Calcified Tissue International

, Volume 73, Issue 1, pp 15–20

Low Bone Mineral Density in the Hip as a Marker of Advanced Atherosclerosis in Elderly Women


    • Center for Clinical and Basic Research A/S, Aalborg and Ballerup
  • Y. Z. Bagger
    • Center for Clinical and Basic Research A/S, Aalborg and Ballerup
  • C. Christiansen
    • Center for Clinical and Basic Research A/S, Aalborg and Ballerup
Clinical Investigations

DOI: 10.1007/s00223-002-2070-x

Cite this article as:
Tankó, L., Bagger, Y. & Christiansen, C. Calcif Tissue Int (2003) 73: 15. doi:10.1007/s00223-002-2070-x


Previous studies indicate that low bone mineral density (BMD) in the hip is a useful predictor of cardiovascular mortality among the elderly. The objective of this study was to investigate whether low hip BMD is directly associated with the severity of atherosclerosis. The per-protocol population consisted of 963 women aged 60-85 years. Study variables were aortic calcification (AC) graded on lateral lumbar radiographs, BMD at various anatomic sites (distal radius, lumbar spine, proximal femur) measured by DXA, information on various risk factors, and medical history. After adjustment for age, BMD at the proximal femur, but not at the radius or spine BMD, showed statistically significant association with the severity of AC (r = −0.12−17, P < 0.001). Age, years since menopause, BMI, level of education, current and previous smoking, and weekly fitness activity were significant common risk factors (all P < 0.05) with contrasting influence on AC and hip BMD. In a multiple regression model, AC contributed significantly and independently to the variation in hip BMD (β = −0.10, P = 0.004). Impaired blood flow represented by 40 women with documented history of intermittent claudication was not an independent contributor and did not alter the association between AC and hip BMD. However, AC and demineralization in the hip was particularly severe in women with intermittent claudication accompanied by a higher prevalence of coronary heart disease compared with age-matched controls (all P < 0.001). In conclusion, severe osteoporosis in the hip may indicate advanced atherosclerosis and thereby an increased risk for not only hip fractures but also for coronary heart disease. The results further emphasize that osteoporosis in the hip and peripheral vascular disease are linked by common risk factors and pathomechanisms.


Aortic calcificationCardiovascular diseaseElderly womenHip BMDRisk factors

Several studies indicate that increased rate of bone loss, low bone mass, and osteoporotic fractures are associated with increased cardiovascular morbidity and mortality among the elderly [1, 2]. A recent study by Johansson et al. [3] noted that low bone mineral density (BMD) in the calcaneus is an independent predictor of mortality among elderly men and women and its predictive value is even better than that of blood pressure and serum cholesterol. Trivedi and Khaw [4] recently demonstrated that low BMD at the hip was also a useful predictor of cardiovascular mortality in elderly men. However, limited information is available on whether low hip BMD is directly related to the severity of atherosclerosis, particularly regarding elderly women.

Studies on men suggest that severe atherosclerosis, by impairing blood circulation in the lower extremity, may alter bone metabolism in the hip that may culminate in osteoporosis [5]. In line with this, several recent studies found significant cross-sectional association between the ankle/arm index and BMD at the femoral neck [6, 7]. However, Vogt et al. [7] found significant age-adjusted association between the ankle/arm index and BMD not only at the hip, but also at the distal radius. This apparent contradiction may, on one hand, point out the importance of other linking factors. On the other hand it may indicate that the ankle/arm index is not a specific indicator of atherosclerosis in the lower extremity. In support of the latter notion, studies confirm that the ankle/arm index is an indicator of overall cardiovascular health, a predictor of not only cardiovascular but also cancer-related mortality. Thus, application of this hemodynamic parameter in cross-sectional settings may result in significant but artifactual correlations reflecting confounding with other factors. Due to these uncertainties, the role of atherosclerosis and related impaired blood flow in the pathogenesis of hip osteoporosis cannot be considered fully clarified.

The present study investigated whether application of another surrogate of peripheral atherosclerosis could offer advantages in obtaining further insights into the association between hip BMD and cardiovascular disease. Therefore, we investigated the independent association between BMD at various sites and the severity of atherosclerosis in a large population of postmenopausal women aged 60-85 years using aortic calcification as a marker of cardiovascular disease.[8, 9, 10, 11]

Subject and Methods

The study population consisted of 1456 postmenopausal women aged 60-85 years, who participated in a follow-up examination of women previously screened for risk factors for osteoporosis at the Center for Clinical and Basic Research in Aalborg, Denmark (The Prospective Epidemiological study [PERF] focusing on risk factors for osteoporosis and cardiovascular disease) [12]. Those who reported any chronic metabolic bone disease (n = 36) or use of medication (n = 457) known to directly influence bone mass (HRT, SERM, bisphosphonate) were excluded from the study. Demographic characteristics and risk profile of the remaining per-protocol population (n = 963) is described in Table 1. All women gave informed consent to participation and the study was carried out in accordance with the Helsinki Declaration II and the European Standards for Good Clinical Practice. The Regional Ethical Committee approved the study protocol.

