Wu F, Mason B, Horne A, Ames R, Clearwater J, Liu M, Evans MC, Gamble GD, Reid IR. Fractures Between the Ages of 20 and 50 Years Increase Women's Risk of Subsequent Fractures. Arch Intern Med. 2002;162(1):33–36. doi:10.1001/archinte.162.1.33
Perimenopausal and postmenopausal fractures are well-recognized, strong, independent predictors of subsequent fractures. However, it is unknown whether premenopausal fractures are predictive of postmenopausal fractures.
To determine whether self-reported fractures sustained before the age of 50 years are associated with fractures after this age.
Subjects and Methods
Cross-sectional study of 1284 women (mean ± SD age, 73 ± 4 years) who were 10 or more years' postmenopausal and who were recruited from electoral rolls in Auckland, New Zealand. Detailed information on their fracture, medical, menstrual, alcohol, and smoking histories was obtained using a standardized questionnaire.
Nine percent of the women reported fractures before the age of 20 years; 7%, between the ages of 20 and 50 years; and 29%,after the age of 50 years. Fractures sustained between the ages of 20 and 50 years were associated with a 74% increase in the risk of fractures after the age of 50 years (odds ratio, 1.74; 95% confidence interval, 1.12-2.70), while fractures occurring before the age of 20 years were not (odds ratio, 1.01; confidence interval, 0.66-1.56). Multivariate analysis showed that after bone density, age, maternal history of hip fractures, age at menopause, weight, history of hormone replacement therapy, and smoking and alcohol histories were adjusted for, a history of fractures between the ages of 20 and 50 years remained a significant independent predictor of risk of fractures after the age of 50 years (risk ratio, 1.83; confidence interval, 1.12-2.76).
Any fracture (unrelated to motor vehicle accidents) sustained between the ages of 20 and 50 years is associated with increased risk of fractures after the age of 50 years. Therefore, this is an important clinical risk factor that points to the need for bone density measurement, consideration of lifestyle modification, and antiosteoporosis therapies in these women.
THE RELATIONSHIP between bone density and fracture risk has been well documented in many prospective studies. There is also evidence that in postmenopausal women, the occurrence of a fracture signals an increased risk of subsequent fractures, independent of the bone density.1- 8 Indeed, a history of fracture is the single most important factor in evaluating future fracture risk. However, many women reach menopause with a history of premenopausal fractures, yet it is unknown whether premenopausal fractures are in any way predictive of subsequent postmenopausal fracture risk. The fact that fractures both in childhood9 and in premenopausal life10- 12 are associated with reduced bone density suggests that these women may be at increased risk. If this were true, then asking about premenopausal fracture history would be an important and cost-free addition to the clinical evaluation of women wishing to make decisions regarding the prevention of osteoporotic fractures. The present study addresses this important clinical question by determining whether self-reported fractures sustained before the age of 50 years are associated with fractures after this age.
The study population comprised 1284 women, 10 or more years' postmenopausal, who were recruited by postal invitations, that were sent to names selected from electoral rolls. None of the women had disorders of calcium metabolism; renal, thyroid, or hepatic dysfunction; or other major systemic illness. All women were independently mobile. Prior to entry, they had not been using hormone replacement therapy or other drugs known to affect bone and calcium metabolism in the previous year.
The subjects answered a questionnaire regarding their medical, menstrual, smoking, alcohol, and fracture histories. They were asked to recall the number and site of fractures, the age at which fracture was sustained, and the mechanism of injury. Fractures resulting from motor vehicle accidents were excluded from analysis. Fractures were classified as occurring before the age of 20 years, between the ages of 20 and 50 years, and after the age 50 years. Data from 3 women with premature menopause (natural or surgical) before the age of 40 years were excluded from the analysis.
Bone mineral density was assessed using a dual-energy x-ray absorptiometer (Lunar Expert; Lunar Radiation Corp, Madison, Wis). Scans of the whole body and proximal femur were performed. Height was measured with a stadiometer (Harpenden; Holtain Ltd, Crymych, Wales), and weight (wearing light indoor clothing) was measured with electronic scales.
