differences were analyzed using 2 tests. P values less than 0.05 were considered statistically significant. The nonlinear association between all-cause mortality and serum level of 25(OH)D, albumin-adjusted serum calcium, and serum PTH was analyzed using a Cox proportional hazards model, where serum levels of 25(OH)D, albumin-adjusted serum calcium, and serum PTH were entered in the model as a restricted cubic spline with five knots placed at the 5th, 27th, 50th, 73rd, and 95th percentiles of serum 25(OH)D, albumin-adjusted serum calcium, and serum PTH. A serum 25(OH)D concentration of 50 nmol/liter was used as a reference. For albumin-adjusted serum calcium and serum PTH, the reference levels were set at 2.35 mmol/liter and 4.5 pmol/liter, respectively. The models were adjusted for season of blood sampling (JanuaryMarch, AprilJune, JulySeptember, and OctoberDecember), age, and gender. All statistical analyses were performed using SAS statistical software (SAS Institute, Inc., Cary, NC) or the computing environment R (R Development Core Team, 2005).

Creatinine assay
Creatinine was determined in serum by the commercially available ADVIA Chemistry System (Bayer/Siemens) creatinine (Jaffe) kit according to the instructions of the manufacturer. The interserial CV% was 2.0% (at level 92 mol/liter and level 527 mol/liter).

Results
A total of 247,574 subjects with a serum 25(OH)D measurement were included in the CopD Study. Table 1 shows the characteristics of the study population by serum levels of 25(OH)D. Measures of serum 25(OH)D were obtained in almost twice as many women as men, and in the total study population, an average of 54.4% suffered from vitamin D insufficiency [25(OH)D 50 nmol/liter]. The study included a wide age range, and vitamin D insuffi-

Statistical analyses
Normally distributed variables were shown as mean (SD), and differences between groups were analyzed using unpaired t tests. Nonnormally distributed variables were shown as medians with 5 and 95% percentiles, and Mann-Whitney U tests were used to test for differences. Categorical variables were shown as pro-

TABLE 1. Characteristics of the CopD study population by serum levels of 25(OH)D

Serum level of 25(OH)D (nmol/liter) All n Age (yr), mean (SD) Gender (female/male) Dead Age groups, n (%) 0 15 yr 1530 yr 30 45 yr 45 60 yr 60 75 yr 75 yr PTH (n) Serum PTH (pmol/liter), median [5th; 95th percentile] Ca (n) Serum Ca (mmol/liter), mean (SD) Albumin-corrected Ca (n) Albumin-corrected serum Ca (mmol/liter), mean (SD) Creatinine (n) Serum creatinine ( mol/liter), mean (SD) 110 nmol/liter (%) 247,574 51.0 (20.4) 65.2%/34.8% 6.1% 7,756 (3.1) 35,670 (14.4) 55,919 (22.6) 59,068 (23.9) 54,530 (22.0) 34,631 (14.0) 34,996 4.5 [1.9; 11.2] <12.5 13,885 46.6 (20.3) 60.6%/39.4% 8.1% 529 (3.8) 2,652 (19.1) 3,963 (28.6) 3, 125 (22, 5) 2,003 (14.4) 1,613 (11.6) 2,272 7.0 [2.9; 23.5] 12.525 41,804 47.9 (20.1) 61.6%/38.4% 6.9% 1,492 (3.6) 7,038 (16.8) 11,606 (27.8) 10,004 (23.9) 6,756 (16.2) 4,908 (11.7) 6,560 5.7 [2.4; 14.2] 2550 82,442 51.1 (20.1) 62.6%/37.4% 5.9% 2,742 (3.3) 11,150 (13.5) 18,813 (22.8) 20,473 (24.9) 17,896 (21.7) 11,368 (13.8) 11,543 4.6 [2.0; 10.3] 50 75 67,462 53.0 (20.2) 67.5%/32.5% 5.3% 1,927 (2.9) 8,338 (12.4) 13,478 (20.0) 16,153 (23.9) 17,275 (25.6) 10.291 (15.2) 8,965 4.0 [1.8; 8.5] 75100 29,680 52.9 (20, 7) 71.2%/28.8% 5.9% 764 (2.6) 4,212 (14.2) 5,635 (19.0) 6,721 (22.6) 7,770 (26.2) 4,578 (15.4) 3,960 3.7 [1.6; 7.7] 100 125 8,539 51.6 (21.5) 73.0%/27% 8.0% 221 (2.6) 1,499 (17.6) 1,669 (19.6) 1,771 (20.7) 2,034 (23.8) 1,345 (15.7) 1,169 3.5 [1.5; 7.4] 125150 2,397 50.3 (21.5) 74.0%/26% 8.2% 52 (2.2) 490 (20.5) 478 (19.9) 499 (20.8) 516 (21.5) 362 (15.1) 344 3.2 [1.4; 7.0] >150 1,365 48.9 (20.8) 73.7%/26.3% 8.4% 29 (2.1) 291 (21.3) 277 (20.3) 322 (23.6) 280 (20.5) 166 (12.2) 183 3.1 [1.1; 6.8] P value 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001

