Vit d


Published on

Published in: Health & Medicine
  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

Vit d

  1. 1. Serum 25-Hydroxyvitamin D Concentration and Risk for Major ClinicalDisease Events in a Community-Based Population of Older AdultsA Cohort StudyIan H. de Boer, MD, MS; Gregory Levin, MS; Cassianne Robinson-Cohen, MS; Mary L. Biggs, PhD; Andy N. Hoofnagle, MD, PhD;David S. Siscovick, MD, MPH; and Bryan Kestenbaum, MD, MSBackground: Circulating concentrations of 25-hydroxyvitamin D[25-(OH)D] are used to define vitamin D deficiency. Current clinical25-(OH)D targets based on associations with intermediate markersof bone metabolism may not reflect optimal levels for other chronicdiseases and do not account for known seasonal variation in 25-(OH)D concentration.Objective: To evaluate the relationship of 25-(OH)D concentrationwith the incidence of major clinical disease events that are patho-physiologically relevant to vitamin D.Design: Cohort study.Setting: The Cardiovascular Health Study conducted in 4 U.S. com-munities. Data from 1992 to 2006 were included in this analysis.Participants: 1621 white older adults.Measurements: Serum 25-(OH)D concentration (using a high-performance liquid chromatography–tandem mass spectrometry as-say that conforms to National Institute of Standards and Technol-ogy reference standards) and associations with time to a compositeoutcome of incident hip fracture, myocardial infarction, cancer, ordeath.Results: Over a median 11-year follow-up, the composite outcomeoccurred in 1018 participants (63%). Defining events included 137hip fractures, 186 myocardial infarctions, 335 incidences of cancer,and 360 deaths. The association of low 25-(OH)D concentrationwith risk for the composite outcome varied by season (P ϭ 0.057).A concentration lower than a season-specific Z score of Ϫ0.54 bestdiscriminated risk for the composite outcome and was associatedwith a 24% higher risk in adjusted analyses (95% CI, 9% to 42%).Corresponding season-specific 25-(OH)D concentrations were 43,50, 61, and 55 nmol/L (17, 20, 24, and 22 ng/mL) in winter,spring, summer, and autumn, respectively.Limitation: The observational study was restricted to whiteparticipants.Conclusion: Threshold concentrations of 25-(OH)D associated withincreased risk for relevant clinical disease events center near 50nmol/L (20 ng/mL). Season-specific targets for 25-(OH)D concen-tration may be more appropriate than static targets when evaluat-ing health risk.Primary Funding Source: National Institutes of Health.Ann Intern Med. 2012;156:627-634. www.annals.orgFor author affiliations, see end of text.Vitamin D has attracted increasing attention in clinicalmedicine and research, in part because of its pleiotro-pic effects on biological processes other than calcium andbone homeostasis (1–3). Animal experimental studies dem-onstrate that 1,25-dihydroxyvitamin D, the active vitaminD hormone, suppresses the renin–angiotensin–aldosteronesystem, modulates immune cell function, and suppressesabnormal cell proliferation (4). Epidemiologic studies sug-gest that these actions may have clinical relevance, demon-strating that, in addition to fracture, vitamin D deficiencyis associated with increased risks for coronary heart disease,cancer, and all-cause mortality (5–11).Circulating concentrations of 25-hydroxyvitamin D[25-(OH)D], which reflect total vitamin D intake fromcutaneous synthesis and dietary consumption, are used todefine vitamin D deficiency (1–3). Biological 25-(OH)Dthresholds below which adequate conversion to 1,25-dihydroxyvitamin D cannot be maintained may exist.Optimal concentrations of 25-(OH)D have been pro-posed on the basis of cross-sectional correlations with in-termediate measures of bone and mineral metabolism, suchas parathyroid hormone concentration, bone mineral den-sity, and intestinal calcium absorption (1, 12–15). Thisapproach relates biomarker levels to biological function, animportant strength, but it also has several limitations. First,25-(OH)D concentrations that are optimal for bone andmineral metabolism may not equal those for nonbone vi-tamin D activities. Second, current recommendations fortarget 25-(OH)D concentrations do not account forknown seasonal variation in 25-(OH)D concentration(16–19). Third, existing recommendations are based ondivergent 25-(OH)D assays, and Standard Reference Ma-terials released by the National Institute of Standards andTechnology (NIST) now permit reproducible 25-(OH)Dtesting to enhance external validity (20). In addition, 25-See also:PrintEditors’ Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 628Summary for Patients. . . . . . . . . . . . . . . . . . . . . . . I-36Web-OnlySupplementsConversion of graphics into slidesAnnals of Internal Medicine Original Research© 2012 American College of Physicians 627
