Original Investigation Vitamin D Status and Mortality Risk in CKD: A Meta-analysis of Prospective Studies

Stefan Pilz, MD,1,2* Simona Iodice, MD,3* Armin Zittermann, PhD,4 William B. Grant, PhD,5 and Sara Gandini, PhD3
Background: Vitamin D deciency, assessed as low 25-hydroxyvitamin D (25[OH]D) level, is highly prevalent in patients with chronic kidney disease (CKD) and is associated with various adverse health outcomes. Whether low 25(OH)D levels in patients with CKD are an independent risk factor for mortality remains to be studied in detail, and this was the objective of our work. Study Design: A systematic review and meta-analysis of prospective observational studies. Setting & Population: Patients with CKD. CKD was diagnosed mainly as decreased estimated glomerular ltration rate. Selection Criteria for Studies: We performed a systematic literature search in MEDLINE, ISI, and EMBASE to identify prospective studies reporting on 25(OH)D levels and mortality. Predictor: 25(OH)D serum concentrations. Outcome: All-cause mortality. Results: 10 studies with an overall sample of 6,853 patients with CKD were included. Relative risk of mortality per 10-ng/mL (25-nmol/L) increase in 25(OH)D level was 0.86 (95% CI, 0.82-0.91), with no indication of publication bias or signicant heterogeneity (I 2 15%; P 0.3). Summary estimates for CKD cohorts with and without dialysis treatment showed homogeneous results (P 0.9). Limitations: Results may be limited by heterogeneity, unconsidered confounders, and the observational design of the studies. Furthermore, publication bias by unpublished null ndings on the association of 25(OH)D level and mortality cannot be ruled out and ascertainment of CKD was based largely on estimated glomerular ltration rate. Conclusions: Higher 25(OH)D levels are associated with signicantly improved survival in patients with CKD. Whether treatment of low 25(OH)D level using natural vitamin D supplementation improves survival in patients with CKD remains to be elucidated in randomized controlled trials. Am J Kidney Dis. xx(x):xxx. 2011 by the National Kidney Foundation, Inc. INDEX WORDS: Vitamin D; mortality; meta-analysis.

evels of 25-hydroxyvitamin D (25[OH]D), which are used to classify vitamin D status,1 are less than the target range of at least 30 ng/mL (75 nmol/L) in most patients with chronic kidney disease (CKD).2,3 Limited sunlight exposure in combination with impaired UVB-induced vitamin D synthesis in the skin and disturbed vitamin D metabolism are considered to contribute to low 25(OH)D levels in patients with CKD.2-5 The clinical signicance of vitamin D deciency is emphasized by experimental and epidemiologic data indicating that vitamin D deciency itself may contribute to impaired kidney function.2,6,7 Apart from this, vitamin D deciency is considered a causal risk factor for CKDmineral and bone disorders and also might contribute to cardiovascular and autoimFrom the 1Department of Internal Medicine, Division of Endocrinology and Metabolism, Medical University of Graz, Graz, Austria; 2Department of Epidemiology and Biostatistics and EMGO Institute for Health and Care Research, VU University Medical Center, Amsterdam, the Netherlands; 3Division of Epidemiology and Biostatistics, European Institute of Oncology, Milan, Italy; 4 Clinic for Thoracic and Cardiovascular Surgery, Heart Centre North Rhine-Westphalia, Ruhr University Bochum, Bad Oeynhausen, Germany; and 5Sunlight, Nutrition, and Health Research Center, San Francisco, CA.
Am J Kidney Dis. 2011;xx(x):xxx

mune diseases, cancer, and infections.1,3,8-18 Whether correction of low 25(OH)D levels using natural vitamin D supplementation improves survival in patients with CKD remains to be elucidated. However, it should be noted that a meta-analysis of randomized controlled trials (RCTs) showed decreased mortality using vitamin D supplementation in frail elderly patients.19 In line with this, observational data from patients with CKD largely, but not consistently, show that low 25(OH)D levels are associated with increased mortality.20-29 The 2009 KDIGO (Kidney Disease: Improving Global Outcomes) clinical practice guidelines include a weak recommendation to test for and treat vitamin D deciency in patients with CKD stages 3-5D using treatment strategies recom* S.P. and S.I. contributed equally to this work. Received December 17, 2010. Accepted in revised form March 17, 2011. Address correspondence to Sara Gandini, PhD, Division of Epidemiology and Biostatistics, European Institute of Oncology, Via Ripamonti 435, 20141 Milan, Italy. E-mail: sara.gandini@ieo.it 2011 by the National Kidney Foundation, Inc. 0272-6386/$36.00 doi:10.1053/j.ajkd.2011.03.020
1

Pilz et al

mended for the general population.8 To raise the evidence level for general statements on the clinical signicance of low 25(OH)D levels, we performed a meta-analysis of observational studies reporting 25(OH)D levels and mortality in patients with CKD. Our main aims were to elucidate whether 25(OH)D levels are associated with overall mortality and test whether the risk prediction of 25(OH)D level is modied according to the presence or absence of dialysis treatment.

