ORIGINAL ARTICLE JBMR

Serum 25-Hydroxyvitamin D and Risk of Major

Osteoporotic Fractures in Older U.S. Adults*
Anne C Looker
National Center for Health Statistics, Centers for Disease Control and Prevention, Hyattsville, MD, USA

ABSTRACT
Results from previous prospective studies linking serum 25-hydroxyvitamin D (25OHD) with fracture risk have been inconsistent. The
present study examined the relationship between serum 25OHD and risk of incident major osteoporotic fracture (hip, spine, radius, and
humerus) in older U.S. adults. The study used a pooled cohort of 4749 men and women ages 65 years and older from the third National
Health and Nutrition Examination Survey (NHANES III, 1988–1994) and NHANES 2000–2004. Incident fractures were identified using
linked mortality and Medicare records that were obtained for participants from both surveys. Serum 25OHD values were measured by
radioimmunoassay in both surveys. Cox proportional hazards models were used to estimate the relative risk (RR) of fracture by serum
25OHD level. There were 525 incident major osteoporotic fractures (287 hip fractures) in the sample. Serum 25OHD was a significant
linear predictor of major osteoporotic fracture and significant quadratic predictor of hip fracture in the total sample and among those
with less than 10 years of follow-up, but it was not related to risk of either fracture type among those with
10 years of follow-up. Major
osteoporotic fracture risk was increased by 26% to 27% for each SD decrease in serum 25OHD among those with less than 10 years of
follow-up. Serum 25OHD was significantly related to risk of major osteoporotic fractures as a group and to hip fracture alone in this
cohort of older U.S. adults from NHANES III and NHANES 2000–2004. However, the predictive utility of serum 25OHD diminished after 10
years. In addition, the relationship appeared to be linear when major osteoporotic fracture risk was considered but quadratic when hip
fracture risk was assessed. ß 2013 American Society for Bone and Mineral Research

KEY WORDS: EPIDEMIOLOGY; POPULATION STUDY; VITAMIN D; OSTEOPOROSIS; FRACTURE

Introduction Nutrition Examination Survey (NHANES III).(10) That study was

limited to hip fracture risk in whites and analyses were performed

V itamin D is known to be important for skeletal health, but

several aspects of its relationship with fracture risk remain
unclear. Several prospective studies have examined the
relationship between fracture risk and serum 25-hydroxyvitamin
D (25OHD), the preferred indicator of vitamin D.(1–16) Slightly
more than one-half of the studies published to date have
reported a significant relationship between 25OHD and fracture
risk for at least one of the skeletal sites studied,(2–10,16) but several
others have found no significant relationship.(3,11–15) These
studies have varied in terms of the skeletal sites and race/ethnic
groups considered, and most have been limited to a single
sex.(2–4,6,8,9,12–14) Only one has reported results separately by
race/ethnicity.(4)
A significant relationship between serum 25OHD and hip
fracture was observed in a previous analysis of linked Medicare and
mortality data from respondents in the third National Health and
for both sexes combined. Since completion of that study, an
additional 7 years of follow-up of respondents from NHANES III
have been obtained and from follow-up of respondents from
NHANES 1999–2004 has been conducted. The present study
expands the NHANES analysis by pooling all available mortality and
Medicare-linked data for these two NHANES surveys; similar
pooling of data for cohorts based on NHANES or other nationally
representative health surveys has been successfully employed in
other studies.(17,18) In addition, the present study expands the
fracture outcomes to include major osteoporotic fracture, as
defined by the fracture risk assessment tool (FRAX) model (eg, hip,
medically-diagnosed vertebral, humerus, or radius fracture).(19) The
relationship between serum 25OHD and major osteoporotic
fracture as a group and for hip fracture alone is examined by age,
sex, and race/ethnicity, and also by recently recommended
thresholds for serum 25OHD.(1,20)

Received in original form August 31, 2012; revised form October 26, 2012; accepted November 12, 2012. Accepted manuscript online November 26, 2012.
Address correspondence to: Anne C Looker, PhD, Room 4310, National Center for Health Statistics, 3311 Toledo Rd, Hyattsville, MD 20782, USA.
E-mail: alooker@cdc.gov
Additional Supporting Information may be found in the online version of this article.
This article is a US Government work and, as such, is in the public domain in the United States of America.
*The findings and conclusions in this report are those of the author and do not necessarily represent the views of the Centers for Disease Control and Prevention.
Journal of Bone and Mineral Research, Vol. 28, No. 5, May 2013, pp 997–1006
DOI: 10.1002/jbmr.1828
ß 2013 American Society for Bone and Mineral Research

