Skip Navigation



NDT Advance Access published online on November 21, 2008
Nephrology Dialysis Transplantation, doi:10.1093/ndt/gfn592
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
24/3/963    most recent
gfn592v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Tentori, F.
Right arrow Articles by Port, F. K.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Tentori, F.
Right arrow Articles by Port, F. K.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

© The Author [2008]. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved. For Permissions, please e-mail: journals.permissions@oxfordjournals.org


The survival advantage for haemodialysis patients taking vitamin D is questioned: findings from the Dialysis Outcomes and Practice Patterns Study

Francesca Tentori1, Justin M. Albert1, Eric W. Young2, Margaret J. Blayney1, Bruce M. Robinson1,3, Ronald L. Pisoni1, Takashi Akiba4, Roger N. Greenwood5, Naoki Kimata4, Nathan W. Levin6, Luis M. Piera7, Rajiv Saran3, Robert A. Wolfe1 and Friedrich K. Port1

1 Arbor Research Collaborative for Health 2 Veterans Affairs Medical Center/ University of Michigan 3 Department of Internal Medicine, University of Michigan Health System, Ann Arbor, MI, USA 4 Division of Blood Purification, Kidney Center, Tokyo Women's Medical University, Tokyo, Japan 5 Lister Hospital, Stevenage, UK 6 Renal Research Institute New York, NY, USA 7 Hospital General, Vall d’Hebron, Barcelona, Spain

Correspondence and offprint requests to: Francesca Tentori, Arbor Research Collaborative for Health, 315 W. Huron Street, Suite 360, Ann Arbor, MI 48103, USA. Tel: +1-734-665-4108, Ext. 261; Fax: +1-734-665-2103; E-mail: francesca.tentori{at}ArborResearch.org



   Abstract
Top
 Abstract
Introduction
 Methods
 Statistical methods
 Results
 Discussion
 References
 
Background. Retrospective studies of haemodialysis patients from large dialysis organizations in the United States have indicated that intravenous vitamin D may be associated with a survival benefit. However, patients prescribed vitamin D are generally healthier than those who are not, suggesting that treatment by indication may have biased previous findings. Additionally, no survival benefit associated with vitamin D has been shown in a recent meta-analysis in CKD patients. Because treatment-by-indication bias due to both measured and unmeasured confounders cannot be completely accounted for in standard regression or marginal structural models (MSMs), this study evaluates the association between vitamin D and mortality among participants in the Dialysis Outcomes and Practice Patterns Study (DOPPS) using standard regression and MSMs with an expanded set of covariates, as well as by instrumental variable models to minimize potential bias due to unmeasured confounders.

Methods. Data from 38 066 DOPPS participants from 12 countries between 1996 and 2007 were analysed. Mortality risk was assessed using standard baseline and time-varying Cox regression models, adjusted for demographics and detailed comorbidities, and MSMs. In models similar to instrumental variable analysis, the facility percentage of patients prescribed vitamin D, adjusted for the patient case mix, was used to predict patient-level mortality.

Results. Vitamin D prescription was significantly higher in the USA compared to other countries. On average, patients prescribed vitamin D had fewer comorbidities compared to those who were not. Vitamin D therapy was associated with lower mortality in adjusted time-varying standard regression models [relative ratio (RR) = 0.92 (95% confidence interval: 0.87–0.96)] and baseline MSMs [RR = 0.84 (0.78–0.98)] and time-varying MSMs [RR = 0.78 (0.73–0.84)]. No significant differences in mortality were observed in adjusted baseline standard regression models for patients with or without vitamin D prescription [RR = 0.98 (0.93–1.02)] or for patients in facility practices where vitamin D prescription was more frequent [RR for facilities in 75th versus 25th percentile of vitamin D prescription = 0.99 (0.94–1.04)].

Conclusions. Vitamin D was associated with a survival benefit in models prone to bias due to unmeasured confounding. In agreement with a meta-analysis of randomized controlled studies, no difference in mortality was observed in instrumental variable models that tend to be more independent of unmeasured confounding. These findings indicate that a randomized controlled trial of vitamin D and clinical outcomes in haemodialysis patients are needed and can be ethically conducted.

Keywords: DOPPS; haemodialysis; parathyroid hormone; vitamin D



   Introduction
Top
 Abstract
 Introduction
Methods
 Statistical methods
 Results
 Discussion
 References
 
In a recent guideline document, Kidney Disease: Improving Global Outcomes (KDIGO) defines chronic kidney disease-mineral and bone disorder (CKD-MBD) as a systemic syndrome characterized by abnormalities in serum calcium, phosphorus and parathyroid hormone (PTH) concentrations, in vitamin D metabolism, and in bone turnover that may be associated with soft tissue calcification [1]. This syndrome is common among CKD patients and has been associated with an increased risk for cardiovascular calcifications [2–4], arterial dysfunction [5,6], morbidity and mortality [7,8]. The standard of care suggested by the 2003 National Kidney Foundation Kidney Disease Outcomes Quality Initiative (K/DOQI) guidelines included targeted therapy with dietary phosphorus restriction, phosphorus binders and administration of vitamin D analogues [9].

