Osteoporosis International

, Volume 16, Issue 7, pp 737–742

Alcohol intake as a risk factor for fracture

Authors

    • WHO Collaborating Centre for Metabolic Bone DiseasesUniversity of Sheffield Medical School
  • Helena Johansson
    • Consulting Statistician
  • Olof Johnell
    • Department of OrthopaedicsMalmo General Hospital
  • Anders Oden
    • Consulting Statistician
  • Chris De Laet
    • Department of Public HealthErasmus Medical Center
  • John A. Eisman
    • Bone and Mineral Research Program, Garvan Institute of Medical ResearchSt Vincent’s Hospital and University of NSW
  • Huibert Pols
    • Department of Internal MedicineErasmus University
  • Alan Tenenhouse
    • Division of Bone MetabolismThe Montreal General Hospital
Original Article

DOI: 10.1007/s00198-004-1734-y

Cite this article as:
Kanis, J.A., Johansson, H., Johnell, O. et al. Osteoporos Int (2005) 16: 737. doi:10.1007/s00198-004-1734-y

Abstract

High intakes of alcohol have adverse effects on skeletal health, but evidence for the effects of moderate consumption are less secure. The aim of this study was to quantify this risk on an international basis and explore the relationship of this risk with age, sex, and bone mineral density (BMD). We studied 5,939 men and 11,032 women from three prospectively studied cohorts comprising CaMos, DOES, and the Rotterdam Study. Cohorts were followed for a total of 75,433 person-years. The effect of reported alcohol intake on the risk of any fracture, any osteoporotic fracture, and hip fracture alone was examined using a Poisson model for each sex from each cohort. Covariates examined included age and BMD. The results of the different studies were merged using weighted β-coefficients. Alcohol intake was associated with a significant increase in osteoporotic and hip fracture risk, but the effect was nonlinear. No significant increase in risk was observed at intakes of 2 units or less daily. Above this threshold, alcohol intake was associated with an increased risk of any fracture (risk ratio [RR]=1.23; 95% CI, 1.06–1.43), any osteoporotic fracture (RR=1.38; 95% CI, 1.16–1.65), or hip fracture (RR=1.68; 95% CI, 1.19–2.36). There was no significant interaction with age, BMD, or time since baseline assessment. Risk ratios were moderately but not significantly higher in men than in women, and there was no evidence for a different threshold for effect by gender. We conclude that reported intake of alcohol confers a risk of some importance beyond that explained by BMD. The validation of this risk factor on an international basis permits its use in case-finding strategies.

Keywords

AlcoholHip fractureMeta-analysisOsteoporotic fractureRisk factors

Introduction

Excessive alcohol intake is a well-recognized cause of secondary osteoporosis, particularly in men [1, 2, 3]. The effects of more moderate intakes are not thought to be deleterious to skeletal health. Indeed, intakes of 7 oz (210 g; 26 units) per week may be associated with a higher BMD than in individuals who abstain from alcohol [2, 4, 5] and a lower risk of hip fracture [6] and other atraumatic fractures [7].

Higher intakes appear to be associated with an increased fracture risk [8] and hip fracture risk [6, 9, 10]. The threshold intake for high risk appears to be approximately 100 g per week in women [8], but is higher in men [6, 9]. In one study, hip fracture risk was not increased in women who had less than 14 units per week and in men who took less than 28 units per week [6].

At present assessment guidelines for osteoporosis do not include a high dietary intake of alcohol as a risk factor for case finding strategies [11, 12, 13, 14, 15, 16, 17, 18]. The question arises whether a high alcohol intake might be used as a risk factor in the stratification of fracture risk. The aim of the present study was to quantify the fracture risk associated with alcohol consumption and identify the threshold risk.

Methods

We studied 16,971 men and women from three prospectively studied cohorts drawn randomly from populations in The Netherlands, Australia, and Canada. These comprised the CaMos, DOES, and the Rotterdam Study. Brief details of the cohorts studied are given below and summarized in Table 1.
Table 1

Details of cohorts studied

Cohort

Sex

Sample size

Person-years

Mean age (years)

Age range (years)

Units of alcohol (mean/day)