METHODS
Study Design
We planned, conducted, and reported this meta-analysis by adhering to the MOOSE (Meta-analysis of Observational Studies in Epidemiology) guidelines, which are published standards for meta-analyses of observational studies.30

Search Strategy
A systematic literature search was performed in PubMed, EMBASE, and ISI Web of Science (Science Citation Index Expanded) without restrictions by using the following search terms: (vitamin D or 25[OH]D) and (mortality or death) and (kidney or dialysis). We searched for the key words in headers and the abstract. In addition, we hand searched reference lists of selected articles to identify further studies of potential interest. This literature search was carried out independently by 2 academic investigators. Group discussion resolved any disagreement with article selection.

limited power, we considered statistically signicant heterogeneity at the P 0.1 level of association. A further measure of heterogeneity, I2, a transformation of the square root of the 2 divided by its df, has been considered to compare between heterogeneities for different numbers of pooled studies. Larger I2 values indicate greater heterogeneity.35 Pooled estimates of the associations of 25(OH)D levels with risk were based on a 2-step procedure. First, a linear model was tted within each study to estimate the RR per unit increase in serum 25(OH)D level. When sufcient information was published (the number of participants at each serum level category), the model was tted according to the method proposed by Greenland and Longnecker.36 This method provides the natural logarithm of the RR and an estimator of its standard error, taking into account that the estimates for separate categories depend on the same reference group. When the number of participants at each serum level category was not available from the publications, coefcients were calculated, ignoring the correlation between estimates of risk at the separate exposure levels. Second, the summary RR was estimated by pooling studyspecic estimates with mixed-effects models to be conservative and generalize results. Mixed models were preformed by using an unstructured covariance matrix as indicated by van Houwelingen et al.37 PROC MIXED in SAS (SAS Windows, version 8.02; SAS Institute Inc, www.sas.com) with maximum likelihood estimates was used to take into account between-study variability.37 The nal summary RR expressed the risk of mortality associated with an increment in serum 25(OH)D level by 10 ng/mL (25 nmol/L).

Heterogeneity and Sensitivity Analyses

To assess the inuence of possible sources of bias, we considered the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) checklist proposed for observational epidemiologic studies. According to the STROBE checklist, using meta-regression, we evaluated between-study heterogeneity assessing the inuence of different study features, such as the accurate description of the study population and study design, unbiased assessment of mortality and 25(OH)D status, appropriate statistical methods, and adequate and correct reporting of results. Sensitivity analyses were conducted to evaluate the stability of summary estimates. We also examined changes in results after exclusion of specic studies. Meta-regressions and subgroup analyses were carried out to investigate between-study heterogeneity, looking at possible factors that could inuence estimates (region: Europe and United States; treatment: dialysis and nondialysis; assay company: DiaSorin and IDS; age; and length of follow-up). Furthermore, we quantied the inuence of adjustment, including variables indicating adjustment for confounders in the meta-regression model. Sensitivity analyses were performed to evaluate the effect of excluding the 4D (Die Deutsche Diabetes Dialyse Studie; RCT with atorvastatin) and/or the Rancho Bernardo Study on cardiovascular mortality.25,29 Furthermore, we tested for publication bias according to the technique of Macaskill et al.38

Study Selection
Only reports fullling the following inclusion criteria were included in the meta-analysis: studies should contain the minimum information necessary to estimate the relative risk (RR) of overall mortality associated with serum 25(OH)D level and a corresponding measure of uncertainty (ie, 95% condence interval [CI], standard error, variance, or P value for the signicance of the estimate). We included observational studies published as original articles; ecologic and prevalence studies were excluded. Studies should be independent. In the case of multiple reports of the same population or subpopulation, we considered estimates from the most recent or most informative report. Study populations should be representative of patients with CKD. Articles were reviewed and data were extracted and crosschecked independently by 3 investigators (S.P., W.B.G., and S.I.) and any disagreement was resolved by consensus among these 3 academic researchers. Data extraction was performed using a predened data extraction design. In accordance with the method of Berlin et al,31 dose-response information was recorded as the value of 25(OH)D (in nanograms per milliliter) assigned as the midpoint of the ranges of reported categories.

