| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
University of Connecticut, School of Allied Health, Storrs, CT (J.Z.I., R.A.B., L.T.)
Clinical Medical Centre, Endocrinology Department, Rijeka, Croatia (Z.C.-O.)
Address correspondence to: Jasminka Z. Ilich, PhD, RD,Associate Professor, University of Connecticut, School of Allied Health, 358 Mansfield Rd., U-101, Storrs, CT 06269. E-mail: ernst{at}uconnvm.uconn.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIAL AND METHODSRESULTSACKNOWLEDGMENTSREFERENCES |
|---|
Methods/Design: A cross-sectional study in 136 Caucasian women, mean ± SD age 68.6 ± 7.1 years, all healthy and free of medications affecting bones, including estrogen. Bone mineral density (BMD) of multiple skeletal regions and body composition were measured by dual X-ray absorptiometry. Serum vitamin D (25-OHD) and parathyroid hormone (PTH) were analyzed and used as confounders. Calcium (Ca) intake was assessed by food frequency questionnaire. Alcohol and caffeine consumption was assessed by questionnaires determining frequency, amount and source of each. There were no current smokers, but the history of smoking was recorded, including number of years and packages smoked/day. Past physical activity was assessed by Allied Dunbar National Fitness Survey and used as confounder. Statistical significance was considered at p 
0.05.
Results: In the correlational analysis, alcohol was positively associated with spine BMD (r = 0.197, p = 0.02), 25-OHD and negatively with PTH. Smoking was negatively related to Ca intake, 25(OH)D and number of reproductive years. In subgroup (stratified by Ca intake) and multiple regression analyses, alcohol (average 
0.5–1 drinks/day or 8 g alcohol/day) was favorably associated with BMD of spine and total body. Caffeine (average 2.5 6-fl oz cups/day or 200–300 mg caffeine/day) had negative association with most of the skeletal sites, which was attenuated with higher Ca intake (
median, 750 mg/day). The past smokers who smoked on average 24 years of 1 pack cigarettes/day had lower BMD in total body, spine and femur than never-smokers when evaluated in subgroup analyses, and the association was attenuated in participants with median Ca intake. There was no significant association between past smoking and BMD of any skeletal site in multiple regression analyses.
Conclusion: The results support the notion that consumption of small/moderate amount of alcohol is positively, while caffeine and past smoking are negatively associated with most of the skeletal sites, which might be attenuated with Ca intake above 750 mg/day.
Key words: bone mineral density, calcium, alcohol, caffeine, past smoking, postmenopausal
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTSACKNOWLEDGMENTSREFERENCES |
|---|
The views on the relationship between caffeine consumption and BMD in postmenopausal women are various and equally inconclusive. Caffeine has been found to contribute to increased urinary calcium (Ca) excretion and possibly reduced endogenous Ca absorption leading to a negative calcium balance [9,10], and hence increased requirements. However, both the epidemiological and clinical data addressing the association between caffeine consumption and bone status are quite contradictory. There are those that show detrimental effect or association [7,11,12] and those that do not [13–16]. According to some, the deleterious effect of caffeine becomes most pronounced when dietary Ca is low and less harmful when dietary Ca is adequate [17,18].
Cigarette smoking has long been recognized as a risk factor for many health problems, including bone health. According to the recent meta analysis, smoking leads to lower BMD and the effect is cumulative with age. In that analysis, smoking was also shown to be one of the major determinants of hip fracture, increasing the life time risk by about half [19]. Data from Cornuz et al. [20] showed that risk of hip fracture declined among past smokers compared to current, but not until 10 years after smoking cessation. Not all studies, particularly in the cross-sectional settings, have shown detrimental effect of smoking on BMD. In the study by Krall et al. [21], there was no difference in bone mass at baseline in over 400 elderly men and women, although in a three-year follow-up of the same population, there was increased rate of bone loss and decreased Ca absorption in smokers compared to non-smokers. The effects of past smoking, both duration and quantity, on bones still remain inconclusive.
