Mortality Rates Across 25-Hydroxyvitamin D (25[OH]D) Levels among Adults with and without Estimated Glomerular Filtration Rate <60 ml/min/1.73 m2: The Third National Health and Nutrition Examination Survey

Background Previous studies exploring the association between 25[OH]D levels and mortality in adults with and without kidney disease utilized 25[OH]D thresholds that have recently been scrutinized by the Institute of Medicine Committee to Review Dietary References Intakes for Vitamin D and Calcium. Objective We explored all-cause mortality rates across the spectrum of 25[OH]D levels over an eighteen-year follow-up among adults with and without an estimated glomerular filtration rate (eGFR) <60 ml/min/1.73 m2. Design The study included 1,097 U.S. adults with eGFR <60 ml/min/1.73 m2 and 14, 002 adults with eGFR ≥60 ml/min/1.73 m2. Mortality rates and rate ratios (RR) across 25[OH]D groups were calculated with Poisson regression and restricted cubic splines while adjusting for covariates. Results Prevalence of 25[OH]D levels <30 and <20 ng/ml among adults with eGFR <60 ml/min/1.73 m2 was 76.5% (population estimate 6.2 million) and 35.4% (population estimate 2.9 million), respectively. Among adults with eGFR ≥60 ml/min/1.73 m2, 70.5% had 25[OH]D levels <30 ng/ml (population estimate 132.2 million) while 30.3% had 25[OH]D levels <20 ng/ml (population estimate 56.8 million). Significantly higher mortality rates were noted among individuals with 25[OH]D levels <12 ng/ml compared to referent group (24 to <30 ng/ml): RR1.41 (95% CI 1.17, 1.71) among individuals with eGFR <60 ml/min/1.73 m2 and RR 1.32 (95% CI 1.13, 1.56) among individuals with eGFR ≥60 ml/min/1.73 m2 after adjustment for covariates including co-morbid conditions. Mortality rates were fairly similar across all 25[OH]D groups with levels >20 ng/ml after adjustment for all covariates. Conclusions Regardless of presence of eGFR <60 ml/min/1.73 m2, mortality rates across groups with 25[OH]D levels 20–40 ng/ml are similar.


25-Hydroxyvitamin D (25[OH]D) deficiency is based on
[OH]D levels associated with rickets in children [1] although there currently is no consensus on thresholds that define 25[OH]D deficiency or insufficiency. For example, the Institute of Medicine Committee to Review Dietary References Intakes for Vitamin D and Calcium defines ''risk of deficiency'' as levels ,12 ng/ml [2] and that some, but not all, individuals may be at ''risk for insufficiency'' with 25[OH]D levels between 12 to 20 ng/ml. The Endocrine Society Clinical Practice Guideline defines 25 [OH] deficiency and insufficiency as levels ,20 ng/ml and levels 20 to 29 ng/ml, respectively. [3] The existing controversy over 25[OH]D levels which define ''insufficiency'' or ''risk of insufficiency'' requiring supplementation carries substantial public health relevance because approximately half of the U.S. noninstitutionalized adult population has a 25[OH]D level within the range of 12-29.9 ng/ml.l [4,5] Thus, the majority of 25[OH]D supplementation is currently for treatment of ''insufficient'' rather than ''deficient'' levels of 25[OH]D.
Two previously published studies examined the association between 25[OH]D levels and mortality in adults with CK [4,5] and without CKD [5] using the Third National Health and Nutrition Examination Study linked with the mortality file. These studies utilized either 25 25[OH]D levels whereby parathyroid hormone (PTH) levels are maximally suppressed in adults without CKD. [4][5][6][7][8][9] This treatment threshold has been questioned by the Institute of Medicine Committee to Review Dietary References Intakes for Vitamin D and Calcium, a committee convened by the Canadian and U.S. governments. The Committee concluded that 25[OH]D levels $20 ng/ml are sufficient for the majority of individuals. [2] Observational studies have shown heightened risk for both mortality and cardiovascular outcomes with 25[OH]D levels ,20 ng/ml, but 25[OH]D levels are strongly influenced by sun exposure, skin pigmentation, and body fat. These factors may co-vary with co-morbid conditions that impact overall mortality. Given the recent reappraisal of clinical thresholds for 25[OH]D supplementation, [2] we examined all-cause mortality rates across the range of 25[OH]D levels in adults with and without eGFR ,60 ml/min/1.73 m 2 , which defines stage 3-5 CKD, using a representative sample of U.S. adults. The population was stratified by presence of stage 3-5 CKD due to the strong link between disorders of mineral metabolism, co-morbid conditions which influence vitamin D levels (through decreased outdoor activity and alterations in diet) and reduced glomerular filtration rate. [10] Models were constructed to determine how baseline co-morbid conditions may influence the association between baseline 25[OH]D levels and all-cause mortality.