Statistical Methods
Every measure of association, adjusted for the maximum or nal number of confounding variables, and corresponding CIs were transformed into log RR, and the corresponding variance was calculated using the formula proposed by Greenland.32 When no estimates were given, we generated the necessary statistics from published Kaplan-Meier curves by adopting a hierarchical series of steps according to Parmar et al.33 We used Woolfs formula to estimate the standard error of the log RR.34 The homogeneity of effects across studies was assessed using the large sample test based on 2 statistic. Because the 2 test has
2

RESULTS
Study Selection Our systematic literature search was updated to November 16, 2010. We identied 931 abstracts in PubMed, 1,450 abstracts in EMBASE, and 857 abstracts in ISI Web of Science, and there were no articles of interest in languages other than English. Of these 3,238 references, we excluded 3,209 articles based on screening of titles and abstracts. Hence, after
Am J Kidney Dis. 2011;xx(x):xxx

Vitamin D and Mortality

this rst selection procedure, 29 articles were evaluated further according to their full text. Reading these 29 articles, we did not nd additional studies of potential interest in the reference lists. After further exclusion of 18 full-text articles that did not report estimates for the end point of interest, 11 studies were left that reported data for 25(OH)D levels and mortality in patients with CKD. There was no bias by exclusion of studies with negative ndings because none of the 18 excluded studies reported data or comment for the association of 25(OH)D level and mortality. From 2 reports of NHANES III (Third National Health and Nutrition Examination Survey), we excluded the report by Melamed et al39 because it overlapped with the study population of the work by Mehrotra et al21 and included a smaller study sample with a lower event rate. Thus, we nally included 10 independent studies. A ow chart of selection of included studies is shown in Fig 1. Study Characteristics All studies included in our meta-analysis were published in 2007-2011.20-29 The study by Wolf et al22 was a nested case-control analysis within a prospective cohort study (ArMORR [Accelerated Mortality on Renal Replacement]) and the study by Drechsler et al25 was a prospective analysis of an RCT evaluating effects of atorvastatin (4D). In that study, there

was no effect of treatment on mortality.25 All other studies included in the meta-analysis were prospective cohort studies.20,21,23,24,26-29 The Rancho Bernardo Study reports on only cardiovascular mortality: we included that investigation in our meta-analysis because cardiovascular mortality was the main cause of death.29 Study features of the included cohorts are listed in Table 1, and estimates for the association of 25(OH)D level and mortality from each included study are listed in Table 2. Effects of 25(OH)D on Total Mortality Risk A forest plot of dose-response RRs for 10 ng/mL (25 nmol/L) of increase in 25(OH)D level is shown in Fig 2. Summary dose-response RR estimates suggest a signicant decrease in mortality with increasing serum 25(OH)D levels. An increase of 10 ng/mL (25 nmol/L) leads to a signicant decrease of 14% in mortality risk (RR, 0.86; 95% CI, 0.82-0.91). No signicant between-study heterogeneity was found: 15% and 2 P 0.3. When investigating I2 between-study variability using meta-regression, we found that summary estimates by dialysis treatment were homogeneous (P 0.9). For this analysis, we handled the study by Barreto et al,23 which included 33% hemodialysis patients, as a nondialysis population. Excluding this study or handling it as a dialysis study showed materially unchanged results. Similar trends were found for all other sensitivity analyses and subgroup estimates by region, assay company, age, and length of follow-up. However, even if the difference was not statistically signicant, we found that unadjusted estimates (RR, 0.71; 95% CI, 0.560.91) were greater than adjusted ones (RR, 0.87; 95% CI, 0.83-0.91; P for the difference 0.1). No indication of publication bias was found (P 0.5, Macaskills test). Furthermore, our results did not signicantly change in sensitivity analyses excluding the 4D and/or Rancho Bernardo Study.

Figure 1. Flow chart of selection of included studies. Abbreviation: RCT, randomized controlled trial. Am J Kidney Dis. 2011;xx(x):xxx

DISCUSSION This meta-analysis of patients with CKD shows that increasing 25(OH)D levels are associated with improved survival. This effect was not signicantly different in CKD cohorts with and without dialysis treatment. These ndings may be of clinical relevance considering that current KDIGO guidelines suggest correction of low 25(OH)D levels in patients with CKD.8 Our results support these guidelines and suggest that the clinical signicance of vitamin D deciency for mortality risk prediction may be similar in patients with and without dialysis. It also should be noted that there was no signicant between-study variability with regard to region, age, assay company, adjust3

Pilz et al
Table 1. Characteristics of Included Studies
Source No. of Patients Age (y)a CKD Stageb Accrural Period F/U (y) Died Assay Methods Country Dialysis (%)

Ravani20 Mehrotra21,c Wolf22,d Wang24 Barreto23 Gracia-Iguacel26 Pecovnik-Balon27 Drechsler25,e Drechsler28,f Jassal29,g

