Professional Documents
Culture Documents
ABSTRACT
Smoking increases the concentrations of free radicals, which have been suggested to be involved in bone resorption.
We examined whether the dietary intake of antioxidant vitamins may modify the increased hip fracture risk
associated with smoking. We prospectively studied 66,651 women who were 4076 years of age. Forty-four of the
cohort members who sustained a first hip fracture within 264 months of follow-up (n = 247) and 93 out of 873
age-matched controls were current smokers. Information on diet was obtained by a validated food-frequency
questionnaire. The relative risk of hip fracture for current versus never smokers was analyzed in relation to the
dietary intake of antioxidant vitamins stratified into two categories (low/high), where median intakes among the
controls were used as cut-off points. After adjustment for major osteoporosis risk factors, the odds ratio (OR) for
hip fracture among current smokers with a low intake of vitamin E was 3.0 (95% confidence interval 1.65.4) and
of vitamin C 3.0 (1.65.6). In contrast, the OR decreased to 1.1 (0.52.4) and 1.4 (0.73.0) with high intakes of
vitamin E and C, respectively. This effect was not seen for beta-carotene, selenium, calcium, or vitamin B6. In
current smokers with a low intake of both vitamins E and C, the OR increased to 4.9 (2.211.0). The influence of
the intake of these two antioxidant vitamins on hip fracture risk was less pronounced in former smokers. Our
results suggest a role for oxidant stress in the adverse effects on the skeleton of smoking, and that an insufficient
dietary intake of vitamin E and C may substantially increase the risk of hip fracture in current smokers, whereas
a more adequeate intake seems to be protective.(J Bone Miner Res 1999;14:129135)
INTRODUCTION
MOKING IS RECOGNIZED as a major risk factor for osteoporotic fractures. A recent meta-analysis of cigarette
smoking, bone mineral density, and risk of hip fracture(1)
showed that the more commonly postulated mechanisms
whereby smoking increases postmenopausal bone loss
(such as lower than average body weight, earlier than average menopause, less exercise of smokers, and antiestrogen effects) account for little of the effect on bone density,
suggesting that smoking may have a direct action on
bone.(1,2)
The finding of osteopetrosis in mice lacking NF-B1 and
NF-B2 indicates that NF-B proteins are important for
Department
Department
Department
4
Department
2
3
of
of
of
of
129
130
MELHUS ET AL.
TABLE 1. CHARACTERISTICS
OF
CURRENT
AND
NEVER SMOKERS
IN THE
Cases
Controls
Characteristic
Current smoker
(n = 44)
Never smoker
(n = 161)
Current smoker
(n = 93)
Never smoker
(n = 653)
Age (years)
Weight (kg)
Height (cm)
Age at menopause (years)
Previous fx distal radius (%)
Diabetes (%)
Ever HRT (%)
Dietary vitamin E (mg/day)
vitamin C (mg/day)
beta-carotene (mg/day)
calcium (mg/day)
energy (kcal/day)
65.7 7.6
62.4 10.4
162.1 6.0
47.7 6.0
31.8
9.1
14.3
5.6 2.7
72.4 59.1
4.4 4.6
764 340
1327 407
68.6 6.7
67.1 10.9
164.7 6.2
48.0 6.5
32.0
14.8
6.0
6.5 2.6
84.8 44.8
5.0 4.1
864 341
1539 472
65.4 7.7
65.6 10.5
163.3 6.3
47.5 6.7
12.9
5.4
18.3
6.1 2.2
65.8 34.9
4.1 3.6
802 346
1474 424
68.3 7.1
68.9 11.1
162.7 6.2
48.9 6.7
18.5
6.9
15.0
6.3 2.3
75.4 43.2
4.6 3.6
821 323
1485 421
decided to investigate whether the dietary intake of antioxidant vitamins may modify the risk of hip fracture in
smokers.
