Smoking predicts incident fractures in elderly men: Mr OS Sweden

Abstract

The aim of this study was to investigate the association between smoking and bone mineral density (BMD) and radiographically verified prevalent vertebral fractures and incident fractures in elderly men. At baseline 3003 men aged 69 to 80 years of age from the Swedish Mr Os Study completed a standard questionnaire concerning smoking habits and had BMD of the hip and spine measured using dual-energy X-ray absorptiometry (DXA); 1412 men had an X-ray of the thoracic- and lumbar spine. Radiologic registers were used to confirm reported new fractures after the baseline visit. At baseline, 8.4% were current smokers. Current smokers had a 6.2% lower BMD at the total hip and a 5.4% lower BMD at the lumbar spine (p < .001). Current smoking remained independently inversely associated with BMD at the hip and lumbar spine after adjusting for age, height, weight, calcium intake, physical activity, and centers as covariates. Prevalent vertebral fractures among current smokers were increased in unadjusted analyses [odds ratio (OR) = 1.90, 95% confidence interval (CI) 1.26–2.87] and after adjustment for lumbar BMD (OR = 1.67, 95% CI 1.09–2.55). Smokers had a high risk for two or more prevalent vertebral fractures (OR = 3.18, 95% CI 1.88–5.36). During the average follow-up of 3.3 years, 209 men sustained an X-ray-verified fracture. Incident fracture risk among smokers was calculated with Cox proportional hazard models. Current smokers had an increased risk of all new fractures [hazard ratio (HR) = 1.76, 95% CI 1.19–2.61]; nonvertebral osteoporotic fractures, defined as humerus, radius, pelvis, and hip fractures (HR = 2.14, 95% CI 1.18–3.88); clinical and X-ray-verified vertebral fractures (HR = 2.53, 95% CI 1.37–4.65); and hip fractures (HR = 3.16, 95% CI 1.44–6.95). After adjustment for BMD, including other covariates, no significant association between smoking and incident fractures was found. Current tobacco smoking in elderly men is associated with low BMD, prevalent vertebral fractures, and incident fractures, especially vertebral and hip fractures. © 2010 American Society for Bone and Mineral Research

Introduction

Scandinavian men and women have an extremely high risk for vertebral1 and hip fractures.2 The reason is unknown and contrasts with the general good health and high life expectancy in Scandinavians. The prevalence of male osteoporotic fractures constitutes a serious clinical problem with significant economic consequences. In several meta-analyses, smoking has been recognized as a risk factor for low bone mineral density (BMD) and increased risk of hip fracture.3–5

Most studies of smoking and osteoporosis have focused on postmenopausal women. One of the first published studies of smoking and osteoporosis was a clinical study of vertebral fractures in female smokers.6 In 1982, we published the first community-based population study showing that smoking was associated with low calcaneal bone mineral mass in elderly men.7 One meta-analysis of fracture risk and smoking, published by Vestergaard and colleagues, showed an increased risk for all fractures [relative risk (RR) = 1.26], for hip fractures (RR = 1.39), and for vertebral fractures (RR = 1.76),4 but most of the studies in this meta-analysis included women, not men. Only one study showed an increased risk of vertebral fractures in smoking men.8 The meta-analysis noted that the hip fracture risk related to smoking increased with northern latitude both in Europe and the United States.4 Several studies have reported an increased smoking-related impact of lower body mass index (BMI) with increasing age.9 In one study of young adult men, we found a lower BMD and lower cortical bone mass in smokers even though there were no differences in BMI between smokers and nonsmokers.10

A longitudinal study from Sweden showed that current smoking was a dose-dependent risk factor for fractures with an adjusted relative risk for any fracture of 2.71.11 A recent large meta-analysis of smoking and fracture risk showed an RR of 1.84 for hip fracture in smokers and an RR of 1.60 after adjustment for BMD. In this study, smoking was a stronger risk factor for fractures in men than in women.5 The results from this meta-analysis led to the inclusion of current smoking as a risk factor in the World Health Organization (WHO) guidelines for calculating the absolute risk of the 10-year probability of fracture (FRAX).12