Any fracture

Osteoporotic fracture

Hip fracture

CaMos

M

2,878

8,226

59.9

25–97

0.75

124

58

9

F

6,523

18,430

63.0

25–103

0.28

462

258

33

DOES

M

841

6,485

70.2

60–92

2.05

141

112

28

F

1,321

9,846

71.0

57–96

0.43

390

306

79

Rotterdam

M

2,220

13,257

67.3

55–93

2.06

137

93

29

F

3,188

19,191

68.0

55–94

0.77

499

380

101

Total

16,971

75,433

65.0

25–103

0.79

1,753

1,207

279

CaMos

The Canadian Multicentre Osteoporosis Study (CaMos) is an ongoing prospective age-stratified cohort. The study is documenting the incidence of fractures and risk factors in a random sample of 9,424 men and women aged 25 years or more, selected by telephone listings. The sampling frame is from nine study centers in seven provinces [19]. The participation rate was 33%. Characterization of individuals was by interview. Bone mineral density was measured by DXA at the hip in seven centers with the Hologic QDR 1000, and in two centers with the Lunar-DPX Alpha. Machines were cross-calibrated using the European spine phantom. For this analysis, validated fracture follow-up was available for 9,401 participants (2,878 men) with an average follow-up time of 3 years. Femoral neck BMD was measured in 8,317 individuals (2,590 men).

DOES

The Dubbo Osteoporosis Epidemiology Study (DOES) is a population-based study with multiple assessments of skeletal status in men and women aged 60 years or more from Dubbo, Australia [20]. Participation in the study was 56% of the population. Baseline measurements included BMD at the femoral neck assessed using DXA (GE-Lunar DPX). Fractures were identified through radiologists’ reports from the two centers servicing the region. Validated fracture outcomes at follow-up were available for 2,162 participants (841 men) with an average follow-up time of 8 years. Femoral neck BMD was measured in 2,071 individuals (803 men).

The Rotterdam Study

The Rotterdam Study, begun in 1990, is a prospective cohort study that aims to examine and follow up all residents aged 55 years and older living in Ommoord, a district of Rotterdam, The Netherlands [21]. By 1993, 7,983 residents had been included (response rate 78%). Bone mineral density was assessed at the femoral neck by DXA using a Lunar DPX-L [22]. Fracture follow-up was undertaken using an automated link with general practitioner computer systems and hospital admission data. Fracture data were collected and validated by two independent research physicians. For this analysis, validated fracture follow-up was available for 5,408 participants (2,220 men) with an average follow-up time of 6 years. Femoral neck BMD was measured in 4,979 individuals (2,060 men).

Baseline and outcome variables

The assessment of alcohol intake differed between the cohorts studied. For Rotterdam and DOES, intake was documented as g/day and in CaMos as servings/day. For the present study, we used units/day as the metric and divided the daily intake recorded in Rotterdam and DOES by 8 (the definition of a unit in the United Kingdom).

Prospective fracture ascertainment was undertaken by self-report and verified from hospital central databases. Fractures considered to be due to osteoporosis were analyzed, and in addition hip fracture alone was considered separately. An osteoporotic fracture was one considered to be due to osteoporosis by the investigator. For the CaMos study, osteoporotic fractures comprised fractures of the spine, pelvis, ribs, distal forearm, forearm, and hip. In the other cohorts (Rotterdam and DOES), fractures at sites considered to be characteristic for osteoporosis were extracted [23].

Bone mineral density was undertaken at the femoral neck by DXA at all centers.

Statistical methods

The risk of fracture was estimated by Poisson regression applied to each cohort and each sex separately. Covariates included current time, current age, alcohol intake, and alcohol intake times current age. Intake of alcohol was examined as a continuous or dichotomous variable. Additional models included the covariates above, with bone mineral density, current smoking, and body mass index (BMI).

The β values for each sex in each cohort were determined for each variable as a linear function of age, βkk+1.age. The estimated value of the β-coefficients and their variance was determined for each sex from the age of 50 years. The results of each cohort in the two sexes were weighted according to the variance and merged to determine the weighted mean and standard deviation. The risk ratio of those on a given intake or less, versus those on higher alcohol intakes was equal to emean. There was no significant heterogeneity in risk between cohorts (P>0.30), and a fixed effects model was used.

Results

The total sample studied was 5,939 men and 11,032 women followed for 75,433 person-years (see Table 1). During this time, there were 1,753 fractures, with 1,207 fractures thought to be related to osteoporosis, including 279 hip fractures. Bone mineral density measurements were available in 91% of individuals.

Intake of alcohol was higher in men than in women (Table 2). Seventy-seven percent of women abstained from alcohol, whereas 49% of men took no alcohol. At the other extreme, 8% of men took 5 or more units per day compared with 1% of women. Nineteen percent of men and 4% of women took more than 2 units daily.
Table 2

Units of alcohol in men and women

Units of alcohol/day

Men

Women

No.

%

No.