111,536 2.31 (0.10)

6,487 2.32 (0.11)

18,855 2.34 (0.11)

36,919 2.36 (0.10)

30,258 2.36 (0.10)

13,379 2.36 (0.10)

3,921 2.37 (0.11)

1,119 2.37 (0.11)

598 2.39 (0.13)

0.0001 0.0001

20,512 2.13 (0.10)

1,146 2.27 (0.10)

3,639 2.29 (0.10)

6,855 2.31 (0.10)

5,481 2.32 (0.10)

2,376 2.32 (0.10)

717 2.32 (0.10)

194 2.33 (0.10)

104 2.37 (0.19)

0.0001 0.0001

189,496 89.4 (22.6)

10,200 86.3 (25.4)

31,338 88.2 (24.3)

63,268 89.8 (22.1)

52,016 89.9 (19.7)

23,000 89,9 (19.2)

6,685 89.9 (19.2)

1,917 89.8 (26.2)

1,072 88.1 (18.3)

0.0001 0.0001

91.4%

92.5%

91.9%

90.9%

91.4%

91.3%

91.4%

91.7%

93.5%

0.0001

Ca, Total calcium. To convert to US units, divide by 2.496 for 25(OH)D (nanograms per milliliter), 0.105 for PTH (picograms per milliliter), and 83.3 for serum creatinine (micrograms per deciliter).

Durup et al.

25(OH)D and Mortality in General Practice

J Clin Endocrinol Metab, August 2012, 97(8):0000 0000

ciency existed within all age groups. Furthermore, the study subjects were also well represented in the very lower and higher levels of 25(OH)D): 13,885 subjects had 25(OH)D levels below or equal to 12.5 nmol/liter, and 1,365 had 25(OH)D levels equal to or higher than 150 nmol/liter. More than 90% of the subjects who had a creatinine measurement, had a creatinine level below 110 nmol/liter, both at low and high serum levels of 25(OH)D, indicating functional kidneys. Finally, from February to April, the prevalence of vitamin D insufficiency was significantly higher than from July to September (P 0.0001), both among women (62 vs. 38%) and among men (74 vs. 39%). All-cause mortality The mortality data were extracted at the date the study ended, at which point a total of 15,198 (6.1%) subjects had died (follow-up: median, 3.07 yr). In the study population, increased mortality was observed at low as well as

high serum levels of 25(OH)D, albumin-adjusted calcium, and PTH (Fig. 1, AC). From these figures, the optimum level of 25(OH)D, albumin-adjusted calcium, and serum PTH has been determined and used to conduct Cox survival analysis for values below the optimum level and values above the optimum level (Table 2). All-cause mortality associated with serum 25(OH)D levels The association between all-cause mortality and serum levels of 25(OH)D was reverse J-shaped (Fig. 1A). A serum 25(OH)D level of 50 60 nmol/liter was associated with the lowest mortality risk. The hazard ratios [95% confidence intervals (CI)] of all-cause mortality at very low (10 nmol/liter) and high (140 nmol/liter) serum levels of 25(OH)D were 2.13 (2.022.24) and 1.42 (1.311.53), respectively. In subjects at least 60 yr old, the mortality was, as in the total population, lowest at a serum 25(OH)D level of 50 60 nmol/liter. In this age group, the

FIG. 1. Hazard ratios of all-cause mortality by restricted cubic spline Cox regression analysis. Estimates were adjusted for age, sex, and season of blood sampling according to: A, serum 25(OH)D level with 50 nmol/liter as the reference value; B, albumin-adjusted serum calcium level with 2.35 mmol/liter as the reference value; and C, serum PTH level with 4.5 pmol/liter as the reference value. The horizontal dashed line corresponds to the normal reference hazard ratio of 1.0, values above are associated with increased mortality risk, and values below are associated with decreased mortality risk compared with the reference value.

