Associations Among Disease Conditions, Bone Mineral Density, and Prevalent Vertebral Deformities in Men and Women 50 Years of Age and Older: Cross-Sectional Results From the Canadian Multicentre Osteoporosis Study†
This article was presented in abstract form at the 23rd Annual Meeting of the American Society of Bone and Mineral Research, Phoenix, AZ, October 12-16, 2001.
This cross-sectional cohort study of 5566 women and 2187 men 50 years of age and older in the population-based Canadian Multicentre Osteoporosis Study was conducted to determine whether reported past diseases are associated with bone mineral density or prevalent vertebral deformities. We examined 12 self-reported disease conditions including diabetes mellitus (types 1 or 2), nephrolithiasis, hypertension, heart attack, rheumatoid arthritis, thyroid disease, breast cancer, inflammatory bowel disease, neuromuscular disease, Paget's disease, and chronic obstructive pulmonary disease. Multivariate linear and logistic regression analyses were performed to determine whether there were associations among these disease conditions and bone mineral density of the lumbar spine, femoral neck, and trochanter, as well as prevalent vertebral deformities. Bone mineral density measurements were higher in women and men with type 2 diabetes compared with those without after appropriate adjustments. The differences were most notable at the lumbar spine (+0.053 g/cm2), femoral neck (+0.028 g/cm2), and trochanter (+0.025 g/cm2) in women, and at the femoral neck (+0.025 g/cm2) in men. Hypertension was also associated with higher bone mineral density measurements for both women and men. The differences were most pronounced at the lumbar spine (+0.022 g/cm2) and femoral neck (+0.007 g/cm2) in women and at the lumbar spine (+0.028 g/cm2) in men. Although results were statistically inconclusive, men reporting versus not reporting past nephrolithiasis appeared to have clinically relevant lower bone mineral density values. Bone mineral density differences were −0.022, −0.015, and −0.016 g/cm2 at the lumbar spine, femoral neck, and trochanter, respectively. Disease conditions were not strongly associated with vertebral deformities. In summary, these cross-sectional population-based data show that type 2 diabetes and hypertension are associated with higher bone mineral density in women and men, and nephrolithiasis may be associated with lower bone mineral density in men. The importance of these associations for osteoporosis case finding and management require further and prospective studies.
THE CLINICAL CONSEQUENCES of osteoporosis relate mainly to fragility fractures. These fractures have physical, psychological, and social consequences that can profoundlyinfluence an individual's well being.(1,2) The risk of fracture is inversely related to bone strength and directly associated with the severity of trauma. Factors that modify the risk of fractures may exert their effects through determinants of bone strength such as bone mineral density (BMD) or determinants of trauma such as the tendency for falls.(3)
A number of factors have been found to be associated with BMD and fractures. For instance, thiazide use, weight, height, and calcium intake are positively related to bone mass at the distal radius; whereas increasing age, cigarette smoking, caffeine consumption, and a maternal history of fracture are negatively associated.(4) Unfortunately, our understanding of these variables and their mechanisms is still inadequate. At best, risk factors can predict only about one-third of variability in bone mass.(5) Furthermore, many risk factors need further investigation because of conflicting results or limited evidence.
Many individuals are affected by one or more disease conditions that may be associated with physical disability.(6) However, there have been few studies that have evaluated the association of these disease conditions with the risk of osteoporosis and fractures. Most of these trials have been relatively small(7) or only examined women.(8,9) Using participants from the Canadian Multicentre Osteoporosis Study (CaMos), a national, random sample of the population, we performed a cross-sectional cohort study to determine whether there are associations among various disease conditions and BMD, as well as vertebral deformities, in community-dwelling women and men 50 years of age and older across Canada.
MATERIALS AND METHODS
Details regarding participant recruitment for CaMos have been reported elsewhere.(10) Briefly, CaMos involves nine sites across Canada including St. John's, Halifax, Quebec City, Kingston, Toronto, Hamilton, Saskatoon, Calgary, and Vancouver. CaMos participants were recruited, beginning in 1995, from a randomly selected list of residential telephone numbers from all postal codes within 50 km of each study center. The study population represents an age-, sex-, and region-specific sample of the population of Canada and consists of non-institutionalized individuals, including 6539 women and 2884 men aged 25 years and older. CaMos participants 50 years and older were eligible for inclusion in the current study. Informed consent was obtained from each individual and the study received approval by the institutional review boards at each participating center.
