JOURNAL OF BONE AND MINERAL RESEARCH

Volume 14, Number 1, 1999

Blackwell Science, Inc.
1999 American Society for Bone and Mineral Research

Smoking, Antioxidant Vitamins, and the Risk of

Hip Fracture
LSSON,2 LARS HOLMBERG,3 ALICJA WOLK,4
HKAN MELHUS,1 KARL MICHAE
and SVERKER LJUNGHALL1

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

osteoclastogenesis.(3) NF-B is activated by exposure to

cells to mitogens, cytokines, or oxidative stress, and it appears that reactive oxygen intermediates (ROIs) may represent the final common mechanism for activation of NFB by releasing IB from the stimulatory subunits.(46)
Smoking is associated with increased concentrations of
ROIs and reduced levels of antioxidant vitamins.(7) We hypothesized that the oxidant stress associated with smoking
may contribute to bone loss and increased fracture risk, and
that antioxidants may protect against oxidant-mediated
bone loss by virtue of their ability to scavenge free radicals
and to inhibit the activation of NF-B.(8)
We have previously carried out a large nested casecontrol study of diet and hip fracture risk.(9) We therefore

Internal Medicine, University Hospital, Uppsala, Sweden.

Orthopaedics, University Hospital, Uppsala, Sweden.
Surgery, University Hospital, Uppsala, Sweden.
Medical Epidemiology, Karolinska Institute, Stockholm, Sweden.

129

130

MELHUS ET AL.
TABLE 1. CHARACTERISTICS

OF

CURRENT

AND

NEVER SMOKERS

IN THE

DIETHIP FRACTURE STUDY*

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

* Unless otherwise stated, values are means SD.

decided to investigate whether the dietary intake of antioxidant vitamins may modify the risk of hip fracture in
smokers.

MATERIALS AND METHODS

The Swedish Mammography Cohort(10,11) and the study
design have previously been presented in detail.(9) Briefly,
in 19871990, all women aged 4076 years who were residents in either of two counties in central Sweden (source
population 90,303) were mailed a validated(11) foodfrequency questionnaire. In this food frequency questionnaire, the women were asked to report their usual intake of
60 foods and beverages during the most recent 6 months.
The questionnaire included all common Swedish foods and
permitted calculation of the total energy intake in each individual. For each food item, the women could choose from
eight response categories, and additional open questions
about consumption of dairy products were included (number of glasses and type of milk/day and number of slices and
type of cheese/day). The questionnaire also requested information about age, weight, marital status, education, parity, and the use of a special diet. About 74% (66,651
women) replied. Participants who sustained a first hip fracture after inclusion into the cohort, with 264 months of
follow-up, were defined as cases in a case-control analysis
of hip fracture nested within the cohort. For every case,
four controls, individually matched by age and county of
residence, were selected from the cohort. A second questionnaire concerning potential confounding factors not included in the first questionnaire was mailed to all cases and
controls in 19921993.(9) Questions about the following
items were asked: age at menopause, hormone replacement
therapy (HRT) and oral contraceptives and their respective
duration of use, use of oral cortisone therapy with time of
exposure, diabetes mellitus with type of treatment, vitamin
and trace element supplementation, physical activity in leisure time on a five-point scale at four different ages, athletic

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-

SMOKING, ANTIOXIDANT VITAMINS, AND RISK OF HIP FRACTURE

TABLE 2. CONTINUED

TABLE 2. THE ASSOCIATION OF SMOKING AND

CONFOUNDING VARIABLES WITH HIP FRACTURE RISK

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)

All variables were included in the multivariate model. Relative

risk of hip fracture was estimated as OR with 95% CI. Ranges of
quartiles are shown when applicable.

ture (distal forearm or proximal humerus vs. no fracture;

vertebral fractures were not included since spinal X-rays
could not be performed to confirm the diagnosis); HRT
(never, former, current use); menopausal age (< 49, 4950,
5152, > 52 years); multivitamin and calcium supplements
(ever vs. never use); diabetes (nondiabetes, oral or insulin
treatment); educational level (primary school only vs. a
higher educational level); and marital status (single vs. married or cohabitee). We also examined regression models
that included an interaction term between current smoking
status (current smoker vs. never smoker) and vitamin E and
C, respectively (low vs. high intake).