A Danish study also indicated that smoking tended to be a stronger risk factor for hip fracture in men than in women.13 (However, the Women's Health Initiative reported that current smoking is a strong risk factor for hip fractures in women. 14) This study is a part of the international Mr OS Study established with the purpose of identifying clinically useful risk factors for male osteoporosis in order to inform clinical guidelines. We are aware of the difficulty of calculating the impact of earlier smoking owing to several subgroups among ex-smokers, so this study focuses on current smoking in elderly men. To our knowledge, there are no published studies showing the association between decreased BMD and increased radiographically confirmed prevalent vertebral fractures and incident fractures among current male smokers in the same study. The aim of this study was to determine the association between low BMD and prevalent vertebral fractures and incident fractures among current smokers.

Methods

Study population

The international Mr OS is a multicenter study with men aged 69 to 80 years in the United States, Hong Kong, and Sweden. Both the rationale and the study design have been described previously.15 In brief, Mr OS is a prospective, longitudinal, observational study of risk factors for vertebral and nonvertebral fractures in older men. The aim was to identify men at high risk for fractures and thereby to inform future guidelines.16

The Swedish part of Mr OS consists of three centers (Gothenburg: n = 1010; Malmö: n = 1005; and Uppsala: n = 999) giving a total of 3014 participants. Men were recruited randomly from the Swedish national population registers. Overall, 45% of the men who were contacted participated in the study. To be eligible for the study, the subjects had to be able to walk without assistance, provide self-reported data, and sign an informed consent. Approval to the study was given by the Ethics Committees at Gothenburg, Lund, and Uppsala Universities.15

Assessment of covariates

The international Mr OS questionnaire was used for collecting information about smoking (current smoking, ex-smoking, and never smoking), calcium intake (mg/day, calculated from calcium-containing foods), physical activity (average total daily walking distance, km/day), and falls during the previous 12 months. Height and weight were measured using standard equipment.15 In addition, questions about diseases and glucocorticoid treatment were asked. In Gothenburg, the bone markers osteocalcin and procollagen were analyzed in 1010 men using methods previously described by Kindblom and colleagues.17

Assessment of BMD (DXA)

Standardized areal bone density (aBMD, g/cm², lumbar spine, femoral neck, trochanter, and total hip) was measured with calibrated scanners: Lunar Prodigy DXA (GE Lunar Corp., Madison, WI, USA) in Malmö and Uppsala and Hologic QDR 4500/A-Delphi (Hologic, Waltman, MA, USA) in Gothenburg. The coefficients of variation (CVs) for the aBMD measurements ranged from 0.5% to 3%. To be able to use dual-energy X-ray absorptiometry (DXA) measurements performed with equipment from different manufacturers, a standardized BMD (sBMD) was calculated, as described previously by Mellström and Lewis.15, 18 Data from the two types of DXA scanners were included as independent variables in all regression analyses (ie, total hip, femoral neck, trochanter, and lumbar spine) to minimize the possible confounding effect of using two different scanners (Lunar and Hologic).

Assessment of prevalent vertebral fractures

In Gothenburg and Malmö, 1412 men had an X-ray of the lateral thoracic and lumbar spine at baseline. All vertebral fractures were evaluated by an expert radiologist (IR-J). If the vertebral body had a reduced height of 3 mm or more compared with the vertebra above, it was classified as a vertebral fracture.19