  2. 2. (OH)D targets are highly controversial—the Institute ofMedicine (IOM) recently recommended a threshold of 50nmol/L (20 ng/mL), substantially less than the 75-nmol/L(30-ng/mL) threshold recommended by other professionalsocieties and expert panels (1, 12–15).The goal of this study was to examine the relationshipof serum 25-(OH)D concentration to vitamin D in termsof risk for major clinical disease events of global pathophys-iologic relevance, focusing on threshold concentrations as-sociated with disease risk.METHODSStudy PopulationThe CHS (Cardiovascular Health Study) is a prospec-tive, community-based cohort study designed to examinerisk factors for the development and progression of cardio-vascular disease in people aged 65 years or older (21). Par-ticipants were recruited from 4 U.S. communities: ForsythCounty, North Carolina; Sacramento County, California;Washington County, Maryland; and Pittsburgh, Pennsyl-vania. Eligible participants were sampled by using Medi-care eligibility lists, were not institutionalized, and wereexpected to remain in the area for at least 3 years. Personswho were wheelchair-bound in the home or receiving hos-pice treatment, radiation therapy, or chemotherapy wereexcluded. The original CHS cohort of 5201 participantswas enrolled between 1989 and 1990, with an additional687 predominantly black participants enrolled between1992 and 1993.We measured serum 25-(OH)D concentration at the1992–1993 study visit for 2312 CHS participants who hadno clinical evidence of cardiovascular disease at that timeand who had available frozen serum (11). To expand ourfocus to incident cancer and hip fracture for this study, weadditionally excluded 328 participants with a history ofcancer and 13 participants with a previous hip fracture. Wealso excluded 45 participants with missing data on smok-ing and physical activity (Supplement 1, available at Because 25-(OH)D concentrations andpossibly their associations with health outcomes vary byrace, we focused on the 1621 white persons meetingthese criteria.25-(OH)D ConcentrationFasting serum was collected from CHS participants atthe 1992–1993 study visit and stored at Ϫ70 °C. We mea-sured total 25-(OH)D [25-(OH)D2 and 25-(OH)D3] byusing high-performance liquid chromatography–tandemmass spectrometry on a Waters Quattro micro mass spec-trometer (Waters Corporation, Milford, Massachusetts) in2008. The interassay coefficient of variation was less than3.4%. The assay was validated by using NIST StandardReference Material 972 (accuracy within 5%) (20). 25-Hydroxyvitamin D is known to be stable for long periodsat Ϫ70 °C (22).Composite Clinical OutcomeThe primary study outcome was time to first occur-rence of incident hip fracture, incident myocardial infarc-tion (MI), incident cancer, or death from any cause. Thiscomposite outcome was chosen before analysis to capturepreviously described associations of 25-(OH)D with dis-ease outcomes. We defined hip fracture by the Interna-tional Classification of Diseases, Ninth Revision, codes820.xx without a concomitant code for motor vehicle ac-cident (E810–E819) or pathologic fracture (733.1x) (6).The CHS Events Committee adjudicated cases of MI byusing available hospital discharge summaries, diagnostictest records, and consultation reports (23). Investigators forthe CHS identified incident cancer cases by linking CHSrecords with population-based cancer registries serving the4 CHS regions (24). We omitted outcomes that may becausally related to low 25-(OH)D concentration but have adiagnosis that is imprecisely ascertained or is made largelyon the basis of physical measurements, such as diabetes,hypertension, and impaired muscle function. We definedtime to composite outcome as the time elapsed betweenthe 1992–1993 examination, when serum 25-(OH)D con-centrations were measured (baseline), and either the earliestevent or the end of follow-up for cancer ascertainment(31 December 2005 for the California, Pennsylvania,and North Carolina sites, and 31 December 2006 for theMaryland site).CovariatesCovariates were ascertained at the 1992–1993 CHSstudy visit and were selected on the basis of their suspectedconfounding influence on associations of 25-(OH)D withstudy outcomes. Total physical activity was estimated byusing the Minnesota Leisure Time Physical Activity Ques-tionnaire, which assesses a range of common activities,ContextVitamin D deficiency is defined by its association withmarkers of bone metabolism, not by its association withclinical outcomes.ContributionThis study followed elderly people and found that base-line levels of 25-hydroxyvitamin D less than 50 nmol/L(20 ng/mL) were associated with a composite outcomethat included hip fracture, myocardial infarction, incidentcancer, and death.CautionThis was an observational study of white persons.ImplicationThe threshold identified in this study is closer to the valuerecently recommended by the Institute of Medicine thanto the value recommended by most other professionalsocieties and expert panels (75 nmol/L [30 ng/mL]).—The EditorsOriginal Research Serum 25-(OH)D and Risk for Major Clinical Disease Events628 1 May 2012 Annals of Internal Medicine Volume 156 • Number 9