168 3,011 984 230 140 115 102 1,108 762 233

70 55 63 55 67 60 61 66 59 80

12 1 15 12 12 16 13 8 15 9

2-5 1-4 5 5 2-5 5 5 5 5 3-4

2002-2003 1988-1994 2004-2005 NA 2006-2007 2007 2005-2007 1998-2002 NA 1997-1999

4 9 0.25 3 1.7 1.2 2 4 3 6.8

78 848 244 70 25 20 27 545 213 40

IDS, RIA Dia, RIA Dia, RIA IDS, ELISA Dia, CLIA Dia, CLIA IDS, EIA IDS, CLIA Dia, CLIA Holickh

Italy USA USA Hong Kong France Spain Slovenia Germany Netherlands USA

0 0 100 HD 100 PD 33 HD 100 (81% HD) 100 HD 100 HD 100 (64% HD) 0

Abbreviations: CKD, chronic kidney disease; CLIA, chemiluminescence assay; Dia, DiaSorin; eGFR, estimated glomerular ltration rate; EIA, enzyme immunoassay; ELISA, enzyme-linked immunosorbent assay; F/U, follow-up; HD, hemodialysis; IDS, immunodiagnostic systems; NA, not applicable; PD, peritoneal dialysis; RIA, radioimmunoassay; USA, United States of America. a Mean standard deviation. b CKD denition of nondialysis patients was based on eGFR or (in the study of Mehrotra et al21) eGFR or microalbuminuria. c NHANES III (Third National Health and Nutrition Examination Survey). d ArMORR (Accelerated Mortality on Renal Replacement) cohort. e 4D (Die Deutsche Diabetes Dialyse Studie). f NECOSAD (Netherlands Cooperative Study on the Adequacy of Dialysis). g Rancho Bernardo Study. h Measured using vitamin D competitive binding recognition and chemiluminescence detection in the research laboratory of Dr Michael Holick.

ments for confounders, and length of follow-up. This emphasizes the robustness of the observed association of increasing 25(OH)D level with decreased mortality. Furthermore, during the review process of our report, an additional report was published that reported data for 25(OH)D concentrations and mortality in 648 hemodialysis patients from the regional ARNOS French cohort.40 In line with ndings from our meta-analysis, in their work, Jean et al40 showed that compared with patients with lower 25(OH)D concentrations, the multivariate adjusted hazard ratio for patients with baseline 25(OH)D levels higher than the median value of 18 ng/mL was 0.73 (95% CI, 0.50 0.96; P 0.002). In view of our results, determination of 25(OH)D levels could be a useful tool to identify high-risk patients. However, from our observational data, we cannot draw nal conclusions regarding causality. We therefore remain with the question of whether vitamin D deciency is the cause or consequence of high-risk conditions leading to fatal diseases. However, it should be stressed that a meta-analysis of RCTs designed to assess musculoskeletal outcomes of vitamin D supplementation in frail elderly patients already has shown that vitamin D prescription signicantly decreases total mortality versus placebo.19 Most patients included in that meta-analysis had decreased kidney function, but there is no published RCT that specically addresses the impact of vitamin D supplementation on mortality in CKD cohorts. However, natural
4

vitamin D supplementation in patients with CKD has improved bone and mineral metabolism (eg, decreases parathyroid hormone and increases 1,25dihydroxyvitamin D [1,25(OH)2D] levels), decreases inammatory parameters, and improves measures of myocardial function and glucose metabolism.41-47 Whether these benecial vitamin D effects translate into improved health outcomes remains to be proved. We did not focus on specic events leading to increased mortality in patients with CKD with low 25(OH)D levels, but accumulating evidence suggests that vitamin D deciency may be associated with fatal events caused by cardiovascular diseases, strokes, cancer, or infections.1,9-18,20,22,25 Various molecular effects of vitamin D metabolites and ndings from some RCTs support a causal association with these latter diseases.1,9,13-18,48 Although 1,25(OH)2D is considered the most active vitamin D metabolite, supplementation with 1,25(OH)2D or its analogues is not the correct therapy for low 25(OH)D levels, which can be increased using much safer and less expensive natural vitamin D supplementation.1,45 Substrate-dependent local conversion of 25(OH)D to 1,25(OH)2D occurs in most organs of the human body and seems to be the main determinant of tissue levels of 1,25(OH)2D.1,9,13-15 Hence, natural vitamin D supplementation is a potentially benecial treatment in patients with CKD because it not only lowers parathyroid hormone levels, but also raises 25(OH)D serum levels, as well as
Am J Kidney Dis. 2011;xx(x):xxx

Am J Kidney Dis. 2011;xx(x):xxx 5

Vitamin D and Mortality

Table 2. Estimates for the Association of 25(OH)D and Mortality Reported in the Included Studies
RR (95% CI) Source No. of Deaths 25(OH)D (ng/mL)a At Risk 25(OH)D Category Categorical Analysis Continuous Analysis Adjustments

Ravani20

78

18.1 (13, 26)

58 110

15 15

1.00 (reference) 0.61 (0.36-1.00)