Table 1

Characteristics of the study population stratified into quartiles of hip BMD




Age (years)

71.3 ± 5.3


BMI (kg/m2)

26.1 ± 3.8

















Systolic BP (mmHg)

149 ± 24


Diastolic BP (mmHg)

81 ± 12


Treated hypertension (%)



Cholesterol (mmol/l)

6.59 ± 1.07


Triglyceride (mmol/l)

1.46 ± 0.69


Glucose (mmol/l)

5.31 ± 1.21


White blood cell count ( × 109/l)

6.08 ± 1.68


Diabetes mellitus (%)



Daily coffee (%)



Daily alcohol (%)



Daily milk (%)



Daily vitamin D(%)












Data shown are mean ± SD

Anthropomorphic Measures, Blood Pressure, Lifestyle Factors, and Medical History

Body weight and height were measured to the closest 0.1 kg and 0.1 cm, respectively, on women wearing light indoor clothes and no shoes. Body mass index (BMI) was calculated as weight (kg) divided by height squared (m2). Arterial blood pressure was measured using the same digital blood pressure monitor (UA-777, A&D Instruments LTD, Oxford, UK). Information on level of education (primary/secondary/university), smoking habits (current and previous smoking), daily alcohol and coffee consumption, weekly walking activity of more than 30 min (1: never, 2: once, 3: two to five times, 4: more than five times), weekly fitness activities (1: never, 2: once weekly, 3: twice weekly, 4: more than twice weekly), presence of treated diabetes mellitus, hypertension, or clinically manifest cardiovascular disease (angina pectoris, myocardial infarct, intermittent claudication) were gathered during personal interviews using standardized questionnaires. The reported history of manifest cardiovascular disease was confirmed by hospital discharge summaries.

Metabolic Risk Factors

Serum glucose, total cholesterol, and triglycerides were determined from fasting blood samples using the same automatic blood analyzer (Vitros-250, Johnson & Johnson, Taastrup, Denmark). The Vitros-250 analyzer performs distinct biochemical analysis based on enzymatic assays. White blood cell count (WBC) was determined by Sysmex K-1000 (Sysmex Corporation, Kobe, Japan), which is a fully automated quantitative hematology analyzer. Measurements were carried out according to the manufacturer’s recommendation.

Bone Mineral Density Measurements

Bone mineral density in the left hip (total hip, trochanter, femoral neck, intertrochanteric area) and the lumbar spine (L1-L4) were measured by DXA using a Hologic QDR 4500A scanner (Hologic Inc., Waltham, Massachusetts, USA, software version 9.03D). Bone mineral density in the distal forearm was measured by a DTX200 arm scanner (Osteometer MediTech A/S, Rødovre, Denmark). To ensure reproducibility of the measurement, a control phantom scanning was performed on a daily basis. Precision error of the QDR4500A scanner for BMD measurements is below 1% [13].

Grading of Aortic Calcification

Aortic calcification was graded on lateral lumbar radiographs as previously described by Kauppila et al. [14]. Calcified deposits in the lumbar aorta adjacent to each lumbar vertebra (L1-L4) were assessed separately for the anterior and posterior wall of the aorta using the midpoint of the inter-vertebral space as the boundaries. In the affected segments score (scale 0-4), the number of aortic segments that showed any calcification was calculated. In the anterior and posterior affected score (scale 0-8) the number of aortic segments, both anterior and posterior, which showed calcification, were calculated. Finally, in the antero-posterior severity score (scale 0-24), the scores of individual aortic segments both for the anterior and posterior walls were summed. The same investigator, who was blinded for all other results of the individual participants, carried out the evaluations. Intra-rater correlations were in the range of r = 0.92–0.98, similar to published results [14].

Data Analyses

Data shown are means ± SEM, unless otherwise indicated. Statistical analysis was carried out using the SPSS data analysis software (version 10.01, SPSS Inc, Chicago, USA). Since the severity score of AC was not normally distributed, Spearman rank order correlation was used to determine the bivariate associations between AC or hip BMD and the selected risk factors. Associations independent of age were determined using partial regression analysis. Univariate general linear models determined age-adjusted estimates of BMD as a function of the severity of AC shown in Figure 1. The independent correlations between BMD and AC were also tested in multiple regression models, including all potential confounders as independent variables. Means of women with and without intermittent claudication were compared using Student’s two-tailed t-test for unpaired observations or by Mann-Whitney test. Correlations and differences were considered significant, at P < 0.05.
Figure 1

Age-adjusted associations between the severity of aortic calcification and bone mineral density at various sub-regions of the proximal femur.