  3. 3. such as walking for exercise, jogging, biking, aerobics, golf,tennis, swimming, weight training, mowing the lawn,strenuous household chores, and use of a treadmill or aer-obic machine (25, 26). Current smoking was ascertainedby questionnaire. Time of blood collection was categorizedin 3-month blocks to reflect the 4 seasons and the observedpattern of seasonal variation in 25-(OH)D concentrationin our population.Statistical AnalysisWe tested associations of 25-(OH)D concentrationwith study outcomes by using Cox proportional hazardsmodels with robust SEs, adjusted for age, sex, clinical site,smoking (current or not current), body mass index (incategories), and physical activity (kilocalories per week incategories). This set of covariates was chosen before anal-ysis to include important demographic characteristicsand potential strong confounders while also maintaininga relatively parsimonious model. We censored partici-pants at the time of death in analyses of nonfataloutcomes.We evaluated 25-(OH)D as a dichotomous variable toaddress the clinical utility of a 25-(OH)D threshold forrisk assessment and because we and others have seenthreshold associations of 25-(OH)D concentration withrisks for fracture, MI, and death (6–10). Given knownseasonal variability in 25-(OH)D concentration (16–19)and its associated effect on modeling (27), we plannedbefore analysis to assess the effect of season on the relation-ship of 25-(OH)D concentration and the composite out-come. We compared nested models with and without in-teraction terms for 25-(OH)D concentration by season byusing a multivariate Wald test. When we saw significantheterogeneity, we examined season-specific 25-(OH)Dconcentrations as exposures. This approach has been ap-plied previously and reduces bias (27, 28).To describe the functional form of the association of25-(OH)D concentration with the composite outcome, wefirst calculated unadjusted incidence rates by season-specific decile of 25-(OH)D concentration. Second, wecreated an adjusted penalized spline model with season-specific 25-(OH)D Z score as the flexibly modeled expo-sure variable and graphically displayed the spline at themean values of adjustment covariates (29). The penalizedspline was computed by using the default algorithm of thesurvival package in R 2.12.1 (R Foundation for Statis-tical Computing, Vienna, Austria), which uses evenlyspaced knots, cubic polynomials, and a penalty to re-strict the overall flexibility of the fitted curve (30).Third, we used a simple statistical approach similar to theContal–O’Quigley method to estimate an “optimal” season-specific 25-(OH)D Z score cut-point (31). For each possi-ble threshold from the inner 90% of the season-specific25-(OH)D Z score distribution (0.01 unit increments), wecomputed the Wald statistic with robust SE to describe thestrength of the adjusted association between 25-(OH)Ddeficiency (using that candidate cut-point) and rates of thecomposite outcome. The Z score cut-point that producedthe largest Wald statistic was defined as “optimal” in thesense that it best discriminated between low- and high-riskparticipants with these statistical criteria. We quantified theuncertainty in our estimated optimal threshold by comput-ing approximate CIs based on the observed quantiles of thedistribution of estimates across 2000 nonparametric boot-strap samples (32). Because an optimal cut-point with highstatistical precision is difficult to estimate, we presentboth standard 95% CIs and the narrower 75% CIs. Wecomputed net reclassification improvement to assesswhether season-specific 25-(OH)D concentration im-proved prediction of the primary composite outcome(cumulative incidence at 10 years, through whichfollow-up was 100% complete) compared with static25-(OH)D concentration (50 nmol/L [20 ng/mL] re-gardless of season) (33).All P values are 2-sided. Statistical analyses were com-pleted using R 2.12.1 and STATA 10.1 (Stata Corp, Col-lege Station, Texas).Role of the Funding SourceThe National Institutes of Health provided fundingfor this study. The funding source had no role in the de-sign, conduct, or analysis of this study or the decision tosubmit the manuscript for publication.Figure 1. Box plot of 25-(OH)D concentration by season,showing the 25th, 50th, and 75th percentiles ofdistributions, with outliers not shown.25-(OH)DConcentration,nmol/L20406080100120Winter Spring Summer AutumnMean 25-(OH)D was 56 nmol/L (SD, 24), 63 nmol/L (SD, 24), 74nmol/L (SD, 25), and 69 nmol/L (SD, 26) (22 ng/mL [SD, 10], 25ng/mL [SD, 10], 30 ng/mL [SD, 10], and 28 ng/mL [SD, 11]) in winter(January–March), spring (April–June), summer (July–September),and autumn (October–December), respectively. 25-(OH)D ϭ 25-hydroxyvitamin D.Original ResearchSerum 25-(OH)D and Risk for Major Clinical Disease 1 May 2012 Annals of Internal Medicine Volume 156 • Number 9 629