997

Materials and Methods
Sample
The baseline data for this study came from NHANES III and
NHANES 2000–2004, which were conducted by the National
Center for Health Statistics (NCHS), Centers for Disease Control
and Prevention, to assess the health and nutritional status of a
large representative sample of the non-institutionalized, civilian
population of the United States. In NHANES III, a nationally
representative sample was obtained in two 3-year cycles
between 1988 and 1994. The NHANES became a continuous
survey starting in 1999, and from that date data were collected
from a nationally representative sample every year. Although a
representative sample is collected annually, data are released in
2-year periods starting with 1999 to protect confidentiality and
increase statistical reliability. The present study is based on data
from the continuous NHANES for the years 2000–2004 because
serum 25OHD data were not collected in 1999. All procedures in
each NHANES were approved by the NCHS Institutional Review
Board, and written informed consent was obtained from all
subjects.(21,22)
In each NHANES, data were collected via household interviews
and direct standardized physical examinations conducted in
specially equipped mobile examination centers.(21,22)
NHANES III and NHANES 2000–2004 were designed to provide
reliable estimates for three race/ethnic groups: non-Hispanic
whites (NHW), non-Hispanic blacks (NHB), and Mexican Amer-
icans (MA). Race and ethnicity were self-reported in both surveys.
Both surveys were linked with data from the mortality files
created for each survey by NCHS and with Medicare enrollment
and claims records in order to provide a longitudinal component
to the survey. Vital status of study participants through 2006 was
determined from the NHANES III and NHANES 2000–2004 Linked
Mortality Files.(23) These files contain mortality follow-up data
based on a probabilistic match between the eligible NHANES III
and NHANES 2000–2004 samples and the National Death Index
(NDI). Vital status data are available for individuals from the date
of their participation in the NHANES III and NHANES 2000–2004
surveys through December 31, 2006. In addition to data on vital
status, information on underlying and multiple causes of death
are available in this file. Vital status for the year 2007 was based
on status from the Medicare Denominator file.(24)
Medicare enrollment and utilization data were available for
NHANES respondents who agreed to provide personal identifi-
cation.(24) Of those who were age 65 years and older at the time
of the Medicare linkage, 97% (n ¼ 8024) of the 8303 eligible
respondents from NHANES III and 98% (n ¼ 4103) of the 4189
eligible respondents from NHANES 2000–2004 were successfully
validated and matched with Medicare administrative records.
Medicare claims data were provided from respondents who
participated in fee-for-service care from 1991 through 2007 for
NHANES III and for 1999 through 2007 for NHANES 2000–2004.
Claims data were not available for those who enrolled in
Medicare managed care plans. In the present study, claims data
from the following files were used: Medicare Provider Analysis
and Review (MEDPAR) Inpatient Hospital Stay File/Skilled Nursing
Facility (SNF) File, Carrier Standard Analytic File, and Outpatient
Standard Analytic File. The Medicare Denominator files were
998 LOOKER
15234681, 2013, 5, Downloaded from https://asbmr.onlinelibrary.wiley.com/doi/10.1002/jbmr.1828 by SEA ORCHID (Thailand), Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
used to restrict the analytic sample to those who were
successfully matched to the Medicare claims data, to obtain
information about enrollment in managed care plans and to
provide vital status for the year 2007.
The analytic sample in this study was restricted to individuals
ages 65 years and older at the time of their NHANES interview at
baseline, because Medicare provides comprehensive health care
for roughly 98% of the U.S. population in this age range.(25) In
the Supplementary Appendix 1, Table A shows the number of
persons excluded from the analytic sample and reason for
exclusion for each survey. After excluding a total of 1146
individuals, 2946 (71%) of the original 4092 eligible interviewed
and examined individuals from NHANES III were included in the
final analytic sample. After excluding a total of 1562 individuals,
1803 (53%) of the original 3365 eligible interviewed and
examined individuals from NHANES 2000–2004 were included in
the final analytic sample. Approximately 16% of the eligible
interviewed and examined sample from both surveys was
excluded as a result of prior fracture at baseline. A relatively
large number of respondents in NHANES 2000–2004 also had to
be excluded because they were either ineligible for linkage to the
Medicare denominator file or were enrolled in a health
maintenance organization (HMO) at the time of their baseline
examination. Descriptive characteristics and risk factors were
compared between respondents who were retained versus
excluded from the analytic sample in order to assess the
potential for nonresponse bias in our results. The excluded
respondents were older, more likely to be women, have blood
drawn between November and April, have shorter height, and
reported smoking, having poorer health and less activity than
respondents who were included. Serum 25OHD levels were
slightly, albeit not significantly, lower in excluded respondents.
Fracture case identification
Major osteoporotic fracture cases were identified using an
approach based on methods developed by Ray and collea-
gues,(26) Taylor and colleagues,(27) and Curtis and colleagues.(28)
Cases were defined using relevant International Classification of
Disease (ICD), Healthcare Common Procedure Coding System
(HCPCS), or Current Procedural Terminology (CPT) codes for
the years 1991–2007.(29,30) Cases identified from the MedPAR
Hospital Inpatient/Skilled Nursing Facility files were based on
ICD-9 diagnosis codes (primary diagnosis only for spine; any of 10
discharge diagnoses for hip, radius, or humerus). Cases identified
from the Carrier file were based on ICD-9 diagnosis codes
(principal diagnosis only for spine; any of five diagnoses for hip,
radius, or humerus) and a concurrent relevant HCPCS/CPT
procedure code. Cases identified from the Outpatient File were
based on ICD diagnosis codes (principal only for spine; any of
11 diagnoses for hip, radius, or humerus) or ICD-9 surgical
procedure code (hip, radius, humerus) and a concurrent relevant
HCPCS/CPT code for any of 45 procedure variables. Respondents
with codes indicating care of previous fracture or other bone
diseases, neoplasm, or hip arthroplasty for arthritis were
excluded from the analyses. Details regarding the definition of
cases from Medicare records, including specific codes, are
provided in Supplementary Appendix 2.
Journal of Bone and Mineral Research

Cause of death information from the NHANES III and NHANES
2000-2004 Linked Mortality Files were also used to identify hip
fracture. Persons with an ICD-9 code 820 or ICD-10 code S72.0–
S72.2 listed as an underlying or multiple cause of death on their
death certificate were also considered to be hip fracture cases
(n ¼ 47 from NHANES III; n ¼ 8 from NHANES 2000–2004).
A total of 525 major osteoporotic fracture cases were
identified in the analytic sample used in the present study.
These cases consisted of 287 hip fractures (55%), 105 radius
fractures (20%), 83 medically diagnosed (clinical) spine fractures
(16%), and 50 humerus fractures (10%). This pattern of fractures
in adults over age 65 years is similar to that seen in other studies
of fracture based on Medicare data from older adults.(27,31) The
majority of fracture cases (n ¼ 328, or 62% of major osteoporotic
cases; n ¼ 221, or 77% of hip fracture cases) in the analytic
sample had diagnoses consistent with fracture on records from
more than one source (eg, inpatient hospital/SNF, carrier,
outpatient, and/or death records). Of the 274 hip fracture cases
that were identified via inpatient hospital records, 89% also had
relevant surgical codes for hip fracture.
Variables
Serum 25OHD was assayed with a radioimmunoassay (RIA) kit
(DiaSorin, Stillwater MN, USA) at the National Center for
Environmental Health (CDC, Atlanta, GA, USA) in both surveys.
Based on quality control pools that passed specification limits,
the interassay coefficient of variation (CV) was 15% to 25% for
lower values (20–62 nmol/L) and 14% to 18% for higher values
(86–143 nmol/L) during NHANES III, and 8.3% to 11% for lower
values (24–58 nmol/L) and 10% for higher values (102–112 nmol/
L) during NHANES 2000–2004. Long-term performance of this
laboratory in the international Vitamin D External Quality
Assessment Scheme (DEQAS) proficiency testing program was
excellent. Details of the assay methods have been published
elsewhere.(32,33) A previous study indicated that serum 25OHD
values from NHANES III and NHANES 2000–2004 were not
directly comparable even though the same assay had been
used, so serum 25OHD from NHANES III was adjusted to be
comparable to serum 25OHD from NHANES 2000–2004 using a
previously published equation.(34) Serum 25OHD was examined
both as a continuous and categorical variable using thresholds
that were recently recommended by the Institute of Medicine
(IOM)(1) and Endocrine Society.(20)
Several variables were used to make exclusions from the
sample or were evaluated for use as confounders in multivariate
models. Only variables that were measured comparably in the
two surveys were used. For example, because bone mineral
density was not measured at the same skeletal site in the two
surveys, it could not be included as a co-variable. Variables that
were measured during the physical examination included body
weight and height, which were used to calculate body mass
index (BMI, equal to body weight in kilograms, divided by height
in meters squared).(20) Variables obtained by interview at
baseline included age, self-reported race-ethnicity, calcium from
food (mg/d) from a 24-hour recall, self-reported hip, wrist, or
spine fracture, smoking status (current, former, never), alcohol
consumption (drinks per day), self-reported physical activity level
Journal of Bone and Mineral Research
15234681, 2013, 5, Downloaded from https://asbmr.onlinelibrary.wiley.com/doi/10.1002/jbmr.1828 by SEA ORCHID (Thailand), Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
compared to others of the same age and sex or compared to
the respondent’s activity level 10 years prior to baseline (same,
higher, lower), self-rated health status (excellent/very good/good
versus fair/poor), and maternal history of hip fracture. Because
NHANES examinations are performed in southern latitudes
during the winter and northern latitudes during the summer,
season of blood collection (November–April versus May–
October) was also evaluated as a potential confounder.
Finally, use of prescription drugs to treat osteoporosis was
evaluated. In both surveys, respondents showed the containers
for all current prescription medications to the interviewer,
who recorded the name of the product. The December 2007
Multum Lexicon Drug Database (Cerner Multum Inc, Denver, CO,
USA; http://www.multum.com/Lexicon.htm) was used to assign
generic drug names and codes in both surveys. These
medications included sex hormones (estrogen and testoster-
one), bisphosphonates, selective estrogen receptor modifiers
(SERMs), teriparatide, calcitonin, calcitriol, ergocalciferol, sodium
fluoride, calcium acetate, tibolone, and strontium ranelate.
Statistical analysis
All analyses were performed using SAS 9.2 (SAS Institute, Cary,
NC, USA) and SUDAAN software(35) for analysis of data from
complex sample surveys. Descriptive characteristics and risk
factors at baseline were compared between fracture cases and
non-cases using linear regression models and chi-square
analyses. To control for all risk factors simultaneously and to
account for unequal length of follow-up, Cox proportional
hazards models were used to model time to event and to
calculate estimates of the relative risk (RR) of major osteoporotic
fractures as a group and hip fracture alone by serum 25OHD. For
cases, length of follow-up was calculated as the time from date of
examination to date of diagnosis or procedure for the relevant
fractures identified by Medicare records or date of death for hip
fractures identified by death certificates. For non-cases, follow-
up time was calculated as time from baseline exam to date of
death for decedents, date of entry into managed care for those
who enrolled in a Medicare managed care program after their
baseline examination, or end of follow-up on December 31, 2007
for those who did not fall into the first two categories.
Analyses to assess the validity of pooling results for NHANES III
and NHANES 2000–2004 supported the combination of the
analytic sample from the two surveys. Specifically, after adjusting
for age, sex, and race/ethnicity, the RR per SD decline
(1 SD ¼ 21.1 nmol/L) in serum 25OHD did not differ significantly
between surveys for major osteoporotic fracture (RR ¼ 1.2 and
1.3 for NHANES III and NHANES 2000–2004, respectively;
p > 0.05) or hip fracture alone (RR ¼ 1.8 and 1.2 for NHANES III
and NHANES 2000–2004, respectively; p > 0.05). In addition, the
survey  serum 25OHD interaction term included in the Cox
model applied to the pooled sample from the two surveys was
not statistically significant for either major osteoporotic fracture
(p ¼ 0.55) or hip fracture (p ¼ 0.10), which indicates that the
relationship between serum 25OHD and fracture risk did not
differ in the two surveys.
The proportional hazard assumption was tested by examining
the consistency of the RR for major osteoporotic fracture and hip
SERUM 25 OHD AND MAJOR OSTEOPOROTIC FRACTURE RISK 999