J Clin Endocrinol Metab, August 2012, 97(8):0000 0000

jcem.endojournals.org

TABLE 2. Stepwise Cox multivariate regression model analysis for mortality for low and high levels of serum 25(OH)D, albumin-adjusted serum calcium, and serum PTH
Variables in equation 25(OH)D 50 nmol/liter Alive/dead, 126,857/8,743 Sex (male vs. female) Season (summer vs. winter) Age (per 1 yr) 25(OH)D (per nmol/liter decrease) 25(OH)D 50 nmol/liter Alive/dead: 105,519/6,455 Sex (male vs. female) Age (per 1 yr) 25(OH)D (per nmol/liter increase) Alb-Ca 2.37 mmol/liter Alive/dead: 14,630/1213 Alb-Ca (per mmol/liter decrease) Sex (male vs. female) Season (summer vs. winter) Age (per 1 yr) Alb-Ca 2.37 mmol/liter Alive/dead: 4,131/538 Alb-Ca (per mmol/liter increase) Sex (male vs. female) Age (per 1 yr) PTH 4.6 pmol/liter Alive/dead: 16,342/843 Sex (male vs. female) PTH (per pmol/liter decrease) Age (per 1 yr) PTH 4.5 pmol/liter Alive/dead: 16,446/1,365 Sex (male vs. female) Age (per 1 yr) PTH (per nmol/liter increase) ln (estimate) ln (SE)
2

P value

Hazard ratio (95% CI)

0.57112 0.17478 0.09558 0.02030 0.43660 0.10672 0.00470 1.75381 0.45992 0.13061 0.09877 2.21423 0.51963 0.08954 0.59541 0.21775 0.10264 0.53462 0.09907 0.02399

0.02216 0.02167 0.0008145 0.0008833 0.02636 0.00112 0.0005051 0.32594 0.05921 0.05772 0.00238 0.41028 0.09599 0.00373 0.07373 0.03727 0.00280 0.05697 0.00222 0.00203

664.4419 65.0417 13771.6912 528.0449 274.3168 9030.0349 86.5633 28.9521 60.3284 5.1206 1729.2257 29.1261 29.3035 576.4659 65.2156 34.1402 1340.0913 88.0626 1986.8131 139.4755

0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0236 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001

1.770 (1.6951.849) 1.191 (1.1411.243) 1.100 (1.099 1.102) 1.020 (1.018 1.022) 1.547 (1.470 1, 629) 1.113 (1.110 1.115) 1.005 (1.004 1.006) 5.780 (3.049 10.989) 1.584 (1.410 1.779) 1.140 (1.018 1.276) 1.104 (1.099 1.109) 9.154 (4.096 20.458) 1.681 (1.3932.029) 1.094 (1.086 1.102) 1.814 (1.570 2.096) 1.244 (1.156 1.33) 1.108 (1.1021.114) 1.707 (1.526 1.908) 1.104 (1.099 1.109) 1.024 (1.020 1.028)

Overall mortality. The covariables are listed according to significance in the final model. The variables that did not reach significance in a model are not included. Alb-Ca, Albumin-adjusted serum calcium.

hazard ratios (95% CI) of all-cause mortality at low (10 nmol/liter) and high (140 nmol/liter) serum levels of 25(OH)D were 2.04 (1.94 2.17) and 1.44 (1.321.56), respectively. To account for the possibility that many of the deaths during the first year after vitamin D measurement could be due to illnesses already present at baseline, the data were analyzed excluding from the all-cause mortality analysis those with death within 1 yr of the blood collection. This left a population with a total of 10,104 deaths, and the J-shape of the curve remained the same although the hazard ratio for all-cause mortality both at the highest and lowest serum levels of 25(OH)D was slightly decreased. The same is true when excluding those with deaths within 2 yr; then, the hazard ratio and power further decreased (data not shown). A similar J-shaped correlation was found for a subgroup consisting of the 23,271 subjects who had serum PTH, calcium, and creatinine measurements performed at the same time as their 25(OH)D measurement. However, due to the limited number of subjects, approximately one

tenth of the original population, the CI were wider (data not shown). All-cause mortality associated with serum levels of albumin-adjusted calcium and PTH Increased mortality was also observed at low and high serum levels of albumin-adjusted calcium and PTH (Fig. 1, B and C). All-cause mortality risk was lowest at an albumin-adjusted serum calcium level of 2.352.37 mmol/liter. The hazard ratios (95% CI) of all-cause mortality at low (1.5 mmol/liter) and high (3.2 mmol/liter) serum levels of albumin-adjusted calcium were 5.9 (3.310.5) and 6.1 (3.0 12.0), respectively. In subjects at least 60 yr old, the mortality was also lowest at an albumin-adjusted serum calcium level of 2.352.39 mmol/liter (data not shown). In this age group, the hazard ratios (95% CI) of all-cause mortality at low (1.5 mmol/liter) and high (3.2 mmol/liter) albumin-adjusted serum calcium were 5.8 (3.310.4) and 3.2 (3.0 11.2), respectively. All-cause mortality risk was lowest at a serum PTH level of approximately 4.5 4.8 pmol/liter. The hazard ra-

Durup et al.