Non-mineral effects of vitamin D therapy have been described, including effects on the immune and inflammatory system [10,11], effects on the renin–angiotensin system [12], apoptosis [13] and cellular proliferation [14] and differentiation [15]. These biological actions are generally beneficial and may contribute to the improved survival reported for non-CKD patients with malignancies and sepsis who receive vitamin D analogues [16,17]. Observational studies have reported a survival benefit for end-stage renal disease (ESRD) patients on maintenance haemodialysis (HD) who received vitamin D analogues [18–22]. A study of Japanese patients receiving oral vitamin D demonstrated a decreased risk for cardiovascular but not for all-cause mortality [18]. In a retrospective analysis of patients starting HD at facilities operated by Fresenius Medical Care North America, Teng et al. reported a 20% survival advantage for patients receiving intravenous (IV) vitamin D versus those who did not [19]. Similar results were subsequently reported in analyses of administrative databases from other large dialysis organizations in the USA [20,21]. Given the suggestion of the beneficial biological effects of vitamin D, nephrologists may be tempted to consider vitamin D as a panacea for HD patients and extend its prescription beyond current clinical practice guidelines. However, no large clinical outcomes trial has been performed, and a recent meta-analysis of clinical trials in patients with CKD concluded that beneficial effects of vitamin D on patient-level outcomes are unproven [23]. The different results reported by observational studies may be explained by the fact that—in the absence of randomization—patients prescribed vitamin D tended to be overall healthier compared to those not on vitamin D. It is possible that model adjustments failed to fully account for treatment-by-indication bias, and therefore the reported survival benefit may be due to underlying health status rather than treatment with vitamin D. Specifically, the role of unmeasured confounders that could not be accounted for in model adjustments may have led to biased results.

This study explores the relationship between vitamin D therapy and mortality in the international Dialysis Outcomes and Practice Patterns Study (DOPPS) [24,25]. In addition to standard regression and marginal structural model analyses, to address treatment-by-indication bias and specifically the effect of unmeasured confounding factors, we also applied an instrumental variable approach that used the percentage of patients prescribed vitamin D at each facility as the predictor [26–28].



   Methods
Top
 Abstract
 Introduction
 Methods
Statistical methods
 Results
 Discussion
 References
 
The DOPPS design has been previously described [24,25]. The current study analysed data from a representative sample of 38 066 HD patients in 12 countries that participated in the three phases of the DOPPS. In DOPPS I (1996–2001), 8609 patients were randomly sampled from 308 dialysis facilities in France, Germany, Italy, Japan, Spain, the UK and the USA. The prevalent cross-section in DOPPS II (2002–2004) and DOPPS III (2005–present) included 9103 and 7757 patients randomly sampled from 322 and 295 facilities, respectively, from the original seven countries, plus Australia, Belgium, Canada, New Zealand and Sweden. Institutional review boards in each country approved the study, and informed consent was obtained in accordance with local requirements.

Patient data were obtained by medical record abstraction. Data regarding demographic factors, comorbid conditions, laboratory values, type of vascular access and prescription of IV and oral vitamin D analogues (collectively referred to as ‘vitamin D’) were collected upon study entry. Longitudinal data, including laboratory measurements and vitamin D prescription, were collected at 4-month intervals.



   Statistical methods
Top
 Abstract
 Introduction
 Methods
 Statistical methods
Results
 Discussion
 References
 
Standard descriptive statistics described prevalent cross-sections of participants. In DOPPS I, II, and III, the adjusted odds of a patient being prescribed vitamin D at enrolment given his/her demographic and clinical characteristics were assessed in multivariable regression analyses. Multivariable discrete-time Poisson regression models were utilized to identify factors associated with vitamin D initiation among patients not prescribed vitamin D at baseline (n = 16 456), and with vitamin D cessation among patients prescribed vitamin D at baseline (n = 14 785).

The relationship between vitamin D prescription and mortality was assessed using five different modelling approaches. First, baseline standard regression models assessed the association between vitamin D prescription (yes/no) at the time of enrolment and risk of death over the course of follow-up.

Longitudinal data were collected in DOPPS every 4  months. Time-varying standard regression was used to assess the association between mortality risk in a 4-month interval and vitamin D prescription (yes/no) in the last week of the prior 4-month interval. Patients were allowed to switch both into and out of the vitamin D treatment category. Separate sets of models were created based on route of vitamin D administration (IV versus oral). Since results were not meaningfully different, we only reported results combining any vitamin D versus none. Standard regression models were run both unadjusted and adjusted for expanding sets of patient characteristics including baseline demographics and dialysis dose (age, sex, race, years with ESRD, diabetes, single pool Kt/V and year of enrolment in DOPPS), comorbid conditions [diabetes mellitus, coronary artery disease, congestive heart failure, cerebrovascular disease, peripheral vascular disease, other cardiovascular disease, hypertension, lung disease, cancer, GI bleeding, neurologic disease, psychological disorders, prior parathyroidectomy, cellulitis/gangrene, HIV/AIDS and vascular access type (AV fistula, graft or catheter)] and either baseline or time-varying laboratory values from the same 4-month interval as the vitamin D prescription (serum albumin, haemoglobin, calcium, phosphorus, PTH and dialysate calcium).