Assessment of incident fractures

Participants were followed for an average of 3.32 years after baseline. The follow-up interval was recorded from the date of the baseline visit to the date of the first fracture, date of death, or end of the present follow-up interval. When a participant sustained a first fracture at different sites during the follow-up, the various fractures and the follow-up time for each respective first fracture type were included in the analyses. Complete follow-up was possible because central registers covering all Swedish citizens were used to identify all participants who died during the study and their date of death. These analyses were performed after the time of fracture validation. At the time of fracture evaluation, the computerized X-ray archives in Malmö, Gothenburg, and Uppsala were searched for new fractures occurring after the baseline visit using the unique personal registration number that all Swedish citizens have. All fractures reported by the participants after the baseline visits were confirmed by physician review of radiology reports. Fractures reported by the study participants but not possible to confirm by radiology report were not included in this analysis. All validated fractures were included in the main analyses, followed by subanalyses of fracture type. In the latter, we studied the associations between current smoking and validated fractures divided into four main groups: (1) X-ray-verified clinical vertebral fractures, (2) nonvertebral osteoporosis fractures at the major osteoporosis-related sites (defined as hip, distal radius, proximal humerus, and pelvis), (3) other fractures (ie, radius/ulna, hand, fingers, humerus, elbow, skull, cervical vertebrae, clavicle, scapula, rib, femoral shaft, patella, upper tibia, ankle, foot, and toes), and (4) subanalyses of the associations between current smoking and hip fracture risk. Fracture rates were expressed as the number of men with first fractures per 1000 person-years

Statistical analysis

Differences in baseline characteristics were compared by ANOVA for continuous variables. Linear regression models including weight, height, physical activity, calcium intake, and age were used to determine current smoking as an independent predictor of standardized BMD in all sites. Prevalent vertebral fractures were analyzed using binary logistic regression models, chi-square test, and Fisher's exact test. The incident fracture risk in smokers was calculated using Cox proportional hazard ratios adjusted for age, center, physical activity, calcium intake, weight, height, diseases (eg, cancer, chronic obstructive pulmonary disease, stroke, myocardial infarction, and diabetes mellitus), treatment (eg, glucocorticoid) and BMD as covariates.

Results

Descriptive data

A total of 3014 men participated in the study, but complete data concerning smoking were lacking for 11 men, leaving 3003 men for the present analyses. Among these, 252 men (8.4%) were current smokers. A total of 1693 (56.4%) were ex-smokers, and 1058 (35.2%) were never-smokers. There were 2751 (91.6%) nonsmokers, defined as ex-smokers + never smokers. Smokers were 0.68 year younger, 4.9 kg lighter and their body mass index (BMI) was 5.6% lower than nonsmokers. Physical activity was 27.5% lower and calcium intake was 7.9% higher in smokers compared with nonsmokers (Table 1). Myocardial infarction and cancer were the most frequent diseases (14.2% and 15.5%, respectively), and glucocorticoid treatment among the subjects was uncommon (1.9%) (Table 2).

Table 1. Descriptive Data

	Smokers (n = 252)	Nonsmokers (n = 2751)	Ex-smokers (n = 1693)	Never-smokers (n = 1058)	All subjects (n = 3003)
Values are given as mean ± SD. sBMD = standardized BMD. Nonsmokers = ex-smokers + never-smokers. * p < .05; ^**p < .01; ^***p < .001 versus nonsmokers.
Age (years)	74.8 ± 3.1^***	75.5 ± 3.2	75.5 ± 3.2	75.5 ± 3.2	75.4 ± 3.2
Height (cm)	174.0 ± 6.6	174.8 ± 6.5	174.9 ± 6.5	174.7 ± 6.7	174.8 ± 6.5
Weight (kg)	76.3 ± 13.2^***	81.2 ± 11.9	82.2 ± 12.0	79.6 ± 11.7	80.8 ± 12.1
BMI (kg/m^2)	25.1 ± 3.8^***	26.6 ± 3.5	26.8 ± 3.5	26.1 ± 3.5	26.4 ± 3.6
Calcium intake (mg/day)	961 ± 476*	891 ± 430	886 ± 438	900 ± 417	897 ± 435
Physical activity (km/day)	2.9 ± 2.8^***	4.0 ± 3.1	4.1 ± 3.2	3.9 ± 3.0	3.9 ± 3.1
DXA analyses
Total hip BMD (g/cm^2)	0.88 ± 0.16^***	0.94 ± 0.14	0.94 ± 0.14	0.94 ± 0.15	0.94 ± 0.15
Femoral neck BMD (g/cm^2)	0.79 ± 0.13^***	0.84 ± 0.13	0.84 ± 0.13	0.83 ± 0.13	0.83 ± 0.13
Trochanter BMD (g/cm^2)	0.74 ± 0.14^***	0.79 ± 0.14	0.79 ± 0.14	0.79 ± 0.14	0.79 ± 0.14
Spine L1–L4 BMD (g/cm^2)	1.09 ± 0.20^***	1.15 ± 0.20	1.15 ± 0.20	1.14 ± 0.21	1.14 ± 0.20

