| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Department of Internal Medicine (H.D., J.C.P.-S., B.O.-P., E.M., P.M., G.R., A.F.-P.), Division of Endocrinology and Nuclear Medicine, and Center for Medical Research (A.T., A.S.), Medical University of Graz, A-8036 Graz, Austria; and Division of Gastroenterology and Endocrinology (C.B.), Philipps University, D 35033 Marburg, Germany
Address all correspondence and requests for reprints to: Harald Dobnig, M.D., Professor of Internal Medicine, Department of Internal Medicine, Division of Endocrinology and Nuclear Medicine, Medical University of Graz, Auenbruggerplatz 15, A-8036 Graz, Austria. E-mail: harald.dobnig{at}meduni-graz.at.
 |
Abstract |
|---|
 Top Abstract IntroductionPatients and MethodsResultsDiscussionReferences |
|---|
Objective: The objective was to investigate whether quantitative bone ultrasound measurements and/or markers of bone turnover/metabolism help in predicting which patients will incur hip or nonvertebral fractures.
Design, Setting, and Participants: In this prospective study, mobile teams enrolled 1664 female patients from 95 nursing homes in Austria.
Main Outcome Measures: Calcaneal stiffness (n = 1117), radial speed of sound (SOS) (n = 1332), and phalangeal SOS (n = 1498) measurements were performed at baseline. Serum samples (n = 960) were analyzed for serum calcium and phosphate, 25 hydroxyvitamin D, PTH, osteocalcin, C-terminal telopeptide crosslinks, and osteoprotegerin (OPG). Patients were prospectively followed for hip and other nonvertebral fractures for 2 yr.
Results: A total of 117 hip fractures and 269 nonvertebral fractures developed during a mean observation period of 2 yr. Prevalence of vitamin D deficiency and secondary hyperparathyroidism was high. A history of a past fracture was significantly associated with a hazard ratio (HR) of 1.47 (95% confidence interval, 1.01–2.15) and 1.65 (1.26–2.16) for the development of hip and nonvertebral fractures, respectively. Cox regression analysis revealed a multivariate adjusted elevation in both hip [HR 1.30 (1.12–1.43)] and nonvertebral [HR 1.14 (1.02–1.25)] fracture risk for each SD decrease in calcaneal stiffness. Patients in the lowest quartile for calcaneal stiffness Z-score had 2.5 and 1.2 times higher rates of hip and nonvertebral fractures when compared with patients in the highest quartile. Fracture rates were not statistically associated with baseline radial or phalangeal SOS measurements or with serum osteocalcin, C-terminal telopeptide crosslinks, and OPG concentrations. When adjusted for bone mass, higher serum OPG levels were associated with fewer hip as well as nonvertebral fractures [HR 0.85 (0.73–0.99) and 0.89 (0.80–0.99) per increment of 1]. Higher serum phosphate levels indicated an increased hip [HR 1.54 (1.07–2.21)] and nonvertebral fracture risk [HR 1.40 (1.10–1.78) per increase of 1 mg/dl]. Body mass index was protective of hip fractures [HR 0.94 (0.90–0.98) per increase of 1] as well as medication with acetylsalicylic acid [HR 0.59 (0.36–0.95) for hip and 0.72 (0.52–0.99) for nonvertebral fractures]. In contrast, current use of glucocorticoids [HR 5.65 (1.77–18.0)] and opiates [HR 1.85 (1.18–2.92)] exerted a negative effect on prospective hip fracture risk.