In order to account for confounding without inappropriately adjusting for intermediate variables between vitamin D dosing and outcomes [29], baseline and time-varying marginal structural models (MSMs) were constructed. MSMs used patient vitamin D prescription to predict mortality risk over the course of the study (baseline) or in the subsequent 4-month interval (time-varying models), with time-at-risk beginning at Month 4. In the MSMs, Cox models were weighted using each patient's inverse probability of vitamin D treatment at study entry (for baseline MSMs) and at the start of each 4-month interval (for time-varying MSMs), given each patient's baseline case mix, comorbid conditions, laboratory values in the current and prior intervals, vitamin D prescription in the prior interval, DOPPS study phase and region (North America = US + Canada; Europe/ANZ = European countries + Australia and New Zealand; and Japan) using a logistic regression model. Using this approach, each patient's weight was standardized by his/her adjusted probability of vitamin D prescription, given his/her baseline case mix, comorbidities and the vitamin D prescription in the prior interval [30].

Standard regression and MSMs cannot overcome bias due to unmeasured confounding. In order to address bias in vitamin D prescription due to unmeasured confounders [31], the association between vitamin D and mortality was assessed in models similar to instrumental variable models in assumptions and construction to the grouped treatment variable approach [32]. A facility's propensity to prescribe vitamin D was modelled as the adjusted proportion of patients on vitamin D at a facility, estimated by linear mixed regression with vitamin D as the outcome and all case-mix, comorbid and baseline laboratory factors as predictor variables with facility treated as a random effect. The intercept for the random effect was considered the expected level of treatment at each facility, or the percentage of patients at a facility who were prescribed vitamin D and were adjusted for the facility case mix and other measured factors that could influence the prescription of vitamin D. Thus, the adjusted facility percentage of vitamin D use, which was applied to each patient at the corresponding facility, was not influenced by an individual patient's characteristics, which could have influenced their mortality risk. The instrumental variables used as predictors of patient mortality were: (a.) a facility being in the highest versus lowest quartile of facility percentage of vitamin D use and, (b.) the adjusted percentage of patients at a facility receiving vitamin D. The facility-level predictors were applied to all patients at that facility and used to predict each patient's mortality risk in unadjusted and adjusted Cox models.

Sensitivity analyses used baseline standard regression models to compare the relationship between mortality risk and IV versus oral vitamin D compounds among all patients, and between mortality risk and the most commonly prescribed oral (alphacalcidol, calcitoriol and other) and IV (calcitriol, paricalcitol, doxercalciferol and maxacalcitol) vitamin D compounds among DOPPS III patients (where information was available on the types of oral vitamin D analogues). In a second sensitivity analysis, time-varying vitamin D prescription was redefined to include patients with a prescription for vitamin D at study entry to remain in the treated category, while those with no prescription for vitamin D at baseline were allowed to switch into the treated category at the start of the first interval when he/she was prescribed vitamin D [19]. An additional sensitivity analysis restricted baseline, standard regression models to patients with similar levels of serum calcium (< and ≥10 mg/dl), phosphorus (< and ≥5.5 mg/dl) and PTH (< and ≥150 pg/ml) at study entry.

All Cox models accounted for facility clustering effects using robust standard error estimation techniques and were stratified by the study phase and geographic region of the DOPPS to account for differences in the availability of vitamin D analogues and survival rates across different countries over time. Patients were considered to be at risk from study enrolment until the date of death (within 60 days of departure from a facility), loss to follow-up, or study end. Results were expressed as relative rates/ratios (RR) with corresponding 95% confidence intervals (CI). P-values <0.05 were considered statistically significant.

Statistical analyses were performed with SAS software, version 9.1 (SAS Institute; Cary, NC, USA).



   Results
Top
 Abstract
 Introduction
 Methods
 Statistical methods
 Results
Discussion
 References
 
International patterns of vitamin D prescription
The practice of vitamin D prescription over time and the route of administration varied widely among countries (Figure 1). Vitamin D was prescribed for 54% of participants in the cross-sectional sample of DOPPS III. The percentage of patients on vitamin D ranged from 33% in France to 66% in the USA. With the exception of an 8% increase in vitamin D prescription from DOPPS II to DOPPS III in the USA, the percentage of patients on vitamin D remained stable over the follow-up time in all countries. While the IV route was largely preferred in the USA (62% of patients), in all other countries oral administration was more common (34–67% of patients). Among patients on IV vitamin D, the most commonly prescribed compounds were calcitriol (66.8%), paricalcitol (25.9%) and doxercalciferol (2.4%).