Table 2. Descriptive Statistics (Diseases and Treatment)

Disease/corticoid	n	%
Note: The number of subjects with different diseases and glucocorticoid treatment.
Diabetes mellitus	287	9.5
Stroke	192	6.4
Myocardial infarction	426	14.2
COPD	254	8.5
Cancer	467	15.5
Glucocorticoid treatment	59	1.9

Smoking and BMD

Among smokers, BMD was lower at all sites investigated (total hip: −6.2%; femoral neck: −5.6%; trochanter: −6.7%; and lumbar spine: –5.4% compared with nonsmokers; p < .001; (Table 1). When adjusting for BMI, physical activity, calcium intake, age, and center using linear regression models, smoking was associated with low BMD at all measured sites with a β coefficient of 0.05 to 0.06 (p < .03; Table 3).

Table 3. Multivariate Analyses of the Relation Between Current Smoking and BMD

	β Coefficient	Significance
Note: Linear regression analyses adjusted for age, height, weight, calcium intake, physical activity, and center. Smokers are coded as 2 and nonsmokers as 1.
Total hip BMD (g/cm^2)	−0.06	.0052
Femoral neck BMD (g/cm^2)	−0.05	.0261
Trochanter BMD (g/cm^2)	−0.05	.0116
Lumbar spine L1–L4 BMD (g/cm^2)	−0.05	.0197

Smoking and accidental falls

Falls during last year of observation were reported in 17.7% of current smokers compared with 16.1% of ex-smokers and 16.4% of never-smokers. A logistic model with age and center showed the risk for accidental falls in the previous year among smokers [odds ratio (OR) = 1.53, 95% confidence interval (CI) 0.61–2.19].

Smoking and prevalent vertebral fractures

In the two centers, Gothenburg and Malmö, overall 1412 men had a spine X-ray at baseline, which identified prevalent vertebral fractures in 224 men (15.9%). Current smokers, ex-smokers, and never-smokers (24%, 16.1%, and 12.7%, respectively) had one or more prevalent vertebral fractures (Table 4). Current smokers, when compared with nonsmokers, had an increased risk for prevalent vertebral fractures (crude OR = 1.90, 95% CI 1.26–2.87), and this association remained significant after adjustment for lumbar BMD (OR = 1.67, 95% CI 1.09–2.55; Table 5). The risk of two or more prevalent vertebral fractures (excluding men with one vertebral fracture) was even higher among smokers (crude OR = 3.18, 95% CI 1.88–5.36) and BMD adjusted (2.63, 95% CI 1.52–4.52; Table 5).

Table 4. Prevalent Vertebral Fractures Among Men with an X-Ray at Baseline

	No. fracture	≥1 Fractures
Never-smokers	397 (87.3%)	58 (12.7%)
Ex-smokers	677 (83.9%)	130 (16.1%)
Current smokers	114 (76.0%)	36 (24.0%)
Total	1188 (84.1%)	224 (15.9%)

Table 5. Association Between Smoking Status and Prevalent Fractures Among Men With an X-ray at Baseline (n = 1412)