sports by type and duration, fractures other than hip fracture after the age of 40, use of thiazides, and smoking with
amount in each decade. The lifetime tobacco consumption
for each individual was calculated by pack-years. One packyear means smoking one pack of 20 cigarettes/day during 1
year. In the analysis, the smokers were divided into former
and current smokers, with concomitant subdivision into
smoking of more or less than 10 pack-years. None of the
cases stopped smoking in the time period between the hip
fracture and the second questionnaire. Of those eligible,
92.5% of the cases and 89.1% of the controls returned the
second questionnaire, and 247 cases and 873 controls were
finally included in the analysis. There were 44 current and
42 former smokers among the cases compared with 93 current and 127 former smokers among the controls. In our
previously published study,(9) several established hip fracture risk factors were identified. However, there were no
significant differences between cases and controls concerning the use of multivitamin or calcium supplements, alcohol,
caffeine, or thiazides.(9)
Data analysis
The relative risk for hip fracture among current and
former smokers compared with never smokers was estimated as odds ratio (OR) with 95% confidence intervals
(CIs) by unconditional logistic regression, to avoid loss of
matched sets after stratification by high/low dietary vitamin
or mineral intakes. Median intakes among controls were
used as cut-off points. In the presented unconditional logistic regression analyses, we included the matching variables,
age, and county of residence, but this had no influence on
our estimates. The matching for age was very close, i.e., the
controls for each case were born in the same month and
year as their case.
Data were analyzed in a multivariate model including the
following covariates: age; body mass index; total energy
intake; intake of fat, protein, calcium, alcohol, and caffeine;
life-time leisure physical activity condensed to a single score
(all by quartiles of the controls); previous osteoporotic frac-
Variable
Smoking
never smoker
current smoker
former smoker
Body mass index (kg/m2)
< 23.0
23.025.2
25.327.9
> 27.9
HRT
never use
current use
former use
Physical activity, leisure time
Q1 (lowest activity)
Q2
Q3
Q4
Diabetes mellitus
nondiabetes
diabetes, insulin treatment
diabetes, oral treatment
Menopausal age (years)
< 49
4950
5152
> 52
Energy intake (kcal/day)
< 1183
11831437
14381695
> 1695
Fat (g/day)
< 39
3948
4960
> 60
Protein (g/day)
< 44
4552
5365
> 65
Calcium (mg/day)
< 585
586786
7871000
> 1000
Alcohol (g/day)
< 0.9
0.91.2
1.32.4
> 2.4
Coffee
Q1 (lowest intake)
Q2
Q3
Q4
n cases/
controls
Multivariate
model*
OR (95% CI)
161/653
44/93
42/127
1.0 (ref)
2.1 (1.33.2)
1.5 (1.02.3)
93/206
50/206
50/210
39/207
1.0 (ref)
0.6 (0.40.9)
0.6 (0.40.9)
0.4 (0.30.6)
229/741
3/53
15/80
1.0 (ref)
0.2 (0.10.6)
0.6 (0.31.1)
76/185
57/151
68/324
40/175
1.0 (ref)
1.1 (0.71.8)
0.6 (0.40.9)
0.7 (0.41.1)
229/833
9/11
17/30
1.0 (ref)
3.9 (1.510.4)
2.3 (1.24.5)
77/268
107/305
26/129
37/172
1.0 (ref)
1.3 (0.91.8)
0.7 (0.41.2)
0.9 (0.51.4)
56/218
69/129
54/218
68/219
1.0 (ref)
1.3 (0.72.4)
0.9 (0.41.9)
1.2 (0.52.9)
61/218
55/219
74/218
57/219
1.0 (ref)
0.7 (0.41.2)
0.8 (0.51.5)
0.5 (0.31.1)
55/218
57/219
73/218
62/219
1.0 (ref)
1.3 (0.72.3)
1.9 (0.94.0)
1.3 (0.53.4)
57/218
59/219
56/218
75/219
1.0 (ref)
1.0 (0.61.6)
0.9 (0.51.6)
1.5 (0.73.0)
126/441
18/61
42/152
61/220
1.0 (ref)
1.0 (0.51.8)
0.9 (0.61.4)
0.9 (0.61.4)
50/213
72/233
56/190
65/232
1.0 (ref)
1.2 (0.81.8)
1.1 (0.71.8)
1.2 (0.71.8)
131
Variable
Calcium supplementation
(ever vs. never use)
Multivitamin supplements
(ever vs. never use)
Previous osteoporotic
fracture (yes vs. no)
Menopause (yes vs. no)
Married vs. single
Primary school vs.