Figure 1
View larger version (36K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 1 Percentage of patients prescribed vitamin D and route of vitamin D administration in the DOPPS countries at each study phase. DOPPS I (1996–2000), II (2002–2004) and III (2005–2006) data combined, among the prevalent cross-sections of patients within each study phase, by country. Shown in the figure is the patient-level distribution of vitamin D prescription within countries and study phases of the DOPPS. Darkly shaded portions of the bars represent the percentage of patients not prescribed vitamin D, unshaded portions represent the percentage of patients prescribed an oral preparation of vitamin D and the lighter shaded portions represent the percentage of patients prescribed an intravenous (IV) preparation of vitamin D.

 
Distributions of the percentage of patients at each facility prescribed vitamin D by country and DOPPS phase are shown in Figure 2. Similar to the patient-level findings, the practice of facility vitamin D prescription in the USA tended to increase over time. No significant differences in prescription patterns were observed over time in other DOPPS countries.


Figure 2
View larger version (25K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 2 Facility-level prescription of vitamin D by region and DOPPS study phase. DOPPS I (1996–2000), II (2002–2004) and III (2005–2006) data combined. The facility-level practice of vitamin D prescription is expressed as the percentage of patients within each facility prescribed any type of vitamin D, and calculated among the prevalent cross-sections of patients within each facility, at the start of each study phase by country. Shown in the figure is the distribution (as 5th, 25th, 50th, 75th and 95th percentiles) of average facility vitamin D prescription across the facilities participating within each respective country and study phase of the DOPPS.

 
Characteristics of patients prescribed vitamin D
Table 1 shows demographic and clinical characteristics of participants according to baseline vitamin D prescription. Patients prescribed vitamin D were younger, more likely to be black, had longer time on dialysis prior to study entry and were more likely to have a fistula (60.9% versus 49.3%) or a graft (20.2% versus 17.5%) (P < 0.01). The prevalence of comorbidities, including diabetes (35.7% versus 40.3%), coronary artery disease (40.7% versus 43.0%) and peripheral vascular disease (23.2% versus 26.8%), was significantly lower among patients on vitamin D. Concentrations of serum albumin, calcium, phosphorus and PTH were higher in patients prescribed vitamin D (P < 0.001). Similar to the findings across all DOPPS countries, within each region patients on vitamin D were also younger and had fewer comorbid conditions than those not prescribed vitamin D (data not shown).


View this table:
[in this window]
[in a new window]

  		Table 1 Patient characteristics and baseline laboratory values according to vitamin D prescription at study entry and adjusted odds ratio (AOR) for vitamin D prescription

 
Predictors of initiation and cessation of vitamin D use (Table 2)
Participants in any DOPPS country were less likely to be started on vitamin D compared to participants in the USA (P < 0.0001). Factors significantly associated with vitamin D initiation were black race, time on dialysis <1 year or >6 years, low haemoglobin, low serum calcium and high PTH. Cessation of vitamin D was more likely for patients outside of the USA, of any race other than black, with <1 year on dialysis, having had a prior parathyroidectomy, with lower haemoglobin, serum albumin, or PTH, and with higher serum calcium, phosphorus or dialysate calcium.


View this table:
[in this window]
[in a new window]

  		Table 2 Factors significantly associated with vitamin D initiation and cessation

 
Patient characteristics at facilities with vitamin D prescription below or above the regional median (Table 3)
Overall, patients in facilities where vitamin D prescription was above the regional median (48% in Europe/ANZ, 59% in Japan and 54% in North America) did not appear to have characteristics associated with better or worse health compared to patients in facilities with below-median vitamin D prescription. Although the distribution of patient characteristics was not entirely balanced, no systematic differences were evident, indicating that the facility pattern of vitamin D prescription tended to be independent of the facility case mix.


View this table:
[in this window]
[in a new window]

  		Table 3 Patient demographic and clinical characteristics at facilities with vitamin D prescription below or above the regional mediana

 
Mortality risk associated with vitamin D prescription (Table 4)
The median follow-up time was 1.3 years (interquartile range: 0.7–2.0 years). The overall unadjusted mortality rate was 16.0 deaths per 100 patient-years. Among patients prescribed and not prescribed vitamin D, the rates were 14.0 and 17.9 deaths per 100 patient-years, respectively.


View this table:
[in this window]
[in a new window]

  		Table 4 Mortality risk associated with vitamin D prescription

 
Standard regression analyses.
In baseline models, the only significant survival advantage for patients receiving vitamin D compared to those not receiving vitamin D was found in the unadjusted analysis [RR = 0.86 (95% CI 0.82–0.90)] or when adjusted for only basic demographic covariates [RR = 0.91 (0.87–0.95)]. However, no differences in survival were observed after adjustment for comorbid conditions or laboratory values. When vitamin D was treated as a time-varying variable, a 21% lower RR of death was observed for patients on vitamin D [0.79 (0.75–0.82), P < 0.001] in the unadjusted model. The survival advantage decreased with progressive adjustment, but remained statistically significant in the fully adjusted model [0.89 (0.85–0.94), P < 0.001]. Similar results were found in sensitivity analyses, where patients were allowed to switch into the vitamin D treatment group during follow-up, with RR ranging from 0.80 (0.76–0.84) in the unadjusted to 0.92 (0.87–0.96) in the fully adjusted models.

