The authors have no conflict of interest
Association Between Vertebral Fracture and Increased Mortality in Osteoporotic Patients†
Version of Record online: 1 JUL 2003
Copyright © 2003 ASBMR
Journal of Bone and Mineral Research
Volume 18, Issue 7, pages 1254–1260, July 2003
How to Cite
Jalava, T., Sarna, S., Pylkkänen, L., Mawer, B., Kanis, J. A., Selby, P., Davies, M., Adams, J., Francis, R. M., Robinson, J. and McCloskey, E. (2003), Association Between Vertebral Fracture and Increased Mortality in Osteoporotic Patients. J Bone Miner Res, 18: 1254–1260. doi: 10.1359/jbmr.2003.18.7.1254
- Issue online: 2 DEC 2009
- Version of Record online: 1 JUL 2003
- Manuscript Accepted: 6 JAN 2003
- Manuscript Revised: 13 NOV 2002
- Manuscript Received: 27 FEB 2002
- vertebral fracture;
- bone mineral density;
Determinants of mortality were studied in a prospective study of 677 women and men with primary or secondary osteoporosis. Prevalent vertebral fractures were associated with increased mortality, but other known predictors of mortality explain a significant proportion of the excess risk.
Introduction: In population studies, prevalent vertebral fractures are associated with increased mortality. It is unknown whether this excess mortality is related to low bone mineral density or its determinants or whether there is an additional component associated with fracture itself.
Methods: We studied 677 women and men with osteoporosis, 28–88 years old, of whom 352 had morphometrically determined vertebral fracture, to examine the risk and causes of mortality in patients with osteoporosis (defined densitometrically as a spine bone mineral density T-score < −2.5 and −3.0 for women and men, respectively, and/or one or more prevalent vertebral fractures without a history of significant trauma). The participants had enrolled in a double-blind placebo-controlled study in osteoporosis and were comprised of 483 women with postmenopausal osteoporosis, 110 women with secondary osteoporosis, and 84 men with osteoporosis of any cause. Demographics, medical history, and other measures of skeletal and nonskeletal health status were assessed at entry.
Results: During a median follow-up of 3.2 years, 37 (5.5%) participants died, with 31 of these deaths occurring in those with prevalent vertebral fractures. Compared with participants who did not have a prevalent vertebral fracture, those with one or more fractures had a 4.4-fold higher (95% CI, 1.85, 10.6) mortality rate. After adjustment for predictors for poor health—including number of medications, number of diseases, use of oral corticosteroids, alcohol intake, serum albumin and erythrocyte sedimentation rate (ESR), renal function, height, weight, gender, and age—the point estimate of risk remained elevated but was no longer statistically significant (hazard ratio, 2.4; 95% CI. 0.93, 6.23).
Conclusions: Prevalent vertebral fractures in osteoporotic patients are associated with increased mortality. Other known predictors of mortality can explain a significant proportion of the excess risk.
Osteoporosis-related fractures have important health consequences, including disability and increased mortality. Although it is widely recognized that hip fractures are associated with increased mortality,(1–5) the effect of other fractures on mortality is less well studied. However, recent investigations have shown that individuals with prevalent vertebral deformities, whether clinically apparent or not, have decreased survival. Excess mortality after vertebral fractures have been reported to range from 16% to 60% among early or late postmenopausal women with prevalent morphometric vertebral fractures.(6–8) The mechanism of increased mortality is unclear, and it has been hypothesized that it may reflect poor underlying general health status in addition to the acute effects of the fracture, especially in elderly people,(1,7) but the results are still conflicting.(9)
Our study aimed to further explore the impact of vertebral deformities on mortality in patients with postmenopausal and secondary osteoporosis (including men). Additionally, we wanted to investigate whether any observed excess in mortality could be explained by the presence of confounding factors linked to osteoporosis or by other known predictors of mortality.