  4. 4. RESULTSBaseline CharacteristicsBaseline 25-(OH)D concentration varied strongly byseason (Figure 1). It was lowest in January through March(“winter”), highest in July through September (“summer”),and intermediate in April through June (“spring”) and Oc-tober through December (“autumn”). Low season-specific25-(OH)D concentration, defined as less than the season-specific 29th percentile, was more common among womenand participants at more northerly study sites and was as-sociated with higher body mass index, hypertension, re-duced physical activity, and higher circulating concentra-tions of parathyroid hormone (Table 1).EventsMedian follow-up for the 1621 participants was 11years (interquartile range, 6 to 13 years). The compositeclinical outcome occurred in 1018 participants (63%)(Supplement 2, available at The qualify-ing event was hip fracture for 137 participants (8%), MIfor 186 participants (11%), cancer for 335 participants(21%), and death for 360 participants (22%). Qualifyingevents are tabulated by cause in Supplement 3 available of 25-(OH)D With EventsWe first evaluated the association of 25-(OH)D withthe composite outcome by using the previously published25-(OH)D threshold of 50 nmol/L (20 ng/mL) (1, 3).Using this approach, we saw borderline statistical evidenceof heterogeneity by season (P ϭ 0.057). Deviations fromstrong associations occurred in winter and summer, theextremes of seasonal variation in 25-(OH)D (Supplement4, available at next evaluated the associations of 25-(OH)D withthe composite outcome by season. Participants in the low-est 2 to 3 deciles of 25-(OH)D concentration (lowest 20%to 30%) tended to have increased risk for the compositeclinical outcome, compared with those in the highest 7 to8 deciles (Supplement 5, available at of a season-based 25-(OH)D Z score similarlysuggested that elevated risk for the composite outcome wasTable 1. Characteristics of Participants in 1992 to 1993*Characteristic Overall (n ‫؍‬ 1621) Normal 25-(OH)D (n ‫؍‬ 1126) Low 25-(OH)D (n ‫؍‬ 495)†Demographic dataAge, y 74.0 (4.6) 73.7 (4.5) 74.5 (4.7)Men 491 (30) 406 (36) 85 (17)SiteForsythe County, North Carolina 450 (28) 330 (29) 120 (24)Sacramento County, California 370 (23) 277 (25) 93 (19)Washington County, Maryland 457 (28) 295 (26) 162 (33)Pittsburgh, Pennsylvania 344 (21) 224 (20) 120 (24)Medical history and lifestyleDiabetes‡ 162 (10) 90 (8) 72 (15)Hypertension‡ 906 (56) 609 (54) 297 (60)Current smoking 151 (9) 93 (8) 58 (12)Current alcohol use 740 (46) 521 (46) 219 (44)Physical activity categoryϽ500 kcal/wk 453 (28) 261 (23) 192 (39)500–1000 kcal/wk 329 (20) 214 (19) 115 (23)1000–2000 kcal/wk 393 (24) 298 (26) 95 (19)Ͼ2000 kcal/wk 446 (28) 353 (31) 93 (19)Physical examinationBMI categoryϽ25 kg/m2664 (41) 484 (43) 180 (36)25–30 kg/m2660 (41) 466 (41) 194 (39)30–35 kg/m2229 (14) 143 (13) 86 (17)Ͼ35 kg/m268 (4) 33 (3) 35 (7)Laboratory dataEstimated GFR, mL/min per 1.73 m2‡ 75.3 (17.8) 75.9 (17.7) 74.0 (17.8)Parathyroid hormone, ng/L 54.7 (27.3) 50.9 (25.4) 63.1 (31.1)Bone alkaline phosphate, ␮g/L 14.6 (6.9) 14.1 (6.7) 15.8 (7.1)Calciummg/dL 9.5 (0.4) 9.5 (0.4) 9.5 (0.4)mmol/L 2.4 (0.1) 2.4 (0.1) 2.4 (0.1)Phosphate, mmol/L 1.2 (0.2) 1.2 (0.2) 1.2 (0.2)Total 25-(OH)Dnmol/L 66.2 (25.8) 77.8 (21.5) 39.8 (11.1)ng/mL 26.5 (10.3) 31.2 (8.6) 15.9 (4.5)25-(OH)D ϭ 25-hydroxyvitamin D; BMI ϭ body mass index; GFR ϭ glomerular filtration rate.* Values are means (SDs) for continuous variables or numbers (percentages) for categorical variables.† Defined as less than the lowest season-specific 29th percentile (43, 50, 61, and 55 nmol/L [17, 20, 24, and 22 ng/mL] in winter, spring, summer, and fall, respectively).‡ Diabetes was defined as use of insulin or oral hypoglycemic agents or fasting blood glucose level Ն6.99 mmol/L (Ն126 mg/dL). Hypertension was defined as systolic bloodpressure Ն140 mm Hg, diastolic blood pressure Ն90 mm Hg, or use of an antihypertensive medication. Serum cystatin C was measured by using a BNII nephelometer (NLatex Cystatin C; Dade Behring, Deerfield, Illinois) and used to estimate GFR with the following equation: GFR ϭ 76.7 ϫ [cystatin C]Ϫ1.18.Original Research Serum 25-(OH)D and Risk for Major Clinical Disease Events630 1 May 2012 Annals of Internal Medicine Volume 156 • Number 9
  5. 5. greatest below a Z score of approximately Ϫ0.5 (near 30%of the normal distribution) (Figure 2).Based on a simple statistical approach, the season-specific 25-(OH)D Z score that best separated low- andhigh-risk 25-(OH)D groups with respect to the compositeoutcome was Ϫ0.54 (29th percentile of the normal distri-bution). This threshold corresponded to season-specificcut-points of 43, 50, 61, and 55 nmol/L (17, 20, 24, and22 ng/mL) under the normal approximation to the distri-bution of 25-(OH)D concentration for winter, spring,summer, and autumn, respectively (mean threshold of 52nmol/L [21 ng/mL]). The observed season-specific distri-butions of 25-(OH)D concentrations in this population,although slightly right-skewed, suggested that the normalapproximation was reasonable. A 25-(OH)D concentrationbelow the season-specific 29th percentile was associatedwith a 24% increased risk for the composite outcome inthe adjusted model (95% CI, 9% to 42%) and with simi-larly increased risks for each component of the