	Crude	BMD-adjusted
Note: Age-adjusted odds ratios are given with 95% CIs within parentheses. Crude = adjusted for age and center; BMD-adjusted = adjusted for age, center, and lumbar spine BMD.
≥1 Fractures:
Current smokers vs. nonsmokers	1.90 (1.26–2.87)	1.67 (1.09–2.55)
Current smokers vs. never-smokers	2.27 (1.42–3.62)	1.97 (1.22–3.18)
Ex-smokers vs. never-smokers	1.30 (0.93–1.82)	1.28 (0.91–1.81)
≥2 Fractures:
Current smokers vs. nonsmokers	3.18 (1.88–5.36)	2.63 (1.52–4.52)
Current smokers vs. never-smokers	2.74 (1.51–4.97)	2.32 (1.25–4.33)
Ex-smokers vs. never-smokers	0.80 (0.49–1.30)	0.81 (0.49–1.35)

Smoking and incident fractures

Overall, 209 men had at least one validated incident fracture after an average follow-up of 3.32 years. Fracture distribution is shown in Table 6. Fracture risk was calculated with Cox proportional hazard ratio (HR). Current smokers, compared with nonsmokers, had an increased risk of all fractures (HR = 1.76, 95% CI 1.19–2.61), nonvertebral osteoporosis fractures (HR = 2.14, 95% CI 1.18–3.88), clinical and X-ray-verified vertebral fractures (HR = 2.53, 95% CI 1.37–4.65), and hip fractures (HR = 3.16, 95% CI 1.44–6.95; Table 7, Model 1).

Table 6. Subjects With Validated Incident Fractures

All fractures	209
Note: The numbers of subjects with first fractures are given. Some subjects included in the group of “All fractures” had more than one type of first fracture, and therefore, these subjects were included in more than one of the different subtypes of fractures. Nonvertebral osteoporosis fractures are defined as fractures in hip, distal radius, proximal humerus, and pelvis. Other fractures include all validated fractures minus nonvertebral osteoporosis fractures and clinical vertebral fractures.
Nonvertebral osteoporosis fractures	83
Hip	38
Distal radius	28
Proximal humerus	17
Pelvis	10
Vertebral fractures, clinically and X-ray-verified	67
Other fractures	77

Table 7. Smoking and Risk of First Incident Fracture

	Model 1	Model 2	Model 3	Model 4
Note: Age-adjusted hazard ratios are given with 95% CIs within parentheses. Model 1 = Adjusted for age and center. Model 2 = Adjusted for age, center, physical activity, calcium intake, weight, and height. Model 3 = Adjusted for age, center, physical activity, calcium intake, weight, height, cancer, COPD, stroke, myocardial infarction, diabetes mellitus, and glucocorticoid treatment. Model 4 = Adjusted for the same covariates as in Model 3, including BMD.
All fractures	1.76 (1.19–2.61)	1.51 (1.00–2.29)	1.54 (1.01–2.32)	1.25 (0.82–1.92)
Nonvertebral fractures	2.14 (1.18–3.88)	1.88 (1.01–3.47)	1.88 (1.01–3.50)	1.38 (0.72–2.64)
Vertebral fractures	2.53 (1.37–4.65)	2.18 (1.16–4.12)	2.23 (1.17–4.24)	1.63 (0.84–3.19)
Hip fractures	3.16 (1.44–6.95)	2.93 (1.30–6.65)	3.07 (1.35–6.98)	2.34 (0.97–5.65)

Also, after adjustment for physical activity, calcium intake, weight, and height, current smokers had a two- to threefold increased risk for vertebral fractures (HR = 2.18, 95% CI 1.16–4.12) and hip fractures (HR = 2.93, 95% CI 1.30–6.65; Table 7, Model 2). With further adjustment (additional covariates such as cancer, chronic obstructive pulmonary disease, stroke, myocardial infarction, diabetes mellitus, and glucocorticoid treatment), there remained a significant association between current smoking and vertebral fracture risk (HR = 2.23, 95% CI 1.17–4.24) and current smoking and hip fracture risk (HR = 3.07, 95% CI 1.35–6.98) but only weak significance between current smoking and nonvertebral fractures and current smoking and all fractures (Table 7, Model 3).