Study Population
The study population source was the Third National Health and Nutrition Examination Survey (NHANES III) linked with the National Death Index with up to 18 years of follow-up (1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006). The NHANES III was designed as a probability sample of the total civilian non-institutionalized population, 2 months of age or older, in the United States and collected health and nutritional data on 33,994 men, women, and children from 1988-1994. Certain subgroups were over-sampled such as young children, older persons, non-Hispanic blacks, and Mexican Americans. The survey was approved by the National Center for Health Statistics Institutional Review Board and all participants provided written informed consent. Details of the survey design may be found in the NHANES III operations manual. [11] All NHANES III participants who were 17 years and older at the time of the survey were eligible for mortality follow-up. This study was approved by the Loyola University Medical Center Institutional Review Board. Figure 1 outlines the selection of NHANES participants included in the final analysis. A total of 17, 683 adults age 20 years and older with known vital status were available for analysis and 864, 344 and 263 were excluded for missing information on 25[OH]D levels, serum creatinine and pregnancy status, respectively. An additional 58 individuals were excluded due to missing information on covariates. The analysis was then limited to 1,097 participants with eGFR ,60 ml/min/1.73 m 2 (defined below) and 14,002 with eGFR $60 ml/min/1.73 m 2 .

Estimated Glomerular Filtration Rate
Serum creatinine was measured by a Roche/Hitachi 737 analyzer (Roche Diagnostics, Indianapolis, IN using the kinetic alkaline picrate reaction. [12] Serum creatinine values calibrated to standardized creatinine values [13] [14].where k is 0.7 for females and 0.9 for males, a is 20.329 for females and 20.411 for males, min indicates the minimum of Scr/k or 1, and max indicates the maximum of Scr/k or 1.

Mortality
Mortality data was obtained from the NHANES III Linked Mortality File Public-use file.
[15] This file contains mortality status through December 31, 2006 for participants 17 years of age and older who participated in the NHANES III survey. Mortality status was ascertained from the National Death Index or Social Security Administration records. In some cases, death of NHANES III participants was also verified by reviewing death certificates and comparing them with survey records to confirm the matching of a death record of a survey participant with the National Death Index record. Deaths occurring during years 1988-1998 were coded according to the International Classification of Diseases, 9 th revision, Clinical Modification. [16] All other deaths were coded according to the International Classification of Diseases, 10 th Revision. [17] Cause of death was grouped according to ICD-10.

Vitamin D Levels
In NHANES III, higher latitude regions (northern states) were sampled during the summer, whereas lower latitude regions (southern states) were generally sampled during the winter. Serum 25[OH]D measurements were completed at the National Center for Environmental Health, Centers for Disease Control using a radioimmunoassay kit (DiaSorin, Stillwater, MN) [18].  Covariates Age was defined as the age at the time of the interview, and race/ethnicity was self-reported as non-Hispanic white, non-Hispanic black, or Mexican-American. Other race/ethnicities were grouped into ''Other''. Trained examiners measured blood pressure using standardized protocols and the average of all six blood pressure measurements collected during the interview and physical exam was used for this analysis. Diabetes was defined as self-reported previous physician diagnosis other than during pregnancy or the current or past use of glucose lowering agents. Presence of congestive heart failure, non-skin related cancer, stroke, and myocardial infarction were based on self-report. Participants provided a single spot urine specimen and urine albumin and creatinine were measured using a solid-phase fluorescence immunoassay and the Jaffe rate reaction (Beckman Astra, Brea, CA), respectively. Urine albumin/creatinine ratios in mg/g were calculated for all participants. Serum phosphate was measured using a Hitachi model 737 multi-channel analyzer (Boehringer Mannheim Diagnostics, Indianapolis, IN). Use of anti-hypertensive medications or anticonvulsants, estrogens, or glucocorticoids was based on self-report. The season in which the examination took place was dichotomized as winter (October-March) or summer (April-September). Education was defined as less than high school, high school completion or education past high school. Cigarette smoking was defined as current smoking (yes/no).