Data source and physical measurements
Data collection at cohort entry was extensive and included an in-person, interviewer-administered questionnaire and a number of physical measurements. The questionnaire covered the following areas: socio-demographic and anthropometric measurements, prescription and nonprescription medication use, dietary intake, medical and fracture history, family history of osteoporosis, reproductive and obstetrical histories, and lifestyle data.
BMD of the lumbar spine, femoral neck, and trochanter were measured by DXA using Hologic (Hologic, Inc., Bedford, MA, USA) or Lunar (Lunar GE Medical Systems, Madison, WI, USA) densitometers (measured in g/cm2). As recommended by the DXA manufacturers, daily and weekly quality assurance data were collected from each center to assess the performance of the scanners. A European Spine Phantom was used at each of the nine CaMos sites at the start of the study to establish instrument cross-calibration values.
Spinal radiographs (lateral thoracic and lumbar) were used to determine whether vertebral deformities were present at baseline. These radiographs were quantitatively assessed for vertebral deformities using a digital graphics tablet.(11) Vertebral bodies were evaluated by measuring the anterior (HA), middle (HM), and posterior (HP) heights of lateral thoracic and lumbar spine radiographs. Height ratios were calculated as HA/HP, HM/HP, HP/HP lower, and HP/HP upper. The presence of a vertebral deformity was determined by a height difference of 3 SD or more from vertebral height ratios. Both sex- and level-specific norms were extracted from a subset of the CaMos data.(11)
Disease conditions were determined based on self-reports (to a trained interviewer) and were not verified. CaMos participants were asked “has a doctor ever told you that you have any of the following conditions?” Both previous and current diseases may have been included in the analysis. We examined 12 disease conditions, which included diabetes mellitus types 1 and 2, nephrolithiasis (reported as kidney stones), hypertension, coronary disease (reported as heart attack), rheumatoid arthritis, thyroid disease, breast cancer, inflammatory bowel disease (IBD), neuromuscular disease, Paget's disease, and chronic obstructive pulmonary disease (COPD). In the case of diabetes mellitus, its duration was also recorded.
Potential confounding variables
Potential confounding variables collected from the questionnaire included age (and polynomials of age, e.g., age5, because the relationship between age and BMD is nonlinear); height; body mass index (BMI; weight in kilograms divided by height in meters squared); CaMOS center; menopausal status; current medication use (thiazide diuretics, nonthiazide diuretics, calcium channel blockers, corticosteroids, bisphosphonates, ovarian hormone therapy, or anticonvulsant therapy); number of years since the diagnosis of type 1 or type 2 diabetes; number of falls during the last month; number of kilocalories spent per week for strenuous, vigorous, and moderate activities (such as jogging, shoveling, and brisk walking); number of sedentary hours spent per day (includes sitting and sleeping); number of alcoholic beverages consumed per week during the previous year; total calcium intake per day in the past year from dairy, nondairy products, and supplements (estimated by a food frequency questionnaire); number of multivitamin supplements per day during the previous year; caffeine intake per day during the past year from coffee, tea, and colas; tobacco use status (yes = greater than 6 months of use); and the lifetime number of 20 cigarette packs smoked (calculated as the product of the number of days of cigarette use with the number of cigarettes smoked per day divided by 20); self-reported previous fracture status (yes/no); the number of previous fractures at either the hip, pelvis, spine, wrist, or ribs as a result of minimal trauma (fragility fractures); and the number of prevalent vertebral deformities (quantitatively assessed as described(11)).
All analyses were conducted separately for women and men. Multivariate linear regression analyses were performed to determine the association between disease conditions and BMD of the lumbar spine, femoral neck, and trochanter. From these analyses, we determined regression coefficient estimates (which represent BMD [g/cm2] differences between participants with and without the clinical condition under study) as well as 95% confidence intervals (CI) of the estimates. The association of these various common clinical conditions with the prevalence of vertebral deformities was investigated using logistic regression analyses. Adjusted odds ratios were calculated along with 95% CI. All variables listed in methods above were assessed for potential confounding in all multivariate analyses. The Bayesian Information Criterion was used for model selection. After model selection, each disease condition was separately forced into the analysis to obtain regression coefficients adjusted for all possible confounders.
All statistical analyses were performed on Sun Workstations using SAS/STAT (version 8.1; SAS Institute Inc., Cary, NC, USA) and Splus (version 5.0; MathSoft, Inc., Seattle, WA, USA) software packages.