The purpose of this study was to determine the relationship between alcohol, caffeine, past cigarette smoking and bone mineral density of different skeletal sites in healthy, postmenopausal women, accounting for important confounders, such as dietary intake of Ca, hormonal factors (vitamin D and parathyroid hormone), body composition and past physical activity. The areal BMD (g/cm2), used as the main outcome and measure of bone quantity in this study, is generally accepted as an appropriate mode of expressing bone status and predicting fracture risk [22].
|
MATERIAL AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTSACKNOWLEDGMENTSREFERENCES |
|---|
Weight in kg and standing height in cm were recorded in light, indoor clothing without shoes, from which body mass index (BMI) (kg/m2) was calculated to screen for obesity. Areal bone mineral density (g/cm2) was measured by dual X-ray absorptiometry (DXA) with a Lunar DPX-MD instrument (GE Medical Systems, Madison, WI) using specialized software for different skeletal regions, as described earlier [25]. The measured skeletal sites were total body, which yields the analysis of body composition, lean and fat tissue, lumbar spine (L1–L4), femur (neck, trochanter, Ward’s triangle, shaft and total femur) and forearm (ulna and radius at ultradistal and 1/3 distal region measured from the styloid process and total forearm). In our preliminary analysis in part of this population, we found a significant influence of lean body mass and past physical activity on BMD of various skeletal sites. Body composition, lean and total body fat, showed stronger relationship with BMD than either weight, height and/or BMI [26]. Therefore, we felt compelled to include these variables as the confounders when evaluating bone mineral status of this population.
The quality assurance of a densitometer was performed daily and in-vitro and in-vivo stability, as well as the coefficients of variation in our laboratory were reported previously [25].
Serum calcium, circulating 25 hydroxy vitamin D (25-OHD) and parathyroid hormone (PTH) are major factors involved in Ca homeostasis, bone metabolism and the pathogenesis of age-related bone loss. In order to evaluate their status and detect possible abnormalities otherwise common in elderly, such as low circulating vitamin D [27] or hyperparahyroidism [28], we analyzed them in serum samples and the results were reported previously [29]. In this study, both serum PTH and 25-OHD were used as confounders.
For the assessment of typical Ca intake over a longer period of time (at least for a past year), participants completed a food frequency questionnaire [30] with the help of a registered dietitian. We also assessed the consumption of Ca supplements, as described previously [24], and examined separately the relationship of Ca from food and total (supplements + food) with bones, alcohol, caffeine and smoking.
Alcohol and caffeine consumption was assessed using questionnaires designed to determine long-term (at least for a last year and extending back to several years) frequency, amount and source of each, with the help of the same dietitian. Both alcohol and caffeine consumption were expressed as drinks per day, from which g/day and mg/day, respectively, were calculated. The amounts of alcohol and caffeine assigned to approximate servings of each are as follows: serving of beer (12 fl oz, 360 mL) = 13 g alcohol, wine (4 fl oz, 120 mL) = 12 g alcohol, liquor (1.5 fl oz, 45 mL) = 14 g alcohol, a serving of brewed coffee (6 fl oz, 180 mL) = 103 mg caffeine, brewed tea (6 fl oz, 180 mL) = 36 mg caffeine and caffeinated drinks (12 fl oz, 360 mL) = 46 mg caffeine [31]. There were no current smokers (as per inclusion criteria of a larger study), but the history of smoking was recorded, including number of years and average number of packages smoked per day. The product of the two was calculated and expressed as smoke exposure.
Past physical activity was assessed using interview format with a modified version of the Allied Dunbar National Fitness Survey for older adults [32] and was completed with the help of a dietitian. It was assessed as percent of adult life, from age 18 years to the present, engaged in sport and recreational activities (such as low impact aerobics, tennis, cycling) of an intensity of at least 4 kcal/minute on a regular basis, at least once a week for three months of the year (with 3–12 months of regular activities equaling one year). Past physical activity was used as a confounder, since it showed strong influence on BMD in this population [26].
Data Analysis
All data are presented as mean ± SD (unless otherwise noted) and were calculated using Data Desk® (Data Description Inc., Ithaca, NY) and StatisticaTM (StatSoft Inc. Tulsa, OK). Pearson’s correlation coefficient, r, was calculated as a part of preliminary analysis to identify association among various variables. The participants were divided into groups of past smokers and never smokers, as well as coffee and/or alcohol consumers and non-consumers, and bone variables were compared between the groups. Ca intake was used as a stratification variable in the above subgroup analyses, and subjects were divided below and above the median intake. For the evaluation of 25-OHD and PTH the subjects were stratified into groups according to age (69.9 and 70 years) and season during which the blood was drawn. ANOVA, with post-hoc two sample t tests, was conducted to determine the group differences in all cases.