Further more, after adjustment for the same additional covariates as above, including BMD, no significant associations between smoking and incident fractures were found (Table 7, Model 4). Ex-smokers when compared with never-smokers had no increased risk of first incident fracture (all fractures) when adjusted for age and center (HR = 1.05, 95% CI 0.78–1.42) and when adjusted for age, center, height, and weight (HR = 1.09, 95% CI 0.81–1.48).

The bone markers osteocalcin and procollagen were analyzed in 1010 men from Gothenburg. No significance difference was found between serum osteocalcin levels among current smokers [26.2 ± 8.7 µg/L (±SD)] versus nonsmokers (26.8 ± 13.8 µg/L), and no significance difference was found in procollagen between current smokers (41.8 ± 14.5 kU/L) and nonsmokers (40.5 ± 18.6 kU/L).

Discussion

In this study we found that current smoking in elderly men is associated with low BMD, prevalent vertebral fractures, and predicts incident fractures, especially vertebral and hip fractures. In Gothenburg, Sweden, current smoking in 75-year-old men decreased from 40% in 1976 to 8% in 2005.7 Smoking has been recognized as a major risk factor for premature death from all causes and in several cancer forms, cardiovascular disease, and chronic obstructive pulmonary disease (COPD). Smoking in the elderly also has been related to lower BMI, muscle function, and lung function, as well as periodontitis.

We have presented data previously that indicate that low BMD in elderly men is associated with increased mortality when adjusted for smoking and other risk factors.20 In this study we found that current smokers were younger than nonsmokers, presumably owing to selective mortality. Despite an increased mortality, smokers had a higher risk of prevalent fractures, incident hip fractures, and other osteoporotic fractures in the Swedish Mr OS Study (n = 3003).

We have shown recently that current smoking in young men (18 to 20 yearrs)—the GOOD cohort—was associated with lower BMD at all sites but also lower cortical thickness at both radius and tibia as judged by peripheral quantitative computed tomography (pQCT).10 The differences in BMD between smokers and nonsmokers in the GOOD cohort were −4.3% at the lumbar spine and −6.6% at the trochanter. This difference is similar to our present findings in elderly men in the Swedish Mr OS Study, where we found in smokers a 5.4% lower BMD at the lumbar spine and a 6.7% lower BMD at the trochanter. In both studies we used age, BMI, physical activity, and calcium intake as covariates, and both studies indicated that smoking was an independent risk factor for low bone mass. These findings indicate that current smoking is a risk indicator for low BMD both at the age of peak bone mass and in elderly men.

Vertebral fracture was one of the first fractures that was associated with smoking.6 However, the literature is contradictory in this field, especially for men. The European Osteoporosis Study (EPOS) showed that smoking or physical inactivity did not increase the risk of vertebral fractures in women or men.21 The US Mr OS Study has a very low smoking prevalence (about 3%). Current smoking was not related to BMD, but 6.6% of men with clinical vertebral fractures were current smokers compared with 3.4% of men without clinical vertebral fractures (NS).22 However, current smoking was not related to nonspine fractures in the US Mr OS Study.18 The Hong Kong Mr OS Study showed that current smokers had an age-adjusted 5% lower lumbar spine BMD compared with nonsmokers.23 The Rotterdam study found that prevalent vertebral fracture and low BMD were associated with increased incidence of vertebral fractures in both women and men, but smoking increased the risk only in women.24 The longitudinal Framingham study with a high smoking prevalence in men (35%) showed at a second lumbar radiograph the smoking-related risk for vertebral fractures (OR = 3.09, 95% CI 0.84–11.35) over a 25-year interval.25 However, only 252 men had a second radiograph in the study, indicating a low power.