%

0

2,982

49.4

8,692

77.2

1

1,250

20.7

1,598

14.2

2

605

10.0

479

4.3

3

433

7.2

292

2.6

4

292

4.8

109

1.0

5

179

3.0

52

0.5

6

92

1.5

19

0.2

7

52

0.9

14

0.1

8

57

0.9

4

9

25

0.4

2

10

22

0.4

0

>10

47

0.8

4

Total

6,036

11,265

When alcohol intake was assessed as a continuous variable, high intakes of alcohol were associated with an increased risk of osteoporotic fracture or of hip fracture. For example, in men and women combined, the risk of hip fracture increased by 7% for each additional unit of intake above 1 unit daily (Table 3). The increase in risk was of borderline significance (P=0.09) for hip fracture risk in women. There was no significant difference in risk ratio between men and women (see Table 3).
Table 3

Risk ratio of fracture per unit increase in alcohol intake in men and women. Gradient of risk is not adjusted for BMD. The reference base is 1 unit per day

Outcome

Sex

RR

95% confidence interval

Osteoporotic fracture

M

1.04

1.01–1.07

F

1.08

1.02–1.14

M+F

1.05

1.02–1.08

Hip fracture

M

1.07

1.00–1.13

F

1.11

0.98–1.26

M+F

1.07

1.02–1.14

When the risk ratio was assessed according to units of alcohol consumed (using 1 unit as the referent), risk ratio increased with more than 2 units per day in both men and women (Table 4), but was not increased below this level. When the data for alcohol intake was dichotomized by intake of more than 2, more than 3 or more than 4 units daily (vs those who took less), the risk of any osteoporotic fracture or of hip fracture was somewhat, but not significantly, higher in men than in women (Fig. 1). In both men and women, risk ratios increased with the higher categories of intake and were significantly increased in men and women with intakes of more than 2 units daily. There was no effect on risk ratio when BMD was added to the model (Table 5). There was also no difference in femoral neck BMD in individuals who abstained from alcohol (Z-score = −0.03± SD 1.02), from those taking 1–2 units daily (Z-score = 0.02 ± 0.99) and from those taking >2 units daily (Z-score = 0.01 ± 1.00). There was no significant interaction of age on risk ratio.
Table 4

Risk ratio (RR) for fracture of the type indicated (and 95% confidence interval [95% CI]) according to intake of alcohol in men and women

Alcohol intake (units/day)

Men

Women

Osteoporotic fracture

Hip fracture

Osteoporotic fracture

Hip fracture

RR

95% CI

RR

95% CI

RR

95% CI

RR

95% CI

0

1.06

0.83–1.34

0.94

0.58–1.54

0.96

0.85–1.08

0.98

0.75–1.27

1

1.00

-

1.00

-

1.00

-

1.00

-

2

1.05

0.92–1.20

1.21

0.92–1.59

1.07

0.99–1.16

1.09

0.91–1.29

3

1.38

0.87–2.18

1.91

1.21–3.03

1.20

0.91–1.58

1.33

1.01–1.75

4

1.81

1.24–2.64

2.84

1.21–6.64

1.38

1.12–1.69

1.72

1.08–2.73

Fig. 1

Risk ratio (RR) and 95% confidence interval for any osteoporotic fracture or hip fracture in men and women according to the intake of alcohol

Table 5

Risk ratio (RR) for fracture and 95% confidence interval (95% CI) according to intake of alcohol in men and women combined, with and without BMD

Categorization (units/day)

Without BMD

Adjusted for BMD

RR

95% CI

RR

95% CI

Any fracture

  >2

1.23

1.06–1.43

1.24

1.06–1.45

  >3

1.33

1.10–1.60

1.34

1.11–1.62

  >4

1.51

1.20–1.91

1.51

1.19–1.93

Any osteoporotic fracture

  >2

1.38

1.16–1.65

1.36

1.13–1.63

  >3

1.55

1.26–1.92

1.53

1.23–1.91

  >4

1.70

1.30–2.22

1.64

1.24–2.17

Hip fracture

  >2

1.68

1.19–2.36

1.70

1.20–2.42

  >3

1.92

1.28–2.88

2.05

1.35–3.11

  >4

2.26

1.35–3.79

2.39

1.39–4.09

When intake was dichotomized at >2 units daily, there was no confounding effect of smoking or BMI on the association (Table 6).
Table 6

Risk ratio (RR) for fracture and 95% confidence interval (95% CI) associated with a consumption of >2 units daily of alcohol with and without adjustment for smoking, body mass index (BMI), and bone mineral density (BMD)