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

Daily dietary intake

Low vitamin E ( 6.2 mg/day)
never smoker
current smoker
High vitamin E (> 6.2 mg/day)
never smoker
current smoker
Low vitamin C ( 67 mg/day)
never smoker
current smoker
High vitamin C (> 67 mg/day)
never smoker
current smoker
Low beta-carotene ( 3.5 mg/day)
never smoker
current smoker
High beta-carotene (> 3.5 mg/day)
never smoker
current smoker
Low selenium ( 24.7 mg/day)
never smoker
current smoker
High selenium (> 24.7 mg/day)
never smoker
current smoker
Low calcium ( 786 mg/day)
never smoker
current smoker
High calcium (> 786 mg/day)
never smoker
current smoker
Low vitamin B6 ( 1.7 mg/day)
never smoker
current smoker
High vitamin B6 (> 1.7 mg/day)
never smoker
current smoker

IN

RELATION TO DIETARY INTAKE

CURRENT SMOKERS

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.

relative risk for hip fracture in individuals with a low intake

of both vitamins (Table 4). The adjusted OR for current
smokers versus never smokers then increased to 4.9 (95%
CI 2.211.0). In contrast, no increased risk was observed for
current smokers with a high dietary intake of vitamins E
and C. To investigate whether the smoking effect is transient or if some of it is sustained and modified by antioxidants, we included former smokers in another analysis
(Table 5). A more modest effect of vitamin E and C intake
was found among former smokers.

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,

SMOKING, ANTIOXIDANT VITAMINS, AND RISK OF HIP FRACTURE

TABLE 4. RISK

OF

HIP FRACTURE

IN

Daily dietary intake

Low vitamin E + low vitamin C
never smoker
current smoker
Low vitamin E + high vitamin C
never smoker
current smoker
High vitamin E + low vitamin C
never smoker
current smoker
High vitamin E + high vitamin C
never smoker
current smoker

RELATION

133

TO THE COMBINED DIETARY INTAKE

VERSUS NEVER SMOKERS

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

RELATION TO DIETARY INTAKE

SMOKING STATUS

OF

VITAMIN E

OR

AND

Daily dietary intake

Low vitamin E
never smoker
current smoker
< 10 pack-years
10 pack-years
former smoker
< 10 pack-years
10 pack-years
High vitamin E
never smoker
current smoker
< 10 pack-years
10 pack-years
former smoker
< 10 pack-years
10 pack-years
Low vitamin C
never smoker
current smoker
< 10 pack-years
10 pack-years
former smoker
< 10 pack-years
10 pack-years
High vitamin C
never smoker
current smoker
< 10 pack-years
10 pack-years
former smoker
< 10 pack-years
10 pack-years

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)

Risk of hip fracture was estimated as in Table 3.

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

found a protective effect of vitamins E and C, but not of

beta-carotene or selenium.
Smoking may have a direct action on bone. There is clinical evidence consistent with the hypothesis that cigarette

134

MELHUS ET AL.

smoking is associated with increased bone resorption.