MATERIALS AND METHODS
Indices of skeletal and nonskeletal health were assessed at entry in 677 osteoporotic patients, 28–88 years old, who participated in a double-blind placebo-controlled trial of clodronate. The methods are described in detail in elsewhere.(10) The patients were randomized into three strata: postmenopausal osteoporosis (I, n = 483), women with secondary osteoporosis (II, n = 110), and men with osteoporosis of any cause (III, n = 84). All patients had osteoporosis defined densitometrically (spine bone mineral density [BMD] T-score ≤ −2.5 and −3.0 for women and men, respectively) and/or had one or more prevalent vertebral fractures without a history of significant trauma. The causes of secondary osteoporosis in women included corticosteroid use (n = 73), previous thyrotoxicosis (n = 7), previous nonrecurrent malignancy (breast cancer, n = 11; other cancers, n = 6), hyperparathyroidism (n = 2), and miscellaneous causes (n = 11). The men with osteoporosis included 48 with primary/idiopathic osteoporosis, 14 with corticosteroid exposure, 12 with osteoporosis secondary to alcohol excess, and 10 with miscellaneous other causes.
Patients excluded from the study included those receiving treatment for a malignancy; those currently taking medication likely to influence skeletal metabolism or the interpretation of results (i.e., >500 mg daily calcium supplements, estrogens, progestogens, calcitonin, anabolic steroids, and bisphosphonates); those who had received bisphosphonates up to 1 year before enrollment, and those known to have malabsorption. Patients in strata II and III were included if they were past or current users of corticosteroids.
The protocol was approved by the local Research Ethics Committees at each participating study center, and all patients gave written informed consent before participating in the trial.
After randomization, the participants received the study medication that was comprised of either clodronate as a single daily oral dose of two 400-mg capsules (Bonefos, Leiras Oy, Finland) or an identical placebo (given at least 1 h before breakfast with water on an empty stomach). All patients were additionally given a calcium supplement of 500 mg daily.
Assessment of vertebral deformities
Standardized lateral thoracic and lumbar radiographs were taken at baseline for vertebral morphometry. The determination of a prevalent vertebral deformity at baseline was made at the coordinating center (WHO Collaborating Centre, Medical School, Sheffield, UK) using the semiquantitative vertebral morphometry method.(11) A patient was classified as having a prevalent vertebral deformity if any one of the ratios of morphometric vertebral heights was more than 3 SDs below the mean population norm for that vertebral level.(11) The prevalent fractures in eight patients with history of cancer were not excluded from analyses. All patients had lateral spine radiographs repeated at annual intervals, regardless of symptoms, for the detection of incident vertebral deformities. An incident vertebral deformity was defined as using a similar definition (i.e., it had to be detected as a prevalent fracture at a previously normal level) but with a minimum decrease in height of 15% and 4.6 mm from the baseline radiograph.
Clinical and biochemical markers of skeletal and nonskeletal health and ascertainment of mortality
BMD was measured at the spine and hip by DXA using two Hologic QDR1000 densitometers (Sheffield, Manchester, UK), two QDR2000plus densitometers (Sheffield, Newcastle, UK), a Lunar DPX-L (Manchester, UK), and a Lunar Expert 1129 (Liverpool, UK). The European Spine Phantom was circulated to all centers to allow computation of standardized BMD (sBMD) and allow comparison of between-center differences at study entry.
Morning blood and urine samples were obtained from the patients after an overnight fast. Levels of 25 hydroxyvitamin D [25(OH)D] and 1,25 dihydroxyvitamin D [1,25(OH)2D] were determined using an automated high-performance liquid chromatographic method.(12,13) Serum calcium, albumin, alkaline phosphatase activity, and erythrocyte sedimentation rate (ESR) were carried out at hospital clinical laboratories using routine techniques, as were assessments of creatinine clearance and fasting urinary excretion of hydroxyproline (expressed as a ratio to urinary creatinine). Each patient's past medical history, including the use of concomitant medication and lifestyle factors such as drinking (yes or no), smoking (current/previous versus never), and exercise exposure (lasting one-half hour or more during a week versus less) were also recorded at entry.