25(OH)D and Mortality in General Practice

J Clin Endocrinol Metab, August 2012, 97(8):0000 0000

tios (95% CI) of all-cause mortality at low (1 pmol/liter) and high (120 pmol/liter) serum levels of PTH were 1.7 (1.4 2.1) and 8.2 (5.8 11.5), respectively. In subjects at least 60 yr old, the mortality was, as in the total population, also lowest at a serum PTH level of approximately 4.5 4.8 pmol/liter (data not shown). In this age group, the hazard ratios (95% CI) of all-cause mortality at low (1 pmol/liter) and high (120 pmol/liter) serum levels of PTH were 1.8 (1.4 2.1) and 7.5 (5.210.8), respectively. Secondary hyperparathyroidism (SHPT)prevalence and all-cause mortality A total of 23,271 subjects had serum PTH, calcium, and creatinine measurements performed at the same time as their 25(OH)D measurement. A total of 227 suffered from primary hyperparathyroidism (defined as serum PTH 7.6 nmol/liter and calcium 2.52 mmol/liter), 1886 of the subjects had impaired renal function (defined as serum creatinine 110 mol/liter), and 37 subjects suffered from both primary hyperparathyroidism and impaired renal function. These subjects were excluded from the SHPT analysis, which then consisted of 21,195 subjects. In this subgroup, 10.5% suffered from SHPT, which was defined as serum 25(OH)D below 50 nmol/liter; serum PTH higher than 7.6 pmol/liter; normal or reduced serum calcium ( 2.52 mmol/liter), and normal serum creatinine ( 110 mol/liter). SHPT was found in all age groups and for both sexes. All-cause mortality among subjects suffering from SHPT was compared to subjects in the same subgroup but not suffering from SHPT (model 1, Table 3). In this comparison suffering from SHPT was significantly associated with higher mortality (P 0.05) in all age groups except for the age groups 0 15 and 30 45 yr (Table 3). All-cause mortality among subjects suffering from hyperparathyroidism independent of serum 25(OH)D levels was also compared to subjects in the same subgroup but not suffering from hyperparathyroidism (model 2, Table 3). Fi-

nally, all-cause mortality among subjects suffering from vitamin D insufficiency was compared to subjects in the same subgroup but not suffering from vitamin D insufficiency (model 3, Table 3). The models demonstrated that mortality was the same among subjects suffering from hyperparathyroidism or SHPT. Thus, compared with vitamin D insufficiency, the mortality was increased in models 1 and 2 (Table 3). The importance of PTH as a possible predictor of mortality was further demonstrated in another analysis showing that subjects suffering from hypovitaminosis D but without an increase in serum levels of PTH had decreased mortality compared with subjects suffering from hypovitaminosis D and raised serum PTH (Table 4).

Discussion
In the CopD Study based on 247,574 subjects from the primary care sector, both low (hazard ratio, 2.13) and high (hazard ratio, 1.42) serum levels of 25(OH)D were associated with increased all-cause mortality. Data exhibited a reverse J-shaped association between all-cause mortality and serum level of 25(OH)D, with the lowest mortality rate at 50 60 nmol/liter when adjusted for age, sex, and season of blood sample. Additionally, both low and high serum levels of albumin-adjusted calcium and PTH were associated with an increased mortality risk. The lowest mortality risk observed in these two graphs corresponded to the normal reference range for albumin-adjusted serum calcium (2.172.53 mmol/liter) and serum PTH (1.57.6 pmol/liter). The controversial findings that high as well as low levels of 25(OH)D are associated with increased risk of all-cause mortality support the concerns stated by the IOM when they reviewed the literature (24). The present study indicates that the association between vitamin D status and mortality is driven by individuals in the very low and high