fracture across follow-up time for the pooled sample and
separately by survey, because the two surveys differed in the
amount of follow-up time. The risk of major osteoporotic fracture
as a group and hip fracture alone differed significantly with
increasing follow-up time in the pooled sample when based on
the entire follow-up period (p < 0.001) but not when follow-up
time was divided into <10 years versus
10 years. Results were
similar when analyzed separately by survey. As a result, results for
major osteoporotic fracture and hip fracture were analyzed by
these follow-up time categories as well as for the combined
follow-up period.
The best fitting model to describe the relationship between
serum 25OHD and fracture risk was assessed by using an
approach suggested by Korn and Graubard,(36) in which a series
of Cox models were calculated that successively included linear,
quadratic, and cubic terms for serum 25OHD in addition to
adjusting for age, sex, race/ethnicity, and survey period. The best
model was chosen based on the statistical significance of the
linear, quadratic, or cubic term for serum 25OHD. The predicted
relative risk per 1 nmol/L change in serum 25OHD was calculated
based on the best-fitting model and a smoothed plot was
created using locally weighted regression (PROC LOESS) after
truncating the serum distribution at 135 nmol/L in order to avoid
extrapolation of fracture risk for serum 25OHD in a range where
no cases were observed. Truncating the serum 25OHD
distribution at 135 nmol/L for this graph resulted in excluding
seven individuals, or less than 0.5% of the study population.
Cox regression models were also used to explore the
relationship between serum 25OHD and risk of major osteopo-
rotic fracture and hip fracture in more detail after grouping
serum 25OHD into categories that reflected recently recom-
mended thresholds (eg, 30 nmol/L, 50 nmol/L, 75 nmol/L). RR
estimates were calculated after adjusting for age, sex, race/
ethnicity, and survey period. Finally, Cox models were used to
compare the relationship between serum 25OHD and major
osteoporosis fracture risk by age (65–79 years versus 80þ years),
sex, and race/ethnicity. The analyses by age, sex, and race/
ethnicity were limited to major osteoporotic fracture because
there were insufficient hip fracture cases in some of the
subgroups to permit calculation of reliable statistical estimates.
Secondary analyses were performed to assess the potential
impact of lack of Medicare data prior to 1991 by estimating the
number of major osteoporotic fracture cases that might have
occurred among adults ages 65 years and older in the NHANES III
sample examined in 1988 to 1990 using age-adjusted mortality
rates for the U.S. population for these years and age-adjusted
incidence of these fractures from Medicare in the late
1980s.(31,37,38)
Results
Baseline characteristics of major osteoporotic and hip fracture
cases versus non-cases are compared for the pooled analytic
sample in Table 1. Results were similar when fractures were
grouped as major osteoporotic fracture status and when hip
fracture status was considered alone. Fracture cases were older,
weighed less, were shorter, and had a lower BMI than non-cases.
1000 LOOKER
15234681, 2013, 5, Downloaded from https://asbmr.onlinelibrary.wiley.com/doi/10.1002/jbmr.1828 by SEA ORCHID (Thailand), Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Hip fracture cases consumed less calcium than non-cases but
calcium intake did not differ between major osteoporotic cases
and non-cases. Conversely serum 25OHD levels were signifi-
cantly lower among major osteoporotic cases than non-cases,
but did not differ by hip fracture status. Both major osteoporotic
and hip fracture were more common in women and in non-
Hispanic whites. Both major osteoporotic and hip fracture cases
were less likely to report using prescription osteoporosis drugs
but more likely to report having excellent, very good, or good
health status than non-cases. Respondents with major osteopo-
rotic fracture were also more likely to smoke and to be less
physically active than non-cases, whereas hip fracture cases were
more likely to drink milk than non-cases. None of the other
variables tested differed significantly by fracture status at
baseline. Mean follow-up time was 7 years (range, 0.01–19.1
years). Mean age at fracture was 81.7 years for major
osteoporotic fracture and 83.3 years for hip fracture.
Results of the analysis to assess whether serum 25OHD was
related to fracture risk in a linear or nonlinear fashion are shown
in Table 2 for the entire follow-up period and by follow-up time
category (<10 years versus
10 years). Serum 25OHD was
significantly related to both major osteoporotic and hip fracture
in the total sample and in those with <10 years of follow-up, but
not in those with 10 years of follow-up or more. The relationship
between serum 25OHD and risk of major osteoporotic fracture
was linear (since neither the quadratic or cubic regression
coefficients were significant), whereas the relationship between
serum 25OHD and hip fracture risk appeared to be quadratic
(because the linear and quadratic coefficients were significant
but the cubic coefficient was not). These relationships are
illustrated graphically in Fig. 1, which shows the smoothed risk of
fracture by serum 25OHD truncated at 135 nmol/L as predicted
by these models for respondents with less than 10 years of
follow-up after adjusting for age, sex, race/ethnicity, and survey.
Results in Fig. 1 for hip fracture suggest a possible small increase
in risk for serum 25OHD values
110 nmol/L, but these results
are based on very sparse data: only 1.8% of the total sample,
including five hip fracture cases, had serum 25OHD values
between 110 and 135 nmol/L.
Additional detail about the relationship between serum
25OHD and risk of major osteoporotic fracture or hip fracture
was analyzed using categories based on recently recommended
thresholds by the IOM(1) and Endocrine Society(39) from
respondents with <10 years of follow-up. After adjusting for
age, sex, race/ethnicity, and survey, fracture risk was significantly
increased among respondents with serum 25OHD <30 nmol/L
(major osteoporotic fracture RR ¼ 2.09, 95% confidence interval
[CI] ¼ 1.32–3.32; hip fracture RR ¼ 2.63, 95% CI ¼ 1.60–4.32)
when compared to respondents with serum 25OHD

