Cardiovascular diseases and future risk of hip fracture in women
- First Online:
- Cite this article as:
- Sennerby, U., Farahmand, B., Ahlbom, A. et al. Osteoporos Int (2007) 18: 1355. doi:10.1007/s00198-007-0386-0
We used a population-based case-control study in women and linkage to the Swedish in-patient register to examine if there is an increased risk of hip fracture after a cardiovascular disease. There was a substantially increased risk of hip fracture after a diagnosis of a cardiovascular disease.
Recent data have indicated that cardiovascular diseases (CVDs) might have a relationship to osteoporosis, which may explain the increased risk of mortality after hip fracture. It is uncertain, however, whether there is an increased risk of fracture after any cardiovascular disease and in subgroups of CVDs. The objective of this study was to determine whether there are associations between CVD and future hip fracture risk. Knowledge of the risk pattern would lead to better understanding of common pathologic pathways of osteoporosis and CVD.
We conducted a population-based case-control study of 1,327 incident hip fracture cases and 3,170 randomly selected population controls among women 50–81 years old in Sweden. Information on cardiovascular and other diseases before the fracture was obtained by linkage to the Swedish National Inpatient Register. Odds ratios (OR) and 95% confidence intervals (CI) where calculated by unconditional logistic regression.
Before study entry, CVDs were diagnosed more than twice as commonly among fracture cases (25%) as among controls (12%). Also, after adjustment for variables including several chronic diseases, we found a doubled risk of hip fracture after a CVD event (OR 2.38; 95% CI 1.92–2.94). There was a gradient increase in risk of hip fracture with increasing number of hospitalizations for CVD and highest fracture risk occurred the first year after the CVD event. Hypertension, heart failure, and cerebrovascular lesions remained independent risk factors, with 2- to 3-fold increases in odds ratios, even after mutual adjustments for other CVDs.
There was a substantially increased risk of hip fracture in women after a diagnosis of a CVD, a finding compatible with the concept of common pathologic pathways for osteoporotic fractures and CVD.
KeywordsCardiovasculare diseaseCerebrovascular laesionHeart failureHip fractureHypertensionOsteoporosis
It is well known that mortality is increased after hip fracture [1–3]. The etiologic and clinical implications of the increased risk remain controversial; however, since the underlying cause of the increased mortality risk and the role of co-morbidity among these patients have not been clarified. In the few studies in which causes of death have been recorded, deaths have largely been attributed to chronic diseases rather than to acute fracture-related complications .
Some studies have reported associations between cardiovascular diseases (CVD) and bone mineral loss [4–7]. Osteoclast regulatory factors can affect vascular calcifications [8–12], and a high blood pressure can induce abnormalities in calcium metabolism and increase bone mineral loss in women . Low bone mineral density is not only an important predictor of osteoporotic fracture , but is also a risk factor for mortality .
Taken together, there are data indicating that CVD might be associated with osteoporosis and that this may partly explain the increased risk of mortality after hip fracture. It is uncertain, however, whether there is an increased risk of fracture after a cardiovascular disease and whether this risk is present only after specific subtypes of cardiovascular events. Knowledge of this risk pattern might be important in attempts to better understand common pathologic pathways of osteoporosis and cardiovascular diseases. The aim of this study was to test the hypothesis that a history of cardiovascular disease leads to an increased risk of hip fracture in women.
Subjects and methods
This population-based case-control study was conducted in six counties of south-central Sweden: Stockholm, Uppsala, Västmanland, Örebro, Göteborg, and Malmöhus [16, 17]. These counties included nearly half of the 8.6 million inhabitants of Sweden during the study period. The study protocol was approved by the local ethics committees in the areas involved.
We aimed to ascertain all fractures of the cervical, intertrochanteric, or subtrochanteric regions of the proximal femur that occurred during the study period October 1993 to February 1995 among native-born Swedish women resident in the study area who were born after 1913. All hospital records and x-ray reports were scrutinized to confirm eligibility and ascertain the types of hip fracture. We identified 2,597 possible hip fracture patients. By chart reviews we excluded patients with a fracture due to malignancy (n = 26) or high-energy trauma (mainly traffic accidents, n = 4) patients with an incorrect diagnosis (n = 41), previous fracture (n = 10), blindness (n = 5) and severe alcohol abuse and psychosis or senile dementia (n = 576), and also patients born outside Sweden (n = 202) and those who had died within 3 months of the fracture (n = 123). The latter group was excluded because of the questionnaire collection of covariate information. After these exclusions, 1,610 cases were eligible for the study. These subjects were approached with a comprehensive questionnaire at a mean interval of 95 days after the fracture.