Statistical Analysis
NHANES III data are weighted to account for the probability of selection and to adjust for non-response to the interview and physical exam. All the analyses were conducted using STATA 12 (STATA Corporation, College Station, TX) ''Survey'' procedures which incorporate the weights, strata and cluster of the complex study design. Descriptive characteristics of adults with and without eGFR ,60 ml/min/1.73 m 2 were reported across the seven groups of 25    subjects in the group,12-15.9 ng/ml compared to the group 24-29.9 ng/ml. Similar calculations were carried out for the other groups.
After testing the assumptions of the proportional hazards model by using log minus log survival plots, analyses were repeated using Cox-proportional hazard models and compared with results using Poisson regression. The model adjusted for for age. sex, race,season, co-morbidities (diabetes, CHF, MI, stroke, cancer) body mass index (BMI), systolic blood pressure (SBP), eGFR, smoking status, use of anti-hypertensive medications, anticonvulsant medications, estrogens or glucocorticoids, spot urine albumin/creatinine ratios, serum phosphorous and education. The    Table 3 shows the number of deaths and reported cause of death by 25[OH]D groups among adults with and without eGFR ,60 ml/min/1.73 m 2 . The majority of deaths for adults with and without eGFR ,60 ml/min/1.73 m 2 were attributed to cardiovascular diseases while cancer was the second leading reported cause. Figures 2 and 3 show the mortality rates per 1,000 personyears by 25(OH)D groups for adults with and without eGFR ,60 ml/min/1.73 m 2 , respectively, after adjustment for age, sex, race/ethnicity and season of 25 Among adults with eGFR $60 ml/min/1.73 m 2 , fully adjusted mortality rates per 1000 person-years were over seven-fold lower compared to adults with eGFR ,60 ml/min/1.73 m 2 for all 25

Discussion
In a representative sample of the US adults with and without eGFR ,60 ml/min/1.73 m 2 , 25[OH]D levels ,12 ng/ml, consistent with 25[OH]D deficiency, [2] are associated with significantly increased risk of all-cause mortality after adjustment for all covariates. These findings are consistent with previous studies which examined the association between 25[OH]D levels and mortality using data from the Third National Health and Nutrition Examination Survey. [4,5] The persistence of the association between 25[OH]D deficiency and mortality after adjustment for co-morbid conditions and the consistent shape of the overall distribution of all-cause mortality across 25[OH]D groups after eliminating participants who died during the first three years of follow-up argues against 25[OH]D deficiency simply reflecting overall poor health. These findings are supported by several observational studies which show that 25[OH]D deficiency is associated with an increased risk of mortality in adults with and without CKD. [4,5,[20][21][22][23] A meta-analysis of observational studies showed that the association between 25[OH]D and mortality appears non-linear with no significant decreases in mortality once 25[OH]D levels exceed 35 ng/ml compared to levels ,11 ng/ml. [23] In the meta-analysis which included 62, 548 individuals from 14 prospective studies, baseline 25[OH]D levels were ,20 ng/ml in the majority of these studies. [23] Most, [4,5,23,24] but not all studies [25] [3]) is associated with a relative risk reduction of 7%. [33] However, the association between 25[OH]D supplementation and mortality differs by baseline serum 25[OH]D levels, which is not consistent across trials. [33] Thus, optimal levels for treatment goals remain undetermined. Moreover, 25[OH]D supplementation may increase the risk of cancers [34,35] and kidney stones. [36] Given the results of previous supplementation trials of other vitamins or minerals, [37][38][39] higher levels may not necessarily be beneficial and may heighten risk despite suggestion of clinical utility from observational studies. Clinical trials are needed to ascertain the risks and benefits of vitamin D supplementation for adults with and without kidney disease.
The strengths of this study include the representative sample of non-institutionalized U.S. adults with known survey weights, allowing us to extrapolate to the general population. However, these findings are not generalizable to adults living in nursing homes or those receiving dialysis or who have received a kidney transplant. Measurement of 25[OH]D occurred during the warmer months in the Northern states, which may potentially underestimate the prevalence of 25[OH]D deficiency and insufficiency in this region. Optimal thresholds where mortality rate is lowest could not be determined due to low numbers of individuals in this group. Although 25(OH)D exists in two forms [25(OH)D2 and 25(OH)D3], the Diasorin assay measures only total 25(OH)D. [40] Finally, because our study is observational, we cannot rule out confounding by unmeasured factors such as social deprivation. [41].
In conclusion, while significantly higher mortality rates are noted with 25[OH]D levels ,12 ng/dl, mortality rates are fairly similar across the range of 25[OH]D levels 20-40 ng/dl among adults with and without eGFR ,60 ml/min/1.73 m 2 .