TABLE 3. All-cause mortality related to SHPT, hyperparathyroidism, and vitamin D insufficiency by age
Model 1: SHPT (PTH > 7.6 pmol/liter, 25(OH)D < 50 nmol/liter, calcium < 2.51 mmol/liter, and creatinine < 110 nmol/liter) No SHPT/ having SHPT 592/83 2,452/217 4,250/594 5,113/562 4,316/426 2,243/347 18,966/2,229 All-cause mortality: no SHPT/ having SHPT 0.17%/0% 0.04%/0.46% 0.38%/0.51% 2.01%/3.38% 6.28%/11.74% 25.90%/31.70% 5.13%/8.21% Model 2: Hyperparathyroidism (PTH > 7.6 pmol/liter, calcium < 2.51 mmol/liter, creatinine < 110 nmol/liter) No HPT/ having HPT 589/86 2,437/232 4,190/654 4,982/693 4,178/564 2,167/423 18,543/2,652 All-cause mortality: no HPT/ having HPT 0.17%/0% 0.04%/0.43% 0.38%/0.46% 1.97%/3.46% 6.20%/10.99 25.75%/31.44% 5.03%/8.41% Model 3: Vitamin D insufficiency [25(OH)D < 50 nmol/liter, calcium < 2.51 mmol/liter, and creatinine < 110 nmol/liter] No insufficiency/ having insufficiency 288/387 1,137/1,532 1,789/3,046 2,559/3,116 2,632/2,110 1,316/1,274 9,730/11,465 All-cause mortality: no insufficiency/having insufficiency 0%/0.26% 0.09%/0.07% 0.22%/0.49% 1.72%/2.50% 5.74%/8.06% 24.32%/29.12% 5.34%/5.55%

Age (yr) 0 15 1530 30 45 45 60 60 75 75 All

P value NS 0.05 NS 0.05 0.0001 0.05 0.0001

P value NS 0.05 NS 0.05 0.0001 0.05 0.0001

P value NS NS NS 0.05 0.005 0.01 NS

No SHPT, Subjects not suffering from SHPT; having SHPT, subjects suffering from SHPT; no HPT, subjects not suffering from hyperparathyroidism; having HPT, subjects suffering from hyperparathyroidism. NS, Not significant; HPT, hyperparathyroidism.

J Clin Endocrinol Metab, August 2012, 97(8):0000 0000

jcem.endojournals.org

TABLE 4. The effect of PTH on all-cause mortality at different vitamin D levels

Mortality 25(OH)D 75 nmol/liter AND PTH 7.6 nmol/liter (n 23,905) PTH 7.6 nmol/liter (n 5,217) 25(OH)D 50 nmol/liter AND PTH 7.6 nmol/liter (n 15,589) PTH 7.6 nmol/liter (n 4,478) 25(OH)D 25 nmol/liter AND PTH 7.6 nmol/liter (n 5,769) PTH 7.6 nmol/liter (n 2,755) 25(OH)D 12.5 nmol/liter AND PTH 7.6 nmol/liter (n 1,262) PTH 7.6 nmol/liter (n 1,010) 4.99% 11.31% 5.02% 10.88% 4.70% 10.31% 4.04% 10.99% P value 0.0001 0.0001 0.0001 0.0001

levels of 25(OH)D, but many of the existing studies on vitamin D and mortality do not evaluate the very low and especially not the very high 25(OH)D levels because most of the of focus has been directed toward the effects of vitamin D insufficiency and deficiency. However, when the IOM reviewed the studies by Jia et al. (11) and Visser et al. (12), a U-shaped or reverse J-shaped association between all-cause mortality and 25(OH)D levels was demonstrated, with an optimum 25(OH)D level at 50 75 nmol/liter (11, 12, 24). Similar trends can be seen in data from the studies by Ford et al. (3) and Cawthon et al. (16) where the lowest mortality was found at serum 25(OH)D levels at approximately 60 75 nmol/liter and in the study by Hutchinson et al. (15) with lowest mortality at serum 25(OH)D levels at approximately 60 80 nmol/liter among smokers, but not among nonsmokers (3, 15, 16). In some of the published studies on vitamin D and mortality, the study designs have not allowed for investigating the effect of high 25(OH)D levels on mortality because the highest exposure categories were pools of subjects with 25(OH)D levels above 50 or above 64 nmol/liter (13, 19, 20). Furthermore, in the study by Virtanen et al. (13), only a few of the study subjects (approximately 55) had 25(OH)D levels above 75 nmol/liter, and among these only four died in the follow-up period. These results highlight the importance of sufficient study size to analyze a possible effect in the very low and high levels of 25(OH)D. One study has not shown an association between allcause mortality and 25(OH)D levels, but this study included very frail old institutionalized persons, and 40% of the study population died during the 31-month follow-up (18). Melamed et al. (14) and Michaelsson et al. (17) both demonstrated a U-shaped or reverse J-shaped association with lowest mortality risk at 75 nmol/liter and 45 80 nmol/liter, respectively. Our results indicate an optimal 25(OH)D level between 50 60 nmol/liter, which is consistent with the above literature, and our results are in line with the statement from the U.S. Food and Nutrition