Conclusion: Calcaneal stiffness measurements proved to be useful in predicting hip fractures and to a lesser extent nonvertebral fractures in nursing home residents. Radial and phalangeal bone ultrasound measurements and baseline markers of bone turnover, however, were not indicative of future fracture risk in this population.
|
Introduction |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionReferences |
|---|
Two studies evaluated the prediction of nonvertebral fractures in institutionalized individuals using bone mineral density (BMD) measurements at the hip (4) and radial site (4, 5). In one study including 252 women and men with a mean age of 88 yr, both hip and radial BMD were predictive of fractures, and subjects in the lowest age-specific quartile of femoral neck BMD had over four times the incidence of nonvertebral fracture as those in the highest quartile (4). In women, however, radial BMD was not associated with fractures. In a second study that enrolled 1427 patients with a mean age of 85 yr, the hazard ratio of any osteoporotic fracture was 1.4 for each SD decrease in radial BMD. Hip fracture prediction was not significant in the overall study cohort, but among residents independent in transfer, the hazard ratio was 2.5 (95% confidence interval, 1.5–4.2) when patients below and above median were compared. The obvious disadvantage with DXA technology, however, is the requirement for transportation, patient agility, and cooperation, prerequisites that often cannot be met by nursing home patients.
Because QUS and DXA measurements appear to be equally predictive of hip fracture risk in community-dwelling individuals (3, 6) and because QUS devices are easily portable, application of this technique in a nursing home environment could provide valuable additional information in terms of fracture risk assessment. A recent case-control study with 151 frail elderly patients with low trauma fractures reported a significant difference in calcaneal broadband ultrasound attenuation and speed of sound (SOS) measurements (7). On a prospective basis, QUS measurements have not, however, been validated in a population of institutionalized patients. Studies of this special cohort are so far limited to reports of noticeably low average calcaneal and phalangeal SOS T-scores of –4.8 and –5.0 (8). Using QUS, almost all institutionalized women were proposed to be osteoporotic (8). In nursing home residents, ultrasound measurements at the calcaneus were further shown to decline with age (9, 10, 11) and correlate with body mass index (BMI), serum calcium, PTH, alkaline phosphatase levels (10) as well as quadriceps strength (9). In patients with a history of a past fracture, the mean calcaneal stiffness index was shown to be 18.9% lower than in nonfractured controls (11).
Besides bone mass, bone turnover plays a significant role in modulating fracture risk (12, 13), and this may especially be the case in institutionalized patients where instances of low bone mass often coincide with high bone turnover that is fostered by immobility (14), malnutrition (15), and vitamin D deficiency (16).
Accordingly, the objectives of the present study were to investigate whether baseline peripheral QUS or bone turnover measurements may help, alone or in combination, to predict hip and other nonvertebral fractures in nursing home residents.
|
Patients and Methods |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionReferences |
|---|
The recruitment process proceeded as follows. Referring to the approval granted by our institutional ethics committee, we asked politicians responsible for "nursing home affairs" on a country level for a letter of recommendation for the present project. We then presented the study to the manager in each nursing home as well as to the persons responsible for home care of the residents. Posters were printed and posted at each nursing home; flyers were distributed to patients and their relatives with information on the problem of osteoporosis and the intention of the study. Often, there was also a general presentation of the project for interested residents where we also invited attending physicians and relatives to participate and ask questions. All involved parties responded very positively to our appeal for participation in this study. Except for five smaller institutions, all nursing homes chose to participate. In the next step, we asked nurses (who knew their residents the best) together with the local attending physicians to prepare a list of persons who in their opinion would fulfil the criteria of being able to sign an informed consent form and have suitable life expectancy. The list of possible candidates was then given to our study teams upon their arrival at the study site. They screened the medical charts of those individuals, obtained informed consent, and drew a small blood sample that was analyzed for the parameters on which inclusion in or exclusion from the study depended. Upon a second visit, blood sampling and bone QUS measurements were performed on subjects who agreed to those procedures.
Follow-up
Depending on the size of the nursing home, at least two staff nurses were responsible for reporting fracture incidents by fax to the study center at the beginning of each month. For 2 yr, all participating centers were additionally visited by one and the same physician on a regular basis. This physician searched the medical charts for new fracture cases and radiology reports. All fatalities that had occurred since the previous visit were also recorded.