0.76 (0.56-1.05) per 10 ng/mL 25(OH)D

Age, HF, current or past smoking habit, CRP, serum albumin, phosphate, and ACEi or ARB

Mehrotra21

848

11.7 22.6 39.9

0.2 0.2 0.5

361 1,566 1,084

15 15-30 30

1.56 (1.12-2.18) 1.17 (0.99-1.38) 1.00 (reference)

Age, sex, race, current smoking, hypertension, history of CVD, BMI, DM, family history, non-HDL cholesterol 160 mg/dL, CRP, month of test, CKD stage, UACR, Hb, albumin, phosphorus, medical insurance, education, income, and whether the participant went to a particular place for care

Wolf22

244

21

13b

187 594 203

10 10-30 30

1.6 (1.0-2.4) 1.3 (0.9-1.8) 1.0 (reference)

Age, sex, race, cause of kidney failure, standardized mortality rates, BP, vascular access, albumin, creatinine, PTH, calcium, phosphorus, Hb, and history of CAD, stroke, malignancy, or CHF

Wang24

70

18.3 (14.3, 24.3)

116 114

18.3 18.3

0.9c; higher mortality for 18.3

1.06 (0.59-1.89) per log 25(OH)D in nmol/L 0.582 (0.368-0.923) per 10 ng/mL of 25(OH)D

None

Barreto23

25

20.5

13.6

140

Aortic calcication score and PWV

Gracia-Iguacel26

18

NA

46 48 115

15 15 15 15

6.96 (1.44-33.64) 1.00 (reference) Log rank 2 5.5; P 0.01; higher mortality for 15 (Continued)

Age, sex, erythropoietin resistance index DM, and CVD None

20 IU/kg/wk/Hb,

20

6 Table 2 (Contd). Estimates for the Association of 25(OH)D and Mortality Reported in the Included Studies
RR (95% CI) Source No. of Deaths 25(OH)D (ng/mL)a At Risk 25(OH)D Category Categorical Analysis Continuous Analysis Adjustments

Pecovnik-Balon27

27

23.3

14.3

49 53 177 607 210 114

20 20 10 10- 20 20- 30 30

Kaplan-Meier: P 0.033; higher mortality for 20 1.65 (1.14-2.38) 1.20 (0.86-1.68) 1.16 (0.80-1.70) 1.00 (reference) 1.36 (1.13-1.64) per log 25(OH)D

None

Drechsler (4D)25

545

8.0 (7, 9)d 14.0 (12, 17)d 24.5 (22, 27)d 37.7 (33, 42)d

Age; sex; atorvastatin treatment; season; CAD; CHF; systolic BP; smoking; duration of dialysis; ultraltration volume; BMI; levels of LDL & HDL cholesterol, CRP, and HbA1c; use of -blockers, ACEi, and diuretics; PTH, calcium, and phosphate

Drechsler (NECOSAD)28

213

8 18 40

2 5 10

193 469 100

10 10- 30 30

1.25 (0.71-2.18) 0.84 (0.52-1.38) 1.00 (reference)

Age, sex, dialysis modality, ethnicity, primary kidney disease, DM, CVD, BMI, systolic BP, smoking, cholesterol, vitamin supplements, albumin, Hb, creatinine, and seasonal variation of vitamin D

Jassal29

40

41

13

233

0.67 (0.43-1.05) per SD of 25(OH)D

Sex, BMI, systolic BP, LDL, glucose, exercise, logUACR, CVD, season, diuretics, CCBs, -blockers, and ACEi

Note: 25(OH)D levels are presented in ng/mL unless otherwise stated (conversion factor for 25(OH)D in ng/mL to nmol/L, 2.496). Abbreviations: 4D, Die Deutsche Diabetes Dialyse Studie; 25(OH)D, 25-hydroxyvitamin D; ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blockers; BMI, body mass index; BP, blood pressure; CAD, coronary artery disease; CCB, calcium channel blocker; CHF, congestive heart failure; CI, condence interval; CKD, chronic kidney disease; CRP, C-reactive protein; CT, computed tomography; CVD, cardiovascular disease; D, dialysis; DM, diabetes mellitus; Hb, hemoglobin; HDL, high-density lipoprotein; HF, heart failure; LDL, low-density lipoprotein; NA, not applicable; NECOSAD, Netherlands Cooperative Study on the Adequacy of Dialysis; PTH, parathyroid hormone; PWV, pulse wave velocity; RR, relative risk; SD, standard deviation; UACR, urine albumin-creatinine ratio. a Values are presented as mean standard deviation or median (25th, 75th percentile) of either the entire study population or the respective 25(OH)D category. b Values for a subsample of 825 of 984 patients. c Log-rank test. d Overall median (25th, 75 percentile) for total population is 15.6 (11, 22).