A total of 5566 women and 2187 men 50 years of age and older participated in the cross-sectional study. The participant ages ranged from 50 to 103 years in women (median age, 67 years) and 50 to 97 years in men (median age, 66 years). Table 1 presents study entry characteristics for women and men. The medical conditions we studied varied in prevalence in Canada from 0.1% (Paget's disease) to 29.9% (hypertension) in men and 0.2% (Paget's disease) to 33.8% (hypertension) in women. Thiazide, corticosteroid, and anticonvulsant medications were taken by 478 (8.6%), 129 (2.3%), and 102 (1.8%) women and 110 (5.0%), 40 (1.8%), and 44 (2.0%) men, respectively. Osteoporosis medications including bisphosphonates and ovarian hormone therapy were taken by 146 (2.6%) and 1336 (24.0%) women, and 5 (0.2%) and 1 (0.1%) men, respectively. Table 2 summarizes the unadjusted BMD measurements and the prevalent vertebral deformity data for groups of women and men with each of the medical conditions being studied. On average (SD), women and men with deformities were 69.8 (8.5) and 67.8 (9.4) years of age compared with 64.1 (8.3) and 64.4 (8.7) years of age in those without deformities. Disease conditions were not strongly associated with vertebral deformities (Table 3).
Table Table 1.. Mean Baseline Characteristics of Women and Men Over 50 in CaMos
Table Table 2.. Unadjusted Mean Baseline BMD Measurements and Prevalent Vertebral Deformity Status in Women and Men With Various Disease Conditions
Table Table 3.. Adjusted Regression Coefficient Parameter Estimates, Odds Ratio, and 95% Confidence Intervals in Women and Men With Various Disease Conditions
Table 3 presents adjusted BMD coefficient estimates and odds ratios along with the 95% CI of whether men or women have diabetes mellitus. The average (SD) number of years since the diagnosis of type 1 or type 2 diabetes was 15.6 (11.4) and 9.4 (9.8) for women and 15.0 (11.2) and 10.2 (9.2) for men. When BMD was compared between those with and without diabetes mellitus, those with type 2 diabetes had higher BMD measurements for both women and men. The differences were most notable in women at the lumbar spine (+0.053 g/cm2; adjusted for CaMOS center, current ovarian hormone use, menopausal and hypertension status, age, age2. height, BMI, number of alcoholic beverages consumed per week during the previous year, the number of previous fractures, and the number of prevalent vertebral deformities). Differences between women with type 2 diabetes at the femoral neck showed a mean of +0.028 g/cm2 (adjusted for CaMOS center, current bisphosphonate and ovarian hormone use, menopausal status, age, height, BMI, number of kilocalories spent per week for strenuous, vigorous, and moderate activities, number of prevalent vertebral deformities, and previous fracture status). The difference in BMD in women with type 2 diabetes was +0.025 g/cm2 at the trochanter (adjusted for CaMOS center, current ovarian hormone use, menopausal status, age3, height, BMI, number of kilocalories spent per week for strenuous, vigorous, and moderate activities, the number of previous fractures, the number of prevalent vertebral deformities, and the lifetime number of 20 cigarette packs smoked). In men with type 2 diabetes, BMD was higher at the femoral neck (+0.025 g/cm2; adjusted for CaMOS center, height, BMI, number of prevalent vertebral deformities, number of previous fractures, number of kilocalories spent per week in strenuous, vigorous, and moderate activities, and age5.
After adjustments were made for confounding variables, hypertension was associated with higher BMD measurements in both women and men compared with normotensive participants (Table 3). The differences were most pronounced in women at the lumbar spine (+0.022 g/cm2; adjusted for CaMOS center, current ovarian hormone use, menopausal status, diabetes type 2 status, age, age2, height, BMI, number of alcoholic beverages consumed per week during the previous year, the number of previous fractures, and the number of prevalent vertebral deformities) and femoral neck (+0.007 g/cm2; adjusted for CaMOS center, current bisphosphonate and ovarian hormone use, menopausal status, type 2 diabetes status, age, height, BMI, number of kilocalories spent per week for strenuous, vigorous, and moderate activities, number of prevalent vertebral deformities, and previous fracture status), and in men at the lumbar spine (+0.028 g/cm2; adjusted for CaMOS center, previous fracture status, age, height, BMI, and the number of prevalent vertebral deformities). Thiazide use was not a confounder of the association of BMD and hypertension.