All subjects were followed from their baseline visit until death or first hip fracture or last follow-up visit, whichever occurred first. Patients without fractures during the observation period formed the "control" (CTR) group, and patients with hip fractures or any kind of nonvertebral fracture formed the "hip fracture" or "nonvertebral fracture" group, respectively.
Bone ultrasound measurements
QUS measurements were performed using Achilles Express (GE LUNAR Corp., Madison, WI) at the calcaneus (primarily trabecular bone) and Sunlight Omnisense ultrasound bone sonometer (Sunlight Ultrasound Technologies Ltd., Rehovot, Israel) at the distal one third of the radius (primarily cortical) and the proximal phalanx of the third finger (cortical) on the nondominant side (17). The Achilles Express system provides a normative database derived from a U.S. female Caucasian population covering an age range between 20 and 100 yr (Achilles Express operator's manual). The reference database for the Omnisense system covers an age range between 20 and 89 yr and is derived from Caucasian women residing in Israel (18). Three Achilles Express and three Omnisense systems were used by the study teams at a given time. Both systems were cross-calibrated at the beginning of the study and ran stable over the time used.
Handling and processing of blood samples and laboratory analysis
Nonfasting blood samples were drawn between 0900 and 1200 h, and specimens were allowed to clot at room temperature for 15 min. They were then immediately centrifuged at the nursing homes using a Heraeus Biofuge (Kendro Laboratories, San Juan, Puerto Rico), aliquoted, and placed in an ice-water bath before they were finally transferred to the electric cooler. When the study teams returned to the study center, serum and plasma samples were immediately stored at –70 C until further analysis.
Serum 25-hydroxyvitamin D was measured by RIA (Immunodiagnostic Systems, Boldon, UK) after extraction. Sixty-two original samples together with 40 samples from patients participating in an osteoporosis study at our department were also analyzed for 25-hydroxyvitamin D by HPLC. Ten of these patients were taking additional vitamin D3 supplements. Overall, there was a correlation of r = 0.96 and R2 = 0.92 between the Immunodiagnostic Systems RIA and the HPLC technique. More details regarding measurements of serum parameters are given elsewhere (17).
Statistical analyses
Power analysis was performed using an online service of DSS Research, Inc. (Fort Worth, TX) (see www.dssresearch.com). All other analyses were conducted using SPSS version 12.0 (SPSS Inc., Chicago, IL). Our prestudy assumption of hip fracture incidence was 7% over the observation period. With a final observed hip fracture rate of 7.3 to 7.5% for the three QUS groups, the statistical power was 90% (calcaneus), 85% (radial), and 95% (phalanx) to detect at least a Z-score difference of 0.5 between the hip fracture and CTR groups.
Normality of data was checked by Kolmogorov-Smirnov and Shapiro-Wilk testing for the overall population as well as for both fracture groups. All parameters were normally distributed. Potential differences between the CTR group and the hip or nonvertebral fracture group were calculated using the t test and, in the case of categorical variables, the 
2 test. The significance of parameters to predict either hip or nonvertebral fractures was calculated by Cox proportional regression analysis with adjustment for important confounders. Hazard ratios estimated from Cox regression models are reported as relative risks, with corresponding 95% confidence intervals (CI). For subgroup analyses, i.e. the patient group using opiates, analysis of covariance was used to look for differences in fracture rates or bone mass, with important clinical or biochemical variables set as covariates. Multiple linear regression models were applied to quantify the explainable percentage of variation of bone turnover and bone ultrasound measurements by the determinants included. These variables had to be significantly correlated to the dependent variable in univariate correlation analysis. P values < 0.05 were considered significant. Data are presented as mean (±SD).