compositeoutcome (Table 2).We did several analyses to evaluate whether the iden-tified optimal threshold was robust. Statistical significanceof the association of low season-specific 25-(OH)D con-centration with risk for the composite outcomes decreasedmarkedly when a threshold Z score above Ϫ0.44 was used;this corresponds to season-specific cut-points of 45, 52, 63,and 57 nmol/L (18, 21, 24, and 23 ng/mL) for winter,spring, summer, and autumn, respectively (mean thresholdof 54 nmol/L [22 ng/mL]) (Supplement 6, available In bootstrap analyses, 95% of optimal25-(OH)D Z score thresholds fell between Ϫ1.48 and0.13 (mean season-specific thresholds of 29 and 69 nmol/L[12 and 28 ng/mL], respectively), whereas 75% fell be-tween Ϫ1.38 and Ϫ0.40 (mean thresholds of 31 and 55nmol/L [12 and 22 ng/mL], respectively).ReclassificationNine percent of participants were reclassified compar-ing low 25-(OH)D concentration defined by season-specific thresholds (29th percentile) versus the staticthreshold of 50 nmol/L (20 ng/mL) (Table 3). When lowseason-specific 25-(OH)D concentration was comparedwith a concentration less than 50 nmol/L (20 ng/mL), netreclassification improvement was 2.4% (95% CI, Ϫ0.6%to 5.3%; P ϭ 0.118).DISCUSSIONWe characterized associations of NIST-verified serum25-(OH)D concentration with risk for adverse clinicalevents that are pathophysiologically relevant to pleiotropicvitamin D actions in a community-based population. Theassociation of 25-(OH)D with a composite clinical out-come of hip fracture, MI, cancer, and death varied by sea-son, supporting use of season-specific 25-(OH)D thresh-olds. In our study population, threshold 25-(OH)Dconcentrations optimally associated with risk for the com-posite outcome were 43 nmol/L (17 ng/mL) in wintermonths, 50 nmol/L (20 ng/mL) in spring months, 61nmol/L (24 ng/mL) in summer months, and 55 nmol/L(22 ng/mL) in autumn months.Figure 2. Association of season-specific 25-(OH)D Z scorewith the risk for incident myocardial infarction, cancer,hip fracture, or death (composite outcome) among 1621participants in the Cardiovascular Health Study, evaluatedusing a penalized spline.LogHazardRatio––1.5 –0.525 50 7575502550 75 1001001007550–1 0.5 1.50 1WinterSpringSummerAutumnSeason-Specific 25-(OH)D Z Score25-(OH)D Concentration, nmol/LProportional hazards model adjusts for age, sex, clinical site, body massindex, physical activity, and smoking. The shaded area represents Z scoreless than Ϫ0.54 (29th percentile of the normal distribution), which bestdiscriminated risk for the composite outcome. The x-axis is displayed asseason-specific Z score (uppermost x-axis, reflecting the primary methodof analysis) and as corresponding season-specific absolute 25-(OH)Dconcentrations (lower 4 axes). 25-(OH)D ϭ 25-hydroxyvitamin D.Table 2. Associations of Low Season-Specific 25-(OH)DConcentration With Rates of the Composite Outcome of MI,Cancer, Hip Fracture, or Death Among ParticipantsOutcome Events (Incidence Rate), n* Hazard Ratio(95% CI)†Normal25-(OH)DLow25-(OH)D‡Composite 681 (6.4) 337 (7.7) 1.24 (1.09–1.42)MI 154 (1.2) 67 (1.3) 1.24 (0.91–1.70)Cancer 259 (2.3) 111 (2.3) 1.13 (0.90–1.42)Hip fracture 118 (0.9) 72 (1.4) 1.34 (0.97–1.84)Death 539 (4.0) 287 (5.3) 1.32 (1.14–1.53)25-(OH)D ϭ 25-hydroxyvitamin D; MI ϭ myocardial infarction.* Participants may be included in more than 1 event category, but only the firstevent for each participant was used in analysis of the composite outcome. Inci-dence rates are unadjusted event rates per 100 person-years of follow-up.† Adjusted for age, sex, clinical site, smoking, body mass index, and physicalactivity.‡ Defined as less than the lowest season-specific 29th percentile (43, 50, 61, and55 nmol/L [17, 20, 24, and 22 ng/mL] in winter, spring, summer, and autumn,respectively).Original ResearchSerum 25-(OH)D and Risk for Major Clinical Disease 1 May 2012 Annals of Internal Medicine Volume 156 • Number 9 631
  6. 6. The IOM recently evaluated the clinical application of25-(OH)D testing in the context of vitamin D supplemen-tation (1). In reviewing available data, it concluded thatinadequate vitamin D can contribute to bone disease, vita-min D supplementation can decrease risk for bone diseasein at-risk populations, and 25-(OH)D concentration lessthan 50 nmol/L (20 ng/mL) identifies persons at increasedrisk. The proposed threshold of 50 nmol/L (20 ng/mL)was lower than that of 75 nmol/L (30 ng/mL) recom-mended by many professional societies and vitamin D re-searchers (12–15). The IOM noted a lack of high-qualitydata about the effects of vitamin D supplementation onrisk for nonbone health outcomes, including MI, cancer,and death, and it did not, therefore, base its estimate oftarget 25-(OH)D concentration on these outcomes. Thesefindings were echoed in an updated clinical practice sum-mary on vitamin D deficiency (3).In comparison with existing literature and recommen-dations, we have 2 principal findings. First, 25-(OH)Dthresholds associated with risk for diverse major clinicaldisease events in our work