Smokers have lower levels of parathyroid hormone and
25-hydroxyvitamin D3 than nonsmokers,(1820) both expected consequences of an increased release of calcium
from resorbed bone. Perhaps because of these hormonal
characteristics, smoking seems to lead to a decreased calcium absorption.(21)
It has been suggested that ROI may be involved in the
bone-resorptive process.(2224) ROI play a central role in
cellular activation in response to a variety of stimuli, by
activation of the transcription factor NF-B,(4,5) that regulate the production of osteoclastogenic cytokines, such as
IL-6 which has been implicated in osteoporotic bone
loss.(25) Vitamin E derivatives inhibit NF-B activation(6,26,27) and vitamin E protects against cellular lipid peroxidation in cartilage and bone cells.(28)
Smoking is associated with increased concentrations of
ROI as well as reduced levels of antioxidant vitamins.(7)
Thus, our results are consistent with the hypothesis of a role
for oxidant stress in the adverse effects on the skeleton of
smoking. Our data also suggest that an adequate antioxidant status may provide a useful approach in protecting
against oxidant-mediated bone loss. One would expect that
smoking should not be a risk factor in populations with a
high intake of antioxidant vitamins, such as in Southern
Europe.(29) Interestingly, this was the case in the Mediterranean Osteoporosis Study.(30)
A less likely possibility would be that vascular effects of
smoking adversely affect bone,(31,32) and that antioxidant
vitamins reduce these effects by, e.g., protecting against the
cytotoxic effects of oxidized low-density lipoprotein.(16)
The strengths of the present investigation are its large
size, population-based design, prospective dietary assessment, and high response rate. The detailed questionnaires
enabled us to consider several confounding factors. A limitation of this study may be that the included variables are of
self-reported origin. Actual measures of exposure would
have reduced misclassification, but to what extent this influences our estimates is uncertain.

ACKNOWLEDGMENTS

5.

6.
7.
8.
9.

10.

11.

12.
13.

14.
15.

16.
17.

18.

19.

This study was supported by grants from the Swedish

Medical Research Council.
20.
21.

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.