|
Results |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionReferences |
|---|
The mean observation period was 24 months, and 594 (35.6%) patients died during that period. The total number of patient years was 3424. During the observation period, 7.0% of the patients acquired a hip fracture (annualized ratio, 3.4%), and 13.8% (annualized ratio, 6.7%) acquired at least one nonvertebral fracture. Thirty-three fractures were not considered to be of osteoporotic nature (24 fractures of the skull, the remainder fractures of the toes and fingers) and thus were not included in the analysis. A total of 269 fractures were considered osteoporotic and were sustained by 231 residents. Thirty-eight persons had more than one nonvertebral fracture; three women had two hip fractures each. The respective fracture categories are as follows: hip, 120; humerus, 33; radius, 31; pelvis, 26; rib, 24; other fractures of upper extremity, 16; other fractures of lower extremity, 13; vertebral, 4; and clavicle, 2. Baseline characteristics of patients with hip and nonvertebral fractures are given in Table 1
. Significantly more patients in both fracture groups were using a walking aid than were CTRs.
|
summarizes relevant comorbidities and classifies medications according to important pharmacological groups. Significantly more patients in both fracture groups were treated with opiates; in hip fracture cases, psychotic disorders as well as corticoid treatment were also more frequent than in CTR subjects.
|
Phalangeal SOS measurement could be performed in all patients who chose to participate in the bone ultrasound protocol (90%). Additional radial and calcaneal QUS measurements were obtained from 89 and 75% of them, respectively. Reasons for drop-outs were acroedema, leg ulcers, significant varicosis, or in a few cases, instable measurements. The histogram of the measurements for all three methods is given in Fig. 1. It demonstrates that despite the heterogeneous patient cohort, different degrees of mobility, various comorbidities, and a high prevalence of medicalization, the data were normally distributed, although they did stretch over a broad range of 9 to 12 T-score units. Ninety-seven percent of all patients had a T-score below 0 at all three sites and 69% (calcaneus), 62% (radius), and 71% (phalanx) had a measurement below a T-score of –2.5 or less. Z-score measurements for CTR and fracture groups are given for all three sites in Fig. 2. The only significant difference could be obtained at the calcaneus. Here, the mean difference in Z-score between the CTR group and the hip fracture group was 0.51, and for the nonvertebral fracture group, 0.23.
|
|
Table 3 gives all the data that relate to calcium and vitamin D metabolism as well as to bone turnover. Small, significant differences were found for mean serum phosphate concentration, which was higher in both fracture groups. The advanced age of the nursing home residents was also reflected in low creatinine clearance rates. According to the chronic kidney disease (CKD) Kidney Disease Outcomes Quality Initiative guidelines, 5% of the patients fulfilled CKD stage 2 [glomerular filtration rate (GFR) 60–89 ml/min/1.73 m2], 68% CKD stage 3 (GFR 30–59) and 27% CKD stage 4 (GFR 15–29) criteria. Vitamin D deficiency was very common, with 15% of patients having 25-hydroxyvitamin D levels below 5 ng/ml, 74% below 10 ng/ml, and 88% below 15 ng/ml; only 7% had values of 20 ng/ml or above. PTH levels were elevated above 65 pg/ml in 50% of the patients, and one in every five patients had levels above 100 pg/ml. As a consequence, bone turnover was also elevated, and 65% of all women had osteocalcin levels above the upper limit of normal for postmenopausal women. Osteocalcin was significantly correlated to ß-C-terminal telopeptide crosslinks (CTx) levels (r = 0.66; P < 0.0001), serum creatinine clearance (r = –0.29; P < 0.0001), PTH (r = 0.55; P < 0.0001), 25 hydroxyvitamin D (r = –0.15; P < 0.0001), and serum osteoprotegerin (OPG) levels (r = 0.12; P < 0.0005).
|
Results of Cox regression analyses are given separately for hip as well as for nonvertebral fractures and are depicted in Table 4.
|
All parameters that were significantly correlated to bone ultrasound measurements or serum osteocalcin levels in univariate analyses were included in a multiple regression analysis and are given in Tables 5 and 6. From 8 to 15% of the variance of QUS measurements could be explained by the parameters given. Serum osteocalcin was correlated to all three measurement sites to a similar degree; however, body weight and muscle strength were only correlated to calcaneal measurements.