Patients were reviewed at 3, 6, and 12 months and every 6 months thereafter. Deaths were confirmed against hospital records. Codes from the WHO's Adverse Reaction Dictionary (1994) were used to classify the cause of death.
Comparison of baseline characteristics between patients who died and those that survived was carried out using the χ2 test for categorical variables and either independent sample t-test or Mann-Whitney's test for continuous variables. Poisson regression models were used to analyze the differences in mortality rates in patients with and without prevalent vertebral fracture in different age groups (Egret for Windows, 2.03; Cytel Software Corp., Cambridge, MA, USA). Univariate Cox proportional hazards analysis was used to determine significant associations with mortality. All significant variables in the univariate analyses were then entered to a multivariate proportional hazards model to determine independent associations with mortality. The survival was tested by the log-rank test.
All results are expressed as means and SD unless stated otherwise. Two-sided p values 0.05 or below were considered as significant. Statistical analyses were done by SPSS version 10.0 for Windows (SPSS Inc., Chicago, IL, USA).
During a mean 3.2 years of follow-up, 37 (5%) of the 677 patients died. Eleven patients died from respiratory disease, seven from cancer related illnesses, nine from cardiovascular disease such as myocardial infarction and cardiac failure, six from extracardiac vascular disorders such as cerebrovascular accidents, and seven from a variety of other causes. There was no discernible difference in the causes of death between patients with and without baseline vertebral fractures, but the numbers within each group were relatively small. For example, 10 (91%) of the 11 patients who died because of respiratory disorders had prevalent vertebral fractures compared with 21 (81%) of the 26 patients who died because of other reasons (p = 0.64).
Of the patients enrolled, 352 (55%) had prevalent radiographic vertebral fractures at entry. The baseline prevalence of vertebral fractures was significantly higher (84%) in patients who subsequently died during follow-up than in those patients who remained alive (50%, p < 0.0001; Table 1). The mortality was higher among osteoporotic male patients (9/84, 11%) than in female patients (28/593, 5%) with established osteoporosis (p = 0.036). Furthermore, the deceased patients had lower mean values of serum albumin (p = 0.012) and 25(OH)D (p = 0.046) and higher values of ESR (p = 0.037) at entry. They also had a higher number of concomitant diseases (p < 0.003) and treatments (p < 0.0001) and more impaired renal function than those patients who remained alive (Table 1). Patients dying during the study had smoked (p = 0.022) and/or used oral corticosteroids (p = 0.003) more often than those who remained alive (Table 1). Exclusion of the eight patients with prior nonrecurrent malignancy and baseline vertebral fractures had no impact on the study outcome.
As expected, the death rates were higher in the older age groups than in the younger age groups, but the mortality rates were about 3- to 6-fold higher among patients with prevalent vertebral fractures compared with those without fractures across the age groups (p = 0.0028; Figure 1). The mortality rates increased from 2.1 per 100 patients-years in patients with prevalent vertebral fractures 59 years old or less to 3.16 per 100 patient-years in patients 69 years old or more compared with 0.37 and 0.90 per 100 patient-years in patients without prevalent vertebral fractures (p < 0.0001), respectively.
Incident vertebral fractures did not seem to be associated with increased mortality. New vertebral fractures occurred in 114 (18%) of the 636 patients with follow-up radiographs available. Altogether, 30 (5.3%) patients without incident vertebral fractures died during the follow-up period compared with 7 (6.7%) patients with incident vertebral fractures. The study medication did not have any impact on patients' survival. The survival after 3.2 years of follow-up did not differ between bisphosphonate- and placebo-treated patients (p = 0.9927). However, interestingly, 6 of 7 patients with incident vertebral fractures who died were in the placebo group.