30 nmol/L. Use of finer categories of serum 25OHD for values
>30 nmol/L (as shown in Fig. 2) resulted in a significant
downward linear trend overall (p < 0.009 for major osteoporotic
fracture; p < 0.002 for hip fracture). Risk was significantly
elevated among those with serum 25OHD <30 nmol/L for both
fracture outcomes (eg, the 95% CIs for the RR did not include
1.00). However, risk for either fracture outcome among those
with serum 25OHD in the two categories between 30 and
74 nmol/L did not differ from that seen in those with serum
Journal of Bone and Mineral Research
 15234681, 2013, 5, Downloaded from https://asbmr.onlinelibrary.wiley.com/doi/10.1002/jbmr.1828 by SEA ORCHID (Thailand), Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 1. Pooled NHANES III and NHANES 2000–2004 Cohort Aged 65 Years and Older: Selected Baseline Characteristics of Analytic
Sample by Fracture Status

Major osteoporotic fracture Hip fracture

Yes No Yes No

n Mean or % n Mean or % p n Mean or % n Mean or % p

Age (years) 525 75.2 4223 73.3 <0.001 287 75.9 4463 73.4 <0.001
Body mass index (kg/m2) 509 26.3 4111 27.4 0.001 276 25.5 4344 27.3 <0.001
Body weight (kg) 518 69.6 4158 75.0 <0.001 283 68.4 4393 74.8 <0.001
Height (cm) 509 162.5 4130 165.3 <0.001 276 163.4 4363 165.1 0.03
Dietary calcium from food (mg) 494 710.3 4026 738.0 0.24 269 671.6 4251 738.6 0.05
Serum 25-hydroxy vitamin D (nmol/L) 525 57.5 4223 60.1 0.02 287 57.6 4461 60.0 0.12
Sex <0.001 0.002
Men 173 25.7 2259 46.9 120 34.4 2312 45.0
Women 352 74.3 1964 53.1 167 65.6 2149 55.0
Race/ethnicity <0.001 0.001
Non-Hispanic white 407 92.5 2581 84.5 232 91.7 2756 85.0
Non-Hispanic black 45 3.7 753 8.1 27 4.2 771 7.8
Mexican American 64 1.7 746 2.6 22 1.1a 788 2.6
Other 9 2.1a 143 4.9 6 b
146 4.7
Season of blood draw 0.11 0.43
November–April 200 28.6 1924 33.9 105 29.7 2019 33.6
May–October 325 71.4 2299 66.1 182 70.3 2442 66.4
Alcohol intake
3 drinks per day 0.85 0.83
Yes 30 6.4 297 6.6 16 6.2a 311 6.6
No 449 93.6 3429 93.4 244 93.8 3635 93.4
Prescription osteoporosis drug use
Yes 26 5.0 267 9.6 0.002 10 3.4a 283 9.4 0.005
No 499 95.0 3956 90.4 277 96.9 4178 90.6
Mother had hip fracture 0.92 0.47
Yes 44 10.8 344 10.5 26 12.9 362 10.4
No 459 89.2 3700 89.5 245 87.1 3914 89.6
Drink milk 0.30 0.01
Yes 432 83.1 3365 80.8 244 86.8 3553 80.7
No 93 16.9 844 19.2 43 13.2 894 19.3
Health status 0.005 0.003
Excellent/very good 313 62.6 2232 55.5 183 65.0 2362 55.8
Good 170 29.7 1451 31.4 86 29.0 1535 31.3
Fair/poor 41 7.8 534 13.1 18 6.0a 557 12.9
Smoking status 0.02 0.28
Current 56 13.0 490 10.5 33 13.7 513 10.6
Former 185 33.6 1706 41.6 102 35.9 1789 41.0
Never 284 53.4 2024 47.9 152 50.4 2156 48.4
Physical activity compared to others 0.12 0.09
More 233 45.4 1870 48.1 124 42.9 1979 48.1
Less 67 11.1 644 13.7 30 10.8 681 13.6
About the same 208 43.5 1594 38.1 124 46.3 1678 38.3
Physical activity compared to 10 years ago 0.02 0.76
More 25 4.6 265 6.3 17 6.8a 273 6.1
Less 351 63.6 2933 68.6 194 65.2 3090 68.2
About the same 148 31.8 1020 25.1 75 28.1 1093 25.7
NHANES ¼ National Health and Nutrition Examination Survey.
a
May be statistically unreliable, relative standard error (SE/percent) is 30% to 39%.
b
Not shown, relative standard error (SE/percent) is
40%.

Journal of Bone and Mineral Research SERUM 25 OHD AND MAJOR OSTEOPOROTIC FRACTURE RISK 1001
 15234681, 2013, 5, Downloaded from https://asbmr.onlinelibrary.wiley.com/doi/10.1002/jbmr.1828 by SEA ORCHID (Thailand), Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 2. Pooled NHANES III and NHANES 2000–2004 Cohort Aged 65 Years and Older: Cox Proportional Hazard Model Regression
Coefficients for Different Models of the Relationship Between Serum 25OHD and Fracture Risk

All follow-up years <10 years follow-up
10 years follow-up

Fracture type Beta p Beta p Beta p

Major osteoporotic fracture cases (n) 525 418 107
Model
Linear (serum 25OHD) 0.0082 0.004 0.0101 0.001 0.0004 0.97
Quadratic (serum 25OHD2) 0.000027 0.69 0.000018 0.82 0.000180 0.25
Cubic (serum 25OHD3) 0.000004 0.11 0.00001 0.33 0.000003 0.18
Hip fracture cases (n) 287 220 67
Model
Linear (serum 25OHD) 0.0095 0.03 0.0175 <0.001 0.0073 0.36
Quadratic (serum 25OHD2) 0.000190 0.01 0.000174 0.05 0.000053 0.71
Cubic (serum 25OHD3) 0.000003 0.29 0.000002 0.30 0.000002 0.45
All models were adjusted for age, sex, race, and survey. The quadratic model included serum 25OHD; the cubic model included serum 25OHD and serum
25OHD2.
NHANES ¼ National Health and Nutrition Examination Survey.