Stepwise multiple regression models were utilized to assess the relationships between alcohol, caffeine, smoking and bone mineral density or content of various skeletal sites. Diagnostics and residual plots for these models were analyzed and no heteroskedascisity and/or non-normality were detected. Partial regression to remove collinearity among predictors and Hadi’s influence method with potential-residual-plots analyses to identify influential data points were performed [33]. Measures of smoking, caffeine and alcohol consumption were mutually adjusted—each for the effects of other two. Each model was also adjusted for age, components of body composition (lean and fat tissue), to control for body size, calcium intake, serum PTH and 25-OHD and past physical activity. The overall accepted level of significance was set at p 0.05.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTSACKNOWLEDGMENTSREFERENCES |
|---|
|
|
Pearson’s r was calculated to examine simple relationship among alcohol, caffeine, smoking and various other variables, including BMD of all measured skeletal sites, as well as to identify all possible confounders. Caffeine (mg/day) and alcohol (either g/day or number of drinks/day) were positively related, with the highest r = 0.216, p = 0.01 and so were caffeine and measures of smoking (years of smoking, packs/day or smoke exposure), with the highest r = 0.177, p = 0.04. The relationship between measures of smoking and alcohol consumption was not statistically significant, although the trend was positive.
There was a negative relationship between all measures of past smoking and Ca intake with r ranging from -0.196 (p = 0.03) to -0.254 (p = 0.00). Past smoking was also negatively related to serum vitamin D (r = -0.269, p = 0.00) and number of reproductive years (r = -0.216, p = 0.01), but positively to weight and BMI (r = 0.208, p = 0.02, for latter). Measures of alcohol consumption were positively related to spine BMD (r = 0.197, p = 0.02), serum 25-OHD and negatively to PTH. Although there was no significant relationship between 25-OHD and PTH, and neither of them showed significant relationship with BMD or BMC of any skeletal sites analyzed in simple regression/correlation [29], since they were associated to alcohol and smoking, we included them as confounders in multiple regression models.
To examine the relationship between BMD and alcohol, caffeine and smoking simultaneously with Ca intake, we conducted the subgroup analyses. We stratified subjects into groups below and above median of 750 mg/day for Ca intake and into consumers and non-consumers of alcohol and caffeine, as well as past and never-smokers and separately examined the sets of two variables (e.g., alcohol and Ca, caffeine and Ca, smoking and Ca, adjusting for other confounders) and their interaction with BMD of each skeletal site. Fig. 1 depicts the interaction between calcium and alcohol intake on BMD of lumbar spine (L1–L4). Although the relationship was at the border of statistical significance (p = 0.07), there was a trend for the augmented effect of alcohol on BMD in both calcium-intake groups (< and > median Ca). A similar relationship was found for the TBBMD (p = 0.09), but not for the regions of hip or forearm.
|
0.05. There was no statistical difference in BMD between caffeine non-consumers in <median Ca group and caffeine consumers or non-consumers in >median Ca group. Similar relationship was found for the BMD of all other sites of femoral region, TBBMD and, at a lower magnitude, in ultradistal regions of forearm.
|
|
|
|
DISCUSSION |
|---|
The beneficial effect of moderate alcohol intake could be due to several reasons. It was shown earlier that alcohol stimulates aromatization of androgens to estrogens, which in postmenopausal women are the only source of estrogen [34]. Alcohol might inhibit osteoclasts, and by that bone resorption, particularly important for the elderly [35]. A study by Rapuri et al. gave even more insight into this issue by showing that moderate alcohol consumption tempers bone remodeling, as evidenced by decreased markers of bone turnover (serum osteocalcin and urinary N-telopeptides), as well as decreased PTH [6]. Similarly, we found in our study alcohol to be negatively related to serum PTH and positively to 25-OHD, the latter being an indicator of vitamin D (1,25 dihydroxy) hormonal status. This could have indirectly contributed to the positive relationship between alcohol and BMD in both subgroup (Fig. 1) and multiple regression analyses (Table 3). It is interesting to note that in a subgroup analysis alcohol tended to augment the effect of Ca and in a multiple regression model alcohol had a stronger relationship with spinal BMD than even Ca.