Univariate and multivariate analysis of associations with increased mortality
Mortality hazard ratios (HRs) for the skeletal and nonskeletal parameters measured at entry were calculated in separate univariate analyses adjusted for strata (Table 2). This analysis demonstrated that mortality was 4.4-fold (95% CI, 1.9, 10.6) higher in the patients with prevalent vertebral fractures. In addition to age and prevalent vertebral fractures, other factors associated with increased mortality included increased numbers of concomitant diseases or treatments, low albumin, increased ESR, renal impairment, and use of oral corticosteroids. Vitamin D levels and prior peripheral fractures were not significantly associated with increased mortality, although the latter had a similar HR to that for prevalent vertebral fractures (Table 2). Each SD decrease in standardized hip BMD was associated with a 38% increase in mortality; however, this was not statistically significant (p = 0.079). Greater height, weight, and alcohol use were all associated with a lower rate of mortality.
To determine whether the excess mortality associated with vertebral fractures could be explained by other skeletal and nonskeletal factors, the parameters that were significantly related to the mortality in the univariate analyses (p < 0.05) were included in multivariate models (Table 3). Complete baseline data for the parameters included in the multivariate model were available from a total of 583 patients (86.1%). The baseline characteristics of patients with or without complete data did not differ, except that the 94 patients without complete data were less likely to drink alcohol (52% versus 69%, p = 0.008) and had slightly lower serum albumin values (mean values 40.8 versus 42.0 U/liter, p = 0.001).
The adjustment for other health factors reduced the association between mortality and prevalent vertebral fractures in the whole study population (HR, 2.40; 95% CI, 0.93, 6.23; Table 3), and the association was no longer statistically significant. The multivariate Cox regression model revealed that in osteoporotic patients the increased mortality was significantly and independently associated with increased serum ESR (HR, 1.03; 95%CI, 1.00, 1.06), an increase in the number of concomitant treatments (HR, 1.35; 95% CI, 1.14, 1.60), and male gender (HR, 2.99; 95% CI, 1.02, 8.75; Table 3). An increased ESR was independently associated with an increased risk of mortality regardless of whether the analysis was confined to women only, postmenopausal osteoporosis, or secondary osteoporosis (Table 3). Interestingly, the point estimates of the HRs for mortality associated with prevalent vertebral fractures in all women, postmenopausal osteoporosis, and secondary (female and male) osteoporosis remained elevated (2.41, 3.19, and 1.79, respectively) to a similar magnitude to that observed in the whole study population, but didn't reach statistical significance in any of these subgroups.
This study showed that osteoporotic patients with radiographically defined prevalent vertebral fractures have a significant 4-fold increase in mortality compared with osteoporotic patients without vertebral fractures. Some of this excess mortality is related to other health-related factors, including number of medications, corticosteroid use, gender, serum albumin, ESR, renal function, weight, and age, and is reflected in the reduced HR after adjustment for these factors. It is worthy of note, however, that the point estimates of the HRs still suggested a persisting 2-fold increase in mortality risk. Our results are consistent with the findings of other studies of prevalent vertebral fractures in early and late postmenopausal women with or without low BMD.(7,8,14) For the first time, our analysis show that the excess mortality associated with vertebral fractures also occurs in secondary and/or male osteoporosis, as well as postmenopausal osteoporosis.
The observed excess in mortality among patients with vertebral fractures stresses the public health impact of these fractures. Three other studies have reported increases in mortality among individuals with prevalent morphometric vertebral fractures, ranging from 23% to 90%.(7,8,14) The 3-fold increased risk of mortality in postmenopausal osteoporosis observed in our study is intermediate to the risks observed in postmenopausal women with low bone mass recruited to the FIT study (1.6-fold increase)(8) and that of clinical vertebral fractures in the same study (8- to 9-fold increase).(15) The wide variation in the strength of association between mortality and vertebral fracture may occur for several reasons. First, some morphometric definitions of fracture may have a high noise-to-signal ratio because of relatively poor specificity for fracture that might diminish the relationship with mortality.(11,16) For example, in two population-based studies, the risk of mortality associated with prevalent morphometrically defined fractures was higher in the study using a more specific definition (1.9-fold)(7) than in the study using a less specific method (1.23-fold).(14) The presence of healthier controls in the population studies might have been expected to increase the association of fractures with mortality, and this may have contributed at least in part to the very high risk observed after clinical vertebral fractures in the FIT study, where the osteoporotic population had been derived by a mass screening of 6459 women.(15) An alternative but not mutually exclusive explanation may be that the risk of mortality after a vertebral fracture may be time dependent. Prospective clinically detected vertebral fractures seem to be associated with a marked increased risk of mortality, at least in the short term (first year after fracture),(15) although this was not supported by observations in the Dubbo study.(17) It is important to note, however, that no systematic radiographic assessment was undertaken in the latter study to allow accurate discrimination between prevalent and incident vertebral fractures. If the risk of mortality decreases with time, studies of prevalent fractures would be expected to show a reduction in this excess risk. Our study population, a typical clinical population referred for investigation and treatment of osteoporosis with a relatively high prevalence of vertebral fracture, is likely to include a higher proportion of patients with recent fracture events that stimulated their referral.