25OHD
75 nmol/L, because the RRs for these categories had
95% CIs that included 1.00.
Differences in the relationship between serum 25OHD and risk
of major osteoporotic fracture by age group, sex, and race/
ethnicity among those with <10 years of follow-up are presented
in Table 3. Overall, major osteoporotic fracture risk was increased
by 26% to 27% for each SD decrease in serum 25OHD. These RRs
are only slightly lower than those observed for other important
risk factors, such as age (RR ¼ 1.5 per 5 years), BMI (RR ¼ 1.4 per
SD decrease), or femur neck BMD when predicting all fractures
(RR ¼ 1.6 per SD decrease).(19,40) The RR point estimates varied by
age (RR80þy > RR65-79y), sex (RRmen > RRwomen), and race/ethnicity
(RRNHB < RRNHW < RRMA), but when tests for interactions be-
tween these demographic variables and serum 25OHD were
calculated, only the sex  serum 25OHD interaction was
significant (p < 0.05).
Results of secondary analyses to estimate the number of
major osteoporotic fractures that were potentially classified as
non-cases due to lack of Medicare records for NHANES III
respondents prior to 1991 indicated that 38 major osteoporotic
fracture (18 hip, 12 radius, 6 humerus, and 2 spine fractures) may
have occurred in the analytic sample prior to 1991.

Fig. 1. Smoothed relative risk of major osteoporotic fracture or hip
fracture by serum 25OHD value among persons with less than 10 years
of follow-up, adjusted for age, sex, race/ethnicity, and survey
Fig. 2. Relative risk of major osteoporotic fracture (A) or hip fracture (B)
by serum 25OHD category among persons with less than 10 years of
follow-up, adjusted for age, sex, race/ethnicity, and survey. Note: the
reference category for hip fracture analyses was defined as serum 25OHD
between 75-110 nmol/L in light of the possible increase in hip fracture
risk above this serum 25OHD value.

1002 LOOKER Journal of Bone and Mineral Research

 15234681, 2013, 5, Downloaded from https://asbmr.onlinelibrary.wiley.com/doi/10.1002/jbmr.1828 by SEA ORCHID (Thailand), Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 3. Pooled NHANES III and NHANES 2000–2004 Cohort Aged 65 Years and Older: Relative Risk of Major Osteoporotic Fracture
per 1 SDa Unit Decline in Serum 25OHD by Age, Sex, or Race and Ethnicity Among Those With <10 Years of Follow-Up

Base modelb Full modelc

Cases (n) RR LL UL Cases (n) RR LL UL

Total 418 1.26 1.11 1.43 400 1.27 1.12 1.44
Age (years)
65–79 218 1.18 1.01 1.38 212 1.14 0.97 1.34

80 200 1.32 1.08 1.62 188 1.40 1.13 1.74
Sex
Male 135 1.44 1.13 1.83 129 1.45 1.11 1.89
Female 283 1.17 1.02 1.35 271 1.16 1.01 1.33
Race/ethnicity
Non-Hispanic white 319 1.24 1.08 1.42 308 1.22 1.07 1.39
Non-Hispanic black 35 0.81 0.56 1.17 32 0.91 0.61 1.35
Mexican American 56 1.62 1.14 2.32 53 1.28 0.93 1.78
NHANES ¼ National Health and Nutrition Examination Survey; RR ¼ relative risk; LL ¼ lower limit; UL ¼ upper limit; BMI ¼ body mass index.
a
1 SD ¼ 21.1 nmol/L.
b
Adjusted for age, survey, race/ethnicity, and sex.
c
Adjusted for BMI, height, smoking, health status, prescription osteoporosis drug use, current physical activity compared to activity 10 years ago, and
milk intake in addition to variables used in the base model.

Discussion
Results from this study based on a pooled cohort from NHANES
III and NHANES 2000–2004 add to previous findings from
NHANES regarding the relationship between serum 25OHD and
hip fracture risk in older white U.S. adults(10) by examining risk for
major osteoporotic fracture, expanding the sample to include
older adults of all races, and examining fracture risk by amount of
follow-up time. Serum 25OHD was significantly related to risk of
major osteoporotic fracture in a linear fashion, and, as previously
seen, significantly related to hip fracture in a nonlinear, quadratic
fashion. Results when stratified by amount of follow-up time
suggested that serum 25OHD may lose predictive utility once an
extended period of time has elapsed since its measurement.
Specifically, both major osteoporotic fracture risk and hip
fracture risk were significantly related to serum 25OHD among
respondents with less than 10 years of follow-up time, but not for
those with
10 years of follow-up time.
Previous studies of serum 25OHD and fracture have either
been limited to less than 10 years of follow-up or have not
reported predictive utility by follow-up time. Results of studies
that have examined tracking of serum 25OHD over time are
consistent with the possibility of attenuation in predictive utility
with increasing time. For example, Platz and colleagues(41)
found that the correlation between serum 25OHD values
measured on the same individuals after three years had
elapsed was 0.7, whereas studies that have re-measured
serum 25OHD after either 11 or 18 years found correlations
that ranged from 0.42 to 0.52.(9,42) Other biochemical measure-
ments, such as serum insulin or blood glucose, have been found
to lose predictive utility for cardiovascular events over time.(43)
However, the present study may also have been underpowered
to detect a relationship between serum 25OHD and fracture risk
in the subsample with 10 or more years of follow-up, because
less than 25% of the total sample were followed this long and
the number of fracture cases among these respondents was
small.
The shape of the best-fitting model also differed for hip
fracture and major osteoporotic fracture. Major osteoporotic
fracture risk by serum 25OHD has not been reported previously,
but previous research regarding the shape of the relationship
between serum 25OHD and hip fracture risk has conflicted.
In addition to earlier results from NHANES III,(10) Barbour and
colleagues(11) also reported a nonlinear relationship between
serum 25OH and hip fracture risk in their analysis of the Health
ABC cohort. However, a linear relationship between serum
25OHD and hip fracture risk has been reported in other
studies(2,3,7) Barbour and colleagues(11) hypothesized that
variability in the relationship between serum 25OHD and
fracture risk at different skeletal sites could stem from differences
in the relationship between serum 25OHD and bone type
(cortical versus trabecular) because the proportion of these bone
types varies at different skeletal sites.(44–46) It is also possible that
serum 25OHD may affect fracture risk via its relationship with
other risk factors, such as muscle strength or falls,(20) which in
turn may differ in their relationship with fracture risk at different
skeletal sites.
The shape of the quadratic relationship observed between hip
fracture risk and serum 25OHD was suggestive of a possible
increase in risk at the high end of the serum 25OHD distribution.
Increased fracture risk among those with high serum 25OHD, at
least in some groups, has been observed in a small number of
other prospective studies.(4,47) However, caution is needed when
interpreting the results observed in the present study, because
the data in the range where risk appeared to show a slight
increase (approximately 110–135 nmol/L) was very sparse: only
1.8% of the total sample and 5 fracture cases had values in this
range.