Board at the IOM that a serum 25(OH)D level of 50 nmol/ liter represents adequacy, and above 125 nmol/liter the risk of harm increases (14). The present study contains a very large number of 25(OH)D measurements from a single laboratory covering general practice in Copenhagen. The number of vitamin D measurements analyzed at the Copenhagen General Practitioners Laboratory has grown exponentially during the last decade, and 14% of all requests of blood samples received at CPGL in the study period included serum 25(OH)D, demonstrating that the indication for measuring vitamin D has been very broad. This is possibly due to an increased awareness of vitamin D insufficiency not just among health care professionals but also in the public. Because access to medical care is free to all citizens in Denmark, this heightened public awareness of the consequences of vitamin D deficiency has led to requests for measurements of 25(OH)D from the patients when they visit their general practitioner for whatever reason. The concern about vitamin D insufficiency could be justified because 54.4% of the individuals in the CopD Study suffered from vitamin D insufficiency. It should be noted that in the general population the prevalence of vitamin D insufficiency might be less than 54.4% because the sampling in the present study is based on subjects who had a vitamin D measurement. Nevertheless, the Danish population is vulnerable to vitamin D insufficiency, primarily in the winter season (October to March) due to the geographic location of the country (latitude 55N to 57N), where sunlight exposure is limited. Historically, beverage and food in Denmark have not been fortified; in 2009, sale of fortified milk and bread was allowed, but the producers are still hesitating. Additionally, the Recommended Daily Intake of vitamin D is low compared with the recommendations in the literature, so the finding was not surprising (29 33). Vitamin D and PTH are the two most important hormones in the tight regulation of calcium. 25(OH)D levels below 50 nmol/liter are associated with risk of SHPT (34). In the CopD Study, 10.5% suffered from SHPT, and SHPT was significantly associated with increased mortality. Furthermore, our study showed PTH to be the dominant factor in the SHPT prediction of all-cause mortality (Tables 3 and 4). This is in line with a cohort study including 6307 elderly frail subjects that found SHPT, but not vitamin D, as a risk factor of mortality (18). Other studies have shown elevated PTH to be associated with increased mortality and morbidity (16, 27, 35, 36). One limitation of the present study that must be emphasized is that our study design did not allow inference of causality because the data does not hold information on causes of death or important covariates such as body mass

Durup et al.

25(OH)D and Mortality in General Practice

J Clin Endocrinol Metab, August 2012, 97(8):0000 0000

index, health status, education, smoking, and ethnicity, which may be important predictors of low serum 25(OH)D levels because 25(OH)D may be a marker of general bad health rather than a risk factor for mortality (10, 37, 38). Furthermore, the categorization of the subjects was based on single baseline measurements of serum 25(OH)D, PTH, and calcium; this could lead to misclassification because single measurements might not reflect long-term biochemical status. Another limitation to the study is the inclusion of subjects exclusively from the Copenhagen area, and hence, our results may not apply to subjects from other geographical areas. The major strength of the CopD Study was the very large sample size, probably the largest sample size of 25(OH)D measurements from a single center, allowing robust analyses of single parameters, such as age, season, gender, and survival analysis with narrow CI. The study is also robust in the very low and high levels of 25(OH)D, with 20,666 subjects having 25(OH)D levels below 10 nmol/liter and 2,527 subjects having 25(OH)D levels higher than 140 nmol/liter. The large number of deaths (15,198) enabled investigation of the correlation between mortality and serum concentrations of 25(OH)D, albumin-adjusted calcium, and PTH. Additionally, the present study included a wider age range than has been seen in most studies on this topic and is from the primary care sector, thus not limited to an elderly and/or sick population. In conclusion, the present study of 247,574 subjects from the primary care sector showed a reverse J-shaped relation between serum level of 25(OH)D and all-cause mortality, with the lowest mortality at 50 60 nmol/liter. This finding underscores the importance of including the very low and especially the very high levels of vitamin D in the analysis. High and low levels of serum albumin-adjusted calcium and PTH were also associated with increased mortality. Furthermore, vitamin D insufficiency was prevalent (54.8%), and 10.5% suffered from SHPT. At 45 yr of age and older, a diagnosis of SHPT was associated with higher mortality. The study did not allow inference of causality, and further studies are needed to elucidate a potential causal relationship between 25(OH)D levels and mortality. Furthermore, there is a need for randomized control trials in this research area.

maceutical Sciences, for critical review and editing of the manuscript. Address all correspondence and requests for reprints to: Anne-Marie Heegaard, Department of Drug Design and Pharmacology, Faculty of Health and Medical Sciences, University of Copenhagen. 2100 Universitetsparken 2, Copenhagen, Denmark. E-mail: amhe@farma.ku.dk. Disclosure Summary: D.D., H.L.J., J.C., P.S., A.M.H., and B.L. have nothing to declare.