vation of the NF-B transcription factor and HIV-1. EMBO J

10:22472258.
Schreck R, Albermann K, Baeuerle PA 1992 Nuclear factor
kappa B: An oxidative stress-responsive transcription factor of
eukaryotic cells (a review). Free Radic Res Commun 17:221
237.
Sen CK, Packer L 1996 Antioxidant and redox regulation of
gene transcription. FASEB J 10:709720.
Duthie GG, Arthur JR, James WP 1991 Effects of smoking and
vitamin E on blood antioxidant status. Am J Clin Nutr 53
(Suppl.):1061S1063S.
Conner EM, Grisham MB 1996 Inflammation, free radicals,
and antioxidants. Nutrition 12:274277.
Michaelsson K, Holmberg L, Mallmin H, Sorensen S, Wolk A,
Bergstrom R, Ljunghall S 1995 Diet and hip fracture risk: A
case-control study. Study Group of the Multiple Risk Survey
on Swedish Women for Eating Assessment. Int J Epidemiol
24:771782.
Holmberg L, Ohlander EM, Byers T, Zack M, Wolk A, Bergstrom R, Bergkvist L, Thurfjell E, Bruce A, Adami HO 1994
Diet and breast cancer risk: Results from a population-based,
case-control study in Sweden. Arch Intern Med 154:18051811.
Hunter DJ, Spiegelman D, Adami HO, Beeson L, van den
Brandt PA, Folsom AR, Fraser GE, Goldbohm RA, Graham
S, Howe GR, et al. 1996 Cohort studies of fat intake and the
risk of breast cancerA pooled analysis. N Engl J Med
334:356361.
la Vecchia C, Negri E, Levi F, Baron JA 1991 Cigarette smoking, body mass and other risk factors for fractures of the hip in
women. Int J Epidemiol 20:671677.
Grisso JA, Kelsey JL, Strom BL, OBrien LA, Maislin G, LaPann K, Samelson L, Hoffman S 1994 Risk factors for hip
fracture in black women. The Northeast Hip Fracture Study
Group. N Engl J Med 330:15551559.
Cumming RG, Klineberg RJ 1994 Case-control study of risk
factors for hip fractures in the elderly. Am J Epidemiol
139:493503.
Jarvinen R, Knekt P, Seppanen R, Reunanen A, Heliovaara
M, Maatela J, Aromaa A 1994 Antioxidant vitamins in the diet:
Relationships with other personal characteristics in Finland. J
Epidemiol Community Health 48:549554.
Diaz MN, Frei B, Vita JA, Keaney J JrF 1997 Antioxidants and
atherosclerotic heart disease. N Engl J Med 337:408416.
Salvini S, Hennekens CH, Morris JS, Willet WC, Stampfer MJ
1995 Plasma levels of the antioxidant selenium and risk of
myocardial infarction among U.S physicians. Am J Cardiol
76:12181221.
Gudmundsson JA, Ljunghall S, Bergquist C, Wide L, Nillius SJ
1987 Increased bone turnover during gonadotropin-releasing
hormone superagonist-induced ovulation inhibition. J Clin Endocrinol Metab 65:159163.
Mellstrom D, Johansson C, Johnell O, Lindstedt G, Lundberg
PA, Obrant K, Schoon IM, Toss G, Ytterberg BO 1993 Osteoporosis, metabolic aberrations, and increased risk for vertebral fractures after partial gastrectomy. Calcif Tissue Int
53:370377.
Hopper JL, Seeman E 1994 The bone density of female twins
discordant for tobacco use. N Engl J Med 330:387392.
Krall EA, Dawson-Hughes B 1991 Smoking and bone loss
among postmenopausal women. J Bone Miner Res 6:331338.
Garrett IR, Boyce BF, Oreffo RO, Bonewald L, Poser J,
Mundy GR 1990 Oxygen-derived free radicals stimulate osteoclastic bone resorption in rodent bone in vitro and in vivo. J
Clin Invest 85:632639.
Key LL Jr, Ries WL, Taylor RG, Hays BD, Pitzer BL 1990
Oxygen derived free radicals in osteoclasts: The specificity and
location of the nitroblue tetrazolium reaction. Bone 11:115
119.
Ries WL, Key LL Jr, Rodriguiz RM 1992 Nitroblue tetrazolium reduction and bone resorption by osteoclasts in vitro inhibited by a manganese-based superoxide dismutase mimic. J
Bone Miner Res 7:931939.
Manolagas SC, Jilka RL 1995 Bone marrow, cytokines, and

SMOKING, ANTIOXIDANT VITAMINS, AND RISK OF HIP FRACTURE

26.
27.
28.
29.

30.

bone remodeling. Emerging insights into the pathophysiology

of osteoporosis. N Engl J Med 332:305311.
Suzuki YJ, Packer L 1993 Inhibition of NF-kappa B activation
by vitamin E derivatives. Biochem Biophys Res Commun
193:277283.
Suzuki YJ, Packer L 1993 Inhibition of NF-kappa B DNA
binding activity by alpha-tocopheryl succinate. Biochem Mol
Biol Int 31:693700.
Xu H, Watkins BA, Seifert MF 1995 Vitamin E stimulates
trabecular bone formation and alters epiphyseal cartilage morphometry. Calcif Tissue Int 57:293300.
Cruz JA, Moreiras-Varela O, van Staveren WA, Trichopoulou
A, Roszkowski W 1991 Intake of vitamins and minerals. Euronut SENECA investigators. Eur J Clin Nutr 45 (Suppl.
3):121138.
Johnell O, Gullberg B, Kanis JA, Allander E, Elffors L, Dequeker J, Dilsen G, Gennari C, Lopes Vaz A, Lyritis G, et al.
1995 Risk factors for hip fracture in European women: The
MEDOS Study. Mediterranean Osteoporosis Study. J Bone
Miner Res 10:18021815.

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.

Address reprint requests to:

Hkan Melhus
Department of Internal Medicine
University Hospital
S-751 85 Uppsala, Sweden
Received in original form February 12, 1998; in revised form June
17, 1998; accepted August 18, 1998.