Journal of Bone and Mineral Research SERUM 25 OHD AND MAJOR OSTEOPOROTIC FRACTURE RISK 1003

Results from the present study can add to the discussion
about appropriate serum 25OHD thresholds, which remains
a controversial topic.(48,49) Risks of both fracture outcomes
considered in the present study were significantly elevated
among those with serum 25OHD <30 nmol/L, which is
consistent with the definition of deficiency proposed by the
IOM.(1) However, risk was not significantly elevated for those with
serum 25OHD in the two categories between 30 and 74 nmol/L
when compared to serum 25OHD
75 nmol/L for either fracture
outcome. These results are consistent with the IOM committee’s
observation of a plateau in effect for serum 25OHD levels
between 30 and 40 nmol/L.(49) However, it is important to note
that recent recommendations from both the IOM and the
Endocrine Society were based on consideration of data for more
than one skeletal health indicator.(1,20) In addition, results from
observational cohort studies have been viewed as offering only
supportive and/or confirmatory input to data from randomized
clinical trials.(1) Thus, although the present study offers relevant
data, it cannot resolve the issue of serum 25OHD thresholds in
light of its observational cohort study design and focus on a
single skeletal outcome.
Clear differences in the relationship between serum 25OHD
and major osteoporotic fracture risk by sex or age were not
detected in the present study. There was inconsistent evidence
for a sex difference in the relationship: the sex  serum 25OHD
interaction term was statistically significant, but the 95% CIs for
the RR overlapped between sexes. Several other prospective
studies have also found a similar relationship in both
sexes.(5,7,11,15) Results from the present study did not support
risk differences by age, as the 95% CIs overlapped between the
two age categories studied and the age  serum 25OHD
interaction term was not significant. Van Schoor and collea-
gues(5) reported a significant interaction between age and serum
25OHD for risk of fracture at any skeletal site in Dutch men and
women, but two other prospective studies performed in
older U.S. adults found no significant interaction with age when
examining the serum 25OHD–fracture relationship.(2,11)
There was also inconsistent evidence to support the possibility
of race/ethnic differences in the relationship between serum
25OHD and major osteoporotic fracture risk in the present study.
Specifically, the RR point estimates for major osteoporotic
fracture in NHW and MA’s indicated a significant increase in
fracture risk with decreasing serum 25OHD, whereas the RR point
estimate for NHB fell below 1.00 and was not statistically
significant. These results suggest that the serum 25OHD–fracture
relationship may differ by race/ethnicity, because serum 25OHD
appeared to be unrelated to fracture risk in blacks. However, firm
conclusions are not possible in light of the nonsignificant race/
ethnicity  serum 25OHD interaction term and the fact that the
95% CIs for the RR in blacks overlapped with the 95% CIs in the
other two race/ethnic groups. The wide 95% CIs and failure to
find a significant interaction term may stem from limited power
in the present study to detect differences due to the small
number of fracture cases among blacks (n ¼ 798 blacks with 35
fractures). Two other prospective studies in older U.S. adults also
found no significant interaction in the relationship between
serum 25OHD and fracture risk by race,(7,11) but, similar to the
present study, they were based on a small number of blacks. In
1004 LOOKER
15234681, 2013, 5, Downloaded from https://asbmr.onlinelibrary.wiley.com/doi/10.1002/jbmr.1828 by SEA ORCHID (Thailand), Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
contrast, Cauley and colleagues(4) found that risk of clinical
fractures was increased in black women compared to white
women who participated in the observational study portion of
the Women’s Health Initiative, which had a much larger sample
of blacks (n ¼ 7639 blacks with 379 fractures).(4) More studies
with larger samples of nonwhites are needed to determine
whether the serum 25OHD–fracture relationship differs between
race/ethnic groups.
Comparing results between prospective studies that have
examined the relationship between serum 25OHD and fracture
risk is complicated by methodological differences between the
studies. For example, use of different assays in the different
studies could hamper comparisons because of the lack of assay
standardization.(50) Even the same assay can change over time,
as noted for measurements from NHANES III compared with
NHANES 2000–2004.(34) Use of fractures at different skeletal sites
may also complicate comparison of results from different studies.
The present study examined risk of major osteoporotic fractures
as a group and of hip fracture alone, whereas outcomes used
in other studies have included fracture at any skeletal site,
any fracture other than hand/finger or foot/toe, any nonspine
fracture, hip fracture alone, or vertebral fracture alone.
Advantages of the current study include use of a pooled
cohort that was derived from representative samples of the U.S.
population, expansion of fracture outcomes to include major
osteoporotic fractures as a group, and the ability to explore the
relationship between serum 25OHD and fracture risk separately
by age, sex, and race/ethnicity within the same cohort.
Limitations of the present study include fracture cases that
were identified using administrative medical records without
confirmation by X-ray. In their validation study, Ray and
colleagues(26) found that the positive predictive values for hip,
radius/ulna, and humerus fractures were
95% when based on
Medicare records, which suggests that most fractures at those
skeletal sites were likely correctly identified in the present study.
The majority of fracture cases in the present study also had codes
consistent with fracture on multiple medical records, which
further supports the likelihood that identified cases suffered a
fracture. However, vertebral fracture cases were limited to those
which come to medical attention, which, although consistent
with the approach used in the FRAX model, is an undercount of
all vertebral fractures.(51) In addition, the positive predictive value
for incident vertebral fractures identified with the algorithm used
in the present study was only 61%.(28) Finally, some fracture cases
may have been mistakenly classified as non-cases because
Medicare records prior to 1991 were not available. This number
is likely to be small: an estimated 38 undetected major
osteoporotic fracture cases were projected to have occurred
in the sample during 1988–1990. Furthermore, misclassification
of cases as non-cases would likely tend to attenuate the relative
risk estimates observed between fracture risk and serum 25OHD.
Results from the present study also apply only to the segment
of the population ages 65 years and older that was not
institutionalized at baseline and participated in Medicare fee-for-
service programs because medical records for respondents who
received care from managed care programs or in Veterans
Administration facilities were not available. Furthermore, exclu-
sions for missing data or loss to follow-up were also made. The
Journal of Bone and Mineral Research