Differences in continuous normally distributed variables were tested between those women with fractures after the age of 50 years and those without using t test analysis. Differences in categorical data between fracture groups were sought using χ2 analysis without continuity correction. The presence of at least 1 fracture before the age of 50 years was compared with the presence of at least 1 fracture after the age of 50 years using the McNemar procedure. Odds ratios were also used to assess the strength of association between history of previous fractures and subsequent fracture risk and to provide adjustment for planned subgroup analyses. Cox proportional hazards models were constructed to assess the independent predictive effect of individual clinical risk factors and bone mineral density measurements on fracture risk after the age of 50 years. Cox proportional hazards models were constructed using the follow-up time from the age of 50 years to the year of the present questionnaire for the women without fractures and from the age of 50 years to the year of the first fracture for the women with fractures. Clinical, anthropometric, and lifestyle factors were treated as independent variables and were entered into models with a variety of iterative procedures (stepwise, backward, forward selection) with a P value of .15 for entry into the model. Parsimony, goodness-of-fit statistics, and biological plausibility enabled the choice of the final model.
Except where otherwise stated, data are presented as mean ± SD and 95% confidence interval. A 5% significance level was maintained throughout these analyses. All tests were 2-tailed. The SAS version 6.12 (SAS Institute, Cary, NC) was used for all analyses.
Table 1 presents the characteristics of the study population. The 2 groups of women with and without a history of fractures after the age of 50 years were similar. However, the former group was on average 10 months older at the time of answering the questionnaire and had lower bone mineral densities. Current alcohol intake was similar between the 2 groups and was low as a whole: 23.0% of all the women in the study never drank, and only 3.8% took more than 2 drinks per day. Women who had fractures before the age of 50 years had lower bone mineral densities than those without such a history (total body: 1.02 ± 0.09 g/cm2 vs 1.04 ± 0.10 g/cm2, P = .01; femoral neck: 0.81 ± 0.10 g/cm2 vs 0.83 ± 0.13 g/cm2, P = .07).
Four hundred ninety-four women reported a total of 721 fractures. Of these, 126 fractures were reported by 111 women before the age of 20 years, 95 fractures by 90 women between the ages 20 and 50 years, and 500 fractures by 367 women after the age of 50 years. Three hundred thirty-two women reported sustaining only 1 fracture, and 162 women reported sustaining more than 1 fracture. Table 2 shows the distribution of the fractures by site.
Table 3 shows the odds ratios and 95% confidence intervals (CIs) of sustaining a fracture after the age of 50 years in terms of fracture history. Fractures before the age of 20 years were not associated with fractures after the age of 50 years. Fractures sustained between the ages of 20 and 50 years were associated with subsequent fractures after the age of 50 years, with an odds ratio of 1.74 (95% CI, 1.12-2.70). Post hoc analysis of the data from the 20- to 50-year-old group divided according to the decade in which their original fracture occurred showed similar trends across the decades. The analysis was also performed substituting each woman's age at menopause instead of age 50 years as the cutoff between premenopausal and postmenopausal fractures, and similar results were obtained. Age 50 years as the cutoff was preferred because it circumvents the problem of determining the age at menopause, which can be particularly difficult in women who have undergone a hysterectomy.
To assess the independent predictive value of fracture history before age 50 years on the risk of subsequent fractures, Cox proportional hazards models for the time to first fracture after age 50 years were developed. History of fractures between ages 20 and 50 years and total bone density consistently emerged from the models constructed as independent predictors of fracture risk after age 50 years. The following variables were entered into the final model: the history (yes/no) of fractures between 20 and 50 years, total bone density (expressed as Z score), age, body weight, age at menopause, smoking (previous or current), alcohol consumption of more than 2 drinks per day, history of hormone replacement therapy, and maternal history of hip fractures. Because a history of fractures before the age of 20 years was not significantly associated with fractures after the age of 50 years, it was omitted from the model. Fracture history between the age of 20 and 50 years was associated with a risk ratio of 1.83 (95% CI, 1.12-2.76; P = .003) and the total bone density with a risk ratio of 0.79 (95% CI, 0.68-0.91; P = .005); ie, a history of fractures between the ages of 20 and 50 years was associated with an 83% increase in the risk of fractures after the age of 50 years, and an increase in Z score of 1 decreased the risk by 21%. Hip bone density could be substituted for total body bone density in this model and produces similar result.