higher education
n cases/
controls
Multivariate
model*
OR (95% CI)
15/35
1.5 (0.73.0)
41/158
0.9 (0.61.4)
82/165
218/750
152/585
2.2 (1.63.1)
1.3 (0.82.0)
0.9 (0.61.2)
198/699
1.0 (0.61.4)
RESULTS
The major characteristics of cases and controls are shown
in Table 1. Among the cases, 18% were current smokers
compared with 11% among the controls. The effect of
smoking and confounding variables on hip fracture risk is
shown in Table 2. We found a significant interaction between current smoking status and both dietary vitamin E (p
0.02) and vitamin C (p 0.03) intake. After stratification by low and high dietary intake using the median intake
among controls as cut-off points, the relative risk of hip
fracture among current versus never smokers was analyzed.
As seen in Table 3, current smokers with a low estimated
dietary intake of vitamin E or C, had an OR of 3.0 (95% CI
1.65.4) of hip fracture in multivariate analysis. In contrast,
among current smokers with a high intake of vitamin E or
C, the OR was 1.1 (95% CI 0.52.4) and 1.4 (95% CI 0.7
3.0), respectively. For beta-carotene the association was
rather the opposite with a higher risk for current smokers
among those with a high intake, whereas a high versus a low
intake of selenium, calcium, or vitamin B6 had no or minor
influence on hip fracture risk. Since antioxidants in the diet
tend to act synergistically and the estimated, energyadjusted intake of vitamins E and C were not collinear
(correlation coefficient, r 0.11), we also investigated the
132
MELHUS ET AL.
TABLE 3. RISK
OF
HIP FRACTURE
IN
OF
SOME VITAMINS
AND
MINERALS AMONG
n cases
n controls
Univariate model
OR (95% CI)
Multivariate model
OR (95% CI)
77
31
328
46
1.0 (ref)
3.1 (1.85.3)
1.0 (ref)
3.0 (1.65.4)
84
13
325
47
1.0 (ref)
1.1 (0.62.1)
1.0 (ref)
1.1 (0.52.4)
67
30
326
51
1.0 (ref)
2.8 (1.64.8)
1.0 (ref)
3.0 (1.65.6)
94
14
327
42
1.0 (ref)
1.2 (0.62.3)
1.0 (ref)
1.4 (0.73.0)
76
25
333
50
1.0 (ref)
2.2 (1.33.8)
1.0 (ref)
1.8 (1.03.5)
85
19
320
43
1.0 (ref)
1.8 (1.03.2)
1.0 (ref)
2.6 (1.35.2)
79
21
318
50
1.0 (ref)
1.8 (1.03.2)
1.0 (ref)
2.0 (1.03.9)
82
23
335
43
1.0 (ref)
2.2 (1.23.8)
1.0 (ref)
2.2 (1.14.2)
78
24
318
49
1.0 (ref)
2.0 (1.23.5)
1.0 (ref)
2.3 (1.24.3)
83
20
335
44
1.0 (ref)
1.9 (1.13.4)
1.0 (ref)
2.0 (1.14.0)
65
25
319
57
1.0 (ref)
2.2 (1.33.9)
1.0 (ref)
2.1 (1.14.1)
96
19
334
36
1.0 (ref)
1.9 (1.03.4)
1.0 (ref)
2.4 (1.24.8)
Risk of hip fracture was estimated as OR with CI by logistic regression using never smokers as reference (ref). Dietary intake was
stratified by low/high using median intakes among the controls as cut-off points. The covariates shown in Table 2 were included in the
multivariate model.
DISCUSSION
Previously published case-control studies, including our
own, have reported fairly consistent results. Relative risks
of hip fracture among current smokers are between 1.5 and
2.0 with age adjustment, and somewhat lower after adjustment for weight and other factors.(9,1214) In the present
study, we found that a low intake of vitamins E and C
increased the relative risk almost 5-fold after adjustment for
age, weight, and other osteoporosis risk factors. In addition,
a more adequate intake seemed to protect against the adverse effects of smoking.