MSMs.
Vitamin D prescription was associated with a significant survival advantage in both baseline [0.84 (0.78–0.89); P < 0.0001] and time-varying [0.78 (0.73–0.84); P = 0.0001] MSMs.

Facility-level analyses.
No survival benefit was found for patients treated at facilities where vitamin D prescription was higher. Specifically, the RR for facilities that were one interquartile range higher (75th versus 25th percentile) in the percentage of patients with vitamin D prescription ranged from 0.99 [0.93–1.04] in the unadjusted model to 0.99 [0.94–1.00] in the fully adjusted model (P = 0.6). Similar results were found when a 10% increase in the facility percentage of patients on vitamin D prescription was used as the predictor [RR in the fully adjusted model = 1.00 (1.00–1.00)]. Models that used the observed, non-case-mix adjusted percentage of facility patients receiving vitamin D as the predictor yielded virtually identical results (data not shown).

Sensitivity analyses.
In baseline standard regression models, the association of vitamin D with mortality did not vary by route of administration [RR for IV versus oral = 1.00 (0.89–1.13) among all DOPPS patients in a fully adjusted model]. Among patients receiving IV vitamin D, no significant difference in mortality was observed according to the type of analogue [RR for paricalcitol versus calcitriol = 1.06 (0.87–1.29); RR for doxercalciferol versus calcitriol = 1.10 (0.70–1.73)]. When testing the association between specific vitamin D analogues and mortality in facility-level models, no survival differences were found between facility percentage prescription of calcitriol, paricalcitol or doxercalciferol. Among DOPPS III patients, no significant differences were found between the association of mortality risk and different types of oral vitamin D analogues (data not shown). Additionally, there were no significant differences in the association between vitamin D and mortality in baseline, standard regression models restricted to patients with similar levels of serum calcium (< and ≥10 mg/dl), phosphorus (< and ≥5.5 mg/dl) and PTH (< and ≥150 pg/ml).



   Discussion
Top
 Abstract
 Introduction
 Methods
 Statistical methods
 Results
 Discussion
References
 
Patients prescribed vitamin D had longer survival, but were also younger and had case mix and laboratory characteristics that are generally associated with greater longevity than patients not prescribed vitamin D (Table 1). The observed RRs for survival in standard regression models and MSMs were remarkably similar to those previously reported [19–21]. However, results from instrumental variable approach models suggest that previous studies may have failed to account for unmeasured patient confounders. As in prior studies [19–21], our study found significant differences in measured case-mix characteristics between patients prescribed vitamin D and those who were not. Our results confirm that higher serum albumin concentrations in patients prescribed vitamin D [20] strongly suggest an overall healthier status. With the addition of a comprehensive list of patient demographics, comorbidities and laboratory covariate adjustments, the survival benefit virtually disappeared in baseline models and was substantially attenuated in time-varying models. These results suggest that associations of vitamin D prescription with improved survival are, in part, due to health status and that biologic effects of vitamin D analogues may be non-significant or small.

With MSMs it is possible to adjust for confounding effects of time-varying variables that may be both confounders and intermediate variables, such as serum calcium, phosphorus or PTH levels [29]. This technique has been used in dialysis patient studies to assess associations of vitamin D [19], and iron with mortality [33], and associations of ESA dosing with haemoglobin levels [34]. The 16% and 22% reduction in mortality risk observed in our unadjusted baseline and time-varying MSMs, respectively, are smaller than the 27% reduction reported by Teng et al. [19]. Substantially, greater levels of comorbidity adjustments in our models may partially explain the smaller magnitude. However, it is essential to note that MSMs can only address the effect of measured confounders, but not unmeasured ones. Despite many covariates, it is likely that unmeasured or unknown confounders remain unaccounted for in both standard regression and MSMs analyses.

In our analyses, no difference in survival was found between facilities with higher or lower vitamin D prescription when using an instrumental variable approach. This technique was developed to reduce treatment selection bias and offers the particular advantage of accounting for unmeasured confounders [26,27,31]. This same technique was recently used to estimate relationships between several treatments and clinical outcomes [27,28,35]. In our instrumental variable models, the adjusted facility percentage vitamin D prescription, which is reflective of a facility's likelihood to prescribe vitamin D based on clinical practice preferences and not the case mix of patients at that facility, was used to predict mortality instead of the patient's individual vitamin D prescription, which might be determined by the patient characteristics that could also be associated with altered mortality risk. To this point, demographic distributions, comorbidities and laboratory values were substantially more comparable in facilities with vitamin D prescription above and below the median (Table 3) than in comparisons of patients treated versus not treated with vitamin D (Table 1). This indicates that differences in case mix were accounted for by this facility-practice-based approach. Nonetheless, this approach also adjusts for the same variables at the patient level and considers the facility's propensity to use vitamin D based on its case mix.

The lack of association between vitamin D therapy and improved survival in our facility-level models concurs with a recent meta-analysis of a randomized controlled trial of vitamin D in CKD that concluded that ‘beneficial effects on patient-level outcomes (including death, hospitalization, cardiovascular events or fractures) of vitamin D are unproven’ [23,36]. However, due to the small number of patients included in mortality trials and the methodological limitations of meta-analyses, no definite conclusion should be drawn from these results.