Model

Any fracture

Any osteoporotic fracture

Hip fracture

RR

95% CI

RR

95% CI

RR

95% CI

Base case

1.23

1.06–1.43

1.38

1.16–1.65

1.68

1.19–2.36

  + smoking

1.22

1.03–1.43

1.36

1.13–1.63

1.50

1.05–2.15

  + smoking + BMD

1.24

1.05–1.46

1.38

1.14–1.66

1.54

1.07–2.22

  + BMD

1.21

1.04–1.41

1.35

1.13–1.61

1.64

1.16–2.32

  + BMI + BMD

1.22

1.04–1.43

1.34

1.11–1.61

1.67

1.16–2.38

Discussion

The present study confirms that a high intake of alcohol confers a significant risk of future fracture. In addition, the effect is over and above that which can be explained by variations in BMD. As found in previous studies, there was a threshold effect, and no increased risk of osteoporotic or hip fracture occurred in individuals who took 2 units or less per day of alcohol. Above this threshold, there was a significant risk in men and in women combined, and no significant difference in risk between men and women. The threshold that we identified is similar to that noted in other studies [6, 8], but in contrast to some studies, we found no evidence for a different threshold in men compared with women [6, 9]. Some studies have shown that the relationship between alcohol intake and hip fracture risk or BMD is J-shaped [2, 4, 5, 6, 7, 24] in that abstainers from alcohol have a higher risk than that of individuals consuming 1 or 2 units daily. In the present study, we found no difference in the risk between these categories or any systematic difference in femoral neck BMD.

A particular strength of the present study is that the estimate of risk is from an international setting from prospectively studied population-based cohorts. Calculations were based on the primary data, eliminating the risk of publication bias. Although the reported average intake differed markedly between cohorts, there was no evidence for heterogeneity in risk between cohorts, suggesting that the results are generalizable. A weakness is the validity of self-reported alcohol intake [25]. In the present study, reported consumption was significantly less in both men and women than that assessed in the United Kingdom [26]. This is likely to weaken rather than strengthen the apparent association found.

In common with other studies [6], men drank more than women. In this prospective study, 8% of men had more than 28 drinks per week, whereas only 1% of women had an intake above this limit. In the present study, drinking was more moderate than that reported in the Finnish study. Using the dichotomy of >2 units daily, 20% of men and 4% of women exceeded this intake. If a causal relationship was assumed, the data suggest that 7% and 2% of hip fractures in men and women, respectively, might be attributable to a high intake of alcohol. Causality, however, is conjectural and our results should not form the basis for any recommendations on the intake of alcohol.

Not all studies show an association of alcohol and fracture risk [24, 27, 28, 29, 30, 31]. A weakness of cross-sectional studies may be that individuals who have fractured have higher comorbidity and their intake of alcohol may be less. However, not all prospective studies show an increased fracture risk with intake of alcohol [32, 33, 34, 35, 36, 37, 38, 39]. The discrepancy in results may be related to the low proportion of heavy drinkers, and the lack of power to examine threshold effects.

The present study also quantifies the independent contribution of low BMD to the risk associated with high alcohol intake. Low BMD explained a minority of the total risk. With regard to BMD there are several mechanisms whereby alcohol might adversely affect fracture risk. Alcohol is shown to have direct effects on osteoblasts including those derived from man [40]. Alcohol also increases the endogenous secretion of calcitonin [41]. In addition, heavy drinkers may have poor nutrition with respect to calcium, vitamin D, or protein [42]. The mechanism for the BMD-independent increase in risk could not be determined from this study, but might be due in part to coexisting morbidity that in turn increases the risk of falls [43] or interferes with the protective response to injury [44, 45, 46].

Irrespective of the mechanism, these data indicate that the risk of fractures is greater in individuals who report taking more than 2 units of alcohol daily. This threshold is comparable to that advocated for cardiovascular health. Thus, the use of this risk factor can be used to enhance a case-finding strategy. Since alcohol intake confers a risk over and above that provided by age and BMD, intervention thresholds for BMD can be less stringent [47]. The precise contribution of high alcohol intake to risk assessment has to take into account the independent effects of other risk factors that might be used. Examples include smoking, a prior fracture, prolonged use of corticosteroids, a family history of fracture, and biochemical indexes of skeletal turnover, all of which have been shown to provide independent contributions to fracture risk [48, 49, 50, 51]. The interrelationship for these risk factors will need to be determined before they can readily be used in combination for assessing fracture risk in the general population.

We conclude that a reported history of high intake of alcohol confers a substantial risk for future fractures and that this risk is largely independent of BMD. The risk factor can be used to enhance a case-finding strategy; patients identified can be targeted for treatment at lower BMD thresholds than in individuals with osteoporosis of the same age but who do not consume alcohol above the threshold amount.

Acknowledgements

We are grateful to the International Osteoporosis Foundation, National Osteoporosis Foundation, International Society for Clinical Densitometry, and the European Community (EU FP 3/5) for their support. We gratefully acknowledge G-E Lunar, Lilly, Hologic, Roche, IGEA, the Alliance for Better Bone Health, Novartis, and Wyeth for their unrestricted support of this work.

Copyright information

© International Osteoporosis Foundation and National Osteoporosis Foundation 2004