center close to the 50 nmol/L(20 ng/mL) recommended by the IOM for bone health.We agree with the IOM’s conclusions that high-qualityintervention studies are needed to test whether vitamin Ddeficiency is causally related to nonbone outcomes in hu-mans. Until these data are available, the finding of a similar25-(OH)D threshold for risk for major clinical diseaseevents to that recommended by the IOM for bone health isreassuring and supports generally targeting 50 nmol/L (20ng/mL) over 75 nmol/L (30 ng/mL) when 25-(OH)D test-ing is clinically indicated.In our study, 30.5% of participants had a 25-(OH)Dconcentration less than the season-specific threshold cen-tered near 50 nmol/L (20 ng/mL). This proportion is con-gruent with the prevalence of 25-(OH)D concentrationsless than 50 nmol/L (20 ng/mL) in other populations andemphasizes the large number of people at risk for potentialcomplications of low 25-(OH)D concentration (34). How-ever, the distinction between 50 and 75 nmol/L (20 and30 ng/mL) is important because more than 40% of theU.S. population has concentrations between 50 and 75nmol/L (20 and 30 ng/mL) (34). Our estimate of the 25-(OH)D threshold that best discriminates risk for clinicaldisease events was generated with some statistical uncer-tainty, but a threshold as high as 75 nmol/L (30 ng/mL)was unlikely to be congruent with our data.Second, our data suggest that season-specific targetsare most appropriate for 25-(OH)D concentration. Varia-tion in 25-(OH)D concentration within persons and pop-ulations over the calendar year is well-known to be largerelative to mean concentration (16–19). This is probablydue to seasonal variation in exposure to ultraviolet light. Asa result, clinical decisions about initiation and dose of year-long vitamin D supplementation are likely to be heavilyinfluenced by time of ascertainment, which is often arbi-trary. Combined with this background knowledge, our re-sults that demonstrated heterogeneity of the 25-(OH)D–composite outcome association by season and a trendtoward improved classification of risk using season-specific25-(OH)D thresholds suggest that season-specific targetsfor 25-(OH)D concentration are more appropriate thanthe static targets previously recommended when the needfor year-long vitamin D supplementation is being consid-ered (1, 12–15).We examined a composite end point of clinical diseaseevents that plausibly reflect net pleiotropic vitamin D ac-tions, are supported by existing literature, and have a quan-tifiable time of onset, understanding that this may include1 or more outcomes that are not causally related to 25-(OH)D and may omit some important vitamin D–relatedeffects. Associations of low season-specific 25-(OH)D con-centration with the composite outcome and each of itscomponents were of similar magnitude, enabling this ap-proach. Statistical significance using the standard ␣ level of0.05 was achieved only for the composite outcome and fordeath. However, this study was not powered to detect as-sociations with individual composite outcome compo-nents, and statistical confidence is likely to be inflated bytesting inference in the same data set in which threshold25-(OH)D concentrations were derived. Moreover, theprimary goal of this study was to study the pattern of the25-(OH)D–composite outcome relationship, not its exis-tence, which has been demonstrated in previous studies(5–11).Strengths of this study include the use of acommunity-based population of older adults, who are of-ten targeted for 25-(OH)D testing; the use of NIST-verified 25-(OH)D concentration, which has not, to ourknowledge, been previously applied to large epidemiologicstudies; and the ascertainment of clinical outcomes directlyrelevant to both bone and nonbone vitamin D actions overTable 3. Classification of 10-Year Risk for CompositeOutcome of MI, Cancer, Hip Fracture, or Death AmongParticipantsStatic 25-(OH)D* Season-Specific 25-(OH)DNormal,n (%)Low,n (%)†Total,nParticipants who have an eventՆ50 nmol/L (Ն20 ng/mL) 484 (90) 55 (10) 539Ͻ50 nmol/L (Ͻ20 ng/mL) 18 (8) 199 (92) 217Total 502 254 756Participants who do not have an eventՆ50 nmol/L (Ն20 ng/mL) 598 (93) 48 (7) 646Ͻ50 nmol/L (Ͻ20 ng/mL) 26 (12) 193 (88) 219Total 624 241 86525-(OH)D ϭ 25-hydroxyvitamin D.* Year-long static threshold of 50 nmol/L (20 ng/mL).† Defined as less than the lowest season-specific 29th percentile (43, 50, 61, and55 nmol/L [17, 20, 24, and 22 ng/mL] in winter, spring, summer, and autumn,respectively).Original Research Serum 25-(OH)D and Risk for Major Clinical Disease Events632 1 May 2012 Annals of Internal Medicine Volume 156 • Number 9