References
1. Forrest KY, Stuhldreher WL 2011 Prevalence and correlates of vitamin D deficiency in US adults. Nutr Res 31:48 54 2. Mosekilde L, Nielsen LR, Larsen ER, Moosgaard B, Heickendorff L 2005 [Vitamin D deficiency. Definition and prevalence in Denmark]. Ugeskr Laeger 167:29 33 3. Ford ES, Zhao G, Tsai J, Li C 2011 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 40:998 1005 4. Yin L, Grandi N, Raum E, Haug U, Arndt V, Brenner H 2011 Meta-analysis: circulating vitamin D and ovarian cancer risk. Gynecol Oncol 121:369 375 5. Holick MF 2007 Vitamin D deficiency. N Engl J Med 357:266 281 6. Autier P, Gandini S 2007 Vitamin D supplementation and total mortality: a meta-analysis of randomized controlled trials. Arch Intern Med 167:1730 1737 7. Rosen CJ 2011 Clinical practice. Vitamin D insufficiency. N Engl J Med 364:248 254 8. Gandini S, Boniol M, Haukka J, Byrnes G, Cox B, Sneyd MJ, Mullie P, Autier P 2011 Meta-analysis of observational studies of serum 25-hydroxyvitamin D levels and colorectal, breast and prostate cancer and colorectal adenoma. Int J Cancer 128:1414 1424 9. Kennel KA, Drake MT, Hurley DL 2010 Vitamin D deficiency in adults: when to test and how to treat. Mayo Clin Proc 85:752757; quiz 757758 10. Eaton CB, Young A, Allison MA, Robinson J, Martin LW, Kuller LH, Johnson KC, Curb JD, Van Horn L, McTiernan A, Liu S, Manson JE 2011 Prospective association of vitamin D concentrations with mortality in postmenopausal women: results from the Womens Health Initiative (WHI). Am J Clin Nutr 94:14711478 11. Jia X, Aucott LS, McNeill G 2007 Nutritional status and subsequent all-cause mortality in men and women aged 75 years or over living in the community. Br J Nutr 98:593599 12. Visser M, Deeg DJ, Puts MT, Seidell JC, Lips P 2006 Low serum concentrations of 25-hydroxyvitamin D in older persons and the risk of nursing home admission. Am J Clin Nutr 84:616 622; quiz 671 672 13. Virtanen JK, Nurmi T, Voutilainen S, Mursu J, Tuomainen TP 2011 Association of serum 25-hydroxyvitamin D with the risk of death in a general older population in Finland. Eur J Nutr 50:305312 14. Melamed ML, Michos ED, Post W, Astor B 2008 25-Hydroxyvitamin D levels and the risk of mortality in the general population. Arch Intern Med 168:1629 1637 15. Hutchinson MS, Grimnes G, Joakimsen RM, Figenschau Y, Jorde R 2010 Low serum 25-hydroxyvitamin D levels are associated with increased all-cause mortality risk in a general population: the Tromso study. Eur J Endocrinol 162:935942 16. Cawthon PM, Parimi N, Barrett-Connor E, Laughlin GA, Ensrud KE, Hoffman AR, Shikany JM, Cauley JA, Lane NE, Bauer DC, Orwoll ES, Cummings SR 2010 Serum 25-hydroxyvitamin D, parathyroid hormone, and mortality in older men. J Clin Endocrinol Metab 95:4625 4634

Acknowledgments
The authors thank Jesper T. Andreasen, Ph.D., and Kristi Kohlmeier, Ph.D., University of Copenhagen, Faculty of Phar-