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionReferences |
|---|
For patients belonging to the lowest and highest quartile of calcaneal stiffness measurements, the absolute risk of hip fracture was 10.9 and 4.3%, respectively, over the observation period. The gradient for nonvertebral fracture incidence, however, was less steep, with 18.0% fractures for the lowest and 12.3% for the highest quartiles. The hazard ratio of 1.30 per 1 SD decrease is lower than the recently reported 1.9, which was based on an analysis of four mainly population-based prospective trials (3). In another cohort of elderly female and male participants living in homes for the elderly, the hazard ratios for calcaneal broadband ultrasound attenuation and SOS measurements were 2.3 and 1.6, respectively, and were thus higher as well (19).
Our results suggest that the fracture risk association with decreasing calcaneal bone mass is diminished in institutionalized patients. A reasonable explanation for this flattening of the bone mass/fracture relationship may be a change in bone quality or an increased likelihood of falls. At this stage, when average bone mass is already low, falls apparently determine outcome to a larger extent than do bone mass variations. Recently, Johnell et al. (2) reported a similar decrease in the risk gradient for hip fractures by DXA in an analysis comprising 12 cohorts of community-dwelling women. In line with their conclusions, we feel that the absolute increase in hip fractures across calcaneal stiffness quartiles still represents a significant clinical difference that could aid in defining patients at very high risk along with other known tools to predict falls and fractures (20); however, to set up screening programs for QUS measurements in nursing homes on a large scale does not seem to be justified on the basis of our data.
The second main finding of this study was that neither osteocalcin, ß-CTx, nor OPG levels were helpful in predicting hip or other nonvertebral fractures. This was a somewhat unexpected result because hypovitaminosis D and secondary hyperparathyroidism usually respond well to appropriate calcium and vitamin D supplementation (16), and lowering bone turnover has been shown to be associated with lower fracture risk in institutionalized patients (21, 22). It is unlikely that diminished renal function contributed significantly to this negative finding because analyses were adjusted for creatinine clearance rates. We cannot exclude a significant effect of timing of blood sampling on fracture prediction. On the other hand, bone turnover markers must still have been high in these patients even with possible food-related decreases in ß-CTx levels. A lack of correlation between bone turnover and fracture risk in such an environment may thus still present an important finding, especially because variations in osteocalcin levels have not been shown to be significantly influenced by food intake (23, 24, 25, 26). It may be speculated that the same above-mentioned argument, namely the predominance of falls, may serve as a likely explanation for the negative association between bone turnover and fracture development.
An interesting finding of the study was that serum phosphate levels were significantly higher in patients who developed a fracture even after multivariate adjustments including all relevant parameters of mineral metabolism. There is little information in the literature on phosphate metabolism and possible influences on fracture risk, and serum phosphate levels often are not even given. Lau et al. (27) studied 200 patients with hip fractures and reported higher serum phosphate levels when compared with 427 controls. In our study, phosphate was a small independent contributor to bone turnover (partial R = 0.07; P = 0.01) but was not correlated to bone mass. Phosphate, on the other hand, is part of an important buffer system, and higher phosphate levels may be indicative of changes in acid-base homeostasis that in turn may adversely affect bone quality. Together with carbonate, phosphate is not only released from bone during metabolic acidosis but is also increasingly absorbed from the intestine (28). Our work cannot provide a conclusive answer on whether higher phosphate levels may also have influenced 1-
-hydroxylase activity and thus modulated 1,25-dihydroxyvitamin D levels.
Apart from classic bone turnover markers, serum OPG levels were also determined in this study. Various sources contribute to serum OPG concentrations, but a significant amount has to originate from bone tissue because significant correlations with bone turnover markers have been shown in this as well as in other studies (29, 30). Higher serum OPG levels conferred a protective effect on hip as well as nonvertebral fracture risk in institutionalized patients, but only after adjustment for bone mass. Previous studies either failed to show an association between serum OPG levels and fracture status (31) or, similar to this study, found an association with vertebral fracture status after adjustment for bone mass (29). In our opinion, more attention should be given in future studies to whether bone mass or bone turnover rate is an important dimension when OPG levels are to be interpreted.