respondents who were excluded from the analytic sample were
more likely to have several characteristics that are associated
with greater fracture risk (older, female, poorer health, and less
activity) than respondents who were included. Thus, although
based on a cohort that was derived from nationally representa-
tive samples of the civilian, noninstitutionalized population at
baseline, results from the present study cannot be generalized to
the entire adult population over age 65 years.
In conclusion, serum 25OHD was significantly related to risk of
both major osteoporotic fractures as a group and hip fracture
alone in this cohort of older adults, but the shape of the
relationship differed (linear for major osteoporotic fracture,
quadratic for hip fracture). For both major osteoporotic and hip
fracture, serum 25OHD was a significant predictor of risk for up
to 10 years after measurement, but was no longer significant
after more than 10 years had elapsed. Clear differences in the
relationship between serum 25OHD and fracture risk by age,
sex, and race/ethnicity were not detected, although RR point
estimates varied by these characteristics. The study may have
been underpowered to assess long-term predictive utility or to
detect a significant relationship by demographic characteristics,
so replication of these results in other cohorts with larger
numbers of fracture cases would be desirable.
Disclosures
The author states that she has no conflicts of interest.
Acknowledgments
Author’s role: Study design, study conduct, data analysis, draft-
ing, revising and approving content of final manuscript: ACL. ACL
takes responsibility for the integrity of the data analysis.
References
1. Institute of Medicine. Dietary reference intakes for calcium and
vitamin D. Washington, DC: National Academies Press; 2010.
2. Cauley JA, LaCroix AZ, Wu L, Horwitz M, Danielson ME, Bauer DC, Lee
JS, Jackson RD, Robbins JA, Wu C, Stanczyk FZ, LeBoff MS, Wactawski-
Wende J, Sarto G, Ockene J, Cummings SR. Serum 25-hydroxyvitamin
D concentrations and risk for hip fractures. Ann Intern Med. 2008
Aug; 149(4):242–50.
3. Cauley JA, Parimi N, Ensrud KE, Bauer DC, Cawthon PM, Cummings SR,
Hoffman AR, Shikany JM, Barrett-Connor E. Orwoll E; Osteoporotic
Fractures Men Mr OSR. Serum 25-Hydroxyvitamin D and the risk of
hip and nonspine fractures in older men. J Bone Miner Res. 2010 Mar;
25(3):545–53.
4. Cauley JA, Danielson ME, Boudreau R, Barbour KE, Horwitz MJ, Bauer
DC, Ensrud KE, Manson JE, Wactawski-Wende J, Shikany JM, Jackson
RD. Serum 25-hydroxyvitamin D and clinical fracture risk in a multi-
ethnic cohort of women: the Women’s Health Initiative (WHI). J Bone
Miner Res. 2011 Oct; 26(10):2378–88.
5. van Schoor NM, Visser M, Pluijm SMF, Kuchuk N, Smit JH, Lips P.
Vitamin D deficiency as a risk factor for osteoporotic fractures. Bone.
2008 Feb; 42(2):260–6.
6. Nakamura K, Saito T, Oyama M, Oshiki R, Kobayashi R, Nishiwaki T,
Nashimoto M, Tsuchiya Y. Vitamin D sufficiency is associated with low
incidence of limb and vertebral fractures in community-dwelling
Journal of Bone and Mineral Research
15234681, 2013, 5, Downloaded from https://asbmr.onlinelibrary.wiley.com/doi/10.1002/jbmr.1828 by SEA ORCHID (Thailand), Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
elderly Japanese women: the Muramatsu Study. Osteoporos Int. 2011
Jan; 22(1):97–103.
7. Robinson-Cohen C, Katz R, Hoofnagle AN, Cauley JA, Furberg CD,
Robbins JA, Chen Z, Siscovick DS, de Boer IH, Kestenbaum B.
Mineral metabolism markers and the long-term risk of hip fracture:
the Cardiovascular Health Study. J Clin Endocrinol Metab. 2011 Jul;
96(7):2186–93.
8. Gerdhem P, Ringsberg K, Obrant K, Akesson K. Association between
25-hydroxy vitamin D levels, physical activity, muscle strength and
fractures in the prospective population-based OPRA Study of Elderly
Women. Osteoporos Int. 2005 Nov; 16(11):1425–31.
9. Melhus H, Snellman G, Gedeborg R, Byberg L, Berglund L, Mallmin H,
Hellman P, Blomhoff R, Hagstrom E, Arnlov J, Michaelsson K. Plasma
25-hydroxyvitamin D levels and fracture risk in a community-based
cohort of elderly men in Sweden. J Clin Endocrinol Metab. 2010 Jun;
95(6):2637–45.
10. Looker AC, Mussolino ME. Serum 25-hydroxyvitamin D and hip
fracture risk in older U.S. white adults. J Bone Miner Res. 2008 Jan;
23(1):143–50.
11. Barbour KE, Houston DK, Cummings SR, Boudreau R, Prasad T, Sheu Y,
Bauer DC, Tooze JA, Kritchevsky SB, Tylavsky FA, Harris TB. Cauley JA;
the Health ABCs. Calciotropic hormones and the risk of hip and non-
spine fractures in older adults: the Health ABC Study. J Bone Miner
Res. 2012 Jan 6; 27(5):1177–85.
12. Cummings SR, Browner WS, Bauer D, Stone K, Ensrud K, Jamal S,
Ettinger B. Endogenous hormones and the risk of hip and vertebral
fractures among older women. N Engl J Med. 1998;339(11):733–8.
13. Garnero P, Munoz F, Sornay-Rendu E, Delmas PD. Associations of
vitamin D status with bone mineral density, bone turnover, bone loss
and fracture risk in healthy postmenopausal women. The OFELY
Study. Bone. 2007 Mar; 40(3):716–22.
14. Pramyothin P, Techasurungkul S, Lin J, Wang H, Shah A, Ross PD,
Puapong R, Wasnich RD. Vitamin D status and falls, frailty, and
fractures among postmenopausal Japanese women living in Hawaii.
Osteoporos Int. 2009 Nov; 20(11):1955–62.
15. Roddam AW, Neale R, Appleby P, Allen NE, Tipper S, Key TJ. Associa-
tion between plasma 25-hydroxyvitamin D levels and fracture risk:
the EPIC-Oxford Study. Am J Epidemiol. 2007 Dec 1; 166(11):1327–36.
16. Center JR, Nguyen TV, Sambrook PN, Eisman JA. Hormonal and
biochemical parameters and osteoporotic fractures in elderly men.
J Bone Miner Res. 2000 Jul; 15(7):1405–11.
17. Flegal KM, Graubard BI, Williamson DF, Gail MH. Excess deaths
associated with underweight overweight obesity. JAMA. 2005 Apr
20; 293(15):1861–7.
18. Breslow RA, Chen CM, Graubard BI, Mukamal KJ. Prospective study of
alcohol consumption quantity and frequency and cancer-specific
mortality in the US population. Am J Epidemiol. 2011 Nov 1; 174(9):
1044–53.
19. Kanis JA. WHO Scientific Group on Assessment of Osteoporosis at the
Primary Health-Care Level. Assessment of osteoporosis at the primary
health-care level. Geneva: World Health Organization; 2007.
20. Holick MF, Binkley NC, Bischoff-Ferrari HA, Gordon CM, Hanley DA,
Heaney RP, Murad MH, Weaver CM. Evaluation, treatment, and
prevention of vitamin D deficiency: an Endocrine Society Clinical
Practice Guideline. J. Clin Endocrinol Metab. 2011 Jul; 96(7):1911–30.
21. Centers for Disease Control and Prevention. National Center for
Health Statistics. Plan and operation of the third National Health
and Nutrition Examination Survey, 1988–94. Series 1: programs
and collection procedures. Hyattsville MD: Centers for Disease
Control and Prevention, National Center for Health Statistics; 1994.
p. 1–407.
22. Centers for Disease Control Prevention National Center for Health
Statistics. NHANES Questionnaires Datasets Related Documentation
[Internet]. Hyattsville, MD: Centers for Disease Control Prevention
National Center for Health Statistics 2012 [cited 2012 Nov 28].
SERUM 25 OHD AND MAJOR OSTEOPOROTIC FRACTURE RISK 1005