Demographic characteristics and frequency of potential risk factors in the per-protocol population are described in Table 1. The mean severity of AC was characterized by an anterior-posterior severity score of 3.94 ± 4.39 (n = 963). There was a significant direct linear relationship between age and the severity score of AC (P < 0.001). In contrast, age showed a significant inverse relationship with BMD at the distal radius and proximal femur, including all anatomic sub-regions (all P < 0.001). No significant association was, however, noted with BMD at the lumbar spine.

Table 2 indicates the simple and age-adjusted correlations between the severity score of AC and BMD at the different anatomic sites. The severity score of AC showed statistically significant negative correlation with BMD at the distal radius and various sub-regions of the proximal femur (P < 0.001). However, after adjustment for the influence of age, only the association with the various sub-regions of the proximal femur continued to show statistical significance (all P < 0.001). Figure 1 also illustrates the age-adjusted associations between the severity of AC and BMD in various sub-regions of the proximal femur.

Table 2

Simple and age-adjusted correlations between aortic calcification and BMD at various anatomic sites

Anatomic site

Bivariate correlation

Age-adjusted correlation


Distal radius




Lumbar spine (L1-L4)




Total hip




Femoral neck




Intertrochanteric area








Values indicate Spearman’s rho (bivariate correlation) and partial correlation coefficient (age-adjusted correlation) obtained from 963 postmenopausal women

In search of common pathomechanisms underlying the apparent direct relationship between low BMD in the proximal femur and peripheral atherosclerosis, we analyzed the correlation of AC and total hip BMD with various potential risk factors. Age, years since menopause, BMI, current and previous smoking, level of education, weekly fitness activity (but not weekly walking activity), and history of intermittent claudication were all common correlates (all P < 0.05) (Table 3). The correlations indicated opposing influence on the severity score of AC and BMD at the hip. In a multiple regression model, including all potential confounders as independent variables, age, BMI, glucose, WBC count, level of education, and the severity score of AC, were independent contributors to the variation in total hip BMD (Table 4). Impaired blood flow to the lower extremity (represented by 40 women with a history of intermittent claudication) did not contribute independently to the variation in hip BMD, and did not influence the association between AC and hip BMD. Nevertheless, women with claudication intermittens showed more severe AC (8.85 ± 0.80 vs. 3.63 ± 0.14, P < 0.001) and significantly lower hip BMD (0.743 ± 0.018 vs. 0.778 ± 0.004, P < 0.001) compared to age-matched controls (n = 801). In addition, the prevalence of coronary heart disease was also higher in the former group: angina pectoris (25% vs. 6.5%, P < 0.00l) and MI (12.5% vs. 2.9%, P < 0.001).

Table 3

Common correlates of aortic calcification and total hip BMD


Aortic calcification

Total hip BMD






Age at menopause












Current smoking




Previous smoking




Weekly fitness activity




Intermittent claudication




Values indicate bivariate correlation coefficients (Spearman’s rho) obtained from 963 postmenopausal women

ap < 0.001, bp < 0.01, cp < 0.05

Table 4

Significant contributors in a multiple regression model for the prediction of total hip BMD in elderly women aged 60-85 years (n = 963)


Standardized coefficient





< 0.001




< 0.001














Diabetes mellitus




Aortic calcification





0.528 (0.10)

< 0.001


Additional adjustment was made for age at menopause, serum cholesterol, triglyceride, daily coffee, alcohol, milk and vitamin D consumptions, daily walking activity, weekly fitness activity, current and previous smoking, systolic and diastolic pressure, heart rate, treated hypertension, history of intermittent claudication


The main finding of the present study is that low BMD in the hip and atherosclerosis are not merely age-related processes, but are also linked by common risk factors and pathomechanisms. Thus, these observations provide further support for the notion that women with severe demineralization at the hip are at increased risk for morbidity and mortality related not only to hip fracture, but also to cardiovascular disease.

Results of the study indicated that AC as well as bone loss from the hip and distal radius progresses simultaneously with advancing age. The lack of association with spine BMD can be explained, at least in part, by “methodological” pitfalls related to DXA measurements in the elderly. Spine BMD in this population is often overestimated due to the presence of AC in the lumbar aorta and/or concomitant presence of osteoarthritis [15, 16]. When adjusting for the influence of age, the association between the severity score of AC and BMD turned out to be site-specific for the proximal femur. In contrast to our finding, Vogt et al. [7], who used the ankle/arm index as an estimate of peripheral vascular disease, found a significant correlation with BMD at the distal arm as well. Since AC is a more direct and localized estimate of atherosclerosis than the ankle/arm index, often considered as a marker of cardiovascular health in general terms, our findings seem to confirm the authors’ concern that this latter correlation is artifactual and probably reflects confounding with other variables [7]. Hence, the results suggest that peripheral vascular disease and bone metabolism are site-specifically linked in the proximal femur, suggesting common pathomechanisms.