J Clin Endocrinol Metab, August 2012, 97(8):0000 0000

jcem.endojournals.org

17. Michaelsson K, Baron JA, Snellman G, Gedeborg R, Byberg L, Sundstrom J, Berglund L, Arnlov J, Hellman P, Blomhoff R, Wolk A, Garmo H, Holmberg L, Melhus H 2010 Plasma vitamin D and mortality in older men: a community-based prospective cohort study. Am J Clin Nutr 92:841 848 18. Sambrook PN, Chen JS, March LM, Cameron ID, Cumming RG, Lord SR, Schwarz J, Seibel MJ 2004 Serum parathyroid hormone is associated with increased mortality independent of 25-hydroxy vitamin D status, bone mass, and renal function in the frail and very old: a cohort study. J Clin Endocrinol Metab 89:54775481 19. Semba RD, Houston DK, Bandinelli S, Sun K, Cherubini A, Cappola AR, Guralnik JM, Ferrucci L 2010 Relationship of 25-hydroxyvitamin D with all-cause and cardiovascular disease mortality in older community-dwelling adults. Eur J Clin Nutr 64:203209 20. Semba RD, Houston DK, Ferrucci L, Cappola AR, Sun K, Guralnik JM, Fried LP 2009 Low serum 25-hydroxyvitamin D concentrations are associated with greater all-cause mortality in older communitydwelling women. Nutr Res 29:525530 21. Ginde AA, Scragg R, Schwartz RS, Camargo Jr CA 2009 Prospective study of serum 25-hydroxyvitamin D level, cardiovascular disease mortality, and all-cause mortality in older U.S. adults. J Am Geriatr Soc 57:15951603 22. Dobnig H, Pilz S, Scharnagl H, Renner W, Seelhorst U, Wellnitz B, Kinkeldei J, Boehm BO, Weihrauch G, Maerz W 2008 Independent association of low serum 25-hydroxyvitamin D and 1,25-dihydroxyvitamin D levels with all-cause and cardiovascular mortality. Arch Intern Med 168:1340 1349 23. Kestenbaum B, Katz R, de Boer I, Hoofnagle A, Sarnak MJ, Shlipak MG, Jenny NS, Siscovick DS 2011 Vitamin D, parathyroid hormone, and cardiovascular events among older adults. J Am Coll Cardiol 58:14331441 24. Institute of Medicine 2010 Dietary reference intakes for calcium and vitamin D. Washington, DC: The National Academies Press 25. Lundgren E, Lind L, Palmer M, Jakobsson S, Ljunghall S, Rastad J 2001 Increased cardiovascular mortality and normalized serum calcium in patients with mild hypercalcemia followed up for 25 years. Surgery 130:978 985 26. Palmer M, Adami HO, Bergstrom R, Jakobsson S, Akerstrom G, Ljunghall S 1987 Survival and renal function in untreated hyper-

27.

28.

29.

30.

31. 32. 33. 34.

35.

36.

37.

38.

calcaemia. Population-based cohort study with 14 years of followup. Lancet 1:59 62 Pilz S, Tomaschitz A, Drechsler C, Ritz E, Boehm BO, Grammer TB, Marz W 2010 Parathyroid hormone level is associated with mor tality and cardiovascular events in patients undergoing coronary angiography. Eur Heart J 31:15911598 Larsson TE, Olauson H, Hagstrom E, Ingelsson E, Arnlov J, Lind L, Sundstrom J 2010 Conjoint effects of serum calcium and phosphate on risk of total, cardiovascular, and noncardiovascular mortality in the community. Arterioscler Thromb Vasc Biol 30:333339 Plotnikoff GA, Quigley JM 2003 Prevalence of severe hypovitaminosis D in patients with persistent, nonspecific musculoskeletal pain. Mayo Clin Proc 78:14631470 Holick MF 2002 Vitamin D: the underappreciated D-lightful hormone that is important for skeletal and cellular health. Curr Opin Endocrinol Diabetes 9:8798 Shinchuk LM, Holick MF 2007 Vitamin D and rehabilitation: improving functional outcomes. Nutr Clin Pract 22:297304 Glerup H, Eriksen EF 1999 [Vitamin D deficiency. Easy to diagnose, often overlooked]. Ugeskr Laeger 161:25152521 Holick MF 2003 Vitamin D deficiency: what a pain it is. Mayo Clin Proc 78:14571459 Mosekilde L 2008 Vitamin D requirement and setting recommendation levels: long-term perspectives. Nutr Rev 66(10 Suppl 2): S170 S177 Hagstrom E, Hellman P, Larsson TE, Ingelsson E, Berglund L, Sund strom J, Melhus H, Held C, Lind L, Michaelsson K, Arnlov J 2009 Plasma parathyroid hormone and the risk of cardiovascular mortality in the community. Circulation 119:27652771 Carlstedt F, Lind L, Wide L, Lindahl B, Hanni A, Rastad J, Ljunghall S 1997 Serum levels of parathyroid hormone are related to the mortality and severity of illness in patients in the emergency department. Eur J Clin Invest 27:977981 Scragg R, Camargo Jr CA 2008 Frequency of leisure-time physical activity and serum 25-hydroxyvitamin D levels in the US population: results from the Third National Health and Nutrition Examination Survey. Am J Epidemiol 168:577586; discussion 587591 Rosen CJ, Gallagher JC 2011 The 2011 IOM report on vitamin D and calcium requirements for North America: clinical implications for providers treating patients with low bone mineral density. J Clin Densitom 14:79 84