  7. 7. long-term follow-up. Limitations include the inclusion ofonly older adults; the availability of only white participantsin sufficient numbers to rigorously evaluate relationships of25-(OH)D with study outcomes; the availability of onlyone 25-(OH)D measurement per participant, which maybias magnitudes of association toward the null; and theinability of available statistical methods to precisely deter-mine optimal threshold concentrations with statistical confi-dence. Most important, this study is observational. Ulti-mately, optimal 25-(OH)D concentrations should be definedas the baseline 25-(OH)D concentrations above which vita-min D supplementation does not improve relevant clinicaloutcomes in large, diverse, randomized clinical trials.In conclusion, we found that “optimal” concentrationsof 25-(OH)D, gauged by associations with major clinicaldisease events, centered near 50 nmol/L (20 ng/mL), thelevel recently recommended by the IOM for bone health.We further report that the association of 25-(OH)D withclinical health events varies by season and suggest thatseason-specific targets for 25-(OH)D concentration aremore appropriate than static targets when considering po-tential implications for long-term health.From the University of Washington, Seattle, Washington.Grant Support: By the National Heart, Lung, and Blood Institute (con-tracts N01-HC-85239, N01-HC-85079 through N01-HC-85086, N01-HC-35129, N01 HC-15103, N01 HC-55222, N01-HC-75150, and N01-HC-45133 and grant HL080295), with additional contribution from theNational Institute of Neurologic Disorders and Stroke. Additional sup-port was provided by the National Institute on Aging (AG-023629,AG-15928, AG-20098, and AG-027058); the National Heart, Lung,and Blood Institute (grants R01HL084443 and R01HL096875); andthe National Institute of Diabetes and Digestive and Kidney Diseases(grant R01DK088762). A full list of principal CHS investigators andinstitutions can be found at Conflicts of Interest: Disclosures can be viewed atϭM11-2074.Reproducible Research Statement: Study protocol and statistical code:Available from Dr. de Boer (e-mail, Dataset: Not available.Requests for Single Reprints: Ian H. de Boer, MD, MS, Box 359606,325 9th Avenue, Seattle, WA 98104; e-mail, author addresses and author contributions are available at Ross AC, Taylor CL, Yaktine AL, Del Valle HB, eds; Committee to ReviewDietary Reference Intakes for Vitamin D and Calcium; Institute of Medicine.Dietary Reference Intakes for Calcium and Vitamin D. Washington, DC: TheNational Academies Press, 2011.2. Holick MF. Vitamin D deficiency. N Engl J Med. 2007;357:266-81. [PMID:17634462]3. Rosen CJ. Clinical practice. Vitamin D insufficiency. N Engl J Med. 2011;364:248-54. [PMID: 21247315]4. Dusso AS, Brown AJ, Slatopolsky E. Vitamin D. Am J Physiol Renal Physiol.2005;289:F8-28. [PMID: 15951480]5. Cauley JA, Lacroix AZ, Wu L, Horwitz M, Danielson ME, Bauer DC, et al.Serum 25-hydroxyvitamin D concentrations and risk for hip fractures. Ann In-tern Med. 2008;149:242-50. [PMID: 18711154]6. Robinson-Cohen C, Katz R, Hoofnagle AN, Cauley JA, Furberg CD, Rob-bins JA, et al. Mineral metabolism markers and the long-term risk of hip fracture:the cardiovascular health study. J Clin Endocrinol Metab. 2011;96:2186-93.[PMID: 21508146]7. Wang TJ, Pencina MJ, Booth SL, Jacques PF, Ingelsson E, Lanier K, et al.Vitamin D deficiency and risk of cardiovascular disease. Circulation. 2008;117:503-11. [PMID: 18180395]8. de Boer IH, Kestenbaum B, Shoben AB, Michos ED, Sarnak MJ, SiscovickDS. 25-hydroxyvitamin D levels inversely associate with risk for developing cor-onary artery calcification. J Am Soc Nephrol. 2009;20:1805-12. [PMID:19443637]9. Giovannucci E, Liu Y, Hollis BW, Rimm EB. 25-hydroxyvitamin D and riskof myocardial infarction in men: a prospective study. Arch Intern Med. 2008;168:1174-80. [PMID: 18541825]10. Melamed ML, Michos ED, Post W, Astor B. 25-hydroxyvitamin D levelsand the risk of mortality in the general population. Arch Intern Med. 2008;168:1629-37. [PMID: 18695076]11. Kestenbaum B, Katz R, de Boer I, Hoofnagle A, Sarnak MJ, Shlipak MG,et al. Vitamin D, parathyroid hormone, and cardiovascular events among olderadults. J Am Coll Cardiol. 2011;58:1433-41. [PMID: 21939825]12. Dawson-Hughes B, Heaney RP, Holick MF, Lips P, Meunier PJ, Vieth R.Estimates of optimal vitamin D status [Editorial]. Osteoporos Int. 2005;16:713-6. [PMID: 15776217]13. Vieth R. What is the optimal vitamin D status for health? Prog Biophys MolBiol. 2006;92:26-32. [PMID: 16766239]14. Dawson-Hughes B, Mithal A, Bonjour JP, Boonen S, Burckhardt P,Fuleihan GE, et al. IOF position statement: vitamin D recommendations forolder adults. Osteoporos Int. 2010;21:1151-4. [PMID: 20422154]15. Holick MF, Binkley NC, Bischoff-Ferrari HA, Gordon CM, Hanley DA,Heaney RP, et al; Endocrine Society. Evaluation, treatment, and prevention ofvitamin D deficiency: an Endocrine Society clinical practice guideline. J ClinEndocrinol Metab. 2011;96:1911-30. [PMID: 21646368]16. Bolland MJ, Grey AB, Ames RW, Mason BH, Horne AM, Gamble GD, et al.The effects of seasonal variation of 25-hydroxyvitamin D and fat mass on a diagnosisof vitamin D sufficiency. Am J Clin Nutr. 2007;86:959-64. [PMID: 17921371]17. Sherman SS, Hollis BW, Tobin JD. Vitamin D status and related parame-ters in a healthy population: the effects of age, sex, and season. J Clin EndocrinolMetab. 