In search of the linking pathomechanisms that may explain this direct association, we investigated the correlation of AC and hip BMD with various potential risk factors. Our results, in accordance with previous studies [10, 11, 17], pointed out that age, BMI, level of education, smoking habits, and physical activity are risk factors that may simultaneously promote or inhibit atherosclerosis and demineralization in the hip. However, when including all of these risk factors in a multiple regression model, the independent association of the severity score of AC with BMD at the hip suggests that the association cannot be explained by these risk factors alone.

Based on the site-specificity of the association, it is tempting to speculate that AC may, to some extent, influence blood flow to the distal regions or, alternatively, may mirror atherosclerosis of the more distal muscular arteries affecting blood supply to the hip in a more local fashion. The plausibility of this assumption is supported by clinical investigations that directly measured blood flow in patients with aortic/peripheral atherosclerosis [18, 19]. Another argument supporting the role of impaired blood flow is that in cases of asymmetrical peripheral arterial disease, bone mineral content in the hip is lower in the affected compared to the unaffected leg [5]. Collectively, these results indicate that impaired blood flow to the lower extremity due to or reflected by severe atherosclerosis in the lumbar aorta may, at least to some extent, contribute to the complex multi-factorial pathogenesis of osteoporosis in the hip.

However, since impaired blood flow and the common risk factors could not fully explain the association between AC and hip BMD, it is conceivable to search for other common pathomechanisms possibly linking the two processes. An evident candidate is estrogen deficiency, known to accelerate bone loss and increase the incidence of cardiovascular disease in postmenopausal women [20]. Our observation that years since menopause were significantly associated with both the severity of AC and BMD at the hip seems to be in line with this notion. While cross-sectional studies are not unanimous [21, 22, 23], longitudinal studies clearly indicate that those showing the greatest magnitude of bone loss demonstrate also the greatest progressions in abdominal AC [24, 25]. Furthermore, in support of the role played by estrogen deficiency, the relation between vascular calcification and bone loss seems to be strongest during the early phase of the menopause, characterized by rapid decrease in circulating estrogen [25]. This latter observation may substantiate not only the role of estrogen as a linking pathogenic factor, but also the recently emerging view [26] that primary prevention of cardiovascular disease by HRT might be more beneficial if initiated in the perimenopausal phase or at least in the early years of the menopause.

In support of this, clinical studies described a significant correlation between estrogen status/HRT and the calcium content of coronary atherosclerotic plaques [27, 28, 29]. In addition to its direct effects on bone vascular cells [30, 31], estrogen may modulate factors that have also been implicated in the common pathogenesis of osteoporosis and vascular disease. The proposed factors are parathyroid hormone (PTH), homocysteine osteoprotegerin, and lipids, especially oxidized LDL [32]. While recent reports from the HERS [33] and the Women’s Health Initiative [34] studies failed to provide convincing evidence for the beneficial effect of HRT in the prevention of cardiovascular disease, it remains to be clarified whether primary prevention started in the early years of the menopause could demonstrate the protective effect of estrogens.

Coronary heart disease was more frequently associated with intermittent claudication (characterized with severe AC and demineralization in the hip) compared to controls, suggesting that this condition is often a part of extensive atherosclerotic disease. Indeed, the presence of aortic plaques has been shown to be frequently associated with atherosclerosis in the femoral, carotid, and coronary arteries and its related diseases [35, 36, 37]. Furthermore, several large trials demonstrated that those with decreased ankle/brachial index [38] and/or intermittent claudication [39] are at increased risk of cardiovascular death due to myocardial infarction or stroke. On the other hand, low bone mass in the hip or calcaneus was also a strong predictor of cardiovascular morbidity and mortality among the elderly women [4, 5, 40]. These observations indicate that those with low hip BMD often show severe atherosclerosis and impaired blood flow in the lower extremity. Cluster of these alterations frequently predestines the later incidence of coronary heart disease or stroke.

In summary, the results of the present observational study provided further evidence for the independent association between atherosclerosis and low BMD in the hip. Severe osteoporosis in the hip may thus draw attention to progressive subclinical cardiovascular disease and may initiate targeted clinical examinations and therapeutical interventions aiming to prevent morbidity and reduce mortality due to cardiovascular events and/or hip fracture.


The authors gratefully acknowledge Grethe Hoffmeister and Merete Christensen for dedicated and skillful assistance in establishing the database, and the staff of the Center for Clinical and Basic Research in Aalborg for assistance in the data collection.

Copyright information

© Springer-Verlag 2003