Am J Kidney Dis. 2011;xx(x):xxx

Pilz et al

Vitamin D and Mortality

Figure 2. Forest plot and summary relative risk (RR) for the association of 25-hydroxyvitamin D (25[OH]D) level and mortality in patients with chronic kidney disease. The size of the box is proportional to the weight of the study (1/variance of the estimate). Abbreviations: 4D, Die Deutsche Diabetes Dialyse Studie; NECOSAD, Netherlands Cooperative Study on the Adequacy of Dialysis.

circulating and tissue levels of 1,25(OH)2D.41,44,45,47 Concerning increases in 25(OH)D levels using natural vitamin D supplementation, it should be acknowledged that Wolf et al22 noted a signicant interaction with active vitamin D use in dialysis patients. It was shown that low 25(OH)D levels were predictive of mortality only in patients who did not receive active vitamin D treatment.22 Hence, future studies should consider interactions and possible additive effects of natural and active vitamin D therapy. In this context, it should be emphasized that in our meta-analysis, the impact of 25(OH)D status on mortality was similar in patients with and without dialysis, although active vitamin D therapy frequently is used in patients on dialysis therapy. Despite missing RCTs with fatal events as primary outcomes, we believe that in view of our results, it seems reasonable to prevent and treat vitamin D deciency in virtually all patients with CKD to increase 25(OH)D to target levels of at least 30-40 ng/mL (75-100 nmol/L), which are considered to be optimal for various health outcomes.1,8,9,49,50 In support of this suggestion are the proposed multiple health benets and low costs of and safety data for natural vitamin D supplementation.1,8,9,49,50 We also should consider the possibility that low 25(OH)D levels may be a marker of adverse health outcomes rather than a causal risk factor for mortality. In particular, decreased mobility with subsequently low sunlight-induced vitamin D production and malnutrition or decreasing 25(OH)D levels as a result of disturbed vitamin D metabolism in the course of CKD may underlie the increased mortality in vitamin Ddecient patients. Inclusion of multivariable-adjusted RRs in our meta-analysis argues against this reverse causation, but residual confounding cannot be excluded. Our results also may be limited by heterogeneity
Am J Kidney Dis. 2011;xx(x):xxx

of the analyzed studies because of differences inherent to the study design, statistical analyses, and laboratory methods for 25(OH)D measurement. Even if our sensitivity analyses showed homogenous results, this should be interpreted with caution considering that analyses had limited power due to the low number of included studies. It also should be noted that our results were derived mainly from comparisons of patient groups with 25(OH)D concentrations ranging from 10- 30 ng/ mL. Our results therefore cannot simply be extrapolated to patients with CKD with 25(OH)D levels that are substantially less or greater than the mentioned values. Furthermore, ascertainment of CKD was based largely on estimated glomerular ltration rate, and publication bias by unpublished null ndings on the association of 25(OH)D and mortality cannot be ruled out. However, there was no indication that any study found by our literature search did not report on the association of 25(OH)D and total mortality despite availability of these data. In conclusion, we performed a meta-analysis of observational studies and found that mortality risk of patients with CKD decreases with higher 25(OH)D levels. This association was not modied signicantly by the presence or absence of dialysis treatment. These results support previous recommendations for natural vitamin D supplementation in patients with CKD and point to the urgent need for further RCTs to evaluate the impact of vitamin D therapy on fatal events in patients with CKD.

ACKNOWLEDGEMENTS
Support: None. Financial Disclosure: Dr Zittermann has received lecture fees from DiaSorin, Abbott, and MSD and consulting fees from BASF; Dr Pilz has received lecture fees and consulting fees from DiaSorin
7