Smoking as a risk factor for osteoporosis and fractures was first described for women, and most meta-analyses indicate that smoking is an independent risk factor for most fractures related to osteoporosis.3–5 The Women's Health Intiative (WHI) study showed that current smoking in US women was associated with an increased risk for hip fracture (multivariate OR = 2.33, 95% CI 1.71–3.18).14 Some studies have indicated that current smoking has a stronger effect on fracture risk in men than in women.13

Scandinavian men and women have, from an international point of view, a high risk for hip and vertebral fractures. However, there are no major studies indicating that elderly men or women in Scandinavia have a lower BMD or higher prevalence of WHO-defined osteoporosis. Indeed, the converse is true, at least for peak bone mass.12 Thus Scandinavians appear to sustain fractures at a higher BMD than people in most other countries. The question arises whether widely accepted risk factors such as prevalent fractures, heredity, and smoking have a greater impact on fracture risk in Scandinavia. A meta-analysis of smoking and fractures indicated an increasing risk for smoking-related fracture with increasing northern latitude.4

Several major studies, including the Study of Osteoporotic Fractures (SOF), have shown that smoking is a risk factor for hip fracture in women (RR = 2.1, 95% CI 1.4–3.3). However, in a multivariate model including a large number of confounders, smoking was not a significant risk factor.26 In the WHO study, smoking was found to be a multivariate independent risk factor for the 10-year probability of major osteoporotic fractures and hip fractures in both women and men.27

The Rancho-Bernado study found that current smoking was a major predictor of bone loss together with age and physical inactivity.28 A longitudinal follow-up of the SOF study of repeated BMD measurements over 15 years showed that current smoking and oral glucocorticoid use were significant risk factors for bone loss.29 These two longitudinal studies indicate that smoking is likely to be a major modifiable predictor of bone loss. The effect of smoking on BMD and fractures may be due to toxins that could have complex effects on bone cells,30 antiestrogenic effects,31 thyroid-modulating effects,32 and parathyroid hormone (PTH)–modulating effects.33 Smoking is a major contributor to cadmium exposure in men, and cadmium shows an age-cumulating effect that has been related to an increased risk of fracture.34 Smoking stimulates adrenocorticotropic hormone (ACTH) secretion, resulting in increased serum levels of cortisol.35 Other studies have shown that smoking tends to decrease insulin-like growth factor 1 (IGF-1) and to induce liver enzyme activity, resulting in the increased metabolism of several drugs.

There is good evidence in the literature that smoking has direct effects on bone metabolism and fractures. However, some of the effects may be due to indirect effects on a great number of diseases such as chronic obstructive pulmonary disease and cardiovascular disease.

The strength of the Swedish Mr OS Study is the large number of participants; reliable data, including smoking, nutritional intake, physical activity, and BMD measurements; and also fracture data owing to unique Swedish national registers, including X-ray registers. All significant differences regarding incident fracture risk were lost when adjusting for BMD. However, the HR for hip fracture in smoking men in the Swedish Mr OS Study—adjusted for age, center, physical activity, calcium intake, BMI, diseases (eg, cancer, chronic obstructive pulmonary disease, stroke, myocardial infarction, and diabetes mellitus), glucocorticoid treatment, and BMD—was 2.34, which is somewhat higher than the BMD-adjusted hip fracture risk (HR = 1.60) among smokers found in the large meta-analysis by Kanis and colleagues.5 This meta-analysis had about 5500 prospective fractures compared with our 209, which indicates that our nonsignificant result might be due to low statistical power.

Two weaknesses of our study are the cross-sectional measurement of BMD and the difficulty in calculating the impact of earlier smoking as a risk factor for low BMD, prevalent vertebral fractures, and incident fractures owing to several subgroups among ex-smokers. We preferred to compare the risk of incident fractures among current smokers versus nonsmokers (ex-smokers and never-smokers). We anticipate that our method might to some degree underestimate the fracture risk related to smoking. However, most guidelines choose to use only current smokers in the multivariate 10-year probability of fractures.

When evaluating male osteoporosis in clinical use, information on smoking is valuable. Current smoking is associated with a low BMD and prevalent vertebral fractures and predicts incident fractures, especially vertebral fractures and hip fractures.

Disclosures

All the authors state that they have no conflicts of interest.

Acknowledgements

This study was supported by the Swedish Research Council, ALF/LUA research grants in Gothenburg, and the King Gustav V and Queen Victoria Frimurarestiftelse Research Foundation.