According to some, moderate drinking in US refers to about two drinks per day for men and one for women [36]. Our participants who drank had on average about 0.6 drinks, amounting to about 8 g of alcohol/day. This was similar to the amount that participants in Nurse’s Health Study had [4], as well as to 489 elderly women in the study by Rapuri et al. [6], both reporting higher spinal BMD associated with alcohol. It is hard to speculate whether a higher amount of alcohol would have shown better association in our population, as it was reported in the French EPIDOS study [5]. In that study, the amount of 11–29 g alcohol/day (to which authors refer as moderate), mostly wine, was positively associated with trochanteric BMD (predominantly trabecular bone). Since their population consisted of over 7,000 elderly women, they were able to stratify them into low, moderate and heavy drinkers. The consumption of 10 g/day had no effect on bone, while the consumption of 30 g/day showed negative association with TBBMD. Based on the above stated evidence and our own, it seems that the effect of alcohol is higher on trabecular (spine and trochanter) than on cortical bone. The mechanism for this speculation needs to be examined.
On an opposite note, however, Hernandez-Avila et al. [7] showed that 2 drinks of alcohol/day, which they equate to about 25 g alcohol/day, increased risk of both hip and forearm fracture in their population of almost 85,000 middle aged women. Their study was prospective and evaluated only risk of fractures after six years of follow-up. The women in that study were younger (34 to 59 years), and, although some confounders were accounted for, the strong hormonal influences in these peri- and menopausal women might have obscured the subtle effects of other factors such as alcohol. The higher fracture rate might have been simply due to the higher risk of falling experienced by women drinking 2 drinks/day.
Caffeine, as a mild diuretic, temporarily increases urinary Ca excretion [9], and some evidence shows that there is no compensation for the losses after 2 to 24 hours [37,38], while there is little evidence showing higher Ca losses over a longer period. The negative effect of caffeine, at least on Ca balance, and presumably indirectly on bones, might also be due to the impaired endogenous Ca absorption [10] and, therefore, particularly present in women with low Ca intake [17,18].
In our subgroup analysis, the negative association between caffeine and BMD was particularly present in all regions of hip (Fig. 2), TBBMD and, to a lesser extent, in ultradistal regions of forearm (not shown). The negative association between caffeine and BMD of hip regions was attenuated in the women with higher Ca intake (750 mg/day), thereby confirming the assumptions from previous studies [17,18]. In multiple regression models, there was a negative association between caffeine and BMD of almost all skeletal sites, but most pronounced in the models with total body, spine and 1/3 distance radius BMD (Table 3). Although the average amount of caffeine was relatively low, 268 mg/day and below what is arbitrarily considered to be a moderate consumption (300 mg/day, about three 6 fl oz cups or 540 mL of brewed coffee) [39], the range was high (0.1–922 mg/day) and Ca from food relatively low.
The evidence of the effect of caffeine on either BMD or fracture risk still remains inconclusive, and it is largely dependent on the design of each particular study and what other adjustments were made. For example, Johansson et al. reported a strong negative association between BMD and fracture risk and coffee drinking in older women [15], which almost disappeared when data were adjusted for some osteoporosis risk factors. Lloyd et al. conducted a study in postmenopausal women with high range of caffeine intake (0–1400 mg/day) and assessed its effect on hip and total body BMD at baseline and two years later [13,14]. They found no evidence of detrimental caffeine effect on bones, regardless of the Ca status or some other adjustments. The situation becomes even more complicated when gene interaction is brought into picture, as in the Rapuri et al. study [40]. Although theirs was a Ca intervention study in postmenopausal women, the subjects were assessed at baseline and placebo group evaluated after three years for the effect of caffeine on bones. They found that women who consumed >300 mg/day of caffeine had a significantly higher rate of bone loss in the spine. In addition, those with tt vitamin D receptor genotype, the one defined by TaqI and strongly concordant with BsmI allele BB [41], were more susceptible to the effects of caffeine than TT (concordant with bb) genotypes. The sample size in each genotype category however, was small, making it impossible to examine interaction simultaneously with Ca. The interaction of genotype with caffeine and pronounced bone loss due to caffeine in a more susceptible genotype might explain some of the controversies and inability of other studies to show clearly the effect of caffeine.
The mechanism of detrimental effect of smoking on bone is complex and probably includes a combination of factors. Some explanations indicate direct toxic effect on osteoblasts and collagen formation [42], reduced Ca absorption [21], alteration of gonadal and adrenal cortical hormones and a shortened number of reproductive years causing earlier onset of menopause [43,44]. However, according to Law and Hackshaw, who performed meta analysis of 29 cross-sectional and 19 case-control studies, reporting BMD and fracture rates among smokers and non-smokers, none of these factors alone or in a combination can explain the magnitude of the bone loss and increased fracture risk in elderly and, at the same time, their almost complete absence in premenopausal women [19].