Pilz et al
and Abbott; Dr Grant receives or has received funding from the UV Foundation, the Sunlight Research Forum, the North Coast Cancer Foundation, Bio-Tech-Pharmacal, the Vitamin D Council, and the Danish Sunbed Federation. The remaining authors declare that they have no relevant nancial interests. 19. Autier P, Gandini S. Vitamin D supplementation and total mortality: a meta-analysis of randomized controlled trials. Arch Intern Med. 2007;167(16):1730-1737. 20. Ravani P, Malberti F, Tripepi G, et al. Vitamin D levels and patient outcome in chronic kidney disease. Kidney Int. 2009;75(1): 88-95. 21. Mehrotra R, Kermah DA, Salusky IB, et al. Chronic kidney disease, hypovitaminosis D, and mortality in the United States. Kidney Int. 2009;76(9):977-983. 22. Wolf M, Shah A, Gutierrez O, et al. Vitamin D levels and early mortality among incident hemodialysis patients. Kidney Int. 2007;72(8):1004-1013. 23. Barreto DV, Barreto FC, Liabeuf S, et al. Vitamin D affects survival independently of vascular calcication in chronic kidney disease. Clin J Am Soc Nephrol. 2009;4(6):1128-1135. 24. Wang AY, Lam CW, Sanderson JE, et al. Serum 25hydroxyvitamin D status and cardiovascular outcomes in chronic peritoneal dialysis patients: a 3-y prospective cohort study. Am J Clin Nutr. 2008;87(6):1631-1638. 25. Drechsler C, Pilz S, Obermayer-Pietsch B, et al. Vitamin D deciency is associated with sudden cardiac death, combined cardiovascular events, and mortality in haemodialysis patients. Eur Heart J. 2010;31(18):2253-2261. 26. Gracia-Iguacel C, Gallar P, Qureshi AR, et al. Vitamin D deciency in dialysis patients: effect of dialysis modality and implications on outcome. J Ren Nutr. 2010;20(6):359-367. 27. Pecovnik-Balon B, Jakopin E, Bevc S, Knehtl M, Gorenjak M. Vitamin D and mortality as a novel nontraditional risk factor for mortality in hemodialysis patients. Ther Apher Dial. 2009;13(4): 268-272. 28. Drechsler C, Verduijn M, Pilz S, et al. Vitamin D status and clinical outcomes in incident dialysis patients: results from the NECOSAD Study. Nephrol Dial Transplant. 2011;26(3):10241032. 29. Jassal SK, Chonchol M, von Mhlen D, Smits G, BarrettConnor E. Vitamin D, parathyroid hormone, and cardiovascular mortality in older adults: the Rancho Bernardo Study. Am J Med. 2010;123(12):1114-1120. 30. Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of Observational Studies in Epidemiology: a proposal for reporting. JAMA. 2000;283(15):2008-2012. 31. Berlin JA, Longnecker MP, Greenland S. Meta-analysis of epidemiologic dose-response data. Epidemiology. 1993;4(3):218-228. 32. Greenland S. Quantitative methods in the review of epidemiologic literature. Epidemiol Rev. 1987;9:1-30. 33. Parmar MK, Torri V, Stewart L. Extracting summary statistics to perform meta-analyses of the published literature for survival endpoints. Stat Med. 1998;17(24):2815-2834. 34. Rothman KJ, Greenland S. Modern Epidemiology. 2nd ed. Philadelphia, PA: Lippincott-Raven Publishers; 1998. 35. Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med. 2002;21(11):1539-1558. 36. Greenland S, Longnecker MP. Methods for trend estimation from summarized dose-response data, with applications to metaanalysis. Am J Epidemiol. 1992;135(11):1301-1309. 37. van Houwelingen HC, Arends LR, Stijnen T. Advanced methods in meta-analysis: multivariate approach and metaregression. Stat Med. 2002;21(4):589-624. 38. Macaskill P, Walter SD, Irwig L. A comparison of methods to detect publication bias in meta-analysis. Stat Med. 2001;20(4): 641-654. 39. 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(15):1629-1637. 40. Jean G, Lataillade D, Genet L, et al. Impact of hypovitaminosis D and alfacalcidol therapy on survival of hemodialysis patients:
Am J Kidney Dis. 2011;xx(x):xxx