It has been shown that the association of dietary antioxidant vitamins with sociodemographic characteristics and
smoking can be strong enough to exert a confounding or
modifying effect in studies of diet and cardiovascular diseases.(15) Whether this may also be true for osteoporosis
and hip fracture is not known. We studied women from the
same geographic region, and adjustment for potential confounders such as educational level and marital status did
not change the results. The substantial increase in relative
risk is unlikely to be the result of generally poor nutrition,
OF
HIP FRACTURE
IN
RELATION
133
OF
VITAMINS E
AND
IN
CURRENT
n cases
n controls
Univariate model
OR (95% CI)
Multivariate model
OR (95% CI)
36
23
201
30
1.0 (ref)
4.3 (2.28.4)
1.0 (ref)
4.9 (2.211.0)
41
8
127
16
1.0 (ref)
1.7 (0.74.4)
1.0 (ref)
1.6 (0.74.1)
31
7
125
21
1.0 (ref)
1.3 (0.53.4)
1.0 (ref)
1.6 (0.46.6)
53
6
200
26
1.0 (ref)
0.9 (0.32.2)
1.0 (ref)
0.8 (0.32.3)
The risk of hip fracture was estimated as ORs with 95% CIs with the same multivariate model as in Table 3.
TABLE 5. RISK
OF
HIP FRACTURE
IN
OF
VITAMIN E
OR
AND
n cases
n controls
77
328
9
22
Univariate model
OR (95% CI)
Multivariate model
OR (95% CI)
1.0 (ref)
1.0 (ref)
18
28
2.3 (1.05.3)
3.6 (2.06.8)
1.6 (0.74.1)
4.3 (2.18.7)
7
14
36
36
0.9 (0.42.0)
1.7 (0.93.4)
1.0 (0.32.4)
2.0 (0.94.0)
84
325
1.0 (ref)
1.0 (ref)
3
10
21
26
0.6 (0.21.9)
1.5 (0.73.2)
0.5 (0.12.1)
1.5 (0.63.7)
9
12
29
27
1.2 (0.62.6)
1.7 (0.83.5)
1.2 (0.53.0)
1.6 (0.73.9)
67
326
1.0 (ref)
1.0 (ref)
7
23
22
29
1.5 (0.63.8)
3.8 (2.07.1)
1.3 (0.53.6)
4.8 (2.210.2)
7
10
29
31
1.2 (0.52.8)
1.5 (0.73.3)
1.3 (0.53.7)
2.1 (0.95.3)
94
327
1.0 (ref)
1.0 (ref)
5
9
17
25
1.0 (0.42.8)
1.3 (0.63.0)
1.1 (0.43.5)
1.6 (0.74.0)
9
16
36
32
0.9 (0.42.0)
1.8 (1.03.5)
1.0 (0.42.3)
1.8 (0.93.8)
since the observed effect was not seen for other vitamins
and minerals, and adjustment for energy, fat, and protein
intake had only minor effects on the OR. As in several
studies on antioxidants and cardiovascular disease,(16,17) we
134
MELHUS ET AL.
ACKNOWLEDGMENTS
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
REFERENCES
22.
1. Law MR, Hackshaw AK 1997 A meta-analysis of cigarette
smoking, bone mineral density and risk of hip fracture: Recognition of a major effect. BMJ 315:841846.
2. Fang MA, Frost PJ, Iida-Klein A, Hahn TJ 1991 Effects of
nicotine on cellular function in UMR 10601 osteoblast-like
cells. Bone 12:283286.
3. Iotsova V, Caamano J, Loy J, Yang Y, Lewin A, Bravo R 1997
Osteopetrosis in mice lacking NF-B1 and NF-B2 [In Process
Citation]. Nat Med 3:12851289.
4. Schreck R, Rieber P, Baeuerle PA 1991 Reactive oxygen intermediates as apparently widely used messengers in the acti-
23.
24.
25.
26.
27.
28.
29.
30.
135
31. Daftari TK, Whitesides TE Jr, Heller JG, Goodrich AC, McCarey BE, Hutton WC 1994 Nicotine on the revascularization
of bone graft: An experimental study in rabbits. Spine 19:904
911.
32. Hirota Y, Hirohata T, Fukuda K, Mori M, Yanagawa H, Ohno
Y, Sugioka Y 1993 Association of alcohol intake, cigarette
smoking, and occupational status with the risk of idiopathic
osteonecrosis of the femoral head. Am J Epidemiol 137:530
538.