Potential controls were women who were born in Sweden after 1913, resident in the study area, and randomly selected from a continuously updated population registry the month before the start of the study.
The controls were frequency matched (at least 2 controls per case) to the expected hip fracture age distribution, within each county of residence [16, 17]. Of the 4,872 candidate controls, 4,059 were eligible and 813 were excluded. The exclusions comprised 610 women born outside of Sweden, 157 who died before being approached, 44 who had senility or psychosis, and two who were blind. Questionnaires were sent to eligible controls on six occasions evenly distributed during the study period.
We mailed a questionnaire to eligible subjects asking about information on height and weight (current, one year before and at age 18). We also requested information on education, leisure time physical activity, dietary habits, recent alcohol consumption, and smoking habits. The women were also asked about age at menarche and menopause, parity, and use of oral contraceptives and postmenopausal hormone replacement therapy .
Approximately 50% of the participants were approached by telephone for completion of missing information. Some women refused to answer the postal questionnaire but accepted a less extensive telephone interview. Of those who were eligible, 1,328 cases (82.5%) and 3,312 controls (81.6%) answered the questionnaire; of these, 202 (15.2%) of the cases and 497 (15.0%) of the controls responded solely by telephone.
Using the subjects’ personal identification numbers, we linked all cases and controls to the Swedish National In-Patient Registry, which since 1987 has had complete coverage of all hospitalizations in the country. We, thus, had hospitalization data on the primary and up to five secondary diagnoses prior to study entry for 100% of both cases and controls.
As measures of associations between CVD and hip fracture, odds ratios (ORs) and 95% confidence intervals (CIs) were computed by unconditional logistic regression. In the analysis, any event of CVD was defined as hospitalization with a cardiovascular disease (ICD9 401–405, 410–414, 428, 430–438, 440–448). This definition was also used as the basis for number of CVD events and recency of CVD hospitalization. Index date was defined as the time of fracture among the cases and 95 days before the mailing of the first questionnaire among the controls. We additionally performed separate analyses of subgroups of CVD: hypertension (ICD9 401–405, often a secondary diagnosis), ischemic heart disease (IHD; ICD9 410–414), atherosclerosis (ICD9 440–448), heart failure (ICD9 428) and cerebrovascular lesions (CVL; ICD9 430–438).
A crude and a multivariable model including age in five categories (<60, 60–64, 65–69, 70–74, and 75–81 years), body mass index (kg/m2) by quintiles according to the distribution of the controls, smoking (never, former, current), current intake of alcohol (yes or no), use of hormone replacement therapy (ever or never), leisure-time physical activity in recent years (less than two, or two or more, times per week) was used. Inclusion of adult weight change, calcium intake, ever use of oral contraceptives, menopausal age and parity did not influence our estimates more than marginally and were therefore not included in the final multivariable model. Through the linkage with the In-Patient Registry, it was also possible to adjust for severe previous co-morbidity, which we defined as: malignancies (ICD9 codes 140–208), diabetes mellitus (ICD9 250), dementia and psychosis (ICD9 290–299), drug abuse (ICD9 303–305), pneumonia (ICD9 480–486), chronic obstructive lung disease (ICD9 490–492), liver cirrhosis (ICD9 571), and renal failure (ICD9 580–586). An indicator variable was used for each disease group. Participants claiming natural menstruation were classified as premenopausal (50 controls and one case) and were excluded from the analysis.
Descriptive characteristics of the study participants
N = 1327
N = 3170
Body mass index (kg/m2)
Consumption of alcoholic beverage
Leisure physical activity (h/w)
Ever use of HRTB
Disease prior to index date
Dementia and psychosis
Alcohol or drug abuse
Chronic obstructive lung disease
Any of above disease
CVD and any of above diseases
Odds ratios (OR) and 95% confidence intervals (CI) of hip fracture associated with previous diseases
OR (95% CI)
OR (95% CI)
OR (95% CI)
OR (95% CI)
Any cardiovascular disease
CVD and any other disease
In an attempt to evaluate the independent associations between the subgroups of CVD and hip fracture risk, we included all these subgroups simultaneously in the multivariable model (Table 2). Hypertension, heart failure and CVL still remained as independent predictors of hip fracture risk, with twofold to threefold increases in relative risks. Interestingly, having had an ischemic heart disease did not remain as a risk factor of hip fracture in this particular analysis. On the other hand, a participant could have had two or more of the CVDs and including them in the same model might have introduced a colinearity problem. The Spearman correlation coefficients were modest, however, ranging from 0.05 to 0.30, thus showing low likelihood of distorted estimates.