The finding that low BMI confers a substantial risk for hip fractures (32) could also be confirmed. Each decrease in BMI of 1 U increased hip fracture risk by 6%. The effect of BMI on fracture risk in the present study, however, was largely dependent on bone mass because when calcaneal stiffness was included as a confounder, the significant correlation with BMI was no longer present. This is identical to what others reported using BMD in a large analysis of 12 prospective population-based cohorts (32). The finding that a history of a past fracture was a significant predictor for both hip and nonvertebral fractures is largely of confirmatory nature in this patient cohort, also in terms of effect size (33).
Three groups of medications could be extracted from the medical charts that, to our knowledge, have not yet been reported to influence fracture risk in institutionalized patients. Glucocorticoid-treated patients had a five times greater risk of hip fracture. This finding extends the well-known adverse effects of steroids to this special population as well. An increased fracture risk could be seen in opiate users and supports a recent analysis from a nationwide registry including more than 10,000 opiate users (34). Subgroup analyses of our study revealed that opiate users were significantly less mobile and had less bone mass, a 24% increase in osteocalcin, and a 13% increase in PTH levels, circumstances that point to pain-related effects, increased bone turnover, and bone loss. Patients taking acetylsalicylic acid preparations had 40 and 28% decreases in hip fracture or nonvertebral fracture risk, respectively, and a multivariate adjusted increase in mean calcaneal Z-score values of 0.32 (P = 0.002) compared with nonusers. Similar increases were also found for nonsteroid antiinflammatory drug and aspirin users in the Study of Osteoporotic Fractures and the Rancho Bernardo study (35, 36) and are likely due to inhibition of cyclooxygenase activity that modulates the effects of prostaglandins on bone resorption (37).
Interesting from a pathophysiological point of view is that serum PTH concentration in this vitamin D-deficient cohort made up 67% of the total explainable variation of bone turnover and was thus by far the most important determinant. The remainder was explained by parameters reflecting glucose metabolism, liver, and renal function. Consequences of high turnover on bone loss must be important over time, otherwise a single osteocalcin value would not be correlated to bone mass in a cross-sectional analysis.
In this study, various potentially important predictors of bone strength, namely bone mass and bone turnover, have for the first time been investigated comprehensively for their potential to predict fractures outside a hospital setting. Other important strengths of the study are its size, the number of fracture endpoints, and thus its associated statistical strength. Almost all patients agreed to participate in the QUS study; however, in only 75% of the patients who had a phalangeal measurement could a calcaneal measurement also be performed. Thus, in one quarter of the patients, technical limitations applied, mostly due to edema, varicosis, or ulcerative disease.
This study demonstrates that despite a very high overall fracture rate, calcaneal stiffness measurements are correlated to hip and to a lesser extent to nonvertebral fracture development. In institutionalized patients, the bone mass/fracture relationship as well as bone turnover level becomes less important for bone strength, probably because falls are the predominant cause of fractures. Applied within a case-finding strategy, calcaneal QUS measurements may be a useful additional tool to define patients at the highest risk for fractures. The results, however, do not in our opinion justify large-scale screening programs in institutions.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Disclosure Statement: The authors have nothing to disclose.
First Published Online February 20, 2007
Abbreviations: BMD, Bone mineral density; BMI, body mass index; CI, confidence interval; CKD, chronic kidney disease; CTR, control; CTx, C-terminal telopeptide crosslinks; DXA, dual energy x-ray absorptiometry; GFR, glomerular filtration rate; OPG, osteoprotegerin; QUS, quantitative bone ultrasound; SOS, speed of sound.
Received September 25, 2006.
Accepted February 14, 2007.
|
References |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionReferences |
|---|
Read all eLetters
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