In the current study, no survival advantage was found for newer IV vitamin D analogues (i.e. paricalcitol and doxercalciferol) versus calcitriol. Since the results of other observational studies are controversial [20,37] and no randomized controlled trials on this topic are available, the question remains on which vitamin D analogue (if any) may be best.

Results of our facility-level models are not subject to the bias found in some ecological studies, which are based on aggregates of patients and do not adjust for patient-level characteristics [38]. To avoid the limitations of the ecological design, we analysed outcomes with adjustment for patient characteristics at the patient level while using a facility-level treatment variable (the facility percentage of patients on vitamin D) to further reduce bias that could arise from unmeasured covariates. This approach reduces treatment-by-indication bias while also accounting fully for observed confounders at the individual level [39].

In contrast with other observational studies [19,20], our study population represents a cross-sectional sample of patients at participating facilities. This approach enables us to address the clinically important question of the relationship of vitamin D therapy to outcomes among patients across patient-years on dialysis. Since the DOPPS sample is representative of the worldwide HD population [40], these results are applicable to the practice of nephrology in a variety of countries across the world.

We acknowledge that the current study is susceptible to several limitations inherent to the observational design of DOPPS. Specifically, despite the comprehensive adjustments and the use of an instrumental variable approach, the possibility still remains that some sources of bias may have not been taken into account.

Due to the limited data available, we were unable to account for several aspects of vitamin D dosing that may affect patient outcomes, including vitamin D exposure prior to study entry and vitamin D dose. We also acknowledge the possibility that lack of compliance for patients prescribed an oral vitamin D analogue could bias our results. However, this is unlikely because in analyses restricted to patients who received oral versus IV vitamin D, no difference in mortality was found by route of administration.

The international nature of the DOPPS enabled us to describe differences and document substantial variations in vitamin D prescription patterns in 12 countries over a decade. Percentages of patients at each facility prescribed vitamin D tended to increase from DOPPS I to DOPPS III in the USA but not in other DOPPS regions (Figure 1). Increased vitamin D prescription in the USA may be related to published studies and the renewed interest in vitamin D, particularly in the US nephrology community [19–21]. Most patient characteristics associated with initiation of vitamin D therapy were inversely associated with cessation of vitamin D therapy, and vice-versa (Table 3). The duration of dialysis <1 year before study enrolment (versus 3–5 years) and low haemoglobin (<9.6 versus 10.8–11.8 g/dl) levels were significantly associated with both initiation and cessation of vitamin D. These associations point to the complexities of patient characteristics and provider preferences that influence decisions to prescribe vitamin D in the international HD community.

The established infrastructure and the collaborative support of the DOPPS investigators represent major strengths of the current study. The international design allowed us to assess relationships between vitamin D and mortality across HD patient populations, taking advantage of substantial variations in practice patterns that occur due to provider preferences rather than case mix alone. Finally, the large collection of DOPPS information allowed us to adjust models for many potentially confounding covariates, including 14 major classes of comorbid conditions based on ~70 comorbidity indicators [25].

In conclusion, this study raises questions about the previously reported survival advantage associated with vitamin D therapy. Although unadjusted models agree with prior reports, this advantage was attenuated or lost with more detailed adjustment for patient characteristics. Further analyses based on each facility's practice of vitamin D prescription as an instrumental variable to minimize potential treatment-by-indication bias show no survival benefit associated with vitamin D. These findings indicate that results of prior observational studies should be interpreted with caution and offer a different perspective to the ongoing debate regarding the survival benefits of vitamin D therapy. More importantly, our results indicate that it is ethical to conduct a much-needed clinical trial to clarify the impact of vitamin D on mortality among patients undergoing maintenance HD.



   Acknowledgments
 
The authors wish to thank all the patients and health care providers at dialysis facilities for participating in the DOPPS. The DOPPS is supported by research grants from Amgen and Kirin Pharma without restrictions on publications. Special thanks goes to Margaret Eichleay for her scientific input and to Shauna Leighton for editorial assistance.

Conflict of interest statement. The Corresponding author and all other co-authors have no potential conflicts of interest to declare.



   References
Top
 Abstract
 Introduction
 Methods
 Statistical methods
 Results
 Discussion
 References
 