The major finding of this study is that any fracture (excluding those related to motor vehicle accidents) occurring between the ages of 20 and 50 years is associated with an increased risk of fractures after the age of 50 years. This increased risk is independent of age, bone density, body weight, alcohol intake, and history of smoking. The increase in risk of further fractures is comparable to that associated with the classic osteoporotic risk factors, such as low bone density, low body weight, maternal history of hip fractures, smoking, and prior osteoporotic fractures.13 The central message of our study is that a history of fractures between the ages of 20 and 50 years identifies women at increased risk of having subsequent fractures. Indeed, having a fracture between the ages of 20 and 50 years resulted in a 74% increase in the risk of fractures after the age of 50 years. However, the incidence of fractures between the ages of 20 and 50 years is low (7%). Thus, the presence of fractures between the ages of 20 and 50 years is helpful in identifying only some of those postmenopausal women who will require bone density measurement and treatment. Our study was not designed to elucidate the mechanisms behind this association; factors likely to be involved include bone quality and strength, fall frequency, and protective neuromuscular responses to falls. Nonetheless, this does not detract from the importance of a premenopausal fracture history as a clinical risk factor. It should be an integral part of fracture risk assessment, helping to determine who should undergo bone densitometry and who should be offered therapies to increase bone mass.
While there has been a growing awareness in recent years that fractures in the perimenopausal and postmenopausal periods are predictive of later fractures, the significance of premenopausal fractures has not really been addressed. Honkanen et al14 reported an association between fractures occurring between the ages of 20 and 34 years and those occurring in the perimenopausal period (age, 35-57 years), the hazard ratio being 1.9 (95% CI, 1.6-2.3). Johansson and Mellström15 carried out a study similar to the present one and found that a fracture occurring before the age of 40 years was associated with a relative risk of fracture occurring after the age of 50 years of 1.20 (95% CI, 0.95-1.53). However, the women in that study ranged in age from 46 years upward, so the ascertainment of postmenopausal fractures must have been very incomplete, because most women still had many postmenopausal years before them. The design of the present study has substantially overcome this problem by having an average time of 24 years since menopause in the study population and, possibly as a result, found a much higher risk associated with premenopausal fracture. Together, these findings suggest that the risk of sustaining fractures is a life-long trait, probably reflecting the interaction of an individual's bone mass, bone quality, fall frequency, and neuromuscular protective response to falls.
Fractures occurring in childhood and young adulthood (before the age of 20 years) were not associated with increased fractures after the age of 50 years in this study. This finding may be because the mechanisms of fractures are different in this age group. Children and adolescents probably sustain more fractures through playing, sports, and accidents, so fracture risk may reflect lifestyle and risk-taking behavior as much as skeletal factors. In adults and older women, fractures may be more dependent on other factors, such as bone strength, soft tissue composition, and frailty. These variables are more likely to be consistent between early and later adult life than between childhood and the sixth and seventh decades of life and could explain the lack of association between fractures occurring before the age of 20 years and after the age of 50 years.
This study is based on the retrospective self-report of fractures. The failure to recall a fracture or its year of occurrence might have biased our results. Accuracy of self-reports is site and severity dependent. The positive predictive values are 98% to 100% for self-reported wrist and forearm fractures, 85% to 95% for ankle and lower leg fractures, 100% for hip fractures, and 84% for all fractures. Minor fractures in the hand and foot often remain unreported.16 While the validity of self-reports of past fractures is not perfect, with problems of both overreporting and underreporting, it is relatively accurate for major fractures such as those of the wrist, ankle, and hip. There is evidence that the predictive value of a history of a postmenopausal fracture on subsequent fractures is site dependent.1,4,8 It is possible that the same applies to premenopausal fractures.14 Therefore, those major premenopausal fractures that are reported accurately by women may be more significant than minor fractures, which are less accurately reported.
The present finding that any fracture occurring between the ages of 20 and 50 years is associated with increased risk of fractures after the age of 50 years has important implications for clinical practice. Whatever the underlying mechanisms, such a history should be considered a significant clinical risk factor when decisions are being made regarding the need for bone density measurement and the use of therapeutic interventions in postmenopausal women. Younger women who have had a fracture will be alerted to the need to optimize modifiable risk factors for osteoporosis, such as cessation of smoking, ensuring adequate dietary calcium intake, moderation of alcohol and caffeine consumption, and participation in regular physical activity.
Accepted for publication March 13, 2001.
This study was funded in part by the Health Research Council of New Zealand, Auckland.
Presented as an abstract at the Austalian New Zealand Bone and Mineral Society Annual Scientific Meeting, Hamilton Island, Australia, November 7, 2000.
Corresponding author: Ian R. Reid, MD, Department of Medicine, University of Auckland, Private Bag 92019, Auckland, New Zealand (e-mail: email@example.com).