The regression model showed that a history of nephrolithiasis (kidney stones) may be associated with lower BMD values in men compared with those without nephrolithiasis (Table 3). The BMD differences were −0.022 g/cm2 at the lumbar spine (adjusted for CaMOS center, hypertension status, age, height, BMI, number of prevalent vertebral deformities, and previous fracture status). At the femoral neck, BMD was −0.015 g/cm2 lower after adjustment for CaMOS center, height, BMI, number of prevalent vertebral deformities, previous fracture status, number of kilocalories spent per week for strenuous, vigorous, and moderate activities, and age5/10,000). For men with a history of kidney stones, BMD at the trochanter was −0.016 g/cm2 lower after adjustment for CaMOS center, height, BMI, number of prevalent vertebral deformities, previous fracture status, number of kilocalories spent per week for strenuous, vigorous, and moderate activities, and tobacco use status). Thiazide use was not a confounder of the association of BMD and nephrolithiasis.
Multivariate analysis revealed that Paget's disease was associated with higher BMD values in both women and men compared with those without the condition (Table 3). The differences were most prominent in women at the lumbar spine (+0.108 g/cm2; adjusted for CaMOS center, current ovarian hormone therapy use, menopausal status, hypertension and type 2 diabetes status, age, age2, height, BMI, number of prevalent vertebral deformities, and the number of previous fractures). In men, Paget's disease was associated with higher BMD values at the femoral neck (+0.220 g/cm2; adjusted for CaMOS center, height, BMI, number of prevalent vertebral deformities, previous fracture status, number of kilocalories spent per week for strenuous, vigorous, and moderate activities, and age5/10,000). Paget's disease in men was also associated with higher BMD at the trochanter (+0.241 g/cm2; adjusted for height, BMI, number of prevalent vertebral deformities, previous main fragility fracture status, number of kilocalories spent per week for strenuous, vigorous, and moderate activities, and tobacco use status).
In this population, the associations with BMD values with diabetes mellitus type 1, coronary disease, rheumatoid arthritis, breast cancer, IBD, neuromuscular disease, thyroid disease, and COPD were inconclusive in both women and men (Table 3). Further research will need to be conducted to determine their relationship with BMD.
Because a variety of medical problems have been reported to relate to osteoporosis risk, this cross-sectional study examined 12 disease conditions in women and men over 50 years of age in a randomly selected sample of the adult Canadian population. The results show that type 2 diabetes mellitus and hypertension were associated with higher BMD measurements and nephrolithiasis may be associated with lower BMD values in men when controlling for the effects of confounding variables.
Women with type 2 diabetes had higher BMD measurements at all three skeletal sites. This observation has been reported previously.(4,12,13) In men with type 2 diabetes, the association reached statistical significance only at the femoral neck. We may speculate that the BMD elevations in both women and men with diabetes mellitus type 2 are caused by an anabolic effect of insulin on bone tissue.(12) It has been postulated that type 2 diabetes is preceded by a period of insulin resistance leading to hyperinsulinemia.(14) Furthermore, hyperinsulinemia may also have a negative impact on sex hormone-binding globulin, resulting in higher free sex hormone levels that may also protect individuals from age-related bone loss.(15–17)
Our observation that higher BMD measurements are present in both women and men who have high blood pressure is difficult to explain. This difference is not accounted for by the prevention of renal calcium wasting secondary to treatment with thiazide diuretics nor by overweight status. Although the clinical significance of the BMD differences between those with and without hypertension may be questioned, in women, the relationship is present at more than one skeletal site after adjusting for covariate variables, and the trends are similar for women and men. In addition, the higher BMD measurements corresponded to 7.5% and 20.8% risk reductions in the prevalence of vertebral deformities in hypertensive women and men, respectively. This suggests that the BMD associations are not likely to be erroneous. Furthermore, Orwoll et al. showed, in a relatively small number of subjects, a similar difference in BMD between hypertensive and normotensive men that did not reach statistical significance.(7)
As has also been reported by several investigators,(18–20) we found an association between nephrolithiasis and reduced bone mass. While our results cannot speak to mechanism, a clinically relevant relationship between hypercalciuric nephrolithiasis and a negative calcium balance has been proposed. Idiopathic hypercalciuria, caused by high intestinal calcium absorption or renal tubular calcium leak, has been associated with low BMD and increased bone resorption.(21,22) Genetic abnormalities may play a part in the association of hypercalciuria and bone loss.(23) Long-lasting hypercalciuria induces a negative calcium balance that may lead to a decline in BMD. The negative calcium balance may be further increased because clinicians commonly counsel patients with calcium-containing stones to limit their intakes of calcium and vitamin D.(24) However, in this cross-sectional study, we did not observe a clinical relationship between calcium intake and bone mass; therefore, we cannot attribute the association of nephrolithiasis and lower BMD to lower calcium intake. We can, however, say that the lower BMD in kidney stone-forming participants was not because of lower calcium intake.