The participants in our study were only past smokers (n = 61) and about 55% of women (n = 75) never smoked. This could have been a reason for the absence of statistically significant association between measures of smoking and BMD of any skeletal site in our multiple regression analyses, when all confounders were accounted for (Table 3). The subgroup analyses, however, showed a significant negative association between BMD of total body (Fig. 3) and to a lesser extent of spine and femur (not shown) and past smoke exposure. That effect was attenuated in participants with higher Ca intake.
It is worth noting that studies examining the effect of smoking history on BMD or fracture risk in currently non-smoking postmenopausal women, who are not on hormone replacement therapy, are rare [19]. The past smokers in our study were significantly heavier than never-smokers but had a shorter reproductive life, the latter being addressed as one of the potential consequences of smoking and a subsequent cause for lower bone mass [43,44]. There was also a negative relationship between measures of past smoking and Ca intake and serum vitamin D, implying that past smokers might have had poorer nutritional habits, which persisted later.
Two measures of smoking that we examined were highly correlated: r = 0.710 for years of smoking and packs/day. Consequently, their product, smoke exposure, presenting the duration and quantity of smoking, was stronger associated with BMD of various skeletal sites than either one of the former. This is, however, in disagreement with the study by Grainge et al., who showed that BMD was more related to the duration of smoking than to the duration and amount in their sample of postmenopausal women who were current smokers [45]. Surprisingly, their data also showed a higher effect of smoking on BMD in premenopausal than in postmenopausal women, which is contradictory to the conclusions from the Law and Hackhshaw meta-analysis [19].
We want to address some issues that might come under scrutiny. We separately examined the relationship of Ca from food and total (supplements and food) with bone variables. In most instances, the relationship was stronger with the amount assessed just from food. The typical amount of Ca consumed from food probably reflects more constant and regular intake. However, with today’s health-consciousness and influence of dietary and health messages many people are taking various supplements. This phenomenon is a relatively new one, at least in the magnitude we noticed (in over 70% participants), and the amount of any supplement, including Ca, may vary, depending on the regularity and duration of supplement consumption. This variability and irregularity in intake, as well as relatively shorter time of consumption, compared to that of food might be a reason for Ca from food to be more reliable and more strongly associated with other variables, including bones [24].
The limitation of our study is its observational nature and, compared with large epidemiological studies, its relatively small sample size. It particularly might have been a case in subgroup analyses in which, due to stratification, the number in each group was even more reduced. The small number of consumers of alcohol and caffeine and past smokers (86, 105 and 61, respectively) made stratification by the level of consumption statistically unsound. On the other hand, in a smaller study, it is possible to recruit a more uniform sample with strict exclusion/inclusion criteria and measure all array of important biological and life style variables and account for them as potential confounders. Even more so, it is possible to collect subject-based recalls and self-reported data (crucial in nutritional studies) and questionnaires more accurately, as subjects can be individually instructed and interact with researchers, as was the case in our study. Therefore, although epidemiological studies give enough power for high generalizability, the smaller studies might have more accurate presentation of data. However, none of these studies could provide a firm cause and effect, but just an association that sometimes might be strong enough to make us comfortable in accepting it as such.
In summary, the results of this study support the notion that consumption of alcohol of about half to one drink a day (or 8 g alcohol/day), mostly as wine, is favorably associated with BMD of spine and total body BMD in postmenopausal women. Alcohol also augmented the effect of Ca in the same skeletal sites. Caffeine consumption of about two and a half 6-fl oz cups a day (between 200–300 mg caffeine/day) had a negative association with most of the skeletal sites, which was attenuated in women with higher Ca intake (median of 750 mg/day). Although there was no significant association between past smoking and BMD of any skeletal site in the multiple regression analyses, the past smokers (who smoked on an average of 24 years of about one pack of cigarettes a day) had lower BMD in total body, spine and femur than non-smokers when evaluated in subgroup analyses, and the association was attenuated in participants with higher Ca intake (750 mg/day).
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTSACKNOWLEDGMENTSREFERENCES |
|---|
Received April 1, 2002. Accepted July 23, 2002.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTSACKNOWLEDGMENTSREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