Odds ratios (OR) and 95% confidence intervals (CI) of hip fracture associated with number of hospitalizations for cardiovascular disease (CVD) before the fracture
Crude OR (95% CI)
Adjusted OR (95% CI)*
Adjusted OR (95% CI)**
Number of hospitalizations with CVD before entry to study
Odds ratios (OR) and 95% confidence intervals (CI) of hip fracture associated with recency of cardiovascular disease (CVD)
Crude OR (95% CI)
Adjusted OR (95% CI)*
Adjusted OR (95% CI)**
Recency (days between index date and CVD)
>1825 (5 years)
Our case-control study is, to our knowledge, the first population-based investigation to address the risk of hip fracture after different cardiovascular events. There was a substantial and especially pronounced increase in this risk in association with hypertension and heart failure as well as with CVL, whereas after multivariable adjustments ischemic heart disease and peripheral atherosclerosis were not independent predictors of future hip fracture events.
Our results extend the findings in previous epidemiological studies in which the association between cardiovascular disease and fracture risk was examined. In a prospective study in women, Bagger et al.  have verified the cross-sectional finding by Schulz et al.  that severe aortic calcifications increased the relative risk of hip fracture two- to threefold. In the MORE study , the risk of future ischemic heart events and stroke was found to be four to five times higher in women with osteoporosis than in those with osteopenia, a result that was independent of smoking, diabetes, hypertension and hyperlipidemia. Furthermore, women with at least one vertebral fracture had a three times higher risk of a future CVD compared to those without a vertebral fracture. The impact of self-reports of fracture  and CVD  is uncertain in these studies. Furthermore, heart failure and non-cardiovascular co-morbidity were not considered as covariates in the analyses. In our study also, before these adjustments, ischemic heart disease and atherosclerosis increased the risk of fracture.
There could be several different underlying causes of these associations involving cellular effects and mechanisms shared by the vasculature and bone. Some intrinsic and extrinsic factors seem to be mutually involved in deranged bone mineralization and vascular calcifications, such as vitamin D insufficiency, a low calcium intake, estrogen deficiency, oxidative stress, dyslipidemia, chronic inflammation, homocysteinemia, smoking, and sedentary physical behaviour . We were also able to control for anthropometry as well as for regular risk factors such as smoking, physical activity and co-morbidity, but after these adjustments the estimates remained substantially unchanged. Thus, explanations for the observed associations other than these common risk factors need to be found.
Could the presence of a cardiovascular disease lead to an alteration of calcium metabolism causing bone fragility? Bone calcification processes have displayed similarities to those in vascular tissue, where calcifications of arteries have contributed to vascular problems such as heart failure, systolic hypertension and peripheral ischemic disease . Calcium metabolism has a central role both in bone mineralization and in the risk of developing arteriosclerosis and hypertension [23, 24]. Elderly people are especially susceptible to calcium and vitamin D deficiency for several reasons. Low calcium and vitamin D intake, limited solar exposure and impaired renal function leading to insufficient vitamin D biosynthesis , all lead to a decrease in intestinal calcium absorption and impairment of renal conservation of calcium. This could contribute to calcium deficiency and calcium mobilization from the bone, with a consequently increased risk of fracture.
Estrogens might play a role both in the development of CVD and in osteoporosis through their effects on paracrine cytokines, such as IL-1, IL-6 and TNF-alpha and osteoprotegerin (OPG). Low levels of estrogens induce an increase in these cytokines together with a decrease in OPG, a reduction of serum vitamin D, increased signs of inflammation, and diminished nitric oxide production, which can be of importance for the progression of bone loss and in atherogenesis [26, 27]. High circulating levels of osteoprotegerin have in themselves been found to increase cardiovascular mortality, but not to alter the fracture risk or bone mineral density .
Another biological link might be the 12/15 lipo-oxygenase system, which is involved in oxidation of lipids. Patients with heart failure have higher plasma levels of oxidized low-density lipoprotein (LDL) [29, 30]. Oxidized LDL inhibits osteoblast development in vitro and bone formation in vivo. It also serves as a substrate for peroxisome proliferator activated receptor gamma (PPAR-γ), which also has a negative influence on osteoblast development .