1990;71:405-13. [PMID: 2380336]18. Jorde R, Sneve M, Hutchinson M, Emaus N, Figenschau Y, Grimnes G.Tracking of serum 25-hydroxyvitamin D levels during 14 years in a population-based study and during 12 months in an intervention study. Am J Epidemiol.2010;171:903-8. [PMID: 20219763]19. Shoben AB, Kestenbaum B, Levin G, Hoofnagle AN, Psaty BM, Siscovick DS,et al. Seasonal variation in 25-hydroxyvitamin D concentrations in the cardiovascularhealth study. Am J Epidemiol. 2011;174:1363-72. [PMID: 22112344]20. Phinney KW. Development of a standard reference material for vitamin D inserum. Am J Clin Nutr. 2008;88:511S-512S. [PMID: 18689392]21. Fried LP, Borhani NO, Enright P, Furberg CD, Gardin JM, Kronmal RA,et al. The Cardiovascular Health Study: design and rationale. Ann Epidemiol.1991;1:263-76. [PMID: 1669507]22. Agborsangaya C, Toriola AT, Grankvist K, Surcel HM, Holl K, Parkkila S,et al. The effects of storage time and sampling season on the stability of serum25-hydroxy vitamin D and androstenedione. Nutr Cancer. 2010;62:51-7.[PMID: 20043259]23. Ives DG, Fitzpatrick AL, Bild DE, Psaty BM, Kuller LH, Crowley PM,et al. Surveillance and ascertainment of cardiovascular events. The CardiovascularHealth Study. Ann Epidemiol. 1995;5:278-85. [PMID: 8520709]24. Chen C, Lewis SK, Voigt L, Fitzpatrick A, Plymate SR, Weiss NS. Prostatecarcinoma incidence in relation to prediagnostic circulating levels of insulin-likegrowth factor I, insulin-like growth factor binding protein 3, and insulin. Cancer.2005;103:76-84. [PMID: 15540247]25. Taylor HL, Jacobs DR Jr, Schucker B, Knudsen J, Leon AS, Debacker G.A questionnaire for the assessment of leisure time physical activities. J ChronicDis. 1978;31:741-55. [PMID: 748370]Original ResearchSerum 25-(OH)D and Risk for Major Clinical Disease 1 May 2012 Annals of Internal Medicine Volume 156 • Number 9 633
  8. 8. 26. Robinson-Cohen C, Katz R, Mozaffarian D, Dalrymple LS, de Boer I,Sarnak M, et al. Physical activity and rapid decline in kidney function amongolder adults. Arch Intern Med. 2009;169:2116-23. [PMID: 20008696]27. Wang Y, Jacobs EJ, McCullough ML, Rodriguez C, Thun MJ, Calle EE,et al. Comparing methods for accounting for seasonal variability in a biomarkerwhen only a single sample is available: insights from simulations based on serum25-hydroxyvitamin d. Am J Epidemiol. 2009;170:88-94. [PMID: 19406919]28. Dobnig H, Pilz S, Scharnagl H, Renner W, Seelhorst U, Wellnitz B, et al.Independent association of low serum 25-hydroxyvitamin d and 1,25-dihydroxyvitamin d levels with all-cause and cardiovascular mortality. Arch InternMed. 2008;168:1340-9. [PMID: 18574092]29. Eilers PHC, Marx BD. Flexible smoothing with B-splines and penalties. StatSci. 1996;11:89-121.30. Therneau T. A package for survival analysis in R. R package version 2.36-0.Accessed at on 1 July2011.31. Contal C, O’Quigley J. An application of changepoint methods in studying theeffect of age on survival in breast cancer. Comput Stat Data Anal. 1999;30:253-70.32. Efron B, Tibshirani R. An Introduction to the Bootstrap. New York: Chap-man & Hall, 1994.33. Pencina MJ, D’Agostino RB Sr, D’Agostino RB Jr, Vasan RS. Evaluatingthe added predictive ability of a new marker: from area under the ROC curve toreclassification and beyond. Stat Med. 2008;27:157-72. [PMID: 17569110]34. Looker AC, Pfeiffer CM, Lacher DA, Schleicher RL, Picciano MF, YetleyEA. Serum 25-hydroxyvitamin D status of the US population: 1988-1994 com-pared with 2000-2004. Am J Clin Nutr. 2008;88:1519-27. [PMID: 19064511]Original Research Serum 25-(OH)D and Risk for Major Clinical Disease Events634 1 May 2012 Annals of Internal Medicine Volume 156 • Number 9
  9. 9. Current Author Addresses: Drs. de Boer and Kestenbaum and Ms.Robinson-Cohen: Kidney Research Institute, Box 359606, 325 9th Av-enue, Seattle, WA 98104.Mr. Levin: Department of Biostatistics, Box 357232, 1959 NortheastPacific Street, Seattle, WA 98195.Dr. Biggs: Collaborative Health Studies Coordinating Center, Building29, Suite 310, 6200 Northeast 74th Street, Seattle, WA 98115.Dr. Hoofnagle: Department of Laboratory Medicine, Box 357110, 1959Northeast Pacific Street, Seattle, WA 98195.Dr. Siscovick: Cardiovascular Health Research Unit, 1730 Minor Ave-nue, Suite 1360, Seattle, WA 98101.Author Contributions: Conception and design: I.H. de Boer, G. Levin,B. Kestenbaum.Analysis and interpretation of the data: I.H. de Boer, G. Levin,C. Robinson-Cohen, A.N. Hoofnagle, D.S. Siscovick, B. Kestenbaum.Drafting of the article: I.H. de Boer, G. Levin, C. Robinson-Cohen.Critical revision of the article for important intellectual content:G. Levin, C. Robinson-Cohen, M.L. Biggs, A.N. Hoofnagle,D.S. Siscovick, B. Kestenbaum.Final approval of the article: I.H. de Boer, C. Robinson-Cohen,M.L. Biggs, A.N. Hoofnagle, D.S. Siscovick, B. Kestenbaum.Statistical expertise: G. Levin, C. Robinson-Cohen, A.N. Hoofnagle.Obtaining of funding: I.H. de Boer, B. Kestenbaum.Collection and assembly of data: G. Levin, M.L. Biggs, A.N. Hoofnagle,D.S. Siscovick.Annals of Internal MedicineW-222 1 May 2012 Annals of Internal Medicine Volume 156 • Number 9