Our study cannot exclude a direct causal relationship between prevalent vertebral fractures and mortality. It is certainly possible that radiographic deformities are markers of underlying clinical diseases associated with aging and frailty. In support of this possibility, the excess in mortality associated with vertebral deformities was no longer significant after adjustment for potential confounding factors for mortality including age, serum albumin and ESR, poor renal function, alcohol intake, and other markers of chronic disease. A similar observation was reported within the longitudinal EPOS study, where the increased risk of mortality in women with vertebral deformities was no longer significant after adjustment for smoking, alcohol consumption, previous hip fractures, general health, body mass index (BMI), and steroid usage.(7) In contrast, two earlier prospective studies (8,14) indicated that the relationship of mortality to vertebral deformities was not explained by advanced age, health behavior, chronic medical conditions, low bone density, and poor health status.
Low hip bone density has been shown to be associated with increased mortality in other studies.(18–22) The 36% increase in mortality per 1 SD decrease in hip BMD in univariate analyses was similar than reported in other studies, although the association was not significant. It has been reported that the increased risk of mortality in subjects with low BMD or prevalent vertebral fractures is caused by certain disease progress such as cardiovascular disease,(18,20) pulmonary disease, or cancer.(6,14) In our study, 20 patients died from respiratory and cardiovascular disorders and 17 from other causes. Although the number of deaths was relatively small in our study, the results suggest, similarly to findings in the FIT study,(8) that the increased mortality seen in subjects with low bone mass and vertebral deformities may not be disease specific.
Vitamin D deficiency is a risk factor for low BMD and bone fractures.(23–25) At baseline, mean serum 25(OH)D levels were significantly lower in patients that subsequently died compared with those who remained alive. Despite a more marked impairment in renal function in the same patients, mean values of 1,25(OH)2D were similar in the two groups, suggesting compensatory hyperparathyroidism in the patients who later died. Values of parathyroid hormone (PTH) have not been measured in the current study, but it might be reasonable to assume that any increase in PTH may have contributed to enhanced bone turnover and fracture risk as reflected by the serum alkaline phosphatase activity and fasting urinary hydroxyproline excretion. However, the decrease in 25(OH)D levels was not significantly associated with increased mortality in univariate analyses.
The limitations of this study relate to a limited follow-up period and a relatively small number of events. Thus, for example, we cannot exclude the possibility that incident vertebral fractures are associated with increased mortality, and we couldn't determine whether the bisphosphonate intervention, which has been shown to reduce the risk of vertebral fracture in this study population,(26) could also impact on mortality. On a more positive note, the study included both men and women with secondary osteoporosis and women with postmenopausal osteoporosis. The baseline radiographs were taken from all participants, enabling us to ascertain the mortality risk in a population of osteoporotic patients regardless of whether they had clinically diagnosed fracture or not. Finally, there was an extensive data set of markers for skeletal and nonskeletal health assessed at randomization to enable us to further investigate the possible confounding factors for increased mortality associated with vertebral fractures.
We conclude that our results confirm earlier reports of an increased risk of mortality associated with radiographic vertebral fractures and extend those results to patients presenting clinically with osteoporosis. Other markers of poor health in osteoporotic patients could largely explain the excess of mortality, although a residual excess of mortality seems to persist in those with vertebral fractures. Further mechanism(s) by which vertebral fractures could be directly or indirectly associated with mortality requires more research.