Several studies indicate that besides oxidized LDL other markers of oxidative stress, such as plasma homocysteinemia and chronic inflammation, are also associated with osteoclastogenesis and fracture risk [31–34], as well as with an increased risk of chronic heart failure  and hypertension .
Body weight and total fat mass are strong predictors of BMD in postmenopausal women [37–39] and adipose tissue has come to be thought of as an endocrine organ releasing a variety of mediators such as leptin, adiponectin and estrogen that seem to influence bone density and atherogenesis [40, 41]. While a high body weight reduce the risk of osteoporotic fractures among elderly women, it is a risk factor for atherogenesis, but body mass per se might exert mechanical forces that enhance bone density, thereby preventing osteoporotic fractures.
Apart from their vulnerability to fractures through direct influences on bone, persons with cardiovascular diseases, could also have a higher propensity for falls. This is naturally a plausible additional explanation for the increased fracture risk among cases with a previous CVL. Additionally, balance disturbances such as vertigo and dizziness may also have occurred as side effects of medications, leading to falling accidents. Diuretics, prescribed for treatment of both hypertension and heart failure, can both improve the renal conservation of calcium and thus indirectly reduce bone loss (thiazides) and increase the urinary losses of calcium (loop diuretics). Nevertheless, the nocturnal diuretic effects of these drugs in the elderly can also increase the likelihood of falling accidents during the night. Furthermore, patients with several hospitalizations for CVD had an especially high risk of fracture. This might be explained by both a higher tendency of falling accidents and a more pronounced risk of osteoporosis in a population with higher disease burden.
A diagnosis of CVD several years previously did increase the risk of hip fracture, but recent diagnosis of CVD was especially associated with such a risk. This indirectly implies that the fracture risk is partly attributable to extra skeletal effects of CVD or its treatment that increase the risk of falling accidents. We observed that the first half-year period after a CVD event dramatically increased the risk for hip fracture. Our data, thus, indicate that in perspective for public health, personal suffering and socioeconomic costs, attempts should be made to avoid environmental risks factors that can lead to falling accidents in those who recently have had a cardiovascular event in an effort to prevent hip fractures.
The reliability of our results is dependent on the quality of the register data. The unique personal registration number used in Sweden enabled us to link, for each individual, prospective data regarding diseases treated during hospitalization to the nationwide, complete inpatient register. The register data, however, do not allow identification of laterality of fractures or of the disability side after a stroke, information that could have helped to better understand the causes of fracture after a hemiparesis, since the fracture risk is most pronounced on the affected side . A further limitation that could have led to attenuated estimates is the conceivably different diagnostic criteria for the CVDs applied by the physicians. Furthermore, the categories of CVD subgroups chosen for the analyses also encompass different disease entities, such as acute myocardial infarction and angina pectoris in the ischemic heart disease subgroup and ischemic and hemorrhagic stroke in the CVL category. We have not analyzed these sub-categories of diseases separately. Nevertheless, the overall quality of the inpatient register is considered to be high, and the validity of the data has been evaluated . After the first event of acute stroke the diagnosis could be confirmed in 94% and for heart failure the validity was found to be 95% . Thus, the impact of diagnostic misclassification is probably modest. Our study has some additional strengths, namely a population-based design with chart review for cases, complete follow-up, and extensive covariate data. However, our subjects were women aged 81 years or younger, and our findings may not apply to the very elderly or to men, or to other ethnic groups. The questionnaire information such as body weight, smoking and physical activity level was obtained retrospectively. The case-control design with a restricted study base due to our exclusions might have affected the validity of the results. However, we were able to include in the analysis all hip fracture cases and controls identified during the study period, for example including those who had died or had a report of a psychiatric diagnosis , although without the possibility of adjusting for the information collected by the questionnaire. The age-adjusted odds ratio of hip fracture in that analysis was 2.84 (95% CI 2.49–3.24) after a CVD event and after multivariable adjustment for other severe diseases the odds ratio was reduced to 2.24 (95% CI 1.94–2.59), estimates similar to those presented in Table 2 with the restricted study base. Thus, our findings seem not to be explained by selection bias. In this study it was not possible to adjust our estimates for BMD.
We conclude that that there is a considerably increased risk of hip fracture among women after especially a recent hospitalization for a CVD, in particular heart failure, hypertension and stroke, but the mechanism and the consequences for health in the community remain to be elucidated.