  1. Moe S, Drueke T, Cunningham J, et al. Definition, evaluation, and classification of renal osteodystrophy: a position statement from Kidney Disease: Improving Global Outcomes (KDIGO). Kidney Int (2006) 69:1945–1953.[CrossRef][Web of Science][Medline]
  2. Goodman WG, Goldin J, Kuizon BD, et al. Coronary–artery calcification in young adults with end-stage renal disease who are undergoing dialysis. N Engl J Med (2000) 342:1478–1483.[Abstract/Free Full Text]
  3. Qunibi WY, Nolan CA, Ayus JC. Cardiovascular calcification in patients with end-stage renal disease: a century-old phenomenon. Kidney Int (2002) 62(Suppl):73–80.[CrossRef]
  4. Kestenbaum B, Sampson JN, Rudser KD, et al. Serum phosphate levels and mortality risk among people with chronic kidney disease. J Am Soc Nephrol (2005) 16:520–528.[Abstract/Free Full Text]
  5. London GM, Guerin AP, Verbeke FH, et al. Mineral metabolism and arterial functions in end-stage renal disease: potential role of 25-hydroxyvitamin D deficiency. J Am Soc Nephrol (2007) 18:613–620.[Abstract/Free Full Text]
  6. Krasniak A, Drozdz M, Pasowicz M, et al. Factors involved in vascular calcification and atherosclerosis in maintenance haemodialysis patients. Nephrol Dial Transplant (2007) 22:515–521.[Abstract/Free Full Text]
  7. Block GA, Klassen PS, Lazarus JM, et al. Mineral metabolism, mortality, and morbidity in maintenance hemodialysis. J Am Soc Nephrol (2004) 15:2208–2218.[Abstract/Free Full Text]
  8. Melamed ML, Eustace JA, Plantinga L, et al. Changes in serum calcium, phosphate, and PTH and the risk of death in incident dialysis patients: a longitudinal study. Kidney Int (2006) 70:351–357.[CrossRef][Web of Science][Medline]
  9. National Kidney Foundation. K/DOQI clinical practice guidelines for bone metabolism and disease in chronic kidney disease. Am J Kidney Dis (2003) 42:S1–S202.[Medline]
  10. Tokuda N, Kano M, Meiri H, et al. Calcitriol therapy modulates the cellular immune responses in hemodialysis patients. Am J Nephrol (2000) 20:129–137.[CrossRef][Web of Science][Medline]
  11. Deluca HF, Cantorna MT. Vitamin D: its role and uses in immunology. FASEB J (2001) 15:2579–2585.[Abstract/Free Full Text]
  12. Li YC, Kong J, Wei M, et al. 1,25-Dihydroxyvitamin D(3) is a negative endocrine regulator of the renin–angiotensin system. J Clin Invest (2002) 110:229–238.[CrossRef][Web of Science][Medline]
  13. Elstner E, Linker-Israeli M, Said J, et al. 20-epi-vitamin D3 analogues: a novel class of potent inhibitors of proliferation and inducers of differentiation of human breast cancer cell lines. Cancer Res (1995) 55:2822–2830.[Abstract/Free Full Text]
  14. Morimoto S, Imanaka S, Koh E, et al. Comparison of the inhibitions of proliferation of normal and psoriatic fibroblasts by 1 alpha, 25-dihydroxyvitamin D3 and synthetic analogues of vitamin D3 with an oxygen atom in their side chain. Biochem Int (1989) 19:1143–1149.[Web of Science][Medline]
  15. Dusso AS, Brown AJ, Slatopolsky E. Vitamin D. Am J Physiol Renal Physiol (2005) 289:F8–F28.[Abstract/Free Full Text]
  16. Tangpricha V, Flanagan JN, Whitlatch LW, et al. 25-Hydroxyvitamin D-1alpha-hydroxylase in normal and malignant colon tissue. Lancet (2001) 357:1673–1674.[CrossRef][Web of Science][Medline]
  17. Zaloga GP, Chernow B. The multifactorial basis for hypocalcemia during sepsis. Studies of the parathyroid hormone-vitamin D axis. Ann Intern Med (1987) 107:36–41.[CrossRef][Web of Science][Medline]
  18. Shoji T, Shinohara K, Kimoto E, et al. Lower risk for cardiovascular mortality in oral 1alpha-hydroxy vitamin D3 users in a haemodialysis population. Nephrol Dial Transplant (2004) 19:179–184.[Abstract/Free Full Text]
  19. Teng M, Wolf M, Ofsthun MN, et al. Activated injectable vitamin D and hemodialysis survival: a historical cohort study. J Am Soc Nephrol (2005) 16:1115–1125.[Abstract/Free Full Text]
  20. Tentori F, Hunt WC, Stidley CA, et al. Mortality risk among hemodialysis patients receiving different vitamin D analogs. Kidney Int (2006) 70:1858–1865.[CrossRef][Web of Science][Medline]
  21. Kalantar-Zadeh K, Kuwae N, Regidor DL, et al. Survival predictability of time-varying indicators of bone disease in maintenance hemodialysis patients. Kidney Int (2006) 70:771–780.[CrossRef][Web of Science][Medline]
  22. Wolf M, Betancourt J, Chang Y, et al. Impact of activated vitamin D and race on survival among hemodialysis patients. J Am Soc Nephrol (2008) 19:1379–1388.[Abstract/Free Full Text]
  23. Palmer SC, McGregor DO, Macaskill P, et al. Meta-analysis: vitamin D compounds in chronic kidney disease. Ann Intern Med (2007) 147:840–853.[Abstract/Free Full Text]
  24. Young EW, Goodkin DA, Mapes DL, et al. The Dialysis Outcomes and Practice Patterns Study (DOPPS): an international hemodialysis study. Kidney Int (2000) 57:S74–S81.[CrossRef][Web of Science]
  25. Pisoni RL, Gillespie BW, Dickinson DM, et al. The Dialysis Outcomes and Practice Patterns Study (DOPPS): design, data elements, and methodology. Am J Kidney Dis (2004) 44:7–15.[CrossRef][Medline]
  26. Newhouse JP, McClellan M. Econometrics in outcomes research: the use of instrumental variables. Annu Rev Public Health (1998) 19:17–34.[CrossRef][Web of Science][Medline]
  27. Brookhart MA, Wang PS, Solomon DH, et al. Evaluating short-term drug effects using a physician-specific prescribing preference as an instrumental variable. Epidemiology (2006) 17:268–275.[CrossRef][Web of Science][Medline]
  28. Stukel TA, Fisher ES, Wennberg DE, et al. Analysis of observational studies in the presence of treatment selection bias: effects of invasive cardiac management on AMI survival using propensity score and instrumental variable methods. JAMA (2007) 297:278–285.[Abstract/Free Full Text]
  29. Robins JM, Hernan MA, Brumback B. Marginal structural models and causal inference in epidemiology. Epidemiology (2000) 11:550–60.[CrossRef][Web of Science][Medline]
  30. Hernan M, Brumback B, Robins JM. Marginal structural models to estimate causal effect of zidovudine on survival of HIV-positive men. Epidemiology (2000) 11:561–570.[CrossRef][Web of Science][Medline]
  31. Angrist J, Imbens G, Rubin D. Identification of causal effects using instrumental variables. J Am Stat Assoc (1996) 91:444–455.[CrossRef][Web of Science]
  32. Schmoor C, Caputo A, Schumacher M. Evidence from nonrandomized studies: a case study on the estimation of causal effects. Am J Epidemiol (2008) 167:1120–1129.[Abstract/Free Full Text]
  33. Feldman HI, Joffe M, Robinson B, et al. Administration of parenteral iron and mortality among hemodialysis patients. J Am Soc Nephrol (2004) 15:1623–1632.[Abstract/Free Full Text]
  34. Cotter D, Zhang Y, Thamer M, et al. The effect of epoetin dose on hematocrit. Kidney Int (2008) 73:347–353.[CrossRef][Web of Science][Medline]
  35. Schneeweiss S, Seeger JD, Landon J, et al. Aprotinin during coronary-artery bypass grafting and risk of death. N Engl J Med (2008) 358:771–783.[Abstract/Free Full Text]
  36. Tonelli M. Vitamin D in patients with chronic kidney disease: nothing new under the sun. Ann Intern Med (2007) 147:880–881.[Free Full Text]
  37. Teng M, Wolf M, Lowrie E, et al. Survival of patients undergoing hemodialysis with paricalcitol or calcitriol therapy. N Engl J Med (2003) 349:446–456.[Abstract/Free Full Text]
  38. Piantadosi S, Byar DP, Green SB. The ecological fallacy. Am J Epidemiol (1988) 127:893–904.[Free Full Text]
  39. Johnston SC. Combining ecological and individual variables to reduce confounding by indication: case study—subarachnoid hemorrhage treatment. J Clin Epidemiol (2000) 53:1236–1241.[CrossRef][Web of Science][Medline]
  40. Port FK, Wolfe RA, Held PJ, et al. Random sample (DOPPS) versus census-based (registry) approaches to kidney disease research. Blood Purif (2003) 21:85–88.[CrossRef][Web of Science][Medline]
Received for publication: 14. 5.08
Accepted in revised form: 26. 9.08


Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?


This article has been cited by other articles:


Home page
 J. Am. Soc. Nephrol.Home page
R. Thadhani
Is Calcitriol Life-Protective for Patients with Chronic Kidney Disease?
J. Am. Soc. Nephrol., November 1, 2009; 20(11): 2285 - 2290.
[Full Text] [PDF]

Home page
 Nephrol Dial TransplantHome page
V. S. Stel, F. W. Dekker, D. Ansell, H. Augustijn, F. G. Casino, F. Collart, P. Finne, G. A. Ioannidis, M. Salomone, J. P. Traynor, et al.
Residual renal function at the start of dialysis and clinical outcomes
Nephrol. Dial. Transplant., October 1, 2009; 24(10): 3175 - 3182.
[Abstract] [Full Text] [PDF]

Home page
 CJASNHome page
K. Kalantar-Zadeh and C. P. Kovesdy
Clinical Outcomes with Active versus Nutritional Vitamin D Compounds in Chronic Kidney Disease
Clin. J. Am. Soc. Nephrol., September 1, 2009; 4(9): 1529 - 1539.
[Abstract] [Full Text] [PDF]

Home page
 Nephrol Dial TransplantHome page
S. C. Palmer, J. C. Craig, and G. F. M. Strippoli
Taking aim at targets
Nephrol. Dial. Transplant., May 1, 2009; 24(5): 1358 - 1361.
[Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
24/3/963    most recent
gfn592v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Tentori, F.
Right arrow Articles by Port, F. K.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Tentori, F.
Right arrow Articles by Port, F. K.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?