Lindsay R, Silverman SL, Cooper C, Hanley DA, Barton I, Broy SB, Licata A, Benhamou L, Geusens P, Flowers K, Stracke H, Seeman E. Risk of New Vertebral Fracture in the Year Following a Fracture. JAMA. 2001;285(3):320–323. doi:10.1001/jama.285.3.320
Author Affiliations: Regional Bone Center, Helen Hayes Hospital, West Haverstraw, NY (Dr Lindsay); Osteoporosis Medical Center, Cedars Sinai Medical Center, and Greater Los Angeles VA Health System, Los Angeles, Calif (Dr Silverman); Medical Research Council Environmental Epidemiology Unit, University of Southampton, Southampton, England (Dr Cooper); University of Calgary Health Sciences Centre, Calgary, Alberta (Dr Hanley); Procter & Gamble Pharmaceuticals, Staines, Middlesex, England (Mr Barton); Center for Arthritis and Osteoporosis, Illinois Bone and Joint Institute, Des Plaines (Dr Broy); Cleveland Clinic Foundation, Cleveland, Ohio (Dr Licata); Centre Hospitalier d'Orleans, Orleans, France (Dr Benhamou); Limburg University Center, Diepenbeek, Belgium (Dr Geusens); University of Maastricht, the Netherlands (Dr Geusens); Procter & Gamble Pharmaceuticals, Cincinnati, Ohio (Ms Flowers); Justus-Liebig University, Giessen, Germany (Dr Stracke); and University of Melbourne, Melbourne, Australia (Dr Seeman).
Context Vertebral fractures significantly increase lifetime risk of future fractures,
but risk of further vertebral fractures in the period immediately following
a vertebral fracture has not been evaluated.
Objective To determine the incidence of further vertebral fracture in the year
following a vertebral fracture.
Design and Setting Analysis of data from 4 large 3-year osteoporosis treatment trials conducted
at 373 study centers in North America, Europe, Australia, and New Zealand
from November 1993 to April 1998.
Subjects Postmenopausal women who had been randomized to a placebo group and
for whom vertebral fracture status was known at entry (n = 2725).
Main Outcome Measure Occurrence of radiographically identified vertebral fracture during
the year following an incident vertebral fracture.
Results Subjects were a mean age of 74 years and had a mean of 28 years since
menopause. The cumulative incidence of new vertebral fractures in the first
year was 6.6%. Presence of 1 or more vertebral fractures at baseline increased
risk of sustaining a vertebral fracture by 5-fold during the initial year
of the study compared with the incidence in subjects without prevalent vertebral
fractures at baseline (relative risk [RR], 5.1; 95% confidence interval [CI],
3.1-8.4; P<.001). Among the 381 participants who
developed an incident vertebral fracture, the incidence of a new vertebral
fracture in the subsequent year was 19.2% (95% CI, 13.6%-24.8%). This risk
was also increased in the presence of prevalent vertebral fractures (RR, 9.3;
95% CI, 1.2-71.6; P = .03).
Conclusion Our data indicate that women who develop a vertebral fracture are at
substantial risk for additional fracture within the next year.
Vertebral fractures are a well-recognized consequence of postmenopausal
bone loss and are the most common osteoporotic fractures.1
It is estimated that less than one third of all vertebral fractures are clinically
diagnosed.2 However, all vertebral fractures,
whether symptomatic or radiographically identified, are associated with increased
mortality and morbidity, including back pain and decreased activity, with
consequent increased days of bed rest.3- 5
Vertebral fractures are associated with increased risk of further vertebral
fractures, with resulting height loss and kyphosis, as well as increased risk
of nonvertebral fractures.6- 10
This increased risk remains after correction for bone mineral density (BMD),
itself a potent risk factor for fracture.6,10,11
Since many vertebral fractures are found by chance and it is difficult
to date these fractures, we do not know whether time from fracture modifies
the risk conferred by the fracture. It has been suggested but not confirmed
by data that the greatest risk of a second fracture exists during the time
immediately following the initial fracture.12
If true, this highlights the clinical importance of fracture identification
as soon after the event as possible. To evaluate this issue, we analyzed data
from women in the placebo groups of 4 large clinical trials conducted at 373
study centers in North America, Europe, Australia, and New Zealand to determine
whether risk of vertebral fracture is increased in the period immediately
following a vertebral fracture.
The study population consisted of women who had been randomly assigned
to a placebo group in 4 large 3-year clinical trials conducted from November
1993 to April 1998 evaluating the efficacy of risedronate, a bisphosphonate,
for treatment of postmenopausal osteoporosis.13- 15
These women had either prevalent vertebral fractures (2 studies),13,14 low femoral neck BMD, or risk factors
for hip fracture.15 All subjects received calcium
supplementation (1000 mg/d). Women with serum 25-hydroxyvitamin D levels of
less than 16 ng/mL (40 nmol/L) at baseline also received vitamin D supplementation
(up to 500 IU/d).
Lateral spine radiographs were obtained at baseline for evaluation of
prevalent vertebral fractures and annually thereafter for incident vertebral
fractures, as previously described.13,14
The vertebral fracture analyses included all placebo subjects who had both
baseline and postbaseline evaluable radiographs. Clinical vertebral fractures
were recorded as adverse events and diagnosed by a physician.