We are grateful to Linda Reaney for her technical skills in performing the vertebral height measurements and to all members of the scanning and nursing departments who provided much needed assistance in conducting the study. This study has been funded by Leiras Oy, Finland.
- 11996 Mortality following fractures in older women. The study of osteoporotic fractures. Arch Intern Med 156:1521–1525., , ,
- 21993 Mortality and morbidity after hip fractures. BMJ 307:1248–1250., ,
- 31985 Epidemiology of osteoporosis and osteoporotic fractures. Epidemiol Rev 7:178–208., , ,
- 41991 Hip fracture incidence and mortality in New England. Epidemiology 2:116–122., , , , , ,
- 51992 Race and sex differences in mortality following fracture of the hip. Am J Public Health 82:1147–1150., , , , ,
- 61993 Population-based study of survival after osteoporotic fractures. Am J Epidemiol 137:1001–1005., , , ,
- 71998 Mortality associated with vertebral deformity in men and women: Results from the European Prospective Osteoporosis Study (EPOS). Osteoporos Int 8:291–297., , , , , , , , , , , , , , , , , , , ,
- 82000 Prevalent vertebral deformities predict mortality and hospitalization in older women with low bone mass. J Am Geriatr Soc 48:241–249., , , , , , ,
- 91996 Health status before and mortality after hip fracture. Am J Public Health 86:557–560.,
- 102001 Effects of clodronate on vertebral fracture risk in osteoporosis- a one-year interim analysis. Bone 28:310–315., , , , , , , , , , , ,
- 111993 The assessment of vertebral deformity - a method for use in population studies and clinical trials. Osteoporos Int 3:138–147., , , , , ,
- 121985 Vitamin D metabolism in patients intoxicated with ergocalciferol. Clin Sci 68:135–141., , ,
- 131986 Seasonal changes in the biochemical indices of vitamin D deficiency in the elderly: A comparison of people in residential homes, long-stay wards and attending a day hospital. Age Ageing 15:77–83., , ,
- 141999 Vertebral fractures and mortality in older women: A prospective study. Study of Osteoporotic Fractures Research Group. Arch Intern Med 159:1215–1220., , , , ,
- 152000 Risk of mortality following clinical fractures. Osteoporos Int 11:556–561., , , ,
- 161996 The assessment of vertebral deformity. In: GenantHK, JergasM, Van KuijkC (eds.) Vertebral Fractures in Osteoporosis. University of California San Francisco, San Francisco, CA, USA, pp. 215–233.,
- 171999 Mortality after all major types of osteoporotic fracture in men and women: An observational study. Lancet 353:878–882., , , ,
- 182000 Rate of bone loss is associated with mortality in older women: A prospective study. J Bone Miner Res 15:1974–1980., , , ,
- 191991 Non-trauma mortality in elderly women with low bone mineral density. Study of Osteoporotic Fractures Research Group. Lancet 338:355–358., , ,
- 201999 The association between low bone mass at the menopause and cardiovascular mortality. Am J Med 106:273–278., ,
- 211998 Bone mineral density is a predictor of survival. Calcif Tissue Int 63:190–196., , , ,
- 222001 Bone mineral density at the hip predicts mortality in elderly men. Osteoporos Int 12:259–265.,
- 231995 Vitamin D deficiency in older people. J Am Geriatr Soc 43:822–888.,
- 241999 Occult vitamin D deficiency in postmenopausal US women with acute hip fracture. JAMA 281:1505–1511., , , , ,
- 251998 Endogenous hormones and the risk of hip and vertebral fractures among older women. Study of Osteoporotic Fractures Research Group. N Engl J Med 339:733–738., , , , , ,
- 262000 Oral clodronate significantly reduces fracture risk in women with postmenopausal or secondary osteoporosis. J Bone Miner Res 15:S227., , , , , , , , , ,