REFERENCES
1. Holick MF. Vitamin D deciency. N Engl J Med. 2007;357(3): 266-281. 2. Mehrotra R, Kermah D, Budoff M, et al. Hypovitamonosis D in chronic kidney disease. Clin J Am Soc Nephrol. 2008;3(11441151. 3. Doorenbos CR, von den Born J, Navis G, de Borst MH. Possible renoprotection by vitamin D in renal disease: beyond mineral metabolism. Nat Rev Nephrol. 2009;5(12):691-700. 4. Jacob AI, Sallman A, Santiz Z, Hollis BW. Defective photoproduction of cholecalciferol in normal and uremic humans. J Nutr. 1984;114(7):1313-1319. 5. Helvig CF, Cuerrier D, Hoseld CM, et al. Dysregulation of renal vitamin D metabolism in the uremic rat. Kidney Int. 2010; 78(5):463-472. 6. Zhang Y, Kong J, Deb DK, Chang A, Li YC. Vitamin D receptor attenuates renal brosis by suppressing the reninangiotensin system. J Am Soc Nephrol. 2010;21(6):966-973. 7. Melamed ML, Astor B, Michos ED, Hostetter TH, Powe NR, Muntner P. 25-Hydroxyvitamin D levels, race, and the progression of kidney disease. J Am Soc Nephrol. 2009;20(12): 2631-2639. 8. Kidney Disease: Improving Global Outcomes (KDIGO) CKD-MBD Work Group. KDIGO clinical practice guideline for the diagnosis, evaluation, prevention, and treatment of chronic kidney disease-mineral and bone disorder (CKD-MBD). Kidney Int Suppl. 2009;113:S1-130. 9. Bouillon R, Carmeliet G, Verlinden L, et al. Vitamin D and human health: lessons from vitamin D receptor null mice. Endocr Rev. 2008;29(6):726-776. 10. Pittas AG, Chung M, Trikalinos T, et al. Systematic review: vitamin D and cardiometabolic outcomes. Ann Intern Med. 2010; 152(5):307-314. 11. de Boer IH, Kestenbaum B, Shoben AB, et al. 25Hydroxyvitamin D levels inversely associate with risk of developing coronary artery calcication. J Am Soc Nephrol. 2009;20(8): 1805-1812. 12. London GM, Gurin AP, Verbeke FH, et al. Mineral metabolism and arterial functions in end-stage renal disease: potential role of 25-hydroxyvitamin D deciency. J Am Soc Nephrol. 2007;18(2):613-620. 13. Pilz S, Tomaschitz A, Drechsler C, Dekker JM, Mrz W. Vitamin D deciency and myocardial diseases. Mol Nutr Food Res. 2010;54(8):1103-1113. 14. Shoenfeld N, Amital H, Shoenfeld Y. The effect of melanism and vitamin D synthesis on the incidence of autoimmune disease. Nat Clin Pract Rheumatol. 2009;5(2):99-105. 15. Peterlik M, Grant WB, Cross HS. Calcium, vitamin D and cancer. Anticancer Res. 2009;29(9):3687-3698. 16. Pilz S, Tomaschitz A, Obermayer-Pietsch B, Dobnig H, Pieber TR. Epidemiology of vitamin D insufciency and cancer mortality. Anticancer Res. 2009;29(9):3699-3704. 17. Yamshchikov AV, Desai NS, Blumberg HM, Ziegler TR, Tangpricha V. Vitamin D for treatment and prevention of infectious diseases: a systematic review of randomized controlled trials. Endocr Pract. 2009;15(5):438-449. 18. Urashima M, Segawa T, Okazaki M, Kurihara M, Wada Y, Ida H. Randomized trial of vitamin D supplementation to prevent seasonal inuenza A in schoolchildren. Am J Clin Nutr. 2010;91(5): 1255-1260.
8

Vitamin D and Mortality

results from the French ARNOS Study. Nephron Clin Pract. 2011;118:c204-210. 41. Kooienga L, Fried L, Scragg R, Kendrick J, Smits G, Chonchol M. The effect of combined calcium and vitamin D3 supplementation on serum intact parathyroid hormone in moderate CKD. Am J Kidney Dis. 2009;53(3):408-416. 42. Blair D, Byham Gray L, Lewis E, McCaffrey S. Prevalence of vitamin D [25(OH)D] deciency and effects of supplementation with ergocalciferol (vitamin D2) in stage 5 chronic kidney disease patients. J Ren Nutr. 2008;18(4):375-382. 43. Stubbs JR, Idiculla A, Slusser J, Menard R, Quarles LD. Cholecalciferol supplementation alters calcitriol-responsive monocyte proteins and decreases inammatory cytokines in ESRD. J Am Soc Nephrol. 2010;21(2):353-361. 44. Matias PJ, Jorge C, Ferreira C, et al. Cholecalciferol supplementation in hemodialysis patients: effects in mineral metabolism, inammation, and cardiac dimension parameters. Clin J Am Soc Nephrol. 2010;5(5):905-911. 45. Jean G, Souberbielle JC, Chazot C. Monthly cholecalciferol administration in haemodialysis patients: a simple and efcient strategy for vitamin D supplementation. Nephrol Dial Transplant. 2009;24(12):3799-3805. 46. Kalantar-Zadeh K, Kovesdy CP. Clinical outcomes with active versus nutritional vitamin D compounds in chronic kidney disease. Clin J Am Soc Nephrol. 2009;4(9):1529-1539. 47. Kandula P, Dobre M, Schold JD, Schreiber MJ Jr, Mehrotra R, Navaneethan SD. Vitamin D supplementation in chronic kidney disease: a systematic review and meta-analysis of observational studies and randomized controlled trials. Clin J Am Soc Nephrol. 2011;6(1):50-62. 48. Pilz S, Tomaschitz A, Drechsler C, Zittermann A, Dekker JM, Mrz W. Vitamin D supplementation a promising approach for the prevention and treatment of strokes. Curr Drug Targets. 2011;12(1):88-96. 49. Hathcock JN, Shao A, Vieth R, Heaney R. Risk assessment for vitamin D. Am J Clin Nutr. 2007;85(1):6-18. 50. Souberbielle JC, Body JJ, Lappe JM, et al. Vitamin D and musculoskeletal health, cardiovascular disease, autoimmunity and cancer: recommendations for clinical practice. Autoimmun Rev. 2010;9(11):709-715.

Am J Kidney Dis. 2011;xx(x):xxx