Few participants in this sample had Paget's disease (13 women and 3 men), and the results varied depending on the skeletal site. In addition, although obvious skeletal abnormalities are excluded from our BMD analyses, more subtle Pagetic involvement of the lumbar spine or hip would result in a higher BMD, and we cannot rule out this possibility. Given this, a definitive conclusion cannot be reached from our Paget's disease findings. Further research will need to be conducted in a larger group of individuals with Paget's disease to determine its association with BMD or prevalent vertebral deformities.
Our study did not confirm previously reported associations with low BMD and glucocorticoids or breast cancer and higher BMD in thiazide-users. This study was not specifically designed to examine the relationship between glucocorticoids or thiazide use with BMD. Important variables such as duration of use and drug dosages were not available for the analysis, which may have had an impact on our results. Studies have suggested that women who experience premature menopause as a result of chemotherapy for breast cancer are at increased risk of bone loss and may be at risk for early development of osteoporosis. Women who maintain menses do not seem to be at risk for accelerated trabecular bone loss.(25,26) Given that our study adjusted for menopausal status and age, it is reasonable that breast cancer was not associated with BMD.
This national study is unique for a number of reasons. The participants were selected at random, and we have acquired a large body of information about potential determinants of both BMD and vertebral deformities. The large population we studied contains both men and women, and the participants were evaluated for a number of common medical problems. Furthermore, X-rays were systematically performed and quantitatively evaluated to document all vertebral deformities. This analysis also adjusted for several factors that may influence BMD and vertebral deformities. Nonetheless, our study is a cross-sectional study, and therefore, we cannot infer causal relationships from these data. In addition, the disease classifications were determined by self-reports and thus may lead to a degree of misidentification, which typically leads to a bias toward smaller effects. It is possible that our population may include only individuals with less disabling diseases, because individuals with severe disease forms may have chosen not to participate. Although many potential confounding variables were included in the analysis, it is possible that not all risk factors may have been adequately captured in the CaMOS data set. Because of the small sample size for certain disease conditions, particularly in men, and the low number of prevalent vertebral deformities, those results have wide CIs and are inconclusive. Further research will need to be conducted in a larger group of individuals with these diseases to determine their association with BMD and prevalent vertebral deformities. Although subjects were randomly selected in our study, only 42% of individuals who were contacted agreed to participate fully. However, from a previous study that examined response bias in CaMos participants, results showed little evidence of bias except in the very elderly (80+ years).(27)
In conclusion, our study contributes both new and supportive information to the existing literature. We found that type 2 diabetes mellitus and hypertension were associated with higher BMD values in women and men, and nephrolithiasis may be associated with lower BMD values in men. Further research will need to be conducted to determine the underlying mechanism of actions related to these outcomes. A greater understanding of medical diseases and clinical conditions that are protective for or increase the risk of low BMD may eventually modify our osteoporosis case finding and medical therapy.
The authors thank the rest of the CaMOS Research Group including J Allan, L Blondeau, B Gardner-Bray, P Hartman, J Kedra, S Kirkland, N Kreiger, P Krutzen, S Godmairid, E Lejeune, BC Lentle, B Matthews, N Migneault-Roy, M Parsons, RS Rittmaster, L Robertson, B Stanfield, and Y Vigna. The Canadian Multicentre Osteoporosis Study was funded by the Senior's Independence Research Program, through the National Health Research and Development Program of Health Canada (6605-4003-OS), the Medical Research Council of Canada, MRC-PMAC Health Program, Merck Frosst Canada Inc., Eli Lilly Canada Inc., Procter and Gamble Pharmaceuticals Canada Inc., and the Dairy Farmers of Canada. LJ is a senior scientist funded by the Canadian Institutes for Health Research.