Available from: http://www.cdc.gov/nchs/nhanes/nhanes_
questionnaires.htm.
23. Centers for Disease Control Prevention National Center for Health
Statistics. NCHS data linked to mortality files [Internet]. Hyattsville
MD: Centers for Disease Control Prevention National Center for
Health Statistics, 2012 [cited 2012 Nov 28]. Available from:
http://www.cdc.gov/nchs/data_access/data_linkage/mortality.htm.
24. Centers for Disease Control Prevention National Center for Health
Statistics. NCHS data linked to CMS Medicare enrollment claims
files [Internet]. Hyattsville MD: Centers for Disease Control
Prevention National Center for Health Statistics, 2012 [cited 2012
Nov 28]. Available from: http://www.cdc.gov/nchs/data_access/
data_linkage/cms_medicare.htm.
25. Moon M. What Medicare has meant to older Americans. Health Care
Financ Rev. 1996; Win; 18(2):49–59.
26. Ray WA, Griffin MR, Fought RL, Adams ML. Identification of fractures
from computerized Medicare files. J Clin Epidemiol. 1992 Jul;
45(7):703–14.
27. Taylor AJ, Gary LC, Arora T, Becker DJ, Curtis JR, Kilgore ML, Morrisey
MA, Saag KG, Matthews R, Yun H, Smith W, Delzell E. Clinical
and demographic factors associated with fractures among older
Americans. Osteoporos Int. 2011 Apr; 22(4):1263–74.
28. Curtis JR, Mudano AS, Solomon DH, Xi J, Melton E, Saag KG. Identifi-
cation and validation of vertebral compression fractures using ad-
ministrative claims data. Med Care. 2009 Jan; 47(1):69–72.
29. American M. edical Association CPT (Current Procedural Terminolo-
gy) Assistant Archives 1990–2009 [CD ROM]. Chicago, IL: American
Medical Association
30. Centers for Disease Control Prevention National Center for Health
Statistics. Conversion Table of New ICD-9-CM Codes. 2010; Hyattsville
MD: Centers for Disease Control Prevention National Center for
Health Statistics, 2010 [cited 2012 Nov 28]. Available from:
http://www.cdc.gov/nchs/data/icd9/icdcnvFY11.pdf.
31. Centers for Disease Control Prevention (CDC). Incidence cost
to Medicare of fractures among Medicare beneficiaries aged
(65 years-United States July 1991–June 1992. MMWR Morb Mortal
Wkly Rep. 1996;45(41):877–3.
32. Gunter EW, Lewis BG, Kokickowski SM. Laboratory methods used for
the third National Health and Nutrition Examination Survey (NHANES
III), 1988–1994 Hyattsville MD: Centers for Disease Control, National
Center for Health Statistics; 1996.
33. Centers for Disease Control Prevention National Center for
Health Statistics. Laboratory procedure manual: 25-hydroxyvitamin
Hyattsville D. MD: Centers for Disease Control and Prevention,
National Center for Health Statistics; 2004 [Internet]. Available
from: http://www.cdc.gov/nchs/data/nhanes/nhanes_03_04/l06vid_
c_met_Vitamin_D.pdf.
34. Looker AC, Pfeiffer CM, Lacher DA, Schleicher RL, Picciano MF, Yetley
EA. Serum 25-hydroxyvitamin D status of the US population: 1988–
1994 compared with 2000–2004. Am J Clin Nutr. 2008 Dec; 88(6):
1519–27.
35. Research Triangle Institute. SUDAAN Language Manual Release 10.0.,
Research Triangle Park, NC: Research Triangle Institute, 2008.
36. Korn EL, Graubard BI. Analysis of health surveys. New York: John
Wiley & Sons; 1999.
37. Jacobsen SJ, Cooper C, Gottlieb MS, Goldberg J, Yahnke DP, Melton
LJ. Hospitalization with vertebral fracture among the aged—a
1006 LOOKER
15234681, 2013, 5, Downloaded from https://asbmr.onlinelibrary.wiley.com/doi/10.1002/jbmr.1828 by SEA ORCHID (Thailand), Wiley Online Library on [20/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
national population-based study, 1986–1989. Epidemiology. 1992.
Nov; 3(6):515–8.
38. Centers for Disease Control Prevention National Center for Health
Statistics. Compressed mortality file 1979–1998. CDC WONDER On-
line Database,compiled from Compressed Mortality File CMF 1968–
1988, Series 20,No. 2A, 2000 and CMF 1989–1998, Series 20, No.2E,
2003 [Internet]. Hyattsville, MD: Centers for Disease Control, National
Center for Health Statistics; 2000, 2003 [cited 2012 Nov 28]. Available
from: http://wonder.cdc.gov/cmf-icd9.html.
39. Holick MF, Binkley NC, Bischoff-Ferrari HA, Gordon CM, Hanley DA,
Heaney RP, Murad MH, Weaver CM. Guidelines for preventing and
treating vitamin D deficiency and insufficiency revisited. J Clin Endo-
crinol Metab. 2012 Apr; 97(4):1153–8.
40. Cummings SR, Nevitt MC, Browner WS, Stone K, Fox KM, Ensrud KE,
Cauley J, Black D, Vogt TM. Risk factors for hip fracture in white
women. N Engl J Med. 1995;332(12):767–74.
41. Platz EA, Leitzmann MF, Hollis BW, Willett WC, Giovannucci E.
Plasma 1,25-dihydroxy- and 25-hydroxyvitamin D and sub-
sequent risk of prostate cancer. Cancer Causes Control. 2004;15(3):
255–65.
42. 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 Apr 15; 171(8):903–8.
43. Wannamethee SG, Shaper AG, Whincup PH, Walker M. Role of risk
factors for major coronary heart disease events with increasing
length of follow up. Heart. 1999 Apr; 81(4):374–9.
44. Lauretani F, Bandinelli S, Russo CR, Maggio M, Di Iorio A, Cherubini A,
Maggio D, Ceda GP, Valenti G, Guralnik JM, Ferrucci L. Correlates of
bone quality in older persons. Bone. 2006;39(4):915–21.
45. Barbour K, Zmuda J, Horwitz M, Strotmeyer E, Boudreau R, Evans R,
Ensrud K, Gordon C, Petit M, Patrick A, Cauley J. Osteoporotic
Fractures in Men (MrOS) Research Group. The association of
serum 25-hydroxyvitamin D with indicators of bone quality in
men of Caucasian and African ancestry. Osteoporos Int. 2011;
22(9):2475–85.
46. Einhorn TA. Bone strength—the bottom line. Calcif Tissue Int. 1992
Nov; 51(5):333–9.
47. Bleicher K. Older men with either high or low serum 25-hydroxyvi-
tamin D levels have significantly increased fracture risk: results from
the prospective CHAMP study. Oral Poster Session, Presentation
Number: FR0349. J Bone Miner Res. 2012;27(Suppl 1)[Internet].
Washington, DC: ASBMR; 2012 [cited 2012 Nov 28]. Available
from: http://www.asbmr.org/Meetings/AnnualMeeting/AbstractDe-
tail.aspx?aid¼a864e09b-4996-4541-a9d2-fca4eea69e32.
48. Heaney RP, Holick MF. Why the IOM recommendations for vitamin D
are deficient. J Bone Miner Res. 2011;26(3):455–7.
49. Rosen CJ, Abrams SA, Aloia JF, Brannon PM, Clinton SK, Durazo-Arvizu
RA, Gallagher JC, Gallo RL, Jones G, Kovacs CS, Manson JE, Mayne ST,
Ross AC, Shapses SA, Taylor CL. IOM committee members respond
to Endocrine Society vitamin D guideline. J Clin Endocrinol Metab.
2012 Apr; 97(4):1146–52.
50. Carter GD. 25-Hydroxyvitamin D: a difficult analyte. Clin Chem. 2012
Mar; 58(3):486–8.
51. Kanis JA, Johnell O, Oden A, Borgstrom F, Zethraeus N, De Laet C,
Jonsson B. The risk and burden of vertebral fractures in Sweden.
Osteoporos Int. 2004 Jan; 15(1):20–6.
Journal of Bone and Mineral Research