Demographic and baseline characteristics were summarized using descriptive
statistics for subjects receiving placebo. The incidence of new vertebral
fractures, based on time to first incident fracture, was analyzed using survival
analysis methods. The cumulative incidence was calculated using Kaplan-Meier
estimates. A Cox regression model was used to compare risk of incident vertebral
fracture in subjects with prevalent fracture compared with those without prevalent
fracture; similar methods were used to investigate risk of additional vertebral
fracture within 1 year of a vertebral fracture that occurred during the study.
The effect of potential baseline covariates (age, weight, lumbar spine BMD,
and vitamin D status) was investigated by adjusting for these covariates as
continuous variables in the Cox regression model.
A total of 4356 subjects were randomly assigned to placebo study arms;
of these, vertebral fracture status was known for 2725 (57%). The baseline
characteristics of these subjects (98% white) are shown in Table 1 and were similar across studies. Because of differences
in study design and recruitment among trials, baseline lumbar spine BMD values
were available for 885 subjects (32%); the mean (SD) lumbar spine T score
was −2.6 (1.3).
Over the course of the studies, vertebral fractures were observed in
381 of the 2725 women. The Kaplan-Meier estimate of the vertebral fracture
incidence over 3 years was 16.9%. Of the 381 women who sustained vertebral
fractures, 23% had symptomatic vertebral fractures. Risk of sustaining a vertebral
fracture increased with presence of prevalent fractures (relative risk [RR],
3.7; 95% confidence interval [CI], 2.8-4.9; P<.001).
During the first year, the proportion of women who developed vertebral fractures
was 6.6% (Table 2) and, again,
risk increased with presence of prevalent vertebral fractures (RR, 5.1; 95%
CI, 3.1-8.4; P<.001) (Figure 1A). In subjects with baseline BMD values, risk of incident
vertebral fracture over 1 year increased significantly for each 1-SD decrease
in baseline BMD value below the mean for a young, healthy population (RR,
1.6; 95% CI, 1.1-2.2; P = .007).
Among subjects in whom incident fractures were confirmed, occurrence
of a second incident vertebral fracture within 1 year of the initial fracture
was 19.2% overall (95% CI, 13.6%-24.8%) (Table 3). Risk also increased with prevalent vertebral fractures
(RR, 9.3; 95% CI, 1.2-71.6; P = .03) (Figure 1B). Twenty-four percent of subjects with 2 or more prevalent
fractures at baseline had an incident vertebral fracture within 1 year of
their first observed fracture. The RR did not change when adjusted for age,
weight, or baseline vitamin D status.
Vertebral fractures are a serious and irreversible outcome of osteoporosis.
Previous data have demonstrated that risk of vertebral fracture is increased
among women in whom a prior vertebral fracture is identified. Our data are
consistent with these reports, with the RR of new vertebral fracture increasing
with the number of baseline vertebral fractures. The design of our clinical
trials also allowed for identification of vertebral fractures on an annual
basis (incident fractures). These incident vertebral fractures also increased
risk of future vertebral fractures and this increased risk appeared to be
greatest in the initial year following the fracture. Twenty-three percent
of incident fractures were clinical events, similar to the relationship between
clinical and radiological fractures observed previously.2
Our finding that almost 20% of women will experience another fracture within
1 year of an incident vertebral fracture has important clinical implications.
The increased fracture risk in the immediate period following a fracture demonstrates
the urgency of identification and intervention for this segment of the population
and was observed despite that all subjects received calcium and vitamin D.
The presence of prevalent fractures significantly enhanced risk after
an incident fracture (4% with 0 vs 24% with ≥2). While BMD values were
available only for a subset of the population, there was a 60% increase in
risk of vertebral fracture during the first year of the study for each 1-SD
decrease in baseline BMD value below the mean for a young, healthy population.
Thus, the combination of low lumbar spine BMD and prevalent fractures is the
best predictor of increased fracture risk in the immediate period after a
There are some limitations to our findings. First, clinical trial subjects
may differ from patients commonly seen in clinical practice. The similarity
of our findings to those already reported with regard to the effect of prevalent
fractures in predicting future fractures suggests that our results may be
generalized to the postmenopausal population with osteoporosis. As in observational
studies, we do not know the timing of the fractures observed at baseline.
Some may have been recent, which would lead to a higher-than-expected incidence
in the first year and would be expected to diminish the differences we observed.
Because we do not know the timing of the fractures that existed at baseline,
our data do not allow us to evaluate an important clinical question: whether
an incident fracture, compared with a history of fracture, leads to greater
risk for fracture. Baseline BMD values were available for approximately one
third of our subjects; thus, we cannot completely correct for the effects
of BMD. However, other studies have shown consistently that the effects of
prevalent fractures are independent of BMD.6,9,10
We have confirmed that prevalent fractures increase risk of further
vertebral fractures and have shown for the first time, to our knowledge, that
incident vertebral fractures exacerbate this effect. Our finding that approximately
20% of women will experience another fracture within the first year of a vertebral
fracture justifies a degree of urgency for clinicians in identifying and treating
all patients who present with vertebral fractures. These data indicate that
osteoporosis actually may be a quickly progressing disease once a fracture
occurs. Further research should be carried out to determine whether an incident
fracture, compared with a history of fracture, leads to greater future risk